Keywords: Causal Models, Deep Learning in Robotics and Automation, AI-Based Methods
Abstract: Reinforcement learning has been applied to solve several complex problems in robotics and automation; however, learning optimal policies in continuous state and action spaces takes a very long time. Incorporating causal knowledge helps to focus exploration and avoid unnecessary actions, thus reducing significantly the number of episodes to obtain an optimal solution. Additionally, the causal models can be easily transferred to similar tasks. In this talk I will give an introduction to causal graphical models, including causal reasoning and discovery. I will then explain how to incorporate a causal model into a traditional reinforcement algorithm, and apply it to solve different problems, including robotic manipulation. Finally, I will present our recent work on learning and using a causal model simultaneously.

Keywords: Computer Vision for Transportation, Intelligent Transportation Systems, Logistics
Abstract: Autonomous driving, remote operation, and driver assistance of rubber-tired gantry cranes (RTG) are vital to the automation of container terminals. Robust and reliable lane detection and tracking are crucial links in them. This paper proposes a novel lane detection and tracking method based on traditional image process techniques instead of deep-learning-based approaches that rely heavily on large-scale datasets. It utilizes the structural characteristics of a container yard. First, we propose an adaptive sliding window method to resolve complex issues of lane detection suffering from smudges, occlusions, and breakages. Furthermore, the continuity and structural constraints of inter-frames of lanes are explored and used to improve the stability of lane tracking. Then, extensive experiments are conducted to detect contaminated, occluded, and separate lane lines under different lighting conditions and to test the robustness of lane tracking with severe disturbances. The results show that the proposed method can attain robust lane detection and track for RTG autonomous driving in a container yard.

Keywords: Logistics, Task Planning, Plug-in Electric Vehicles
Abstract: This paper presents an optimization-based approach for the Electric Vehicle Routing Problem considering Smart-Charging methods. The objective, based on the application of the model, is to obtain the shortest route for each of the electric vehicles that have to deliver freight to a set of customers. Based on the Smart-Charging method, in which vehicles can charge/discharge energy from/to the grid, the power grid limits, and balancing needs are considered. In this way, both the charging points and the energy districts are prevented from exceeding the maximum allowed energy peak. A real case study in the Apulia Italian region (Italy) shows the effectiveness of the proposed optimization model.

Keywords: Planning, Scheduling and Coordination, Logistics, Machine learning
Abstract: An order batching and sequencing problem in a warehouse is studied in this work. The problem is proved to be an NP-hard problem. A mathematical programming model is formulated to describe it clearly. To minimize tardiness, an improved iterated local search algorithm based on reinforcement learning is proposed. An operator selecting scheme, which aims to automatically select local search operator combinations instead of simply performing all the operators in each iteration, is designed to reduce the computational cost greatly. Besides, an adaptive perturbation mechanism is designed to improve its global search ability. Extensive simulation experimental results and comparisons with the state of the art demonstrate the high effectiveness and efficiency of the proposed approach.

Keywords: Intelligent Transportation Systems
Abstract: Crack detection is essential for guaranteeing airport runway structural reliability. An efficient solution we take is to employ a robot equipped with a camera to perform inspection task. However, automatic crack detection for airport runway is challenging, as the runway surface is seriously polluted by fuel stain and aircraft wheel mark, and the cracks need to be detected are usually extremely thin. Thus, we propose a CNN model, AggCrack, to perform the crack detection task.AggCrack adopts an innovative semantic-level attention mechanism on the edges of the targets to focus the model on crucial features, and combines edge information and semantic segmentation for more accurate crack detection. We have implemented the algorithm and have it extensively tested on an airport runway dataset collected by our inspection robot from four different airport runways. Compared with four existing deep learning methods, experimental results show that our algorithm outperforms all counterparts. Specifically, it achieves the Precision, Recall and F1-measure at 84.24%, 70.36% and 76.68%, respectively.

Keywords: Intelligent Transportation Systems
Abstract: On-ramp merging is often associated with highly dynamic interactions between ego and other vehicles, which are more challenging in dense traffic. Considering both the overall traffic situation and the individual characteristics of other interacting drivers, we propose a social-aware hierarchical decision algorithm based on level-k game theory. To adapt to dynamic interactive situations, the social value orientation of interacting drivers is estimated on-line, while the right of way and tentative merging attempts are further integrated to improve the social-awareness of the decision model. A drone dataset of naturalistic driving is built to calibrate and validate the model effectiveness. Simulator experiments with drivers in the loop further show that the model can improve the safety and success rate in ramp merging.

Keywords: Motion Control, Intelligent Transportation Systems
Abstract: The aerial transportation system is a kind of nonlinear underactuated mechatronic system, which suspends the cargo beneath the rotorcraft’s fuselage and undertakes two basic missions of rotorcraft positioning and cargo swing suppression. Currently, most available methods need simplifications such as the near hovering hypothesis and dimension reduction operations, which may badly degrade the control performance when state variables get far away from the equilibrium point. In addition, integral terms, which can eliminate the steady errors, are not reflected in controller design and stability analysis processes. To tackle the aforementioned issues, this article provides a novel nonlinear control approach with an elaborately constructed integral term for aerial transportation systems, which not only achieves satisfactory antiswing and positioning performance but also reduces steady errors in practical flight. Meanwhile, the actuating constraint is taken into consideration so as to avoid saturation problems. Without linearization operations, we prove the closed-loop asymptotic stability of the equilibrium by the explicit Lyapunov-based analysis. As far as we know, this article is the first solution for controller design with the consideration of both steady errors elimination and actuating constraints. Finally, several groups of hardware experimental results are provided to validate the effectiveness of the presented control scheme.

Keywords: AI-Based Methods, Machine learning, Big-Data and Data Mining
Abstract: Recently Digital Twin (DT) has attracted much attention from researchers due to its capacity of system monitoring and health management to improve the reliability and availability of systems. This emerging technology has been considered a promising solution for various sectors to enhance the sustainability of business development. In general, DT relies on the living models for simulating system behaviors to monitor the systems or assets in production. Such models could be either mathematical/physics-based or data-driven, which are able to explain, predict, and describe system behaviors timely and accurately. Therefore, the machine learning-enabled modeling technology has become a powerful tool to develop such data-driven living models. However, data-driven models developed with supervised learning techniques carry a fatal deficiency: once the operational environments are changed, the model may hardly work well or even becomes useless due to the distribution shift between the training data and the new dataset. This paper attempts to address this issue by proposing to apply transfer learning techniques to develop lifetime robust living models for real-world DT systems. This paper presents a framework for developing lifetime data-driven living models. A case study, railway digital twin from our on-going research project along with the preliminary results, will be presented to demonstrate the feasibility and usefulness of the proposed modeling framework for digital twin.

Keywords: AI-Based Methods, Deep Learning in Robotics and Automation, Computer Vision in Automation
Abstract: This study aims at providing a conceptualized framework for railway to realize the Digital Twin (DT) beyond traditional structural modelling or information systems. First, we deduce a generic formula that shows that DT estimates the future states and decides actions beforehand. Then, based on this formula, we design a generic framework called RailTwin. The framework combines the insight of current states, the foresight representing the prediction of the future states, and the oversight based on the current and future state to enable automation and actuation. The key enabler of this framework to obtain these states is Artificial Intelligence (AI) technologies, including Deep Learning, Transfer Learning, Reinforcement Learning, and Explainable AI. We present a use case for asset health inspection and monitoring through the proposed framework.

Keywords: Diagnosis and Prognostics, Failure Detection and Recovery, Probability and Statistical Methods
Abstract: For the surface cracks of 304 stainless steel, there is no effective nondestructive testing method for crack detection. In this paper, the weak magnetic detection technology is adopted, and the self-developed high-precision magnetic flux sensor is used to measure the magnetic induction intensity of 304 stainless steel surface, and the magnetic gradient algorithm is used to process the magnetic signal. The defect is located by finding the mutation position of the original signal and calculating the threshold line by differential signal processing.

Keywords: Assembly, Reinforcement, Compliant Assembly
Abstract: This paper proposes an efficient robot assembly skill learning framework based on only a few demonstrations. The assembly skill learning process consists of two phases, e.g.,the pre-training phase and the self-learning phase. In pre-training phase, the assembly networks are initialized from demonstration data. A novel data augmentation model based on state transition model is designed in pre-training. In self-learning phase, the pre-trained assembly networks are further optimized by a hybrid exploration strategy in assembly environment. On par with the learning framework, a fuzzy reward function balancing the efficiency and compliance of assembly is designed to evaluate action’s performance in self-learning process. Series of physical experiments were conducted on assembly platform.

Keywords: Deep Learning in Robotics and Automation, Sensor-based Control, Factory Automation
Abstract: Classical Visual Servoing (VS) relies on handcrafted visual features, which limit their generalizability. Recently, a number of approaches, some based on Deep Neural Networks, have been proposed to overcome this limitation by comparing directly the entire target and current camera images. However, by getting rid of the visual features altogether, those approaches require the target and current images to be essentially similar, which precludes the generalization to unknown, cluttered, scenes. Here we propose to perform VS based on visual features as in classical VS approaches but, contrary to the latter, we leverage recent breakthroughs in Deep Learning to automatically extract and match the visual features. By doing so, our approach enjoys the advantages from both worlds: (i) because our approach is based on visual features, it is able to steer the robot towards the object of interest even in presence of significant distraction in the background; (ii) because the features are automatically extracted and matched, our approach can easily and automatically generalize to unseen objects and scenes. In addition, we propose to use a render engine to synthesize the target image, which offers a further level of generalization. We demonstrate these advantages in a robotic grasping task, where the robot is able to steer, with high accuracy, towards the object to grasp, based simply on an image of the object rendered from the camera view corresponding to the desired robot grasping pose.

Keywords: Reinforcement Learning, Deep Learning Methods, Agent-Based Systems
Abstract: Humans possess the uncanny ability to perceive and manipulate deformable objects with agility. Deformable objects, such as a ziplock bag, with their infinite configurations, are particularly challenging. Manipulating deformable objects is hard as deciding corrective action demands great familiarity with its state transitions. Humans learn to manipulate such objects by leveraging multimodal perception capabilities wherein they build a rich understanding by exploration and active manipulation. For robots, such adaptability is lacking as multimodal perception is still limited. In this work, we emulate human-like learning by extending multimodal perception to assist in exploratory manipulation--to build a rich understanding of the deformable object-- and purposeful manipulation-- to learn to achieve the desired state based on the knowledge of the object. To validate this framework, we undertake the task of closing a ziplock bag as it is a challenging perception task with a simple state structure demanding multi-sensory modes. The bag’s parameters are learned in 10mins during exploratory manipulation. With the multimodal inputs of the three sensors, the accuracy of state detection improves significantly across the different types of bags, illustrating the adaptability. The multimodal perception achieves a maximum state detection accuracy of 95.5%. The bags are closed fully in all 50 trials with the purposeful manipulation framework.

Keywords: Cyber-physical Production Systems and Industry 4.0, Factory Automation, Collaborative Robots in Manufacturing
Abstract: In this paper, we explore a cloud-facilitated collaboration paradigm that supports resources sharing and collaborative development of factory robot applications cyber-physically. This framework is facilitated by a cloud platform, which acts as a collaboration interface between multiple resource owner, developer and user groups. Robot resources are made available on the cloud to facilitate design and modeling. Kinematics representations, dynamic models and control modules are constructed to facilitate design and testing of automation workflow in augmentation to the physical validation using the actual hardware. This hybrid mode of process design combines physical and virtual testing while supporting remote resource sharing and collaboration, which in further leverage the potential of cyber-physical networks for the next-generation smart manufacturing.

Keywords: Cyber-physical Production Systems and Industry 4.0, Intelligent and Flexible Manufacturing
Abstract: Cyber-physical manufacturing networks are composed of real-world manufacturing system and its cyber representation to achieve fast virtualization of real world applications and real-time feedback from cyber models to the real world. As a promising technology, digital twins can improve the efficiency and resilience of a system with feedback loops in which physical processes affect cyber parts and vice versa. In this process, vast amounts of data are collected from both real-world system and cyber model, including product and process design, scheduling, fault detection etc. Knowledge driven technologies have emerged as important tools for improving the intelligence of digital twins and achieving smart manufacturing. This work presents a review of knowledge driven technologies for digital twins in cyber-physical manufacturing networks. It includes the state-of-the-art researches, innovation, applications, challenges and future directions.

Keywords: Computer Vision for Manufacturing, Motion and Path Planning, Mechatronics in Meso, Micro and Nano Scale
Abstract: A vision-based robotic micromanipulator capable of self-calibration is proposed for microscale positioning of tools. We extend a previously developed a functional robotic micromanipulator system successfully applied for cell manipulation towards micropositioning of a tool under the microscopic view of a printed circuit board to demonstrate prospective applications. Smart features of the developed system including its self-calibration mechanism, disruption-resilient tracking capability and modular design for portable deployment are demonstrated to validate the potential for universal task executions for high-precision manufacturing. The long-term goal is to leverage the development of robotic micromanipulation for universal micromanipulation applications contributive towards adaptive manufacturing.

Keywords: Additive Manufacturing, Computer Vision for Manufacturing
Abstract: This paper applies the ResNet convolutional neural network to automatically detect defects from polymer AM parts CT scans. The study considers designed internal de- fects of a nozzle. Detection using our trained classifier equipped with ResNet34 framework reaches a high accuracy. Scalability of the method is also explored.

Keywords: Cyber-physical Production Systems and Industry 4.0, Intelligent and Flexible Manufacturing, Planning, Scheduling and Coordination
Abstract: To remain competitive in a highly dynamic environment, manufacturing companies have to quickly react to disturbances or changing customer requirements. To enable manufacturing systems to cover these dynamics, the concept of reconfigurable manufacturing systems was introduced. From a technical point of view, this concept has been exploited the past 20 years, revealing several different design solutions. However, industrial application is still an exception. Our analysis led to the assumption that this is due to a lack of operator support for reconfiguration management. In addition, mostly individual aspects of reconfiguration are considered instead of exploiting the entire reconfiguration space at the system and machine level. Therefore, in this paper we present a digital twin framework for reconfiguration management considering reconfiguration as a holistic problem and taking advantage of the digital twin concept to integrate heterogeneous data from different sources.

Keywords: Cyber-physical Production Systems and Industry 4.0, Collaborative Robots in Manufacturing, Human-Centered Automation
Abstract: A robot needs multiple interaction modes to robustly collaborate with a human in complicated industrial tasks. We develop a Coexistence-and-Cooperation (CoCo) human-robot collaboration system. Coexistence mode enables the robot to work with the human on different sub-tasks independently in a shared space. Cooperation mode enables the robot to follow human guidance and recover failures. A human intention tracking algorithm takes in both human and robot motion measurements as input and provides a switch on the interaction modes. We demonstrate the effectiveness of CoCo system in a use case analogous to a real world multi-step assembly task.

Keywords: Semiconductor Manufacturing, Computer Vision for Manufacturing, Failure Detection and Recovery
Abstract: Wafer is the crucial raw material of semiconductor devices. In wafer production, impurities cannot be removed entirely, which will cause various wafer map defects. Quickly and precisely classifying wafer map defects can help engineers track failures in the semiconductor manufacturing process. However, different wafer map defect patterns occur randomly and irregularly, and labeling work is labor-intensive and time-consuming. Therefore, the wafer map dataset is usually imbalanced and consists of many unlabeled data. In this paper, we utilize unlabeled data by using semi-supervised learning methods and alleviate the imbalance problem by optimizing the loss function to increase the accuracy of wafer map classification. The performance of the proposed method is illustrated with the WM-811K dataset which consists of real-world wafer maps.

Keywords: Big-Data and Data Mining, Product Design, Development and Prototyping, Machine learning
Abstract: For user experience (UX) analysis in product design, the context of use (such as task, activities and environment) is a valuable element that enables context-awareness in accordance with the available context of use elements of users. Typically, conventional methods such as interviews or questionnaires are used to extract the context of use elements, but they are labour-intensive and time-consuming and thus do not scale well. On the other hand, the automatic extraction approaches from existing studies are not as effective as the extracted context of use phrases are not as informative. In this study, we present an automatic approach to exploit and understand the context of use elements from online customer reviews using BERT. Firstly, the context of use elements from online customer reviews is labelled using a BERT-based approach. Secondly, a syntactic-based post-processing is designed to check the labelled results and form the phrases related to the context of use. Finally, the customers’ preferences related to the contexts of use is analysed by studying and aggregating it with its relevant hedonic quality (such as positive or negative) which can then be used to enrich the UX modelling. For product designers, the modelling results can facilitate the optimisation of product design. A case study was conducted to understand and leverage the context of use elements in UX from online customer reviews to support customer strategy creation and design activities.

Keywords: Diagnosis and Prognostics, Cyber-physical Production Systems and Industry 4.0, Machine learning
Abstract: In recent years, Intelligent Fault Diagnosis (IFD) technology, as a promising method, has been a hot research topic in the field of condition monitoring and diagnosis systems, which is the focus of ensuring industrial production safety and productivity. However, the utilities of many existing IFD methods are limited by the poor-quality monitoring data, most of which are unlabeled, non-stationary, and collected from various working conditions. In addition, the unavailability of the testing data in the IFD model training phase makes the problem more challenging but more practical. In the paper, a simple-structured one-dimensional convolutional neural network(1-D CNN) with a feature extractor, a classifier, and a meta-optimizer is constructed to tackle the tricky cross-domain issues. A novel and scalable meta-learning-based domain generalization strategy is proposed to reduce the gap among the multi-source domains. As a result, the network can learn common fault knowledge from multiple related but different source domains and then be used to analyze new target domains. Two case studies verify the effectiveness, real-time performance, and application prospects of the proposed training strategy in cross-domain fault diagnosis tasks.

Keywords: Failure Detection and Recovery, Deep Learning Methods, Intelligent and Flexible Manufacturing
Abstract: The encoder-decoder network is one of the most common deep learning models for time series representation learning and anomaly detection. However, it is hard to reconstruct time series, which is complex, correlated, and lacking in common patterns. In this paper, we apply the attention mechanism to rescale convolution layers and learn representation in the principal and the residual space. To avoid the reconstruction process, we define the residual space by the omitted segments according to the attention score in the encoder. We introduce the temporal information inside the token level and use sparse penalty to improve representation learning. We apply the proposed model to anomaly classification and fault detection experiments on two datasets, i.e. multivariate bearing fault dataset and UCRArchive profile dataset. The result shows that the representation learned by the proposed model is more likely to cluster by category, especially in the residual space. Compared to the baselines and state-of-the-art models, the proposed model has higher accuracy and recall in the limited-labeled situation, which illustrates the stability of the learned representation and its superiority in the downstream tasks.

Keywords: Process Control, Factory Automation, Intelligent and Flexible Manufacturing
Abstract: In intelligent manufacturing, products' high precision and reliability put forward higher requirements for machining process stability. The machining process's new stability evaluation method is proposed based on the minimum entropy space state (MESS). Four metrics, including entropy, variation of entropy, minimum entropy space state, and hyper entropy, are used to evaluate the machining process. First, the definitions of the four metrics are given, and the relationships between the four metrics are concluded. Then the calculation methods of the four metrics are given. Time-domain features of machining process signals are also extracted. The evolution mechanism of the machining process is expressed through the perspective of entropy. Then the values of the four metrics are used to evaluate the machining process. Extensive empirical evaluations on the turning process of bearing seats validate the applicability and practicability of the proposed method.

Keywords: Planning, Scheduling and Coordination, Intelligent and Flexible Manufacturing, Deep Learning in Robotics and Automation
Abstract: We address a scheduling problem of robotic flow shops with deep reinforcement learning for the first time. Parts follow the same process sequence whereas their processing times are all different on each machine. The objective is to minimize makespan. A deep reinforcement learning approach combined with look-ahead search is proposed.

Keywords: Planning, Scheduling and Coordination, Assembly, Intelligent and Flexible Manufacturing
Abstract: A scheduling approach for the assembly of one airplane for all platforms on its assembly line is designed and analyzed. It uses instructions from a top-level planning model and deals with more details regarding tasks and labor skills. In the proposed model, we consider generalized temporal constraints between tasks, multiple modes, limited resources (space and tools), and work time breaks. For some of the tasks, the platform can be chosen. A heuristic decomposition procedure is designed to obtain high-quality solutions. It contains two steps. First, plat-forms for tasks are chosen where applicable. Second, we solve the resulting single-platform subproblems by a genetic algo-rithm (GA).

Keywords: Cyber-physical Production Systems and Industry 4.0, Manufacturing, Maintenance and Supply Chains, Sustainability and Green Automation
Abstract: High-fidelity mathematical models are essential to implement model-based analysis and control in manufacturing research and practice. Currently, such models are typically conducted manually in an ad hoc manner. This approach presents several limitations, especially to small and medium-sized manufacturers, such as unavailability of equipment status data, inconvenient data collection process, non-standard and non-unique modeling rules, etc. In this paper, we describe a case study at a local small manufacturer of medical devices and apply a novel approach of production system modeling to overcome various practical challenges in collecting up- and downtime data of the operations. Specifically, the parametric model of the production system is identified based on system performance metrics derived from the parts flow data. With the model constructed, system bottleneck is analyzed and then, to enhance system throughput, potential improvement actions including operation speed-up, downtime reduction, and buffer expansion are explored. Finally, model sensitivity is analyzed by comparing the deviation of the model-predicted performance metrics to those produced by a reference nominal model.

Keywords: Factory Automation, Cyber-physical Production Systems and Industry 4.0, Manufacturing, Maintenance and Supply Chains
Abstract: As the manufacturing industry transforms its practice to adapt to the Industry 4.0 era, various new technologies are deployed on the factory floor. With these technologies, more production data are being generated and collected. On the other hand, development of analytics tools to digest these factory floor data and extract useful information for production decision-making and control has been lagging behind. Moreover, due to technological limitations of the sensors and the noisy environment in manufacturing facilities, it is quite common that the raw data collected from the factory floor contain errors and often must go through a time-consuming, labor-intensive data cleaning process before they are ready to be fed to any analytics tools. This is one of the main challenges and concerns about effective use of factory floor production data in practice. In this paper, we focus on the buffer occupancy time series data of a production system (i.e., the number of parts in each buffer at every time point). Specifically, we study the buffer occupancy data in two-machine Bernoulli serial lines that are subject to noise and develop an effective algorithm to automatically detect and correct errors in these data. Numerical experiments show that the proposed method is capable of restoring data integrity and significantly improving system performance estimation accuracy using the corrected data.

Keywords: Cyber-physical Production Systems and Industry 4.0, Discrete Event Dynamic Automation Systems, Intelligent and Flexible Manufacturing
Abstract: Residence time constraints are widely observed in production systems, such as semiconductor manufacturing, food industry and battery production, where the time that a part spends in one or several consecutive buffers is restricted. When the residence time of a part is beyond a certain level, the part might face quality problems and need to be further treated or scrapped immediately. To optimize the production performance such as production rate and scrap rate, one needs to properly manage all machines' behavior according to real-time system states to prevent from producing too many intermediate parts with high risk of scrap. To solve this problem, a simulation-based real-time control method is proposed to perform production control in face of four basic classes of residence time constraints that are widely seen in semiconductor manufacturing. A Markov Decision Process (MDP) model is first built, and a feature extraction method and a feature-based approximate architecture are proposed to deal with the curse of dimensionality. Simulation is applied in the training to estimate parameters of the feature-based approximate architecture, so the lookahead function in the MDP model can be approximately obtained. Simulation experiments suggest that such a method leads to significant system performance improvement with low computation overhead, which makes real-time production control feasible for longer serial lines with different classes of residence time constraints.

Keywords: Motion and Path Planning
Abstract: With the rising of Industry 4.0, robots are expected to perform increasingly complex manipulation tasks in collaboration with humans. However, current industrial robots are still largely preprogrammed with very little autonomy and still require to be reprogramed by robotics experts for even slightly changed tasks. Therefore, it is highly desirable that robots can adapt to certain task changes with motion planning strategies to easily work with non-robotic experts in the production line or manufacturing environments such as warehouse delivery settings. In this paper, we propose a reinforcement learning (RL) based motion planning method that uses a library of a few user demonstrations to generate adaptive motion plans for new tasks. The library of user demonstration can be built based on common tasks adopted for specific application environments. To achieve an adaptive motion plan facing task changes or new task requirements, features embedded in user demonstrations are captured. A new task can be either learned as an aggregation of a few features of common tasks in the library or a requirement for further human demonstration if the library of human demonstrations is insufficient for the new task. We evaluate our approach on a 6 DOF UR5e robot on multiple tasks and scenarios and show the effectiveness of our method with respect to different scenarios.

Keywords: Semiconductor Manufacturing, Intelligent and Flexible Manufacturing, Planning, Scheduling and Coordination
Abstract: Traditionally, process modules in cluster tools can execute a single operation only. With the rapid development of equipment design, recent cluster tools tend to have multi-functional process modules (MPMs) serving for processing multiple operations together in a single module. However, the different function settings of MPMs make the tool configuration various such that it is difficult to determine a schedule plan. Meanwhile, with different wafer processing parameters, a reasonable function setting for MPM can result in an excellent schedule so as to maximize the productivity. Thus, to solve the scheduling problem of a single-arm cluster tool with MPMs, a timed Petri net model is established to describe the behaviors of the system. Based on the model, two algorithms are developed to calculate the makespan for completing a given number of wafers. Then, an adaptive schedule method is presented to set the functions of MPMs to minimize the makespan. Finally, experimental results show the efficiency and effectiveness of the proposed method.

Keywords: Machine learning, Planning, Scheduling and Coordination, Intelligent and Flexible Manufacturing
Abstract: In petrochemical industry, the process conversion between different product types causes the capacity loss, and a larger change of MFI (melt flow index) will generate the transition products (i.e. the products scrapped). This study considers two objectives to improve production scheduling: minimizing transition products to reduce the defective products and minimizing total tardiness to satisfy the customer’s due date. We propose the engineering experience heuristic (EEH) to characterize the production and the constraints to generate a feasible scheduling. Based on EEH, we propose a metaheuristic algorithm embedded with reinforcement learning to solve the scheduling problem. The reinforcement learning is used to generate the optimal policy which guides the hyperparameter tuning in metaheuristic. We validate the proposed algorithms with an empirical study and the results show that it provides better performance than other popular benchmarks.

Keywords: Semiconductor Manufacturing, Hybrid Strategy of Intelligent Manufacturing, Process Control
Abstract: As the Critical Dimension (CD) becomes more and more shrink and the technology of Multiple Patterning extends, the accuracy of the Photolithography becomes strictly. The control of overlay error has become challenge in semiconductor manufacturing. Overlay errors are the displacements of the target position to the shifting position. Overlay errors can be compensated by adjusting the parameters in step-and-scan equipment (scanner). In particular, the effect of compensation is determined by overlay error itself, overlay error model, and sampled overlay. Although most of existing studies have been done on overlay error modeling and sampling, however, little research has been done on compensation of high-order overlay error in nano-technology. This study aims to propose a systematic framework of overlay error compensation by considering the high-order overlay factors and designing a sampling strategy. To consider the sampled overlays have various control specifications, this study adopts constraint programming model to estimate the compensated parameters of overlay error factors. Given the number of sampling overlays, this study also proposes a sampling algorithm to cover the considered overlay error factors as possible. To validate the proposed method, an empirical study from a semiconductor company was conducted. The results show that this study can effectively compensate high-order overlay errors and the proposed sampling strategy with more overlays in spec which are superior to existing method by equipment vendor. The results demonstrated the practical viability of the proposed approach for yield improvement.

Keywords: Machine learning
Abstract: The steel industry is very important for industry, and relies on a high cost and technology. It is regarded as a high cost because of (1) long time of building the factory, (2) cost recovery is very slow, (3) high energy consumption, (4) economic scale of production, and high transportation costs. The steel industry is deeply and widely related to the industry, and it is also a necessary industry for other industries. Both Nickel and Chromium are the main raw materials for the manufacturing stainless steel, and about 75% of the global nickel consumption is used to produce stainless steel. These Nickel and Chromium metals account for about 70% to 80% of the total cost of manufacturing stainless steel. The price of Nickel is relatively unstable. Therefore, price of Nickel is main influence factor for global stainless steel prices. Overall price of stainless steel will be based on the international price of Nickel at that time as the basis for pricing. This research attempts that analyzes the different wavelet decomposition layers, finds the best prediction mode, predicts the price of nickel with long-short memory network (LSTM), and analyzes the trend, season, residual and other indices. Empirical results indicate that the hybrid deep learning model has better performance in terms of forecasting accuracy than the other methods. Therefore, the hybrid deep learning model can efficiently provide credible long-term predictions for price of Nickel forecasting.

Keywords: Planning, Scheduling and Coordination
Abstract: Stainless steel has been widely used in various fields, such as chemical industry, oil refining industry, medical, daily necessities industry, …, etc. According to the statistics of the World Steel Association (WSA), the main region of global crude steel output is in Asia (Total of output is 68.6%). This study develops a job shop scheduling with restricted set-up time for steel manufacturers. The objection function includes total weighted completion time, total weighted tardiness, and work-in-process queue time. Moreover, the restricted set-up time is considered for steel manufacturers. The change process set-up time which considers the thickness and width of steel. This study adopts genetic algorithm to search the optimal scheduling result. The performance shows the job shop scheduling with restricted set-up time model can provide robust and credit scheduling result for steel manufacturers.

Keywords: Intelligent and Flexible Manufacturing, Big-Data and Data Mining
Abstract: Thanks to the rapid development of information technology, the concept of smart manufacturing has been widely promoted in recent years, and gave birth to a large number of industrial software. The large amount of data generated and accumulated in industrial software has the characteristics of distributed, multi format and complex relationship. The traditional relational database cannot complete the unified description of these data and respond to their rapid changes. Dataspace is a new data management method that can deal with multi-source heterogeneous data and update dynamically. Therefore, this paper determines the requirements of industrial system data integration. Based on this, a construction and update process of industrial dataspace is proposed. Metadata extraction is used to complete the unified description of data, and then through the construction of data relations and data-service mapping, multi-source heterogeneous data can be effectively organized. Finally, the effectiveness of this method is evaluated through a case study in a nuclear power enterprise.

Keywords: Manufacturing, Maintenance and Supply Chains, Planning, Scheduling and Coordination
Abstract: High-end equipment is manufactured using a “main manufacturer–supplier” mode. The supply–demand relationship of high–end equipment involves benefit conflicts among the customer, the main manufacturer, and multiple suppliers, so transaction decisions become very complicated. Traditional optimisation methods are inadequate at revealing the interactions among multiple stakeholders. In this study, based on the transaction process of high-end equipment manufacturing, two correlated Stackelberg game models, namely, “main manufacturer–customer” and “main manufacturer–supplier,” are constructed and their Nash equilibria are solved to maximize the profit of each stakeholder. The effects of various parameters on the decision variables of each stakeholder are analyzed through numerical simulation.

Keywords: Manipulation Planning
Abstract: Rotational slippage due to the gravitational torque is common in grasping tasks using a parallel gripper. To pick the object with a specific grasp pose, if the parallel gripper cannot provide enough force to eliminate the gravitational torque, the rotational slippage will cause the manipulation failure. This paper presents an approach to eliminate rotational slippage by pivoting an object on a supporting surface to a pose where it can be picked up, while minimizing the gravitational torque. For evaluation we compare the proposed method with simple grasping, which lifts the object directly, showing that our approach can significantly reduce the rotational slippage problem and help in task completion.

Keywords: Manipulation Planning, Compliant Joints and Mechanisms, Deep Learning in Robotics and Automation
Abstract: Vision-based techniques for grasp planning of robotic end-effectors have been successfully deployed in pick-and-place tasks. However, for computing the optimal grasp, they assume the gripper to be of a rigid parallel-jaw type or a single-point vacuum suction-based design. Planners for soft robotic grippers have used learning from demonstration or heuristics that rely on the compliance of the gripper to achieve a grasp. We demonstrate a model-free reinforcement learning (RL) approach that modifies vision-based grasp plans generated for parallel-jaw grippers and adapts them to grasping with a four-fingered soft gripper. The observed state of the RL model is comprised of the grasp plans from the vision module, the deformation of the fingers, and the pose of the end-effector. The RL model controls each finger separately, discovering grasp synergies during training. This approach is compared to a baseline grasp synergy where all four fingers simultaneously enclose the object. In simulation, we achieve a pick-and-place success rate of 58.4% with the RL model for top-down grasping, compared to 43.2% with the baseline grasp synergy.

Keywords: Manipulation Planning, Planning, Scheduling and Coordination
Abstract: Manipulation of deformable linear objects (DLOs) is a key robot capability in applications such as manufacturing, logistics, and healthcare. DLOs are commonly found in industrial and domestic environments in the form of cables, ropes, and wires. However, dexterous manipulation of these objects autonomously is computationally expensive due to their infinite degrees of freedom in 3-D space. Manipulation of DLOs is also typically constrained by the environment, which poses additional challenges due to restricted robot motions. In this study, we propose a method for automating the manipulation of DLOs in constrained spaces. The approach contains a geometric model of cable-like objects based on multiple features including point cloud, color, and shape. Given the estimated 3-D cable model and environment information, we propose a model-based planning methodology for manipulating DLOs in constrained spaces. We demonstrate the efficiency and robustness of our method by automating a cable threading task with real robot experiments using an assembly task board designed by the National Institute of Standards and Technology.

Keywords: Reactive and Sensor-Based Planning, Computer Vision in Automation, Agricultural Automation
Abstract: For robotic tasks in unknown and large environments, it is important to locate targets to be manipulated by the robot in the environment. If the workspace is too large to capture at a time within the limited measurement field of visual sensors, it is necessary to explore the workspace by moving sensors. However, previous research which models almost the whole environment conduct unnecessarily exhaustive exploration for task execution. The purpose of this research is to enable robots to locate targets in a large workspace in a short time for task execution. Target exploration is formulated as a variation of a 2D black-box optimization problem. A sample-efficient exploration algorithm is proposed based on Bayesian optimization(BO) with Gaussian Process(GP). To utilize the continuity of physical motion of the camera, the algorithm for continuous search is proposed. Furthermore, to utilize prior knowledge about target positions, if available, a method of incorporating prior distribution in GP is proposed, which enables even faster exploration. Experimental results in both simulation and real environments demonstrated the efficiency of the proposed method compared to hand-crafted or naive BO-based exploration algorithms.

Keywords: Factory Automation, Foundations of Automation, Robotics and Automation in Life Sciences
Abstract: Using a deformable sheet to handle objects is convenient and found in many practical applications. For object manipulation with multi-mobile robot team through deformable sheets, modeling of the robots-sheet-object system is a main challenge, since there are complex interactions among the robot team formation, deformable sheet and the object. Thus, in this paper, a virtual variable cables model (VVCM) is proposed to simplify the modeling of the robots-sheet-object system with N robots. With the VVCM, we can compute the kinematics of the robot team and manipulate the object with the changing of robot team formations. We further extend the VVCM to a motion planner with the capabilities of both bypassing and crossing obstacles, since the height of the object is controllable in this case. The motion planner can be formulated as a formation optimization problem. Then, local paths can be planned with the computed formation. Simulation and experimental results with different robot team sizes show the effectiveness and versatility of the proposed VVCM. The motion planner can successfully plan an obstacle crossing path in complex scenarios where other benchmark planners fail.

Keywords: Formal Methods in Robotics and Automation, Foundations of Automation, Optimization and Optimal Control
Abstract: Soft robots show important application prospects because of their infinite degree of freedom, flexibility and dexterity, among which parallel extensible soft robots can elongate and bend at the same time, and are capable of crawling, grasping and manipulation. However, the constraints between pneumatic soft actuators (PSAs) and the coupling of elongation and bending make it harder to build dynamic model than fixed-length soft robots. In this paper, we propose a dynamic modeling method capable of describing soft robots that can elongate and bend at the same time. First, we give the characteristics of parallel extensible soft robot. Then we deduce the dynamic model of PSA. By developing the interactive constraint topological structure (ICTS), we further deduce the complete dynamic model of parallel extensible soft robot. Finally, we carry out simulation experiments and it proves that the established model can well describe the multi-degree-of-freedom motion and the constraint relations of parallel PSAs.

Keywords: Process Control, Foundations of Automation
Abstract: This work proposes an adaptive control methodology for a class of time-varying uncertain systems. The goal of this work is to address problems in which the model is known, but the parameters involved are either unknown, time-varying, or the system is being affected by disturbances. The proposed methodology relies on Lyapunov control concepts and the bound of the time-varying elements (only the boundedness feature is known) to extend traditional adaptive control techniques restricted to linear constant parameterization. In order to show the efficiency of this methodology, we present a comparison simulation and an implementation for speed tracking in a Brush-Less Direct Current (BLDC) motor under parametric uncertainty.

Keywords: Model Learning for Control, Compliant Joints and Mechanisms, Hydraulic/Pneumatic Actuators
Abstract: Soft grippers are gaining momentum across applications due to their flexibility and dexterity. However, the infinite-dimensionality and non-linearity associated with soft robots challenge modeling and closed-loop control of soft grippers to perform grasping tasks. To solve this problem, data-driven methods have been proposed. Most data-driven methods rely on intensive model learning in simulation or offline, and as such it may be hard to generalize across different settings not explicitly trained upon and in physical robot testing where online control is required. In this paper, we propose an online modeling and control algorithm that utilizes Koopman operator theory to update an estimated model of the underlying dynamics at each time step in real-time. The learned and continuously updated models are then embedded into an online Model Predictive Control (MPC) structure and deployed onto soft multi-fingered robotic grippers. To evaluate the performance, the prediction accuracy of our approach is first compared against other model-extraction methods among different datasets. Next, the online modeling and control algorithm is tested experimentally with a soft 3-fingered gripper grasping objects of various shapes and weights unknown to the controller initially. Results indicate high success ratio in grasping different objects using the proposed method. Sample trials can be viewed at https://youtu.be/i2hCMX7zSKQ.

Keywords: Failure Detection and Recovery
Abstract: This work introduces the design of a sliding mode fault diagnoser capable of diagnosing simultaneous faults in a three-wheeled omnidirectional mobile robot. The approach proposes an augmented system comprising the system states and the dynamics of the faults. Based on this, a sliding mode observer is designed to estimate both, the system state and fault magnitudes, allowing the fault diagnosis. This approach leads to light and efficient implementations, which is suitable for real-time applications. The approach has been tested in a real mobile robot showing satisfactory results.

Keywords: Motion Control
Abstract: Compared to wheeled and tracked robots, legged robots like quadruped robots have much more degrees of freedom, which makes them capable to maintain a constant posture on uneven terrains as well as moving platforms. Previous research has focused on the locomotion of quadruped robots in multiple motionless unstructured environments, while the problem of walking on swaying platforms like the shaking cable bridge is rarely studied. The posture stabilization control on swaying surfaces remains an open problem because of the unpredictability of the external forces and torques acting on the feet caused by the movement of the platform. The main work of this article is to present a method that maps static gaits from stationary rigid platforms to swaying rigid platforms while keeping the torso posture level during the whole walking process. The proposed method can greatly suppress the vibration of the fragile items or attached cameras carried by the robot, which is important in situations such as emergency response. Simulation results confirm the effectiveness of the method proposed.

Keywords: Motion Control, Collaborative Robots in Manufacturing, Robust/Adaptive Control
Abstract: In this paper, an innovative manufacturing system based on a dual-arm robot, each arm with 5 degree-of-freedom (DOF), is proposed. Although the proposed system can work in the automation mode or machining mode, only the function of machining is studied here. In this case, the system is equivalent to a 5-axis machine tool. The method of equivalent errors is employed to design a robust contouring controller. It is shown that reducing the equivalent contour error in the joint space is equivalent to reducing both position contour error and tool orientation error in the task space. As a comparison, a conventional distributed controller is also designed. Several cases of experiments have been conducted, including a circular path with 4 different feedrates and an elliptic path. The experimental results show that the proposed method can achieve much better results in all cases, and demonstrate the feasibility of adopting the dual-arm robotic system for machining task.

Keywords: Motion Control, Deep Learning in Robotics and Automation, Computer Vision in Automation
Abstract: Vision-based control provides a significant potential for the end-point positioning of continuum robots under physical sensing limitations. Traditional visual servoing requires feature extraction and tracking followed by full or partial pose estimation, limiting the controller's efficiency. We hypothesize that employing deep learning models and implementing direct visual servoing can effectively resolve the issue by eliminating such intermediate steps, enabling control of a continuum robot without requiring an exact system model. This paper presents the control of a single-section tendon-driven continuum robot using a modified VGG-16 deep learning network and an eye-in-hand direct visual servoing approach. The proposed algorithm is first developed in Blender software using only one input image of the target and then implemented on a real robot. The convergence and accuracy of the results in normal, shadowed, and occluded scenes demonstrate the effectiveness and robustness of the proposed controller.

Keywords: Optimization and Optimal Control, Discrete Event Dynamic Automation Systems, Probability and Statistical Methods
Abstract: Ordinal optimization (OO) is a powerful tool for optimizing discrete-event dynamic systems (DEDS). It seeks to correctly select the best among a finite set of system designs via sampling. In this note, we consider a well-known OO method called optimal computing budget allocation (OCBA). It includes a set of optimality conditions for sample allocations that can intelligently determine the number of samples allocated to each design. Sample allocations that satisfy the optimality conditions have been shown to be the asymptotic optimizer of the probability of correct selection of the best design. We analyze a numerically robust OCBA algorithm and study its convergence rates under different performance measures in the setting of Gaussian samples and known variances. It fills the gap of convergence analysis for the OCBA algorithm. We also provide numerical results to show its empirical performance.

Keywords: Simulation and Animation, Optimization and Optimal Control, Probability and Statistical Methods
Abstract: We study optimization problems on a continuous decision space using the Ordinal Transformation framework. The original method was proposed to exploit the advantages of bi-fidelity simulation models, so observations from the low-fidelity can efficiently guide sampling on the high-fidelity model. However, this framework only worked on a discrete decision space. We extend the framework to continuous cases and through preliminary analysis, we show promising performance and resent future research directions.

Keywords: Planning, Scheduling and Coordination, Machine learning, Discrete Event Dynamic Automation Systems
Abstract: Intelligent manufacturing systems require real-time optimization algorithms for daily operations management. Scheduling rules have been proven to be efficient and commonly used in plenty of practical production scenarios, especially for the large-scale problems. However, almost all the scheduling rules are manually designed, which is time consuming and also results in the large loss of accuracy for complex problems. This paper proposes a new simulation optimization-aided learning method, denoted by SOaL, for design automation of scheduling rules. The proposed SOaL method treats the automated design of scheduling rules as a simulation optimization problem, where we use genetic programming algorithm to guide the rule generation and introduce ranking and selection algorithm to improve the rule evaluation accuracy. Using dynamic job shop scheduling problem as the simulation testbed, numerical results show the superiority of the proposed method.

Keywords: Big-Data and Data Mining, Machine learning, Probability and Statistical Methods
Abstract: Nested simulation is commonly used in estimating risk measures of a portfolio with complicated financial products, e.g., options and swaps, but due to its nested structure the nested simulation needs a certain amount of time to implement, and could not be used in real-time risk monitoring. Recently, Jiang et al. proposed a simulation analytics approach that combines simulation models with machine learning methods to monitor real-time risk of financial portfolio. They argue that the sample paths and derivative prices can be stored in a database and used to estimate online risk measures based on the current values of the underlying risk factors without running additional simulation experiments.Different from the setting in Jiang et al., we consider another real-time monitoring regime in this work. In their work, the time of measuring the risk of the portfolio is fixed at point T. Because they think T may be the end of the fiscal year at which portfolio loss needs to be reported to shareholders, or T is the end of the investment cycle at which bonuses are distributed. Then the managers of the portfolio may be interested in estimating the loss distribution and the portfolio risk measures at time T. However, we take a different view. The time of measuring the risk of the portfolio may not be fixed. Instead, we would like to measure the risk with a fixed time interval T. This will sound more like to be the way of measuring risk in practice.

Keywords: Building Automation, Smart Home and City, Power and Energy Systems automation
Abstract: In this paper, recent intelligent methods to modeling and control of smart buildings in the sense of continuous control theory, such as refrigeration, ventilation and air conditioning systems are reviewed. The models are subdivided into physics-based models, data-driven models, and gray-box models, with more emphasis given to data-driven models. A complete study of the control methods is made. These are divided into conventional control methods, intelligent control methods, and mixed control methods.

Keywords: AI-Based Methods, Machine learning, AI and Machine Learning in Healthcare
Abstract: Recognising the human context in terms of ongoing human activities is of major importance to ensure an efficient context-aware assistance. In this paper, a hybrid approach combining deep learning and probabilistic commonsense reasoning is proposed for anticipating human activities in AmI environments. Deep learning models are exploited for recognising environment objects, human hands and user's indoor locations. To implement probabilistic commonsense reasoning, probabilistic fluents are introduced in the formalism of event calculus formulated in answer set programming (ECASP). The reasoning axiomatization is based on an ontology describing the user's context when performing an activity. Using reasoning based on temporal projection and abduction enables an eXplainable AI (XAI) approach for activity anticipation. Experimental results show the high accuracy of inferences in terms of activity anticipation and a very low computation time in knowledge-intensive scenarios, rendering the system compatible with real-time applications.

Keywords: Industrial and Service Robotics, Collaborative Robots in Manufacturing
Abstract: Robot Learning from Demonstration (RLfD) is a research field that focuses on how robots can learn new tasks by observing human performances. Existing RLfD approaches mainly enable robots to repeat the demonstrated tasks by mimicking human activities, which usually requires efficient demonstrations for human experts. This paper proposes a new Function Object-Oriented Network (FOON) based approach to make robots learn and optimize assembly tasks from non-expert demonstrations. It first proposes an assembly FOON construction approach with automatic subgraph creation and merging algorithms to extract information from multiple non-expert demonstrations. It then proposes an assembly task tree retrieving approach with a robot execution optimization process to make the robot learn and generate the best possible task execution plan from the constructed FOON. The proposed approaches are validated through experiments with a dual-arm YuMi robot and the experimental results illustrate the effectiveness and advantages of the proposed approach.

Keywords: Machine learning, AI-Based Methods
Abstract: Increasing the semantic understanding and contextual awareness of machine learning models is important for improving robustness and reducing susceptibility to data shifts. In this work, we leverage contextual awareness for the anomaly detection problem. Although graphed-based anomaly detection has been widely studied, context-dependent anomaly detection is an open problem and without much current research. We develop a general framework for converting a context-dependent anomaly detection problem to a link prediction problem, allowing well-established techniques from this domain to be applied. We implement a system based on our framework that utilizes knowledge graph embedding models and demonstrates the ability to detect outliers using context provided by a semantic knowledge base. We show that our method can detect context-dependent anomalies with a high degree of accuracy and show that current object detectors can detect enough classes to provide the needed context to show good performance within our example domain.

Keywords: Sustainability and Green Automation
Abstract: Aiming at the problems of high spatial and temporal dispersion, strong quality uncertainty and weak cross-organizational information correlation in the recycling of decommissioned electromechanical products, this paper proposes a knowledge graph-based method for tracing the full life cycle information of decommissioned electromechanical products. First, the structure of the knowledge graph of the full life cycle information of decommissioned electromechanical products is constructed, then the key techniques for constructing the knowledge graph of the full life cycle information of decommissioned electromechanical products are proposed, and finally, the effectiveness of the method is verified through examples, which can provide strong support for the full life cycle information tracing of decommissioned electromechanical products.

Keywords: Big data Analytics for Large-scale Energy Systems, Sustainability and Green Automation, Machine learning
Abstract: To achieve correct operation of wind farms, it is necessary to create accurate wind energy forecasting. Autoregressive integrated moving average (ARIMA) models were combined with artificial neural networks (NN) to obtain acceptable forecasting accuracy. But the forecasting results become worse when there are missing data or local minima in NN.

In this paper, we use the multiple models and transfer-learning techniques to ARIMA. Since different wind farms have some similar features, we can use different ARIMA models and their wind farms' data to get pre-training features. Then we do fine training by using the transfer-learning to combine these ARIMA models. This novel method can solve the low forecasting accuracy problems of ARIMA and NN. We successfully apply this method to wind energy forecasting. Experimental results show the forecasting accuracy of one wind farm is improved using the pre-trained models of the other two farms.

Keywords: Power and Energy Systems automation, Optimization and Optimal Control, Agent-Based Systems
Abstract: A microgrid consists of multiple distributed energy resources and loads and can work at both grid-connected and autonomous modes. The smart microgrid is an ideal way for renewable integration and a key building block for a smart grid. However, a microgrid is very hard to control due to its small inertia, large uncertainties, and wide range of operating conditions. To unlock the potential of microgrids, new control solutions need to be developed to improve stability, flexibility, and energy efficiency. There are significant research opportunities in this highly interdisciplinary research area. During the past decade, Dr. Liu’s team has been actively performing theoretical and experimental studies on microgrid control. His group has designed a series of advanced distributed algorithms for microgrid operation and control. Benefited from his control background, he is able to perform rigorous stability analyses for the designed algorithms. To overcome the limitations of control theory and further improve control performance, he recently extended his research to deep reinforcement learning-based control of dynamic systems. In addition to algorithm designs, his lab also puts significant effort into hardware development for algorithm experimentation. His lab is well equipped with multiple real-time digital simulators and self-developed microgrid test-beds. Based on the advanced equipment, his group can perform both controller-/power- hardware-in-the-loop simulations and hardware experimentations. The experimental study has significantly improved the TRL of his research outcome and inspired future research topics. During Dr. Liu’s presentation, he will introduce his previous and current projects and research plan on microgrid control.

Keywords: Motion and Path Planning, Planning, Scheduling and Coordination, Swarms
Abstract: This paper investigates the scheduling problem related to engaging a swarm of attacking drones with a single defensive turret. The defending turret must turn, with a limited slew rate, and remain facing each of the drones for a dwell time to eliminate it. It must eliminate all the drones in the swarm before any drone reaches the turret. In 2D, this is an example of a Traveling Salesman Problem with Time Windows (TSPTW) where the turret must visit each target during the window. In 2D, the targets and turret are restricted to a plane and the turret rotates with one degree of freedom. In 3D, the turret can pan and tilt, while the drones attempt to reach a safe zone anywhere along the vertical axis above the turret. This 3D movement makes the problem more challenging, since the azimuth angles of the turret to the drones vary as a function of time. This paper investigates the theoretical optimal solution for simple swarm configurations. It compares heuristic approaches for the path scheduling problem in 2D and 3D using a simulation of the swarm behavior. It provides results for an improved heuristic approach, the Threat-Aware Nearest Neighbor.

Keywords: Optimization and Optimal Control, Motion and Path Planning, Automation Technologies for Smart Cities
Abstract: Real-time optimal path planning for robotic manipulations in task space is a very fundamental and important problem. In this paper, the problem of generating robot trajectories in an obstacle-ridden environment is formulated under an optimal control framework using Hamilton-Jacobi-Bellman (HJB) equation. The novel contribution of this paper is that a closed form HJB control solution (a necessary and sufficient condition for global optimality of a control solution with respect to a cost function) has been achieved for generating real-time optimal trajectories for a robot manipulator. In contrast with the decoupled end-effector path planning and subsequent trajectory generation, the proposed scheme can exploit sensory input for real-time trajectory generation where the end-effector path as well as the joint trajectory is recomputed online while satisfying the real-time constraints. The stability and the performance of the proposed control framework is shown theoretically via Lyapunov approach and also verified experimentally using a 6 degrees of freedom (DOF) Universal Robot (UR) 10 robot manipulator. It is shown that a significant saving in cost metrics can be obtained over similar trajectory generation approaches from the state-of-the-art with obstacle-ridden environment and also has better performance in high speed tracking applications. Warehouse applications of the proposed scheme in case of static and dynamic targets with respect to the robot manipulator is also included.

Keywords: Agent-Based Systems, Motion and Path Planning, Energy and Environment-aware Automation
Abstract: We study a coverage and tour planning problem wherein a robot with limited sensor coverage is assigned to serve n known points in an environment. A location is served if it is within the visibility range of the robot’s sensor. However, the robot is equipped with a battery that powers the robot to travel a maximum distance of B. Several charging stations are placed in the environment so that the robot can charge itself (if needed) to complete the mission. The objective is to compute an energy-constrained tour that starts at a given start location, visits a minimum set of service locations for serving the n points of interest, and returns to the start location. We also aim to minimize the distance traveled between any two consecutive service locations in the tour via a subset of charging stations. This problem has applications in search-and-rescue and surveillance missions where such coverage and energy-aware path planning are of utmost importance. We propose a new algorithm for this problem and show its efficacy using experiments with up to 1000 points of interest on a plane. The running time of our algorithm for such a scenario was a negligible 5.33 sec.

Keywords: Behavior-Based Systems, Motion and Path Planning, Intelligent Transportation Systems
Abstract: There is a sudden surge in the number of people ordering items online that are facilitated through warehouses. Modern-day warehouses use robots for most of the tasks including picking and sorting. Efficiency is of prime concern in any warehouse, while current efforts in the literature are restricted to solving the optimization of warehouse processes as an operational research problem using fixed travel costs. As the number of orders and robots increases, congestion is witnessed in the warehouse that invalidates fixed travel costs assumptions. To facilitate research in warehousing we first propose a modular simulator for the warehouse that simulates the business operations of order generation, order fulfillment scheduling, item picking, and sorting. The simulator also models the robots for traveling within the warehouse network, scheduling charging, intersection management, and congestion management. With the increasing demand, the warehouses increase the number of robots making the transportation network operate beyond capacity. In this direction, we analyze the performance of the warehouse from a transportation perspective using fundamental diagrams. The warehouses contain only controllable entities (robots) that enable predicting the congestion levels for solving the planning problem. The results show improvements of around 5% in the order fulfillment time and the number of picks that can significantly increase the profitability of the warehouse.

Keywords: Collaborative Robots in Manufacturing, Motion and Path Planning, Optimization and Optimal Control
Abstract: In today’s manufacturing companies that rely on Human-Robot Collaboration (HRC), ensuring a safe and ergonomic workplace is becoming of pivotal importance. In a collaborative assembly scenario, this paper aims at planning the trajectory of a cobot arm, guaranteeing safety and ergonomics for the operator without neglecting production efficiency requirements. In particular, a multi-objective optimization approach for the trajectory planning in safe and ergonomic Human-Robot Collaboration is defined, with the aim of finding the best trade-off between the total traversal time of the trajectory for the robot and ergonomics for the human worker, while respecting safety requirements. The proposed approach consists of three main steps. First, the Rapid Upper Limb Assessment (RULA) ergonomic index is evaluated on a manikin designed on a dedicated software. The aim is to ensure a high quality of work in the considered HRC scenario with a consequent decrease of the musculoskeletal disorders associated with highly repetitive and dangerous activities. Second, a time-optimal and safety-constrained trajectory planning problem is defined as a second-order cone programming problem. Finally, a multi-objective control problem is formulated and solved to compute the trajectory that ensures the best compromise between time end ergonomics. The method is tested on numerical simulations and the obtained results are discussed, proving the effectiveness of the approach.

Keywords: Motion and Path Planning, Autonomous Vehicle Navigation, Intelligent Transportation Systems
Abstract: In this paper, a robot navigating an environment shared with humans is considered, and a cost function that can be exploited in RRTX, a randomized sampling-based replanning algorithm that guarantees asymptotic optimality, to allow for a safe motion is proposed. The cost function is a path length weighted by a danger index based on a prediction of human motion performed using either a linear stochastic model, assuming constant longitudinal velocity and varying lateral velocity, and a GMM/GMR-based model, computed on an experimental dataset of human trajectories. The proposed approach is validated using a dataset of human trajectories collected in a real world setting.

Keywords: Embedded Systems for Robotic and Automation, Planning, Scheduling and Coordination
Abstract: This work introduces a methodology to consider preemption and migration overhead in hard real-time cyclic executives on multicore architectures. The approach takes two iterative stages. The first stage computes a cyclic executive, after which the number and timing of all preemptions and migrations for every task is known. Then, it includes this overhead by updating the worst-case execution time (WCET) of the tasks. The second stage calculates a new cyclic executive considering the new WCET of tasks. The stages iterate until the preemption and migration overhead keeps constant.

Keywords: Optimization and Optimal Control, Planning, Scheduling and Coordination
Abstract: This paper addresses the non-preemptive unrelated parallel machine scheduling problems with machinedependent and sequence-dependent setup times, machine eligibility restrictions and hard Q-time constraints. The objective is to minimize the makespan. All jobs are assumed to be available at time zero and all times are deterministic. To formulate this complex problem, a mixed integer linear programming (MILP) model is developed and the lower bound is derived to evaluate the proposed algorithms. Due to the complexity of the problem, the nested partition framework is introduced and two new heuristic algorithms are designed based on it. One is embedded with one-step-backtracking rule (NP-O), and the other uses root-backtracking (NP-R). The performance is evaluated by conducting both algorithms on test instances. The computational results demonstrate that both proposed algorithms are efficient and effective at solving small scale instances while NPO outperforms NP-R for large scale problems.

Keywords: Intelligent Transportation Systems, Automation Technologies for Smart Cities, Autonomous Agents
Abstract: Urban mobility solutions such as mobility-on-demand services have become prevalent given the convenience of door-to-door transport. However, a majority of these approaches are user-centric greedy solutions that cause traffic congestion. We propose a near social-optimal routing algorithm which accounts for the overall network traffic congestion. Specifically, we leverage on multi-class mobility options to dissipate traffic congestion while maintaining near social optimal travel time efficiency. We divide each route into three parts with micro-mobility options such as walking or cycling for the first and last parts and on-demand cars for the middle part of the route. In addition, we propose a computational and travel time efficient transit point search algorithm for switching between different modes of travel. We validate our approach by using a diverse set of road networks from different cities. We achieve an average of 84% increase in network utilization by using our proposed multi-class social model compared to single-class user-centric approach. Our proposed transit point search algorithm is on average 68% more computationally efficient with an insignificant maximum average travel time delay of less than 5 seconds compared to an optimal exhaustive routing solution.

Keywords: Planning, Scheduling and Coordination, Task Planning, Reinforcement
Abstract: In this work, we introduce a novel framework to model and solve Job-Shop scheduling problem. In particular, we view the job-shop scheduling as a parameterized sequential decision-making (para-SDM) problem; where the parameters correspond to machine availability time, and the sequential decisions (policy) correspond to the sequences in which the jobs are processed by the machines. The proposed framework results into a discrete state and action space that scales linearly with respect to the number of jobs and machines. Thus, allowing to develop simple and efficient reinforcement learning techniques when the underlying stochasticity and cost functions are unknown. Our framework is also flexible to model and incorporate several inclusion-exclusion, and capacity constraints in terms of the decision variables.

Keywords: Planning, Scheduling and Coordination, Autonomous Agents, Task Planning
Abstract: We address multi-robot geometric task-and-motion planning (MR-GTAMP) problems in synchronous, monotone setups. The goal of the MR-GTAMP problem is to move objects with multiple robots to goal regions in the presence of other movable objects. To perform the tasks successfully and effectively, the robots have to adopt intelligent collaboration strategies, i.e., decide which robot should move which objects to which positions, and perform collaborative actions, such as handovers. To endow robots with these collaboration capabilities, we propose to first collect occlusion and reachability information for each robot as well as information about whether two robots can perform a handover action by calling motion-planning algorithms. We then propose a method that uses the collected information to build a graph structure which captures the precedence of the manipulations of different objects and supports the implementation of a mixed-integer program to guide the search for highly effective collaborative task-and-motion plans. The search process for collaborative task-and-motion plans is based on a Monte-Carlo Tree Search (MCTS) exploration strategy to achieve exploration-exploitation balance. We evaluate our framework in two challenging GTAMP domains and show that it can generate high-quality task-and-motion plans with respect to the planning time, the resulting plan length and the number of objects moved compared to two state-of-the-art baselines.

Keywords: Planning, Scheduling and Coordination, Failure Detection and Recovery, Task Planning
Abstract: This paper takes the first step towards a reactive, hierarchical multi-robot task allocation and planning framework given a global Linear Temporal Logic specification. The capabilities of both quadrupedal and wheeled robots are leveraged via a heterogeneous team to accomplish a variety of navigation and delivery tasks. However, when deployed in the real world, all robots can be susceptible to different types of disturbances, including but not limited to locomotion failures, human interventions, and obstructions from the environment. To address these disturbances, we propose task-level local and global reallocation strategies to efficiently generate updated action-state sequences online while guaranteeing the completion of the original task. These task reallocation approaches eliminate reconstructing the entire plan or resynthesizing a new task. To integrate the task planner with low-level inputs, a Behavior Tree execution layer monitors different types of disturbances and employs the reallocation methods to make corresponding recovery strategies. To evaluate this planning framework, dynamic simulations are conducted in a realistic hospital environment with a heterogeneous robot team consisting of quadrupeds and wheeled robots for delivery tasks.

Keywords: AI and Machine Learning in Healthcare, AI-Based Methods, Big-Data and Data Mining
Abstract: Myocardial infarction (MI), also known as heart attack, is the leading cause of death in the United States. Accurate MI prediction is of critical importance to reduce healthcare costs and save lives. Rapid developments in healthcare data infrastructure and information technology provide an unprecedented opportunity for data-driven MI prediction. However, real-world medical data are generally subject to a high level of uncertainty with imbalanced issue and considerable missing values, which pose significant challenges for reliable disease prediction. Realizing the full potential of medical data calls upon the development of novel machine learning methods that are capable of handling the uncertainty factors in medical data. In this paper, we propose a Multi-Branching Neural Network (MB-NN) framework for robust and reliable MI prediction. First, we implement the weighted K-Nearest Neighbors (wKNN) method to estimate the missing values in the medical data. Second, we develop a Hierarchical Clustering (HC)-based under-sampling approach to create multiple balanced sub-datasets from the original imbalanced data to eliminate the potential bias caused by imbalanced data distribution in model training. Third, we combine a multi-branching architecture with multi-layer perceptron (MLP) to further handle the imbalanced data for robust MI prediction. We evaluate the proposed MB-NN framework on the medical records from the Cohort Component of the Atherosclerosis Risk in Communities (ARIC). Experimental results show that the MB-NN method achieves better performance in MI prediction compared with existing widely used machine learning methods.

Keywords: AI-Based Methods, Big Data in Robotics and Automation, Process Control
Abstract: Causal relationships are commonly examined in manufacturing processes to support faults investigations, perform interventions, and make strategic decisions. Industry 4.0 has made available an increasing amount of data that enable data-driven Causal Discovery (CD). Considering the growing number of recently proposed CD methods, it is necessary to introduce strict benchmarking procedures on publicly available datasets since they represent the foundation for a fair comparison and validation of different methods. This work introduces two novel public datasets for CD in continuous manufacturing processes. The first dataset employs the well-known Tennessee Eastman simulator for fault detection and process control. The second dataset is extracted from an ultra-processed food manufacturing plant, and it includes a description of the plant, as well as multiple ground truths. These datasets are used to propose a benchmarking procedure based on different metrics and evaluated on a wide selection of CD algorithms. This work allows testing CD methods in realistic conditions enabling the selection of the most suitable method for specific target applications. The datasets are available at the following link:

Keywords: AI-Based Methods, Embedded Systems for Robotic and Automation, Intelligent and Flexible Manufacturing
Abstract: Novel Artificial Intelligence (AI) approaches try to process an excessive amount of field-level data. However, challenges arise as network bandwidth is limited, and thus this data cannot be entirely transferred to the cloud for further processing. Edge computing tries to overcome that limitation by bringing the computational resources closer to the data generating sources. However, edge devices are also constraint by both CPU power and memory, and strict real-time requirements of the manufacturing domain have to be met. Thus, the selection of suitable devices for specific AI algorithms poses a severe challenge. Currently, the choice is often made by a trialand- error approach or by selecting more powerful devices than needed. This paper tries to address those challenges by showing relevant aspects for algorithm benchmarking in the manufacturing domain. Selected algorithms, namely Grubbs Test, Butterworth Filter, DBSCAN, Random Forest, Support Vector Machine, Matrix Multiplication, and Matrix Inversion, are examined. Analysis of their theoretical time and space complexity sheds some light on the behaviour of the algorithms with respect to their input data points. In addition, relevant metrics for the manufacturing domain, such as execution time, memory consumption, and energy consumption, are identified. This paper furthermore examines the algorithm behaviour on various heterogeneous hardware devices, such as PLCs, an MCU, IPCs, a single-board computer, and a dedicated edge device. Altogether, this paper can guide selecting suitable algorithms and hardware to equip Cyber-Physical Production Systems (CPPS) with novel data processing solutions thoughtfully. Moreover, the presented metrics can support the creation of novel ML benchmarks for smart manufacturing (SM).

Keywords: AI-Based Methods, Surveillance Systems
Abstract: Although deep learning-based calibration methods can predict extrinsic camera parameters from a single image, the accuracy of these methods is severely degraded due to “data uncertainty” e.g., noisy and outlying input images. To address this problem, we propose a novel Data Uncertainty-Driven Loss (DUD Loss), could derive the uncertainty of input image, during the camera calibration process. Instead of estimating the camera extrinsic parameter as scalar numbers, the proposed method models it as a Gaussian distribution with its variance representing the uncertainty of the input image. Hence, each camera parameter is no longer a deterministic scalar, but a probabilistic value with diverse distribution possibilities. With the help of DUD loss, the network can be trained to alleviate the perturbations caused by noisy input images. Furthermore, in the real-world single image camera calibration process, noisy input images, which yielded larger variance/uncertainty could be effectively omitted without increasing the computation cost. To evaluate the DUD loss, we also propose a new large-scale, Dataset for the Vehicle-Infrastructure Collaborative Autonomous Driving task, which contains millions of objects (e.g., cars, trucks, pedestrians, and so on) along with the camera calibration parameters (e.g., roll, pitch, yaw angles, and height of the camera setting position), dubbed as VICAD. Extensive experiments conducted on the proposed database demonstrate the effectiveness of the proposed method. Experimental results show that our proposed DUD loss has achieved a more accurate camera calibration prediction result than L1 loss or Huber loss functions without further increasing computation cost.

Keywords: Learning and Adaptive Systems, Industrial and Service Robotics, Motion and Path Planning
Abstract: This paper presents an approach for transferring surface finishing behaviors to new surfaces while preserving the quality of the process. The idea is to let a human demonstrate the desired grinding behavior on a planar surface and subsequently generate an equivalent grinding behavior on new surface geometry. The transfer of the process quality is accomplished by imitating the material removal rate of a human. This is achieved with an adaptive control that relies on the online estimation of the material removal rate, which depends on the contact area, normal force, tool speed, and tool wear. The proposed approach is verified in simulation and experimentally validated on the grinding of planar and curved surfaces.

Keywords: Reinforcement, Agent-Based Systems, AI-Based Methods
Abstract: The disaster relief operations during floods require time critical information of the flooded area to save lives. Finding critical regions of the disaster struck area in a limited time frame is crucial for effective relief planning. In this paper, we propose a multi-UAV based system with directed explorations of flooded area to gather time-critical ground information using deep reinforcement learning based controls. We learn an exploration policy for the multi-UAV system with limited battery for autonomous coverage of the flooded region. Further, we integrate D8 flow algorithm that approximates the water flow direction based on image pixel information of a sub-region in the UAVs' exploration strategy. The results show that our proposed method for multi-UAV exploration of flooded area outperforms other methods from the literature. Moreover, the learnt multi-UAV exploration policy is able to generalize to unseen flooded regions without any retraining.

Keywords: Manufacturing, Maintenance and Supply Chains, Intelligent and Flexible Manufacturing
Abstract: The performance of a smart manufacturing system is affected by not only the constituent processes but also their system-level interactions. However, in most of the current studies, process modeling and system-level performance evaluation are independent. This can substantially impact production efficiency. In this paper, utilizing both the system level and process level parameters, an integrated data-enabled model is developed that can seamlessly fuse two conventionally separated analysis. A fast recursive method is developed to evaluate the system yield. Based on the proposed integrated model, the permanent production loss (PPL) concept is defined and is evaluated. Furthermore, PPL attributions due to random downtime and quality issues have been identified. Case studies have shown the high fidelity of the integrated model and the effectiveness of the PPL analysis in identifying the root cause of production yield loss.

Keywords: Manufacturing, Maintenance and Supply Chains, Intelligent and Flexible Manufacturing
Abstract: This paper aims at modeling and real-time control of a flexible manufacturing system (FMS) that is operated by multi-skilled mobile robots. By introducing the idea of a unique ideal clean configuration and the effective disruption event, the concepts of Opportunity Window (OW) and Permanent Production Loss (PPL) are extended to the FMS. The control considered in this paper is a robot assignment problem where individual robots are dynamically assigned to each workstation in real time to improve performance. The problem is formulated as a Markov Decision Process(MDP) and solved using the Double Deep Q-Network (DDQN) reinforcement learning algorithm. PPL is used as the reward setting for training and a space discretization is used between each workstation to capture real time movement of the robots on the plant floor. The effectiveness of the control strategy is then demonstrated by comparing the results to several heuristic control schemes.

Keywords: Probability and Statistical Methods, Intelligent and Flexible Manufacturing, Failure Detection and Recovery
Abstract: Modeling stress-induced processes is challenging and extremely critical in the quality control of advanced manufacturing systems. While residual stresses may be beneficial in some situations, in composite structures assembly, high residual stresses and extreme deformations are crucial and must be accounted for to prevent future catastrophic failures. Currently, conventional approaches to the optimal placement of actuators on composite structures are non-optimal, require ten actuators heuristically, and are insufficient in considering residual stresses. To overcome these limitations, we propose a Stress-Aware Optimal Actuator Placement framework. The stress-aware optimal actuator placement framework is able to achieve significant reductions of at least 39.3% in mean root mean squared deviations (RMSD) and 52% in maximum forces (MF), and only requires eight actuators on average while satisfying the safety threshold of residual stresses.

Keywords: Factory Automation, AI-Based Methods, Intelligent and Flexible Manufacturing
Abstract: Finding the root causes and their definite effects on machining outcomes is challenging when using the complicated and unexplainable weights and adopting the structure of a traditional artificial intelligence (AI) model, e.g., neural networks and random forests (RF). This research proposes a suboptimal and explainable Scheme (SES) for monitoring and reasoning machining outcomes by data-driven decision tree models. After machining, the time-frequency features in the wavelet package are extracted from the sensing data. Then, the key features are selected via a mixed-integer optimization method and served as inputs of the random-forests-based local search tree (RF-LST) for estimating machining outcomes. The estimation results of machining tool wear and surface roughness indicate that this RF-LST has interpretable features and structure. Also, the performance of RF-LST is comparable to those of random forests, gradient boosting, and the long short-term memory model.

Keywords: Sustainability and Green Automation, Hybrid Strategy of Intelligent Manufacturing, Automation Technologies for Smart Cities
Abstract: Due to increased technological advances and the high use of phones, tablets, computers, and other electronics, we continue to see rapid growth in the volume of e-waste. On the one hand, there are millions discarding their older devices adding to the growing volume of e-waste, while on the other hand there are millions who would benefit from receiving such devices. To address this problem, a smart e-waste marketplace has been implemented – this marketplace matches donated e-waste items to potential recipients. The inherent complexity here lies in matching donors with recipients as there are many constraints and optimization criteria (inventory, geo, logistics/costs, reduce bias, maximum allocation). We provide an overview of the “smart” e-waste marketplace and summarize experimental results comparing different optimization criteria, incentives and present a live web service that allows for e-waste supplies to reach schools and nonprofit institutions.

Keywords: Zero-Defect Manufacturing, Intelligent and Flexible Manufacturing, Factory Automation
Abstract: The concepts of Industry 4.1 for achieving Zero-Defect (ZD) manufacturing were disclosed in IEEE Robotics and Automation Letters in January 2016. ZD of all the deliverables can be achieved by discarding the defective products via a real-time and online total inspection technology, such as Automatic Virtual Metrology (AVM). Further, the Key-variable Search Algorithm (KSA) of the Intelligent Yield Management (IYM) system developed by our research team can be utilized to find out the root causes of the defects for continuous improvement on those defective products. As such, nearly ZD of all products may be achieved. However, in a multistage manufacturing process (MMP) environment, a workpiece may randomly pass through one of the manufacturing devices with the same function in each stage. Different devices of the same type perform differently in each stage, where the performances will be accumulated through the designated manufacturing process and affect the final yield. KSA can only identify the influence of univariate variables (i.e., single devices) on the yield, yet it cannot detect the manufacturing paths that have significant influence on the yield. In order to cope with this deficiency such that the golden path with a better yield amongst all the MMP paths can be found, this research proposes the Golden Path Search Algorithm (GPSA), which can plan golden paths with high yield under the condition of the number of variables being much larger than that of samples. As a result, it makes the improvement of manufacturing yield be more comprehensive.

Keywords: Embedded Systems for Robotic and Automation, Control Architectures and Programming
Abstract: Hardware programming skills are essential to develop control software of robot-like-systems. However, a significant percentage of mechanical engineering students struggles, especially if ‘complex’ programming constructs like arrays, loops and pointers are combined. A concept for continuous monitoring of learning success throughout a course on digitalization and C programming using different objective and subjective assessments is introduced. After a pre-analysis of challenges in different learning milestones, an application-oriented approach to teach these concepts using real world engineering examples is proposed. Immediate and continuous feedback is recorded during lectures to check students’ learning progress. Emotional engagement serves as subjective assessment. As objective assessment, three different methods are used: classical web-based assessments, muddy cards and exam rates. The results confirm the benefit: 5 percentage points better exam results and a 15 percentage points reduced dropout rate in web-based assessments. Nevertheless, the concept of pointers, interlinked with dynamic data types such as linked lists remain troublesome in the exam despite very good results in web-based assessments. The mapping of a 2D information like a matrix of sensor values into a 1D storage structure like a data register using the appropriate address type seems to be too abstract to understand for mechanical and process engineers.

Keywords: Software, Middleware and Programming Environments, Optimization and Optimal Control, Robust/Adaptive Control
Abstract: While controllers for modern robots often appear to users as monolithic black-boxes, approaches exist to write custom controllers and tune them with tools from community based frameworks. Using ROS2 as such a framework, single components and parameters can be exchanged and customized easily. Especially in research institutions but also in industrial environments this framework is used increasingly to fulfill the demanding requirements of robotic control. While ROS1 had no real-time capable method of communication, ROS2 promises to give this advantage. Matching the trend to use many distributed, specialized and small systems instead of one higly system-specific, inflexible system, this study assesses tuning of DDS parameters in order to improve communication speed.

Keywords: Human-Centered Automation, Software Architecture for Robotic and Automation
Abstract: The rapid technological leaps of Industry 4.0 require the consideration of human factors, which is one of the key drivers of Industry 5.0. In particular, automation software development for mechatronic systems is becoming increasingly demanding, as both domain knowledge and programming skills are required for high-quality maintainable software. Low-code platforms enable domain experts to write software using visual programming languages, even without in-depth software engineering skills. Especially for small companies from automation and robotics without dedicated software engineering departments, domain-specific low-code platforms become indispensable to enable domain experts with little programming knowledge to develop code by intuitively combining graphical elements, e.g., for tasks such as retrofitting mobile machines. However, for extensive functionalities, visual programs may become overwhelming due to the scaling-up problem. In addition, the ever-shortening time-to-market to remain globally competitive increases the time pressure on programmers. This letter thus presents an assistance system concept that can be implemented by low-code platform suppliers based on combining data mining and static code analysis. Domain experts are supported in writing low-code by providing targeted recommendations, measuring complexity using software metrics, and reducing complexity by encapsulating functionalities. The concept is implemented for the industrial low-code platform HAWE eDesign to program hydraulic components in mobile machines, and its benefits are successfully evaluated in a user study and an industrial expert workshop.

Keywords: Industrial and Service Robotics, Autonomous Agents
Abstract: We present a methodology for a service robot to complete the task of putting away the groceries in a precise and stable configuration that satisfies the desired semantic relationship of grocery and pantry objects. This task is invaluable to the motor impaired and increases the capability of contextualized service robots. We show that augmenting traditional geometric assessment with contextual knowledge of food categorization, allows the robot to complete this task efficiently. We quantitatively validate our approach with a data set of 51 common grocery items, of which 11 objects are used for real world experiments. According to our evaluation, our method is able to successfully shelve objects next to semantically related objects 100% of the time when these relationships exist. We achieve this with an average placement precision of 3.2cm and a standard deviation of 1.1cm. We discuss remaining challenges and needed improvements for robots with these capabilities to be introduced to home environments.

Keywords: Product Design, Development and Prototyping, Factory Automation, Logistics
Abstract: This paper describes the design of a conveyor belt manipulator for repositioning boxes to locate the bar code side in the right position to be sensed in a logistic center. The general rule is that a few pairs of hands touch the package as possible in parcel logistics centers, ideally just three. First for loading, second for unloading the containers, and third for a worker has to place the packages on the conveyor logistic system in the right position, i.e., the bar code must be visible by the vision system. The correct position of the bar code is essential for the vision systems to check the delivery information and manage the package automatically through the correct conveyor in the logistics centers. That process needs to be automated using a collaborative robot (COBOT) and a specific manipulator design for that task. In this work, we present a reposition boxes methodology with a novel conveyor belt manipulator. Our novel design has been achieved by analyzing the manipulation characteristics needed to control, rotate, and flip boxes in a real logistic industry. Then, a prototype was built and tested in the laboratory.

Keywords: Product Design, Development and Prototyping, Motion and Path Planning, Telerobotics and Teleoperation
Abstract: A key factor in the economic efficiency of satellites is their availability in orbit. Replacing standardized building blocks, such as empty fuel tanks or outdated electronic modules, could greatly extend the satellites' lifetime. This, however, requires flexible robots that can locomote on the surface of these satellites for optimal accessibility and manipulation. This paper introduces ReCoBot, a 7-axis walking space manipulator for locomotion and manipulation. The robot can connect to compatible structures with its symmetric ends and provides interfaces for manual teleoperation and motion planning with a constantly changing base and tip. We build on open-source robotics software and easily available components to evaluate the overall concept with an early stage demonstrator. The proposed manipulator has a length of 1.20 m and a weight of 10.4 kg and successfully locomotes over a satellite mockup in our lab environment.

Keywords: Collaborative Robots in Manufacturing, Human Factors and Human-in-the-Loop, Compliance and Impedance Control
Abstract: This letter presents a robotic system for a deep-micro-hole robotic drilling application, used in a glass container mold production industry. The delicate task of robotic micro-drilling is achieved with the developed framework capable of capturing the skills from an experienced human operator and repeating the task with the robotic system. Special mechanical and software setup is proposed for on-site recording of specific tasks performed by a human operator in real-case scenarios. Recorded motions are localized with respect to the product surface, combining the a priori known CAD models of the moulds and the proposed scanning and localization procedure. Finally, the obtained expert skill is applied to the robotic system for deep-micro-drilling. The performance of the proposed methods is experimentally validated.

Keywords: Collaborative Robots in Manufacturing, Human-Centered Automation, Industrial and Service Robotics
Abstract: Mobile manipulators have become increasingly popular in industry because they can be used for a large variety of tasks in a versatile and flexible manner. The perception and planning/control schemes of mobile robots enable them to share a workspace with humans. However, as undesired or unforeseen contacts can generally not be avoided, it is essential to understand the intrinsic safety properties of such systems. Then, collision safety can be systematically accounted for in mechanism design, planning, and control. In this paper, we derive the safety characteristics of wheeled mobile manipulators, more specifically the achievable reflected mass and velocity range, and show how they can be related to human injury data in the previously introduced Safety Map for a model-independent and interpretation-free safety assessment. We investigate two common types of wheeled mobile platforms and the combination of these with the seven-DOF Franka Emika robot. We analyze the influence of the vehicle parameters on the safety performance and derive the Safety Map representations for four practically relevant industrial collision scenarios.

Keywords: Collaborative Robots in Manufacturing, Optimization and Optimal Control, Logistics
Abstract: We consider a fleet of elementary robots that can be connected in different ways to transport loads of different types. For instance, a single robot can transport a small load and the association of two robots can either transport a large load or two small loads. We seek to determine the number of robots necessary to transport a set of loads in a given time interval, at minimum cost. The cost is function of the number of robots and of the distance travelled by robots. The fleet sizing problem can be formulated by an integer linear problem. In the special case of two types of loads and two configurations, closed-form expressions for the minimum number of robots can be derived. Finally, we show how reconfigurability can allow to diminish the number of required robots.

Keywords: Compliance and Impedance Control, Collision Avoidance, Collaborative Robots in Manufacturing
Abstract: Kinesthetic teaching allows human operators to reprogram part of a robot's trajectory by manually guiding the robot. To allow kinesthetic teaching, and also to avoid any harm to both the robot and its environment, Cartesian impedance control is here used for trajectory following. In this paper, we present an online method to modify the compliant behavior of a robot toward its environment, so that undesired parts of the robot's workspace are avoided during kinesthetic teaching. The stability of the method is guaranteed by a well-known passivity-based energy-storage formulation that has been modified to include a strict Lyapunov function, i.e., its time derivative is a globally negative-definite function. Safety Control Barrier Functions (SCBFs) that consider the rigid-body dynamics of the robot are formulated as inequality constraints of a quadratic optimization (QP) problem to ensure forward invariance of the robot's states in a safe set. An experimental evaluation using a Franka Emika Panda robot is provided.

Keywords: Industrial and Service Robotics, Collaborative Robots in Manufacturing
Abstract: In hands-on tasks, operator arm and robot are linked together, forming a unique dynamical system whose stability is crucial to guarantee a safe and comfortable human-robot interaction. To this extent, a passivity-based adaptation strategy is here proposed, allowing to modify the parameters of the robot control system, in accordance with changes in the operator arm impedance, so that stability is guaranteed. Experimental results, including a comparison with the methodology introduced in~cite{bib:Landi2017a}, demonstrate the importance of considering human arm impedance changes in the adaption strategy, in order to guarantee stability without causing excessive changes in admittance filter parameters.

Keywords: Formal Methods in Robotics and Automation, Reinforcement, Collaborative Robots in Manufacturing
Abstract: Complex safety-critical cyber-physical systems, such as autonomous cars or collaborative robots, are becoming increasingly common. Simulation-based falsification is a testing method for uncovering safety hazards of such systems already in the design phase. Conventionally, the falsification method takes the form of a static optimization. Recently, dynamic optimization methods such as reinforcement learning have gained interest for their ability to uncover harder-to-find safety hazards. However, these methods may converge to risk-maximising, but irrelevant behaviors. This paper proposes a principled formulation and solution of the falsification problem by automata constrained reinforcement learning, in which rewards for relevant behavior are tuned via Lagrangian relaxation. The challenges and proposed methods are demonstrated in a use-case example from the domain of industrial human-robot collaboration, where falsification is used to identify hazardous human worker behaviors that result in human-robot collisions. Compared to random sampling and conventional approximate Q-Learning, we show that the proposed method generates equally hazardous, but at the same time more relevant testing conditions that expose safety flaws.

Keywords: Autonomous Agents, Factory Automation, Process Control
Abstract: We have designed, implemented and tested a mobile robot capable of autonomously measuring the flush and gap in car bodies. The selection of the target object, a corresponding measurement plan and the rough position on a map of the environment are the only information needed to trigger the system with full autonomy - a versatile system for quality assurance in industrial environments.

Keywords: Software, Middleware and Programming Environments, Formal Methods in Robotics and Automation
Abstract: A smart manufacturing system is a complex cyber-physical system consisting of a collection of component machines and a floorplan layout defining the spatial relationship between components. Each component may be of different physical behavior with different control software. Simulation and testing on smart manufacturing systems require a software infrastructure that can orchestrate the execution of heterogenous, cyber-physical components besides modeling physical machines in respect to floorplan layouts. Automated simulation as a result is challenging and error-prone. Recent strides in formal modeling of cyber-physical systems and programming languages offer some new techniques for addressing this challenge. In this paper, we present a compositional automata-based modeling formalism and programming abstractions to design coordination logic between heterogeneous robots in different layouts. Our formalism allows us to automatically simulate and compare performance metrics for different floorplan layouts. We implement our proof-of-concept prototype with the challenging simulation environment for 2021 Agile Robotics for Industrial Automation Competition. Our experiment results demonstrate how our simulation can be used to evaluate and compare performance under different layouts and applicable for reconfiguration and virtual commissioning.

Keywords: Semiconductor Manufacturing, Planning, Scheduling and Coordination, Product Design, Development and Prototyping
Abstract: In high-technology industries, fierce competition forces firms to constantly introduce new products into the market to replace older generations in a process known as product transition. This high pace of innovation imposes continuous pressure to invest resources in development activities that will yield revenue in the future. At the same time, the firm must allocate resources to meet the current market demand, causing products from different generations to compete for resources. In addition to financial resources, development activities require manufacturing capacity in an industry where short-term capacity expansion is infeasible. Hence, manufacturing capacity is shared by fully developed products on the market, providing current revenue and products in development that will provide future revenue.

Keywords: Learning and Adaptive Systems, Sustainability and Green Automation, Semiconductor Manufacturing
Abstract: We consider a scheduling problem for a single batch processing machine in semiconductor wafer fabrication facilities (wafer fabs). An integrated objective function that combines the total weighted tardiness (TWT) and electricity cost (EC) is considered. A time-of-use (TOU) tariff is assumed. A genetic pro-gramming (GP) procedure is proposed to automatically discover dispatching rules for list scheduling approaches. Results of computational experiments demonstrate that the learned dis-patching rules lead to high-quality schedules.

Keywords: Robotics and Automation in Construction, Robotics and Automation in Life Sciences
Abstract: In this article, the problem of designing a practical active disturbance rejection control (ADRC) scheme for a class of differentially flat omnidirectional mobile robots is addressed. A custom version of ADRC is proposed that uses a special observer that allows the controller feedforward input to be designed with an anti-peaking functionality, which helps to decrease the peaking phenomenon in the observer response. To further increase the practical appeal of the proposed ADRC design, the control algorithm here is derived using only the robot kinematic model and assumes the robot position and orientation as the only available system information. Experimental results, including a comparison with the proposal without the mechanism and a classic controller are shown using a laboratory robot operating in an irregular terrain, verifying the effectiveness of the proposed governing scheme in terms of trajectory tracking and disturbance rejection.

Keywords: Energy and Environment-aware Automation, Automation in Construction, Learning and Adaptive Systems
Abstract: Pavement sweeping, which is primarily labor-intensive, is essential to keep it clean and hygienic for use. Humans play the role of identifying the litter to pick and adjust the vacuum suction power. In this paper, we propose a framework that consists of two layers, namely, a) the first method is to identify commonly found litter on pavements, b) secondly, an adaptive vacuum suction scheme based on fuzzy logic is implemented for more efficient pick up of the identified litter. Semantic segmentation using Convolution Neural Network (CNN) SegNet was adopted to segment the pavement region from other objects. Then, the Deep Convolutional Neural Network (DCNN) based object detection is used to detect pavement litter. Afterward, the calibrated vacuum suction as per identified litter was selected based on fuzzy-based adaptive actuation. Further, the proposed framework's efficacy is successfully tested on a self-reconfigurable pavement sweeping robot named Panthera. The experimental results showed that our technique identifies the pavement litter with satisfactory accuracy and uses less energy compared to conventional cleaning methods for pavement cleaning task. The inspection framework was configured in Jetson Nano Nvidia GPU and took approximately 132.2 milliseconds to detect litter. The proposed technique is suitable for deploying real-time pavement litter detection. In the experiment conducted, there is a 38.5 % improvement in energy consumption for the pavement cleaning task using a depth-based vision system and a vacuum suction motor.

Keywords: Biomimetics, Reactive and Sensor-Based Planning, Collision Avoidance
Abstract: The paper proposes a bio-inspired approach to the obstacle avoidance problem using analytic wavelet transform to extract perception cues for the reactive control. The small and large scale environment features are perceived using laser scans. Wavelets are then used as the bases of projection of the spatial signals, in order to decompose and observe the environment features of interest. The wavelet-based element analysis is leveraged for this purpose, using Morse wavelets as the analytic wavelet bases. The inspiration of the approach is derived from the principle of sensorimotor convergence that has been studied in many small living organisms. A steering potential function, based on a human walking model, is finally used to close the control loop. The safety of the method is analyzed by computing appropriate barrier certificates. The algorithm is programmed and released as a ROS/C++ package and is thoroughly tested using physical hardware experiments.

Keywords: Motion and Path Planning
Abstract: To solve the autonomous navigation problem in complex environments, an efficient motion planning approach is newly presented in this paper. Considering the challenges from large-scale, partially unknown complex environments, a three-layer motion planning framework is elaborately designed, including global path planning, local path optimization, and time-optimal velocity planning. Compared with existing approaches, the novelty of this work is twofold: 1) a novel heuristic-guided pruning strategy of motion primitives is proposed and fully integrated into the state lattice-based global path planner to further improve the computational efficiency of graph search, and 2) a new soft-constrained local path optimization approach is proposed, wherein the sparse-banded system structure of the underlying optimization problem is fully exploited to efficiently solve the problem. We validate the safety, smoothness, flexibility, and efficiency of our approach in various complex simulation scenarios and challenging real-world tasks. It is shown that the computational efficiency is improved by 66.21% in the global planning stage and the motion efficiency of the robot is improved by 22.87% compared with the recent quintic Bezier curve-based state space sampling approach. We name the proposed motion planning framework E³MoP, where the number 3 not only means our approach is a three-layer framework but also means the proposed approach is efficient in three stages.

Keywords: Motion and Path Planning, Task Planning, Motion Control
Abstract: This paper addresses the problem of transporting a rigid box filled with unknown objects with a dual-arm robotic system. Enforcing non-sliding contact behavior, which guarantees the transportation of the box despite the unknown load’s action, is the main difficulty in this setting. To solve this problem, we propose a high-level model-predictive controller, which uses a nonlinear extended state observer to estimate the external disturbance, and determine the wrench required to the box for tracking a trajectory. A quadratic program transforms the calculated wrench into optimal desired contact forces on the two end-effectors. Finally, a low-level admittance control framework with an inner velocity loop is established to indirectly control the actual contact forces. We verify the effectiveness of the proposed control method with experiments carried out on a real dual-arm robotic system

Keywords: Motion Control, Optimization and Optimal Control, Robust/Adaptive Control
Abstract: Linear motors are widely used in automation industry where both the tracking performance and the productivity are of importance. The inevitable uncertainties in system parameters and process nonlinearities and the existence of hard constraints due to physical limitations of control input and system states often makes the simultaneous improvement of the tracking accuracy and the time efficiency a daunting theoretical problem to solve. In this paper, the constrained time-optimal tracking control problem of the linear motors under uncertainties and hard constraints is addressed through a seamless integration of the constrained optimization methods and the adaptive robust controls. Specifically, the demand for high productivity under state and input constraints is met by solving a constrained time-optimal trajectory replanning problem using the Pontryagin maximum principle (indirect method). The resulting replanned trajectories are then fed into the low-level adaptive robust controller that effectively deals with the parametric uncertainties and uncertain nonlinearities for high tracking performance. Comparative experiments have been carried out and validated the superiority of the proposed method over existing ones.

Keywords: Autonomous Vehicle Navigation, Motion Control
Abstract: An autonomous bikebot (i.e., bicycle-like robot) is an attractive single-track platform for off-road, agile navigation applications. It is challenging for bikebots to navigate at low velocity on off-road, cluttered terrains. In this paper, we design a wheel-leg hybrid bikebot control system. The bikebot control can switch between different actuation modes. At low-speed movement and on off-road, bumpy terrains, the regular steering-induced balance torque itself cannot effectively balance the platform and the leg-assisted balance torque is used. A model predictive control is designed for the leg assistive actuation to take advantage of the leg-ground interaction force and balance torque. By doing so, the bikebot can safely navigate and balance in various off-road environments. High-fidelity simulation results are presented to demonstrate that the wheel leg bikebot can efficiently navigate at low speed in cluttered space and keep balance on bumpy terrains.

Keywords: Motion and Path Planning, Collision Avoidance
Abstract: Many commodity sensors that measure the robot and dynamic obstacle’s state have non-Gaussian noise characteristics. Yet, many current approaches treat the underlying uncertainty in motion and perception as Gaussian, primarily to ensure computational tractability. On the other hand, existing planners working with non-Gaussian uncertainty do not shed light on leveraging distributional characteristics of motion and perception noise, such as bias for efficient collision avoidance. This paper fills this gap by interpreting reactive collision avoidance as a distribution matching problem between the collision constraint violations and Dirac Delta distribution. To ensure fast reactivity in the planner, we embed each distribution in Reproducing Kernel Hilbert Space and reformulate the distribution matching as minimizing the Maximum Mean Discrepancy (MMD) between the two distributions. We show that evaluating the MMD for a given control input boils down to just matrix-matrix products. We leverage this insight to develop a simple control sampling approach for reactive collision avoidance with dynamic and uncertain obstacles. We advance the state-of-the-art in two respects. First, we conduct an extensive empirical study to show that our planner can infer distributional bias from sample-level information. Consequently, it uses this insight to guide the robot to good homotopy. We also highlight how a Gaussian approximation of the underlying uncertainty can lose the bias estimate and guide the robot to unfavorable states with a high collision probability. Second, we show tangible comparative advantages of the proposed distribution matching approach for collision avoidance with previous non-parametric and Gaussian approximated methods of reactive collision avoidance

Keywords: Intelligent and Flexible Manufacturing, Industrial Robots, Software Architecture for Robotic and Automation
Abstract: The research community is puzzled with words like skill, action, atomic unit and others when describing robots' capabilities. However, for giving the possibility to integrate capabilities in industrial scenarios, a standardization of these descriptions is necessary. This work uses a structured review approach to identify commonalities and differences in the research community of robots' skill frameworks. Through this method, 210 papers were analyzed and three main results were obtained. First, the vast majority of authors agree on a taxonomy based on task, skill and primitive. Second, the most investigated robots' capabilities are pick and place. Third, industrial oriented applications focus more on simple robots' capabilities with fixed parameters while ensuring safety aspects. Therefore, this work emphasizes that a taxonomy based on task, skill and primitives should be used by future works to align with existing literature. Moreover, further research is needed in the industrial domain for parametric robots' capabilities while ensuring safety.

Keywords: Industrial Robots, Intelligent and Flexible Manufacturing, Reinforcement Learning
Abstract: In this paper, we propose an approach for collision-free trajectory planning of multiple robot manipulators in a common workspace. In recent years, robot arms are often introduced to factories in place of human beings, and it has become important how efficiently multiple robot arms can be operated in a small space. The problem of trajectory planning for multiple robot arms is often solved by graph search algorithms, however, it is difficult for the conventional approach to provide flexible trajectory planning to cope with unexpected situations such as robot arm failure. Therefore, we propose a combined method for Q-learning and RRT for the trajectory planning problem. The effectiveness of the proposed method is further verified using numerical experiments. The planned trajectories is able to guarantee a certain degree of optimality when the motion trajectory is generated by combining reinforcement learning than by using only the graph search algorithm. The results indicate that the time required to generate the motion trajectory is reduced by the proposed method.

Keywords: Intelligent and Flexible Manufacturing, Sensor-based Control, Robust Manufacturing
Abstract: Process monitoring and quality control for industrial robotic Gas Metal Arc Welding (GMAW) systems are key components in ensuring the reliability of the produced products. While being a widely used process, there is still a lack of a robust, plug and play monitoring solution. In particular, weld bead penetration depth is a crucial factor in many fabrication applications, where substantial bonding strength is crucial. This paper introduces a new penetration depth estimation method using the emitted acoustic signal to monitor the droplet transfer process. By monitoring the droplet transfer, an estimation of the welding energy transferred to the base material can be obtained while accounting for variations in the welding process. Using this method, the penetration depth is able to be measured within an error of +-15%, proving to be a promising solution for online monitoring in robotic welding applications.

Keywords: Intelligent and Flexible Manufacturing, Cyber-physical Production Systems and Industry 4.0, Optimization and Optimal Control
Abstract: This paper studies the following basic flexible manufacturing problem: Given N machines that can perform the same job on a production item (e.g., drilling or tapping) but with different capabilities (e.g., energy requirements and speeds), what is an optimal schedule for the job on these machines? While this is a well-studied problem, the main innovation this paper introduces is the explicit modeling of the underlying process dynamics - i.e., the physical interaction of the item and the machine - using differential equations. The resulting scheduling problem is in a hybrid systems setting that involves determining the transition times between states, where the system evolution in each state is defined by differential equations. To the best of our knowledge, such a cyber-physical systems (CPS) oriented approach to machine scheduling has not been studied before, although it lies at the core of flexible manufacturing in Industry 4.0. We believe that this new formulation might lead to a renewed interest in machine scheduling problems, but now in a hybrid/CPS-oriented setting.

Keywords: Intelligent and Flexible Manufacturing, Factory Automation
Abstract: To ensure stable processing and high-yield production, high-tech factories (e.g., semiconductor, TFT-LCD) demand product quality total inspection. Generally speaking, sampling inspection only measures a few samples and comes with metrology delay, thus it usually cannot achieve the goal of real-time and online total inspection. Automatic Virtual Metrology (AVM) was developed to tackle such problem. It can collect the data from the process tools to conjecture the virtual metrology (VM) values in the prediction model for realizing the goal of online and real-time total inspection. With the advancement of technology, the processes become more and more precise, and better accuracy of VM value prediction is demanded. The CNN-based AVM (denoted as AVMCNN) scheme can not only enhance the accuracy of the original AVM prediction, but also perform better on the extreme values. Nevertheless, two advanced capabilities need to be addressed for its practical applications: 1) effective initial-model-creation approach with insufficient metrology data; and 2) intelligent self-learning capability for on-line model refreshing. To possess these two advanced capabilities, the Advanced AVMCNN System based on convolutional autoencoder (CAE) and transfer learning (TL) is proposed in this work. It is verified that the Advanced AVMCNN System is more feasible for the onsite applications of the actual production lines.

Keywords: Intelligent and Flexible Manufacturing, Manufacturing, Maintenance and Supply Chains, Domain-specific Software and Software Engineering
Abstract: The semiconductor industry is characterized by highly complex manufacturing processes. In order to face these challenges, research is being advanced through funded projects. These have a great intricacy because of the number of partners involved and its complex problem statements, thus, it is essential to achieve an efficient way of communicating knowledge. A possible method is using collaborative ontologies, in particular, the use of DrOWLing for modelling. The term DrOWLing is composed of ‘Drawing‘ and ‘OWL‘. By creating DrOWLings, thinking in triples becomes more intuitive as it omits heavy restructuring, keeping the intuition in the creation of ontologies by avoiding ontology tool based restrictions and facilitate thinking. Using this approach, on the one hand reusable results are obtained in the early stages, thus, facilitating understanding and reporting when dealing with complex and new problem statements for the partners. On the other hand, DrOWLings can easily become an ontology, meaning that this information is in a machine-readable format. DrOWLing development improves the efficiency when working within and across domains or projects – like funded projects. It is a tool that can connect projects, domains with each other tailored to humans understanding and is almost in the machine readable semantic web format. DrOWLings have also found usages in different contexts besides the semantic web one.

Keywords: Prosthetics and Exoskeletons, Machine learning, Human-Centered Automation
Abstract: Industrial workers often perform awkward gaits such as squatting, kneeling, etc. for a prolonged time when conducting skilled tasks. We present a real-time wearable sensing and exoskeleton control design to provide assistance for the industrial workers under awkward gaits. A wearable sensor-based gait activity detection and pose estimation scheme is designed to predict the human motion and lower-limb joint angles in real time during a sequence of walking, standing, squatting, and kneeling gaits. Wearable bilateral exoskeletons provide assistive torques at various gaits under a multi-level controller. Human-subject experiments are presented to demonstrate the gait detection and exoskeleton control performance. The results show that the overall accuracy of human gait recognition is up to 95% and the average detection latency is around 50 ms. The exoskeleton control strategy reduces muscle activation in knee extension/flexion up to 25% during various stationary gaits and posture transitions.

Keywords: Motion Control, Robust/Adaptive Control, AI and Machine Learning in Healthcare
Abstract: Due to high biological adaptability and flexibility, pneumatic artificial muscle (PAM) systems are widely employed in exoskeleton robots to accomplish rehabilitation training with repetitive motions. However, some intrinsic characteristics of PAMs and inevitable practical factors, e.g., high nonlinearity, hysteresis, uncertain dynamics, and limited working space, may badly degrade tracking performance and safety. Hence, this paper designs a new learning-based motion controller for PAMs, to simultaneously compensate for model uncertainties, eliminate tracking errors, and satisfy preset motion constraints. Particularly, when PAMs suffer from periodically non-parametric uncertainties, the elaborately designed continuous update algorithm can repetitively learn them online to enhance tracking accuracy, without employing upper/lower bounds of unknown parts for controller design and gain selections. Meanwhile, some non-periodic uncertainties are handled by a robust term, whose value is only related to the initial states of PAMs, instead of exact upper bounds of unknown dynamics. From safety concerns, we introduce error-related saturation terms to limit initial amplitudes of control inputs within saturation constraints and avoid overlarge errors inducing overlarge acceleration. Meanwhile, the constraint-related auxiliary term is utilized to keep tracking errors within allowable ranges. To the best of our knowledge, this paper presents the first learning-based error-constrained controller for uncertain PAM-actuated exoskeleton robots, to realize high-precision tracking control and improve safety without additional gain conditions. Moreover, the asymptotic convergence of tracking errors is strictly proven by Lyapunov-based stability analysis. Finally, based on a self-built exoske

Keywords: Medical Robots and Systems, Rehabilitation, Prosthetics and Exoskeletons
Abstract: This paper presents an adaptive quasi-static model-based control algorithm for controlling the motion of a soft robotic exo-digit while physically interacting with the human hand for a continuous passive motion physical therapy. Quasi-static analytical models were developed for modeling the soft robot, the human finger, and their coupled physical interaction. A discrete-time state-space representation was derived for the position control system. A control input was designed to linearize the input-output where the input is the actuation pressure and the output is the overall bending angle of the distal end which follows the reference input. The analytical models and controller were examined through experimental testing for two case studies: 1) a step-response with a disturbance attenuation and 2) for tracking the desired angular bending of the distal end while adaptively adjusting the control gain to achieve the desired dynamic performance. The results showed the effectiveness of the adaptive quasi-static controller.

Keywords: Industrial Robots, Factory Automation, Software Architecture for Robotic and Automation
Abstract: Automation has been playing a major role in manufacturing such as in automotive, electronics, and pharmaceutical industries. While robots are able to perform repeatable tasks with speed and accuracy, its use in garment industry has been limited by challenges in grasping and handling soft fabrics using commercially available robot end effectors. This paper considers a common garment manufacturing process, fabric fusing, which combines a piece of woven or knitted fabric with an interface material (interlining) to provide additional firmness and support. Current practice uses human operators to perform fabric pick-up, alignment, and manipulation tasks to feed the combined materials into a fusing machine. We have developed a robotic system with an electroadhesion robot gripper containing actuated pins to alleviate human operators from performing these repeated and laborious tasks in an uncomfortable manufacturing environment. This robotic system can reliably pick-up fabric pieces, place them without wrinkles, align them using machine vision, and feed the combined bundle through the conveyor belt into the fusing machine. The actuated pins interlaced with the electrodes on the gripper can detach the fabric from the gripper with residual charges and to slide the bundle onto the conveyor belt without affecting the fabric alignment. The electroadhesion force depends on the applied voltage, fabric material property, and humidity. The environmental condition needs to be controlled and the applied voltage adjusted based on the type of fabric materials and humidity to achieve reliable performance. The prototype system has been demonstrated in both the laboratory setting and actual garment manufacturing shop floor.

Keywords: Factory Automation, Human-Centered Automation, Industrial Robots
Abstract: In this paper, we describe a method of measuring the force applied to a finger pad by attaching two strain gauges to the nail. We target sewing work at production sites. To measure the dexterous fingertip force work, we determine the position where the strain gauge is attached based on how the nails are distorted upon force application. Additionally, we devise a procedure for proper installation and removal of the proposed sensor. Then, we verify via experiments that the force in the shear direction can be measured, the pressing position of the finger pad can be estimated, and work needing fine fingertip force can be classified. In conclusion, the proposed sensor offers more detailed sensing than conventional sensors.

Keywords: Automation in Life Science: Biotechnology, Pharmaceutical and Health Care, AI and Machine Learning in Healthcare, Clinical and Operational Decision Support
Abstract: Accessibility to eye care in New Zealand is a significant issue that causes cases of preventable blindness. One reason for this issue arises from shortages of specialists, thus presenting obstacles such as inconsistent triaging times and long travel times. Consequently, a patient requiring urgent medical attention may sometimes be unable to access the required care in a timely manner. Moreover, another cause of inaccessibility is the scarcity of specialised equipment in certain regions resulting in insufficient diagnostic capabilities by GPs and specialists alike. Thus, methods are being developed to address these issues. An automatic triaging system using deep learning and image processing algorithms is being explored. This system will use classification models that utilise categorical patient data alongside retinal fundus images to predict a triaging severity level to organise patient referrals. Furthermore, to tackle issues with image quality inconsistencies, the system will include sub-systems with functions that include image quality grading and enhancement. The emergence of affordable portable smartphone-based ophthalmology imaging devices has been one approach to tackling the lack of ophthalmology equipment. However, these devices have issues in a limited field of view and lower quality when compared to expensive standard devices. Thus, retinal fundus mosaicking was developed based on the SIFT algorithm to create an image with a greater field of view using multiple images pertaining to different views of a patient’s retina. However, this method can be improved to enhance its reliability and thus its clinical applicability. Therefore, a deep learning key point detection-based image registration process is being developed.

Keywords: Automation in Life Science: Biotechnology, Pharmaceutical and Health Care, Robotics and Automation in Life Sciences, Industrial and Service Robotics
Abstract: Benzodiazepines are psychoactive agents due to their effect on the central nervous system. They are widely used as psychotropic drugs and pharmaceuticals. Benzodiazepines are characterized by high effectiveness and rapid action. Due to their effect on the nervous system and their high potential for dependency and misuse, researchers are interested in understanding the pharmacology and physiology. Typically, benzodiazepines are detected in human urine, whole blood, plasma and serum as well as in meconium. Such measurement tasks belong to compound-oriented measurements, which often require a labor-intensive sample pretreatment—including sample clean-up, matrix change and reformatting the sample containers. To ensure a stable quality in the entire measurement process, the sample preparation for serum treatment were automated using a Biomek i7 Hybrid Workstation and a positive pressure unit. The benzodiazepines Alprazolam, Clonazepam, Diazepam, Lorazepam, Midazolam, Nordazepam, Oxazepam and Temazepam were determined using liquid chromatography and tandem mass spectrometry (LC-MS/MS). The performance of the automated method—repeatability, recovery rate, within-laboratory precision, measurement precision, as well as limits of detection and quantification—was compared with measurements of benzodiazepine standards in acetonitrile.

Keywords: Process Control, Robotics and Automation in Life Sciences
Abstract: Left ventricular assistive devices (LVAD) have been used for patients who experience congestive heart failures and are waiting for heart transplantation. These devices can help patients to resume their daily life/work until a suitable donor is found. LVAD mainly consists of a rotary pump that is directly attached from the left ventricle to the aorta using two cannulas. To meet patients' needs, a control algorithm should be implemented to adjust the pump speed in different physiological conditions such as exercising, resting, etc. However, the accurate control of LVADs is still a challenging problem due to the unavailability of hemodynamic variables measurements, the requirement for suction avoidance, and the varying nature of the heart in different scenarios. In this study, a supervised adaptive fuzzy control strategy with pulsatility ratio modulation has been implemented to not only provide enough perfusion for the circulatory system but also to prevent suction phenomena. It has been shown that the proposed adaptive fuzzy control scheme presents a faster response compared to the previously developed fuzzy controller for LVAD and at the same time shows a faster response to push the system out of the suction area if happens. Most importantly, it has been demonstrated that the supervised adaptive fuzzy controller could help the system to stay away from the suction bound compared to the conventional methods.

Keywords: Petri Nets for Automation Control, Modelling, Simulation and Optimization in Healthcare, Discrete Event Dynamic Automation Systems
Abstract: Patient flow is the movement of patients through a healthcare facility. Over the past decades, healthcare service researchers and providers have spent tremendous efforts in optimizing patient flow and reducing patient waiting time. Identifying bottleneck in health service is an important step to curb long queues. In this paper, we use timed Petri nets to model the patient flow in typical emergency departments (ED). Waiting areas in an ED are modeled with places in a Petri net. A simulation tool is developed that tracks the number of tokens in those places to estimate the length of queues. The average sojourn time of patients in each waiting area is investigated based on simulation results. Strategies to mitigate the bottleneck impact are discussed.

Keywords: Human Factors and Human-in-the-Loop, Human-Centered Automation, Collaborative Robots in Manufacturing
Abstract: Since the last years, collaborative robots have been experiencing a continuous dramatic diffusion. Notwithstanding, they are often used in human-robot collaboration scenarios that do not fully leverage humans and robots capabilities. This paper aims to investigate how human individuals perceive collaboration with robots, comparing it to collaboration with another human agent. To this end, we design a collaborative task in the form of a joint motor action, where the human agent shares actions in a dyadic interaction with a robot or a human confederate. Our aim is to assess quality of the task and perceived pleasantness or discomfort in the two collaborative situations. The achieved results showed that differences exist when participants collaborated with another human agent or a robot. Specifically, when working with the robot, on average the task was carried out more cautiously, and less errors were made, thus leading to the assumption that participants were aware that the robot is a non-intentional agent. They acted faster and made more errors in the human-robot condition. Moreover, they reported that collaborating with the human federate was more pleasant, although more competitive. Ultimately, the study paves the way on understanding human attitude towards the collaboration with robots and shaping human-robot collaboration around it.