Keywords: Navigation, Guidance and Control, Robot Vision, Autonomous Vehicle Systems
Abstract: We propose tightly-coupled LiDAR thermal inertial odometry for LiDAR and visual odometry degraded environments to deal with LiDAR and RGB-based visual odometry degenerate environments. Our approach exhibits high robustness and accuracy by tightly-coupled sensor fusion between LiDAR edge/planar features and thermal camera point features. Furthermore, to mitigate challenges inherent in thermal vision such as low contrast, we employ a learning- based optical flow trained on both synthetic thermal images generated from RGB images and real-world thermal images. Experimental results demonstrate that our method effectively handles not only the degradation of LiDAR and visual odometry but also challenges inherent in thermal vision.

Keywords: Sensors and Signal Processing
Abstract: Joint detection and tracking of multiple targets in infrared images is critical to video surveillance and has a variety of applications. In this paper, we present a new detection-by-detection tracking method for the online tracking of small moving targets using a labeled random finite set (RFS) framework. The first step is to perform image processing using Laplacian of Gaussian filters and density-based clustering to detect multiple targets. We then use the positions of numerous detected targets and false alarms as a set of measurements to predict valid data in RFS frames. The prediction of the next state is obtained recursively from the last state and the kinematic model. Finally, we solve the data association problem by generating an assignment matrix between the predicted states and the current detected measurements to generate the next states and track labels for multiple targets. The experimental results show that the proposed method can effectively track and recognize small infrared targets in real multi-frame infrared target tracking scenarios.

Keywords: Navigation, Guidance and Control, Artificial Intelligent Systems
Abstract: Performing manipulation tasks with a legged robot in hazardous and confined disaster environments entails significant challenges due to spatial constraints, the risk of collisions with various obstacles, and the need to maintain posture balance. In this paper, we propose a motion planner and a controller designed to navigate the robot to perform tasks by utilizing the legs as manipulators while ensuring tracking and maintaining balance. The motion planner employs a sampling-based method within the configuration space, incorporating the manifold structure. The manipulation controller tracks the generated path, while the balance controller is developed to ensure stability using a three-legged stance. Compared to previous approaches that utilize legs as manipulators, our method expands the reachability space through body path planning and accounts for potential collisions between the robot and obstacles, as well as self-collisions within the robot. Using this framework, we efficiently navigated the robot to designated task points within narrow gaps, utilizing both its body and legs without collisions, thereby demonstrating the capability to reach targets.

Keywords: Information and Networking, Robotic Applications
Abstract: This paper explores the application of Software-Defined x (SDx) principles to robotics, specifically focusing on Software-Defined Robot (SDR), to enhance flexibility, interoperability, and portability in robotic systems. SDR facilitates a significant paradigm shift from traditional, hardware-dependent configurations to more agile, software-defined setups. By abstracting hardware functions into software or separating hardware from software, SDR allows for more dynamic development, rapid updates, and adjustments of robotic applications without the constraints of physical hardware. Utilizing standardized information model of ISO 22166-201 and DIS 22166-202, this paper discusses how SDR facilitates the development, testing, and deployment of robotic functionalities such as mobility services without direct hardware interaction. The implementation of SDR is demonstrated using a Jackal robot and a TurtleBot, which are examples of mobile platforms, with further testing through digital twins in a simulated environment, showcasing the potential of SDR to transform the robotics industry by reducing hardware dependency and enhancing system adaptability.

Keywords: Sensors and Signal Processing
Abstract: The study proposes the combined step-size affine projection Champernowne adaptive filter (CSS-APCMAF), addressing the back-and-forth relationship between convergence rate and steady-state misalignment in fixed step-size adaptive filtering algorithms. Additionally, the study introduces a new approach to active noise control using the combined step-size filtered-x affine projection Champernowne adaptive filter (CSS-FxAPCMAF). The emphasis of this research lies in robustness in environments with correlated input and impulsive noise. Simulation results demonstrate that the proposed algorithm can maintain a convergence rate while exhibiting smaller steady-state misalignment compared to existing algorithms, ensuring superior performance.

Keywords: Industrial Applications of Control, Robot Mechanism and Control, Robotic Applications
Abstract: With advancements in robot manipulator technology, its applications have become widespread across various fields, especially with manipulators operating in shared spaces with humans. This study aims to develop trajectory planning for industrial robots used in flexible manufacturing environments. Given the significant size of these robots, ensuring their safe operation in human-occupied spaces is crucial. Thus, this research proposes the creation of safe and efficient nominal trajectories using optimal control methods. Motion planning for such large robots is particularly hazardous and challenging in dynamic environments. This study emphasizes safety more than other motion planning methods, focusing on the large robots commonly used in manufacturing settings. It addresses the generation of safe trajectories through the application of optimal control techniques, including trajectory scaling to slow down the robot when near obstacles and regenerating paths to avoid potential collisions.

Keywords: Industrial Applications of Control, Robotic Applications, Navigation, Guidance and Control
Abstract: In this study, we developed a simulation environment as part of the Flexible Manufacturing Project to verify the safety of command execution. The ultimate goal of this project is to generate commands that can flexibly adapt to changes in the environment or tasks using visual data from cameras. These commands are often complex and not intuitively understandable, making it difficult to predict the robot’s movements prior to execution. Therefore, visualizing and reviewing these commands before deployment is essential to ensure safe operations. To address this issue, we utilized the Gazebo simulator to create a virtual environment that accurately reflects actual working conditions. Before executing commands on the physical robot, the commands are first tested in this simulated environment using a virtual robot. Based on these observations, operators can decide whether to proceed with executing the generated control commands on the actual robot. This approach ensures both safety and reliability in robotic operations.

Keywords: Control Theory and Applications, Artificial Intelligent Systems, Robotic Applications
Abstract: This paper introduces a method for determining and verifying the gain values of Linear Quadratic Regulator(LQR) controller in the complex task of balancing an inverted pendulum using a quadrotor. In the LQR controller, user must specify the weight matrices for states and control inputs, which becomes challenging as these matrices increase in size. The method presented in this paper utilizes Bayesian optimization with Expected Improvement (EI) to address the difficulty of selecting the weight matrices in the LQR controller. Through Expected Improvement, the global optimization point can be found at a low cost and in a short time. By parameterizing the weight matrices, the optimization method was easily applied. It was confirmed that the controller gain values, composed of 9-dimensional parameters for the given task, could be determined quickly and efficiently. Consequently, the proposed method for selecting the weight matrices demonstrated a reduction in the position error of the inverted pendulum compared to using arbitrarily chosen initial weight matrices. Additionally, the simulation tests confirmed a reduction in settling time.

Keywords: Autonomous Vehicle Systems
Abstract: LiDAR-based place recognition holds significant potential for various applications in autonomous driving, particularly in mapping and localization. This study analyzes which loss functions are most suitable for enhancing the robustness and performance of LiDAR-based place recognition over the long term. Specifically, we train feature extraction neural networks using Triplet loss, commonly used in LiDAR-based place recognition, and NT-Xent loss, which has been rarely used in this context, and compare their results. Unlike traditional pairwise learning approaches, the method utilizing NT-Xent loss reveals the overall relationships between descriptors in the representation space, effectively capturing global relationships and shape similarities among descriptors. Consequently, our experiments demonstrate that training feature extraction neural networks with NT-Xent loss can improve place recognition performance compared to the widely used Triplet loss. Experimental evaluations conducted on the NCLT dataset illustrate the performance variations according to different loss functions, particularly in terms of place recognition accuracy and robustness for long-term place recognition.

Keywords: Artificial Intelligent Systems
Abstract: Monitoring the temperature distribution within the battery pack is crucial for battery pack management systems. The distribution is typically tracked using physics-based battery and thermal numerical models. However, high-resolution models capable of simulating various environments and internal dynamics are limited in their use for real-time monitoring due to their high computational and memory complexity. The physics-informed neural network (PINN) that approximates the simulation using neural networks can be utilized to tackle the limitation. The PINN can dramatically reduce memory and computational complexity compared with the numerical methods. This paper simulates a battery pack's 2D heat transfer model composed of two cylindrical constant heat sources using PINN and commercial software COMSOL. As a result, we observed that PINN can learn a convection boundary condition and heat diffusion equations without a simulation dataset. The PINN shows 11.75 a mean absolute error (MAE) of surface temperature for the entire simulation, a forward computational time improvement of approximately 82 times, and a memory size improvement of about 1 million times compared to COMSOL. This demonstrates the applicability of PINN to the problem of monitoring temperature distribution within a battery pack.

Keywords: Artificial Intelligent Systems, Sensors and Signal Processing, Industrial Applications of Control
Abstract: The black box problem of neural networks (NNs) is a significant issue in various industries. This problem affects reliability, troubleshooting in case of issues, and debugging during malfunctions. In the field of fault diagnosis (FD) using vibration data, model explainability is even more crucial because it is not as intuitive as image data. NNs for FD process vibration data similarly to networked FIR filters. Therefore, visualizing which frequency components in the input data contribute to the output inference has developed to enhance the model explainability. However, existing methods only evaluate the frequency-domain decision criteria of trained models without discussing the practical use of these insights. We approach the application of frequency-domain decision criteria by assessing the robustness of models in different expected environments, thereby pre-testing for robustness and selecting or combining more robust models. Model robustness is critical in FD because varying operating conditions alter the power spectral densities of vibration data. However, the robustness of these models is often unknown to users due to their limited explainability. To this end, we introduce an FD framework that uses a power-perturbation-based decision boundary analysis (POBA) to elucidate the decision boundaries of vibration classification models. In POBA, perturbed data are generated from training data by applying power perturbations to frequency bands centered on dominant class-discriminative frequencies. The decision boundary of a model is then evaluated and visualized using these perturbed data. Additionally, the decision boundary information can be used to define a robustness score for each class. By ensembling trained models based on their class-specific robustness scores, a more robust model can be achieved. Demonstrations with two vibration datasets validate the explainability and robustness of the proposed FD framework.

Keywords: Control Theory and Applications
Abstract: Takagi-Sugeno (T-S) fuzzy systems, characterized by a convex combination of linear subsystems weighted by nonlinear membership functions (MFs), are highly valued for modeling and analyzing nonlinearities. They are widely used because they allow nonlinear systems to be addressed through linear system theory, facilitating easier analysis and control. Consequently, expanding the stability region to reduce conservatism is considered the most challenging and crucial objective for T-S fuzzy systems. Recent research has aimed to reduce conservatism by introducing fuzzy Lyapunov functions. However, these methods still tend to be conservative due to the use of multiple linear matrix inequalities (LMIs). This study addresses stability analysis and H∞ control synthesis for T-S fuzzy systems using a quadratic fuzzy Lyapunov function, which incorporates second-degree information of MFs. The Lyapunov function and controller are developed within a membership-quadratic framework using a quadratic form of membership-dependent outfactors. A structural relaxation lemma is formulated based on zero-equality conditions, leveraging matrix quadratic properties and orthogonal complements. Stability conditions for both the analysis and H∞ synthesis are derived through LMIs. Conceptual foundations utilizing high-degree membership functions are provided to extend the structural relaxation approach. Numerical examples illustrate the effectiveness of the proposed method [1].

Keywords: Control Theory and Applications
Abstract: This note illustrates how weak synaptic coupling can achieve rapid synchronization in heterogeneous networks. A motivating example will highlight the combination of nodal excitability and synaptic coupling, the essential mathematical properties for this biophysical phenomenon. Finally, the main message of the original paper will be given.

Keywords: Control Theory and Applications
Abstract: Neural networks, inspired by the human brain's neurons, are pivotal mathematical models for handling complex systems. Their successful applications in fields like signal and image processing highlight the importance of ensuring their stability. However, inherent communication delays between neurons can lead to system instability. Consequently, research on stability analysis for neural networks with time-varying delays (NNTVDs) has garnered significant attention. The Lyapunov-Krasovskii functional (LKF) method is commonly used for stability analysis, offering flexibility. Two major approaches are pursued to obtain less conservative stability criteria: designing proper LKFs and employing integral inequalities. While existing studies have mainly focused on constant or single integral LKFs, recent advancements emphasize cross-information between state and derivative terms. Emerging trends involve integral inequalities based on free matrices for tighter estimations with increased variables. Notably, recent research explores generalized integral inequalities to incorporate higher-order multiple integral terms. However, existing approaches often overlook the information on nonlinear activation function of NNTVDs, prompting the need for novel methodologies. This paper introduces an extended generalized integral inequality based on free matrices (EGIIFM) and applies it to the stability analysis of NNTVDs. The EGIIFM estimates an upper bound for a quadratic form of a positive definite matrix with an augmented vector staked not only with the state and its derivative but also with the nonlinear activation function. By reflecting the correlated cross-information among the terms in the augmented vector as free matrices, the EGIIFM provides a tighter upper bound and encompasses various existing single integral inequalities as special cases. In addition, by establishing a new double integral LKF including the correlated cross-information, a less conservative stability criterion is obtained. Through three well-known numerical examples, the effectiveness of the EGIIFM is evaluated.

Keywords: Control Theory and Applications, Robotic Applications, Robot Mechanism and Control
Abstract: This paper deals with a robust trajectory tracking controller for articulated manipulators that are subject to model uncertainties and external disturbances. The proposed controller employs the disturbance observer based controller which can effectively estimate and compensate for the effect of model uncertainties and the disturbances. It is assumed that the model uncertainty is bounded with known bounds, and that the disturbance is composed of two parts; the one, called modeled disturbance, is a summation of sinusoids with known frequencies, and the other, called unmodeled disturbance, is unknown time-varying with known bounds. To deal with the modeled disturbance, we embed its internal model into the proposed controller so that the controller can reject this modeled disturbance without using the magnitude or phase. The unmodeled disturbance is approximately rejected by tuning the controller parameter using the bounds. The stability of the closed-loop system is rigorously analyzed and it turns out that all the signals are bounded and the tracking error can be made arbitrarily small by choosing the controller parameters appropriately.

Keywords: Navigation, Guidance and Control, Artificial Intelligent Systems, Robotic Applications
Abstract: In the rapidly advancing field of robotics and precision agriculture, the autonomous Unmanned Ground Vehicles (UGVs) robot in oil palm agriculture environment presents both challenges and opportunities. A map of the oil palm plantation area is required for deployment and autonomous navigation of robots. Creating an occupancy map for this type of area using conventional surveying methods or using range equipment with simultaneous localization and mapping (SLAM) can be time consuming due to the very large area. This paper introduces an approach for semi-automatically creating occupancy grid maps for navigation by leveraging high-resolution aerial image. A comprehensive system is developed comprising tree crown detection, resolution estimation, and map creation modules. The proposed approach can produce reliable occupancy map of two types which are crop map and fenced map, catering to different navigation needs. This approach marks a significant step forward in utilizing remote-sensing data for the application of robotics in precision agriculture.

Keywords: Navigation, Guidance and Control, Robot Vision, Autonomous Vehicle Systems
Abstract: Localization is one of the most crucial tasks for Unmanned Aerial Vehicle systems (UAVs) directly impacting overall performance, which can be achieved with various sensors and applied to numerous tasks related to search and rescue operations, object tracking, construction, etc. However, due to the negative effects of challenging environments, UAVs may lose signals for localization. In this paper, we present an effective path-planning system leveraging semantic segmentation information to navigate around texture-less and problematic areas like lakes, oceans, and high-rise buildings using a monocular camera. We introduce a real-time semantic segmentation architecture and a novel keyframe decision pipeline to optimize image inputs based on pixel distribution, reducing processing time. A hierarchical planner based on the Dynamic Window Approach (DWA) algorithm, integrated with a cost map, is designed to facilitate efficient path planning. The system is implemented in a photo-realistic simulation environment using Unity, aligning with segmentation model parameters. Comprehensive qualitative and quantitative evaluations validate the effectiveness of our approach, showing significant improvements in the reliability and efficiency of UAV localization in challenging environments.

Keywords: Navigation, Guidance and Control, Robotic Applications
Abstract: In this paper, we propose a potential functions-based RH-RRT* (PRH-RRT*) path planning algorithm to increase the safety and optimality of mobile robot navigation. The RH-RRT* is an extension of the RRT* algorithm that improves search efficiency and reduces computational load by optimizing the local path within a receding horizon. However, the variance during random sampling is challenging as it must balance exploration and path optimality. Additionally, the obstacle collision check method that only considers the robot's center point, compromises path safety. To address these limits, we propose the PRH-RRT*, which integrates Artificial Potential Fields (APF) with RH-RRT*. The proposed algorithm leverages attractive and repulsive potentials to explore the state space enough, enhancing path safety and optimality by avoiding obstacles and guiding the robot toward the goal. To verify the performance of the proposed method, numerical simulation is conducted and results are analyzed.

Keywords: Navigation, Guidance and Control, Autonomous Vehicle Systems, Artificial Intelligent Systems
Abstract: Navigating safely in dynamic human environments is crucial for mobile service robots, and social navigation is a key aspect of this process. In this paper, we proposed an integrative approach that combines motion prediction and trajectory planning to enable safe and socially-aware robot navigation. The main idea of the proposed method is to leverage the advantages of Socially Acceptable trajectory prediction and Timed Elastic Band (TEB) by incorporating human interactive information including position, orientation, and motion into the objective function of the TEB algorithms. In addition, we designed social constraints to ensure the safety of robot navigation. The proposed system is evaluated through physical simulation using both quantitative and qualitative metrics, demonstrating its superior performance in avoiding human and dynamic obstacles, thereby ensuring safe navigation. The implementations are open source at: https://github.com/thanhnguyencanh/SGan-TEB.git

Keywords: Navigation, Guidance and Control, Robotic Applications, Robot Vision
Abstract: In this paper, we present the development of a visual navigation and mobile robot platform designed for autonomous driving in outdoor unstructured environments. To address the challenges posed by such environments, where inter-object features are ambiguous and conditions are highly irregular, we incorporate a semantic segmentation technique into the mobile robot's local navigation system. This enables the robot to identify and move within navigable regions. A simplified segmentation structure is used to efficiently process large volumes of visual data, which is integrated with vision-based control strategies to facilitate effective navigation planning. Additionally, we introduce a supplementary method for road segmentation, leveraging depth information to ensure stable and robust driving. Our research also includes the design of a wheeled mobile robot capable of operating in various environments, emphasizing its practical applicability across diverse fields. The potential of this platform is validated through empirical evaluations with real robots in various driving scenarios, demonstrating over 80% accuracy in navigable region classification and over 90% accuracy in road segmentation performance under dynamic conditions.

Keywords: Autonomous Vehicle Systems, Robotic Applications
Abstract: This paper proposes a task reallocation method for multi Unmanned Grounded Vehicle (UGV) systems in sensor and actuator fault situations. The proposed method formulates the task reallocation problem as a Mixed-Integer Linear Programming (MILP) using the distance between UGVs and task priority. When a UGV sensor or actuator fault occurs, the proposed method solves the MILP and finds the optimal task reallocation solution. The solution ensures continuing high-priority tasks for the multi-UGV systems. The proposed method is validated in simulation and experiment.

Keywords: Control Theory and Applications, Industrial Applications of Control
Abstract: In this paper, fault-tolerant control is designed for yaw control of a distributed electric propulsion aircraft such as NASA X-57. To minimize the effect on the aircraft's response to faults in several electric propulsors during the yaw control, an RS-LQR control strategy that equalizes the yaw moments produced by each propulsor is proposed. Firstly, real-time fault detection of the electric propulsors is performed, and upon detecting a fault, the RS-LQR controller is reconfigured to stabilize the aircraft. Computer simulations are conducted to compare the proposed control strategy with a conventional control strategy where the yawing moments of electric propulsors are not necessarily equal to each other. The results demonstrate that using the proposed control strategy significantly minimizes the effect on the aircraft's response to propulsor faults compared to the conventional strategy.

Keywords: Navigation, Guidance and Control, Control Theory and Applications, Industrial Applications of Control
Abstract: 本文的目的是提出一种故障诊断方法 基于改进未知输入观测器（UIO）的方法 用于电液执行器 （EHA） 系统。方法是 旨在提高故障的准确性和可靠性 EHA 系统中的诊断，已知是复杂的和 非线性。改进后的UIO能够有效地 通过优化 观察者增益和状态估计算法。本文 通过以下方式对所提方法进行验证： 使用 MATLAB/Simulink 仿真平台，该 允许在以下情况下测试系统性能 正常工作条件正常和故障。结果 证明增强的 UIO 优于 在估计方面的传统自适应观察器 精度、故障检测率、隔离率、检测 时间和误报率。这在以下方面尤为明显 传感器故障、执行器故障、液压故障的情况 泄漏故障，其中检出率达到95%、90%、 和88%，隔离率达到90%， 分别为 88% 和 85%。这些增强功能 显著提升了故障#

Keywords: Sensors and Signal Processing
Abstract: The research on bearing degradation plays an important role in the health management and maintenance of the system.In response to the problems of environmental noise interference and low accuracy of degradation state evaluation during the service process of gearbox bearings, this paper proposes a degradation state evaluation method based on the VMD-CNN-BiLSTM model. Based on the fault identification of the entire life degradation process of bearings, the performance change law of the degradation process is analyzed, and the classification of degradation stages is achieved by combining fuzzy evaluation method. By optimizing the parameters in Variational Mode Decomposition (VMD), the reconstruction of the entire lifecycle vibration signal has been achieved, which helps to extract subsequent degradation features. At the same time, by constructing a trend degradation index for rolling bearings and analyzing the changes in the health status of bearings using membership functions, the classification of degradation states is achieved, providing a basis for scientific decision-making in diagnosis. Finally, the degradation process evaluation of rolling bearings is achieved, and applied to the gear bearings of port shore container cranes, providing a theoretical and methodological basis for preventive maintenance.

Keywords: Sensors and Signal Processing, Artificial Intelligent Systems, Industrial Applications of Control
Abstract: As a kind of unmanned aerial vehicle (UAV) with new layout, the flying-wing UAVs have received increasing attention with unique advantages. Precise fault diagnosis for its critical sensor system can effectively enhance the safety of flight missions. Without the requirement of precise mechanism models, deep learning-based approaches can automatically excavate valuable information and identify the sensor faults intelligently. However, single deep learning model has deficiency in diagnostic precision and stability. In this study, one enhanced ensemble deep auto-encoder (EEDAE) model is proposed to simultaneously combine the advantages of ensemble strategy and deep learning models. First, different structure parameters are generated and diverse training subsets are randomly bootstrapped to increase the diversity of base models and extract critical features from raw data automatically. Meanwhile, to further enhance the effect of model integration, one enhanced weighted voting (EWV) strategy with threshold is designed to realize selective model ensemble through removing the models with poor performance and assigning the voting weights to the remaining models based on their diagnostic accuracy. Finally, the experimental results indicate that the designed EEDAE can realize prominent performance on sensor fault diagnosis.

Keywords: Sensors and Signal Processing, Artificial Intelligent Systems, Control Theory and Applications
Abstract: To address the challenge of detecting damage in wind turbine blades, this paper proposes a fault diagnosis method combining the unsupervised Local Outlier Factor (LOF) algorithm and the Back Propagation (BP) neural network. The method involves collecting strain signals from blades during fatigue testing and using Fourier transform and wavelet packet analysis to extract time-frequency domain features for model training. To solve the issue of imbalanced training data, a two-stage recognition method is employed. First, the LOF algorithm is used for unsupervised learning to detect outliers and classify samples. These classifications are then compared with the actual sample labels. Misclassified samples are reconstructed into a new training set and fed into the neural network for training, which helps mitigate data imbalance. Experiments demonstrate that the LOF-BP neural network fusion algorithm performs better than single methods, achieving superior fault diagnosis and recognition for wind turbine blades. This approach shows significant potential for application in the wind turbine operation and maintenance industry.

Keywords: Robot Vision, Autonomous Vehicle Systems, Navigation, Guidance and Control
Abstract: Spherical images have a wide field of view and are effective for pose estimation, but they have the problem of inherent distortion.Existing methods to reduce the effects of distortion significantly increase computation time and cannot be used for real-time applications.We propose a method that enables accurate and fast feature point-based two-view pose estimation using spherical images by reducing the effect of distortion in equirectangular images.In our approach, one image is generated by rotating an equirectangular image, and feature point detection and descriptor extraction are performed from the two images: the original image and the generated image.The information is then integrated by adopting the least distorted regions of the two images.Our approach works faster than existing distortion reduction methods because of the small number of projection planes.In experimental evaluation, it was shown that our proposed method is faster and equally accurate compared to state-of-the-art methods in pose estimation.

Keywords: Robot Vision, Autonomous Vehicle Systems, Sensors and Signal Processing
Abstract: The identification of granular materials is a crucial ability for automating construction machinery. Current point clouds segmentation methods struggle to segment shapeless objects, such as sand piles, due to sparse features. A real-time semantic sparse point cloud segmentation framework based on multi-view method is developed using a multi-sensor fusion SLAM algorithm with visual odometry. This approach has inherent semantic mapping errors caused by external calibration, dynamic pose estimation, and image segmentation. To enhance segmentation accuracy, a grid-space optimization algorithm has been proposed. The first step involves finding the incorrect segmented points by checking the pixel-depth gradient in grid space. Secondly, depth density clustering are applied to these points for re-segmentation. Our algorithm was tested on gravel and sand piles in construction scenarios. The experimental results demonstrated that our segmentation strategy can effectively segment granular objects. Furthermore, our two-step scanning and re-segmentation methods can significantly improve the performance of point cloud semantic segmentation.

Keywords: Sensors and Signal Processing
Abstract: Digital Holographic Microscopy (DHM) can obtain a 3D profile of a micro-object. However, the obtained 3D profile includes the noise. One reason for this is the problem of the extraction method of phase information in DHM. Conventional methods cannot extract only the phase information from the Fourier domain, and noise components are included. To solve this problem, we propose a use of the Kalman filter in DHM systems. In this method, multiple frequency components are obtained from the Fourier domain as time-series data for Kalman filtering. 3D profiles are obtained from each frequency component and stored in an array as time-series data. This is used to remove randomly generated noise. In this process, it remains unclear how many 3D profiles obtained from frequency components are needed as time-series data. Therefore, the optimization of the number of windowed sidebands in the Fourier domain is the objective in this paper. The objective is to use the Kalman filter efficiently by performing quantitative evaluation using image quality metrics and optimization based on processing time.

Keywords: Sensors and Signal Processing
Abstract: Digital holographic microscopy (DHM) is used in the fields of microbiological research and disease diagnosis. Especially in the medical field, it is required to obtain the precise shape information of red blood cells and living cells for disease diagnosis. To obtain the highly accurate three-dimensional profile of an object, it is necessary to set parameters accurately on the system and perform the analysis. In DHM, it is especially important to set the refractive index of the medium surrounding the object accurately. However, the　refractive index varies depending on the temperature and the humidity of the experimental environment, as well as on the sample preparation method, and it is difficult to derive the appropriate refractive index. This is a problem in the medical field, which requires the accurate three-dimensional in-formation. Therefore, in this paper, we propose a method to derive the optimum refractive index and to obtain the ac-curate height information of red blood cells by conducting experiments using two types of lasers with different wave-lengths and comparing the height information of red blood cells obtained with each laser. By using the proposed method, the optimum refractive index of the surrounding medium and the accurate height information of the red blood cells are obtained.

Keywords: Sensors and Signal Processing
Abstract: In recent years, imaging techniques are essential in our lives, such as security cameras and drive recorders. However, the visualization quality of these techniques sometimes is affected by scattering media, such as smoke at a fire or fog during driving. To solve this problem, Peplography has been proposed. Peplography is an optical algorithm that emphasizes the object information presented in the scattering media, even when the density of the scattering media varies within the image. On the other hand, Peplography estimates the scattering media in each local region, and when the region is not defined appropriately, the estimation of the scattering media will not be visualized. Therefore, the purpose of this research is to find the optimal processing for each environment by experiment with various situations, such as when the size of the object in the image is small, when the density of the scattering media is heavy, and when the target object is various.

Keywords: Autonomous Vehicle Systems, Robotic Applications
Abstract: Unmanned Aerial Vehicles (UAVs) are increasingly used in fields such as surveillance, reconnaissance, and communication relay due to their flexibility and efficiency. A key challenge in UAV swarm operations is maintaining stable communication between mission UAVs (MUs) and the Ground Control Station (GCS), especially in obstructed environments. Both the GCS and MUs have limited communication ranges, necessitating relay UAVs (RUs) to ensure connectivity for extensive missions. Therefore, relay UAVs task planning involves the appropriate deployment and operation of the minimum number of RUs required to ensure that all MUs can communicate with the GCS. In this research, the relay UAVs task planning, originally a Mixed-Integer Nonlinear Programming (MINLP) problem, has been reformulated as a Mixed-Integer Linear Programming (MILP) problem. Additionally, the task planning for relay UAVs has been segmented into two optimization problems: initial deployment optimization and real-time repositioning optimization. First, the initial deployment optimization aims to minimize the number of active RUs and determine their optimal positions. Second, the real-time repositioning optimization focuses on minimizing the total travel distance of RUs as MUs move. Finally, simulation results demonstrate the effectiveness of these methods in maintaining robust communication networks in dynamic environments.

Keywords: Autonomous Vehicle Systems, Information and Networking, Navigation, Guidance and Control
Abstract: This paper addresses the problem of maximizing coverage in a mission space with both stationary and mobile agents such that effective coverage constraints are satisfied, i.e., each point of the mission space must be covered to a predefined level at least once over a given period. The deployment of the stationary agents may be given in advance or obtained by a classical coverage control algorithm. The motion planning of the mobile agents is designed under maximal speed and acceleration constraints. When there is only one mobile agent, it is shown that its path planning and velocity planning can be designed separately. We first obtain an optimal velocity policy and the corresponding optimal coverage performance for a given path, which provides a criterion to prescribe a good path. Then, path planning is generated to meet the effective coverage constraint by connecting a set of “inspection points”. Inspired by the optimal velocity policy, we propose three methods to generate the inspection points and obtain the optimal order of connecting the inspection points by a Traveling Salesman Problem (TSP)-based method. Finally, we extend the one-mobile-agent motion planning scheme to multiple mobile agents by proposing two methods. Simulation examples are included to compare the performance of the three methods for generating inspection points and compare the performance of the proposed methods for multiple mobile agents.

Keywords: Control Theory and Applications
Abstract: This extended abstract introduces a behavioral conditional generative model for stochastic Linear Time-Invariant (LTI) systems. The model leverages offline trajectories, historical input-output data, and future input to generate future output samples for an online trajectory of an LTI system. We demonstrate that the statistical properties of these generated samples asymptotically align with the conditional distribution of future outputs, given past observations and future inputs, as the number of offline trajectories increases to infinity. Furthermore, we incorporate the proposed model into a Model Predictive Control (MPC) framework, resulting in a robust data-driven predictive controller that can be implemented end-to-end. Simulations on a three-pulley system trajectory tracking problem validate the model's effectiveness in managing system uncertainties and ensuring robust control performance.

Keywords: Control Theory and Applications
Abstract: This paper investigates the stability analysis of delayed neural network (DNN) systems. A novel augmented integral inequality is proposed to derive hierarchical estimation conditions for augmented integral terms. Furthermore, a new quadratic delay-product term and an augmented double integral term are integrated into the Lyapunov-Krasovskii functional (LKF) to incorporate additional delay-dependent information. Finally, stability criteria with less conservatism are established based on the proposed inequality and LKF. The effectiveness and superiority of the proposed stability conditions are demonstrated through comparisons with existing results.

Keywords: Sensors and Signal Processing, Process Control Systems
Abstract: Optimal sensor placement design (SPD) for distributed parameter systems (DPS) is essential for robust control, fault detection, and real-time optimization. The challenge of SPD is intensified by the infinite potential positions and associated cost of the sensors. Most existing methods rely on robust dynamic models of the system, which often do not exist. This study introduces a novel diffusion-attention based algorithm for SPD, which relies solely on data, bypassing the need for a system model. The algorithm was applied to one-dimensional and two-dimensional benchmark problems to evaluate its performance. The algorithm achieved optimal spatio-temporal SPD, ensuring maximum accuracy with a limited number of sensors.

Keywords: Process Control Systems, Industrial Applications of Control
Abstract: Replacing petroleum with bio-gas and bio-oil produced from biomass pyrolysis is getting attention due to its carbon-neutral characteristics. To enhance the economic efficiency of biomass pyrolysis, appropriate modeling of the pyrolysis process and uniformity in the composition of the products are important. This paper presents a study on the kinetic modeling and control of biomass pyrolysis in a conical spouted bed reactor (CSBR). The research utilizes Model Predictive Control (MPC) to regulate the composition of pyrolysis products, optimizing the process for industrial applications. The CSBR enhances particle collisions and heat transfer, improving pyrolysis efficiency. A kinetic model based on lumped products (gas, bio-oil, char) was developed and validated against experimental data. The application of MPC effectively adjusted product composition, demonstrating the potential for improved economic viability and process control in biomass pyrolysis.

Keywords: Process Control Systems
Abstract: Fossil fuels are carbon sources that account for a significant 80% of energy supply. Electricity-based fuel (E-fuel) and biofuel are attracting attention as technologies that produce alternative fuels to these fossil fuels. Among low-carbon eco-friendly materials, dimethyl ether (DME) is of high interest as it can be used as both E-fuel and biofuel. In this study, three feeds, i.e. syngas, biogas, mix of captured CO2 and green H2, are used to compare fossil fuel with biofuel and E-fuel. Furthermore, by adding a comparison of the two synthesis routes, Techno-economic analysis and environmental impact evaluation are conducted for these six processes. As a result, the most eco-friendly process was the direct CO2 and H2-based process, and the most cost-competitive process was the direct syngas-based process. If the production cost of biogas and H2 is lowered than expected, it will be an economical and eco-friendly process. In addition, if the carbon tax credit is applied, production costs will be significantly reduced, so the realization of E-fuel and biofuel production is expected.

Keywords: Process Control Systems, Artificial Intelligent Systems, Industrial Applications of Control
Abstract: In this study, we address the optimization of thermal energy storage (TES) operations in Concentrating Solar Power (CSP) plants through the integration of reinforcement learning (RL) and model predictive control (MPC). CSP plants, known for their ability to stabilize power grids and maximize economic benefits by shifting production times, require effective scheduling to fully leverage their dispatch capabilities. The study utilizes the Double Deep Q-Network (DDQN) algorithm to determine the economic value of energy stored in TES and integrates this value function into a low-level MPC framework. The proposed approach is validated through simulations, which demonstrate superior economic benefits and operational stability compared to traditional MPC methods alone. This research underscores the potential of RL-MPC integration in enhancing the efficiency and profitability of CSP plant operations under dynamic market conditions.

Keywords: Process Control Systems
Abstract: Rapid and accurate detection of anomalies in chemical processes plays a crucial role in ensuring safe operation and maintaining quality. While existing monitoring studies have primarily focused on single operating modes, real-world processes often involve multiple modes due to seasonal changes and fluctuating demand. Moreover, process data frequently lacks labels for each operating mode, making unsupervised learning methods a suitable approach. This study addresses these challenges by utilizing a Deep Autoencoding Gaussian Mixture Model (DAGMM), which effectively handles complex relationships between variables and classifies various operating modes. To overcome the limitations of deep learning models, which can be difficult to interpret, we introduced the Shapley Additive Explanations (SHAP) as part of an explainable artificial intelligence (XAI) technique to analyze the causes of process anomalies. Our methodology was validated using data from six operating modes of the Tennessee Eastman process. The results demonstrate that DAGMM achieves high detection accuracy for multiple anomaly scenarios, and the incorporation of SHAP aids in identifying the causes of process anomalies.

Keywords: Control Theory and Applications
Abstract: In this paper, the exponential stability problem of switched positive impulsive system (SPIS) is considered. With the aid of a discretized switched maximum Lyapunov function approach and the mode-dependent range dwell time (MDRDT) switching signal, a sufficient criterion is derived to ensure the exponential stability of SPIS, even when all subsystems exhibit instability. The work offers two advantages compared to prior research. Firstly, an algorithm is introduced for computing the maximum and minimum dwell times. Secondly, our result is less conservative than that reported in previous literature. A numerical example is presented to demonstrate the effectiveness of the theoretical result.

Keywords: Control Theory and Applications, Industrial Applications of Control, Process Control Systems
Abstract: In increasingly complex systems, it is becoming increasingly difficult to accurately identify the location of failures and the extent of their effects. Chaos engineering is a method to improve the reliability of the system by causing failures to reveal the problems. In this study, we propose a method of reset control that suppresses the failure of a pneumatic control valve by simulating the failure of the valve, incorporating the concept of chaos engineering. In this study, in order to study a new control method for the control parts of the positioner, simulations are shown in which a chaotic signal is superimposed as a simulated disturbance on the input signal of a mathematical model of a pneumatic control valve with an electronic positioner. As an appropriate control method for control parts of the positioner, we apply the reset control, which autonomously generates pulses from the control deviation as a feedback control. The parameters of the reset control are adjusted for this model, and the simulation results show this reset control with these parameters can suppress failures caused by the superimposition of simulated chaotic signals.　

Keywords: Control Theory and Applications
Abstract: This paper investigates the stability of positive switched homogeneous systems (PSHSs) with exogenous inputs. The purpose of the article is to propose a sufficient condition for global exponential stability (GES) of PSHSs by using the novel inequality and comparison principle. Compared with the traditional Lyapunov function method, our approach is more concise, intuitive, and widely applicable. At first, we provide a new inequality that will be used for stability analysis of the system. Then, combining the new inequality and a rigorous proof by contradiction, we establish the stability criterion based on the slow switching technique, and the obtained result is extended to a general degree of alpha in the range of 0 to 1, deepening and developing some previous research findings. Finally, the article further demonstrates the effectiveness of the theoretical discoveries through an important example.

Keywords: Control Theory and Applications
Abstract: This paper investigates adaptive event-triggered output feedback global stabilization for a class of nonlinear systems with parametric uncertainties and unknown non-polynomial growth rate of output. Different from existing research, the system nonlinearities possess unmeasurable states dependent growth, while the growth rate is unknown non-polynomial function of the output. To address this issue, we propose an adaptive event-triggered output feedback control scheme. Specifically, we construct a dynamic gain observer to rebuilt the unmeasurable states, whose dynamic gains are introduced to compensate for uncertainties and counteract the unknown non-polynomial growth rate. Subsequently, an adaptive event-triggered output feedback controller based on the gain observer is constructed to achieve global stabilization of the closed-loop system while improving the utilization of resources, in which the time-varying threshold is key to ensure the validity of the event-triggered controller. A simulation example is given to demonstrate the effectiveness of the proposed control strategy.

Keywords: Control Theory and Applications
Abstract: This paper presents a high-gain observer-based high-order internal model disturbance observer considering a single-frequency sinusoidal disturbance and mth-order polynomial-in-time disturbances. Two Q-filters in the disturbance observer structure serve different roles: one has disturbance compensation capability, and the other performs the implementation of the inverse dynamics. In specific, since the block of the Q-filter with the inverse dynamics has a high-gain observer structure, it is designed as the high-gain observer to facilitate state feedback control design. On the other hand, the disturbance compensation Q-filter incorporates the modeled part of the disturbance to eliminate it completely. Using the singular perturbation theory, it is demonstrated that the overall closed-loop system with the proposed disturbance observer completely compensates for the effect of modeled disturbances and behaves like the nominal closed-loop system. Furthermore, robust stability conditions under model uncertainties are provided. The effectiveness of the proposed disturbance observer is validated through simulation.

Keywords: Control Theory and Applications
Abstract: This paper suggests a new necessary and sufficient condition of bounded real lemma (BRL) for discrete-time descriptor systems (DTDSs). First, a Lyapunov function for DTDSs is designed by using a positive definite matrix whose the dimension equals to the rank of a singular matrix in DTDSs, whereas the existing work in the literature has considered the dimension which equals to the length of the state. By considering the zero constraint which comes from the singular matrix in DTDSs, two slack variables are introduced into the difference of Lyapunov function. Then, a set of linear matrix inequalities (LMIs) is provided to ensure the necessary and sufficient condition of the BRL for DTDSs. Next, an inversion formula is given to apply the proposed BRL to a stabilization problem of DTDSs. Based on the inversion formula, the BRL for the closed-loop system with state-feedback H∞ control is obtained in terms of non- convex conditions. Therefore, a sufficient condition of the non-convex conditions is provided in terms of LMIs. A numerical example shows the effectiveness of the proposed approaches.

Keywords: Control Theory and Applications
Abstract: This paper introduces the L1 controller synthesis problem for piecewise continuous nonlinear systems and its solutions obtained by set invariance principles. The solution of a piecewise continuous nonlinear system is defined in a Filippov’s sense and the L1 performance is defined in terms of the Filippov’s solutions. A generalized version of controlled invariance domain is established based on the external contingent cone. Based on this, a sufficient condition for the existence of the L1 state-feedback controller has been obtained in this paper. It is discussed on the developed results in a comparison fashion with the existing results on nonlinear L1 control.

Keywords: Control Theory and Applications
Abstract: This paper presents an impact angle guidance law of unmanned aerial vehicles(UAVs) with time-varying thrust in a boosting phase. Most current research on impact angle guidance law assumes that UAV speed is constant in terms of controlled thrust. However, UAV speed and acceleration in a boosting phase keep changing because of time-varying thrust. Environmental factors and manufacturing process error may prohibit accurately predicting vehicle thrust profiles. We propose a nonlinear impact angle guidance law by analytically solving a second-order error dynamics with nonlinear time-varying coefficients. The proposed analytic solution enables updating guidance gains according to initial and current states so that desired impact angle is met while miss distance error is reducing. We prove finite-time error convergence of the proposed guidance law with the Lyapunov stability theory. Various simulation study are performed to verify the proposed guidance law.

Keywords: Control Theory and Applications
Abstract: In this paper, we propose a data-driven disturbance observer-based controller for sampled-data system. We first express the inverse dynamics of the nominal system, which is one of the components of the disturbance observer, using the input and output data of the sampled-data system. Based on a model-based disturbance observation-based controller, we propose an overall structure and element-specific design method for a data-driven disturbance observer-based controller. Three Stability conditions are presented through the characteristic equation of the whole closed-loop system based on the transfer function, and simulation is performed to prove the theory.

Keywords: Robot Mechanism and Control, Industrial Applications of Control, Control Theory and Applications
Abstract: This paper introduces a novel stability criterion for biped walking systems using the linear inverted pendulum model in which the dynamics between the center of mass (CoM) and the zero moment point (ZMP) is dealt with. The criterion is based on ensuring that the ZMP error, the difference between the reference and real ZMP, remains within a specified area to maintain walking stability. Considering this criterion, we examine the initial CoM conditions necessary for walking balance and define the stability regions based on these conditions. Toward a more practical significance, we address the impact of unknown disturbances on walking balance of biped walking systems. More importantly, the stability region is shown to be computable through a finite number of calculations, ensuring stability over an indefinite period. Finally, simulation results are provided to validate the proposed method.

Keywords: Control Theory and Applications
Abstract: This paper provides a new method for computing the l_{infty}-induced norm of a multivariable discrete-time linear system. To this end, we reinterpret the l_{infty}-induced norm computation problem as finding the {infty}-norm of an explicit infinite dimensional matrix. To make such a computation feasible, we deal with the infinite-dimensional matrix in a truncated fashion and derive an upper bound and a lower bound on the l_{infty}-induced norm. More precisely, the matrix {infty}-norm of the (infinite-dimensional) tail part is approximated using these bounds, while the (finite-dimensional) head part is calculated exactly. Combining these results allows us to compute the upper and lower bounds of the original l_{infty}-induced norm. Finally, some numerical examples are provided to demonstrate the theoretical validity of our approach.

Keywords: Autonomous Vehicle Systems, Robotic Applications, Navigation, Guidance and Control
Abstract: This paper investigates the efficiency of autonomous indoor exploration utilizing simulation testing environments in Gazebo. Two exploration methods, Floodfill algorithm and Frontier-based algorithm, using the 2D LiDAR sensor are compared. The Floodfill algorithm employs a systematic traversal approach, while the Frontier-based method dynamically detects and navigates towards frontiers. Results indicate that the Frontier-based approach outperforms Floodfill Algorithm in terms of efficiency and map completeness, particularly in complex environments. The study underscores the importance of the Frontier-based strategy for autonomous indoor exploration and paves the way for enhanced robotic applications in diverse domains.

Keywords: Industrial Applications of Control, Navigation, Guidance and Control, Control Theory and Applications
Abstract: This study proposes a novel integrated strategy aimed at enhancing both energy efficiency and safety levels in maritime operations by leveraging advanced ship trajectory optimization and motion control technologies. The paper quantitatively analyzes CO{_2} emissions and fuel consumption using the International Maritime Organization's Energy Efficiency Operational Indicator (EEOI). Additionally, it employs Particle Swarm Optimization (PSO) algorithms to adjust and optimize ship routes, thereby significantly reducing energy consumption. Moreover, the research delves into precise ship motion control under constraints and uncertainties within a Multiple-Input–Multiple-Output (MIMO) nonlinear system environment. It achieves this by utilizing asymmetric barrier Lyapunov functions (ABLF) and adaptive neural networks (NN), which together ensure robust and reliable control performance. By integrating these advanced methodologies, the study provides comprehensive solutions that are applicable for sustainable and safe maritime operations. Simulation results demonstrate the effectiveness of using PSO to design vessel trajectory to reduce vessel fuel consumption and the effectiveness of ABLF and adaptive NN for ship safe motion control.

Keywords: Navigation, Guidance and Control, Sensors and Signal Processing
Abstract: A technique for robust data fusion in autonomous spacecraft navigation is presented. Observations of X-ray pulsar signals and measurements from an optical camera are considered for integration. Data from sensors usually contain outliers (points that lie several standard deviations away from typical observations). This can be due to rare external phenomena or faults within the sensor itself. Unless the effect of these outliers is mitigated, orbit determination solutions will be severely degraded. A nonlinear optimization-based approach that incorporates ideas from robust statistics is presented to address this problem. What results is a filtering algorithm which we name Levenberg-Marquardt Robust Iterated Extended Kalman Filter (LM-RIEKF). We demonstrate the effectiveness of the proposed filter in a case study involving an autonomous approach trajectory toward a planet. Our results show that LM-RIEKF provides resilience even under severe sensor data contamination levels reaching 40%, while the results from the standard approach show significantly reduced performance in such settings.

Keywords: Robot Mechanism and Control, Robotic Applications, Sensors and Signal Processing
Abstract: Target tracking mobile robots still face challenges in terms of localization accuracy and tracking performance. This study presents a novel target tracking strategy for an omnidirectional mobile robot by using dual ultra-wideband (UWB) anchors. The proposed target tracking strategy integrates these sensors with a Mecanum wheeled mobile platform, optimizing motion control and target tracking. Based on the developed dual UWB anchors' geometric localization model, the horizontal relative distance and orientation between the target and the robot were estimated. In tracking scenarios, the robot demonstrates advanced adaptability: While the target is at close-range, the robot rotates in place for orientation tracking to directly face the target, and the designed tracking algorithm allows the robot to dynamically adjust its rotation speed, resulting in smooth tracking movements. While the target is at longer-range, the robot dynamically modifies its tracking speed and orientation according to relative positional data, to ensure effective and continuous target following. Through experimental validation, the lateral error during target tracking is reduced, the tracking accuracy is improved by 50% compared to existing related studies, and the tracking performance is smoother.

Keywords: Robot Vision, Autonomous Vehicle Systems, Sensors and Signal Processing
Abstract: Autonomous robot navigation in social environments, such as crowded hallways, is a challenging task. Robots need to monitor the movement directions of surrounding objects, search for collision-free routes, and update their paths in real-time to navigate safely. To address this challenge, we present dynamic multi-object analysis framework for robust and efficient social navigation. To ensure consistent tracking performance, we adopt an optimized feature-matching algorithm combined with a particle filter to classify static and dynamic points effectively. We evaluate our proposed approach through simulation testing. Additionally, we have released the project’s source code and supplementary materials, including a video demonstrating experimental results on GitHub (https://github.com/iminolee/SCAN).

Keywords: Robotic Applications, Sensors and Signal Processing, Robot Vision
Abstract: In this paper, we introduce the semantic navigation framework for Multi-Robot Systems (MRS). In order for a robot to understand a complex environment, it is necessary for it to comprehend the surrounding environment like a human. Therefore, We have adapted the single robot framework from previous research to be suitable for MRS . The proposed framework consists of Semantic Modeling Framework (SMF), Semantic Autonomous Navigation (SAN), Semantic Information Processing (SIP), and Multi-Robot Task Planner. SMF represents the surrounding environment as a topological graph using Triplet Ontology Semantic Model (TOSM), providing essential information for autonomous navigation planning and processing. SAN is an autonomous navigation module that executes actions based on sequences generated by the Multi-Robot Task Planner. SIP processes information to determine the current state using SMF knowledge data and sensor input. The Multi-Robot Task Planner generates behavior sequences to ensure that multiple robots can perform tasks without colliding. We validated the framework through experiments conducted in both virtual and real-world environments, achieving successful mission completion with an average error of approximately 0.117m.

Keywords: Autonomous Vehicle Systems, Control Theory and Applications, Artificial Intelligent Systems
Abstract: In this work, we demonstrate the effectiveness of nonlinear model predictive control (NMPC) approximation based on deep neural network (DNN). MPC has been widely adopted in autonomous driving control problems to handle multiple objectives and constraints. We first design the implicit NMPC for the forward and backward motions of a truck-trailer (TT) system, which follows the reference path while maintaining safety between the head truck (HT) and the trailer (TR). However, the computational load in implicit MPC makes it a challenge for real-time implementations. To alleviate the computational burden in implicit NMPC online, an NMPC approximation approach based on DNN is adopted in this study to achieve a parametric function approximation. We conduct a comparative study on the proposed approach and a baseline controller for control performance analysis, and the computational load is evaluated on a hardware-in-the-loop (HILs) experimental system.

Keywords: Control Theory and Applications, Process Control Systems, Control Devices and Instruments
Abstract: Robust Model Predictive Control (RMPC) is a powerful control strategy used in various engineering applications to handle system uncertainties and maintain desired performance. Traditional RMPC methods often require substantial computational resources, particularly when solving optimization problems online. This paper introduces a novel offline RMPC approach that leverages linear matrix inequality (LMI) optimization to enhance computational efficiency and robustness. Firstly, the system parameters are considered uncertain and are modeled within a polytope. This representation is common in robust control design, allowing for a comprehensive characterization of uncertainties. Secondly, a set of linear matrix inequalities (LMIs) is formulated to solve for the optimal controller gain. LMIs are a powerful tool in control theory due to their convex nature, which ensures that global optima can be found efficiently. The optimization process involves determining a controller gain that minimizes a quadratic cost function while ensuring robust stability and performance across all possible parameter variations within the polytope. Consequently, by establishing the upper bound of the cost function as a quadratic function of the state variable, the optimal controller gain can be computed offline. This approach significantly reduces the computational burden during real-time operation, making it feasible for systems with stringent performance requirements. Additionally, the proposed RMPC approach enforces constraints on both inputs and outputs to ensure that the system operates within a safe and specified range. This feature is crucial for practical applications where violating these constraints can lead to system failure or unsafe operation. Finally, the proposed approach is validated through extensive simulations. The results demonstrate the effectiveness of the LMI-based offline RMPC method in achieving robust stability and performance under a wide range of uncertainties. The performance of the proposed method is compared with existing RMPC techniques. The comparison highlights the advantages of the new approach, particularly in terms of reduced computational requirements and improved handling of uncertainties.

Keywords: Control Theory and Applications, Artificial Intelligent Systems, Industrial Applications of Control
Abstract: Ensuring accurate and efficient operation of automated surface vessels under uncertainties and environmental disturbances remains technically complex and practically challenging for the safety and reliability of intelligent maritime systems. For such safety-critical operations, this paper proposes the physical-informed neural network (PINN)-based method for automated surface vessel control that integrates deep learning neural networks into the nonlinear model predictive control (NMPC) framework. Compared with the pure neural networks method, PINN integrates the physics information into the loss function and adds control actions and initial conditions to the input of the prediction model. By adopting the PINN as the prediction model of NMPC during the optimization process, better control accuracy has been achieved, and the forward prediction time and computational complexity have been reduced. In conclusion, the proposed method is proven to be efficient through comparison studies with pure deep learning networks and experiments with higher accuracy of prediction and control.

Keywords: Control Theory and Applications, Robot Mechanism and Control
Abstract: In this paper, we propose a RTMPC (Robust Tracking Model Predictive Control) method for the trajectory tracking of a DDMR (Differential Drive Mobile Robot) under conditions of uncertainty. The Proposed control law employs feedback control law to reduce deviations caused by the inherent uncertainty in the model during the trajectory tracking process. The stabilizing gain value of the control law is obtained optimization using LMI (Linear matrix Inequality). Based on the feedback control law and Lyapunov stability, a feasible and invariant set is defined. The proposed control designs the MPC using feasible and invariant set to achieve control performance. The optimization process incorporates constraints related to the control input, cost function, and the feasible and invariant set. The performance of the proposed controller was verified through comparative simulation with the nominal model predictive control. As a result, it was confirmed that the performance of the proposed control method was superior.

Keywords: Industrial Applications of Control, Artificial Intelligent Systems, Control Theory and Applications
Abstract: This paper presents a sparse identification of nonlinear dynamic systems (SINDy) for a diesel engine air path system and nonlinear model predictive control (NMPC) with the SINDy model to attain good control performance. The air path system control is well known as a challenging problem, and many studies have been presented such as traditional model-based control design and machine learning. However, these conventional approaches still have some difficulties including the control performance and design costs. In this paper, we obtain the model of the air path system in a data-driven manner using the SINDy algorithm and construct the offset-free NMPC with the SINDy model. SINDy is a suitable modeling method for controlling a complicated air path system, owing to its characteristics of high computational efficiency, high learning efficiency, high modeling accuracy, and applicability to complex systems. Additionally, NMPC provides high control performance under constraints. The proposed offset-free NMPC with the SINDy model is verified through the simulations. The results show that the coefficient of determination of the SINDy model provided over 90%, and the controller performance of the NMPC was better than that of the traditional robust controller and satisfied the constraints.

Keywords: Artificial Intelligent Systems
Abstract: In this work, we present Curled-Dreamer, a novel reinforcement learning algorithm that integrates contrastive learning into the DreamerV3 framework to enhance performance in visual reinforcement learning tasks. By incorporating the contrastive loss from the CURL algorithm and a reconstruction loss from autoencoder, Curled-Dreamer achieves significant improvements in various DeepMind Control Suite tasks. Our extensive experiments demonstrate that Curled-Dreamer consistently outperforms state-of-the-art algorithms, achieving higher mean and median scores across a diverse set of tasks. The results indicate that the proposed approach not only accelerates learning but also enhances the robustness of the learned policies. This work highlights the potential of combining different learning paradigms to achieve superior performance in reinforcement learning applications.

Keywords: Artificial Intelligent Systems
Abstract: Kolmogorov-Arnold Networks (KANs) have shown potential as an alternative to Multi-Layer Perceptrons (MLPs) in neural networks, providing universal function approximation with fewer parameters and reduced memory usage. In this paper, we explore the use of KANs as function approximators within the Proximal Policy Optimization (PPO) algorithm. We evaluate this approach by comparing its performance to the original MLP-based PPO using the DeepMind Control Proprio Robotics benchmark. Our results indicate that the KAN-based reinforcement learning algorithm can achieve comparable performance to its MLP-based counterpart, often with fewer parameters. These findings suggest that KANs may offer a more efficient option for reinforcement learning models.

Keywords: Robotic Applications, Industrial Applications of Control, Robot Vision
Abstract: This paper presents an efficient deep reinforcement learning (DRL) framework for online 3D bin packing (3DBPP). The 3D-BPP is an NP-hard problem significant in logistics, warehousing, and transportation, involving the optimal arrangement of objects inside a bin. Traditional heuristic algorithms often fail to address dynamic and physical constraints in real-time scenarios. We introduce a novel DRL framework that integrates a reliable physics heuristic algorithm and object rearrangement and stable placement. Our experiment show that the proposed framework achieves higher space utilization rates effectively minimizing the amount of wasted space with fewer training epochs.

Keywords: Robotic Applications, Control Theory and Applications, Navigation, Guidance and Control
Abstract: Quadcopters have been studied for decades thanks to their maneuverability and capability of operating in a variety of circumstances. However, quadcopters suffer from dynamical nonlinearity, actuator saturation, as well as sensor noise that make it challenging and time consuming to obtain accurate dynamic models and achieve satisfactory control performance. Fortunately, deep reinforcement learning came and has shown significant potential in system modelling and control of autonomous multirotor aerial vehicles, with recent advancements in deployment, performance enhancement, and generalization. In this paper, an end-to-end deep reinforcement learning-based controller for quadcopters is proposed that is secure for real-world implementation, data-efficient, and free of human gain adjustments. First, a novel actor-critic-based architecture is designed to map the robot states directly to the motor outputs. Then, a quadcopter dynamics-based simulator was devised to facilitate the training of the controller policy. Finally, the trained policy is deployed on a real Crazyflie nano quadrotor platform, without any additional fine-tuning process. Experimental results show that the quadcopter exhibits satisfactory performance as it tracks a given complicated trajectory, which demonstrates the effectiveness and feasibility of the proposed method and signifies its capability in filling the simulation-to-reality gap.

Keywords: Robotic Applications, Artificial Intelligent Systems, Robot Mechanism and Control
Abstract: In the field of robotics, the efficiency of robotic manipulation tasks poses significant challenges, especially in dynamic and unpredictable environments. This paper introduces an innovative approach to enhance robotic manipulation through the application of reward shaping strategies based on the principles of reinforcement learning (RL). We propose a method that generates a diverse set of actions and automatically creates task sequences and reward functions. This structured approach enables the robot to learn and execute the necessary sequence of actions to achieve complex manipulation goals. Our methodology allows robots to autonomously design their learning processes, enabling them to learn and act like humans across various tasks.

Keywords: Robotic Applications, Artificial Intelligent Systems, Autonomous Vehicle Systems
Abstract: This paper introduces novel deep reinforcement learning (Deep-RL) techniques using parallel distributional actor-critic networks for navigating terrestrial mobile robots. Our approaches use laser range findings, relative distance, and angle to the target to guide the robot. We trained agents in the Gazebo simulator and deployed them in real scenarios. Results show that parallel distributional Deep-RL algorithms enhance decision-making and outperform non-distributional and behavior-based approaches in navigation and spatial generalization.

Keywords: Robot Mechanism and Control, Robotic Applications
Abstract: This paper presents the design of a three-wheeled mobile robot consisting of a three drive wheels and two supporting legs. We have designed a wheel-linkage mechanism for a stair-climbing robot, which is capable of navigating both flat and rugged terrains, as well as climbing steps. We developed the body control algorithm, and the robot was able to maintain the stability in its orientation while moving on different types of surfaces. Furthermore, the robot can overcome steps with a height of 12 cm, which is roughly 2.5 times the radius of its wheel. The experimental results show that the proposed robot, with its compact design, is capable of being used in flat and rugged terrains, as well as for step climbing.

Keywords: Robot Mechanism and Control, Sensors and Signal Processing, Human-Robot Interaction
Abstract: In this study, a robot system has been developed to systematically and effectively support reading activities for children and adolescent users. The robot interacts with users by grasping the user’s emotions through the user’s voice, image, and touch. A total of 8 emotional expressions are designed based on actual dogs’ behavior patterns to implement interaction with users. We design the hardware that can perform various emotional expressions based on language/emotion models with high degrees of freedom. The entire robot system is designed to implement complex emotional expressions through mechanical design and display while designing a high-spec controller module to process multiple communications, sensing, and graphic control in real time. In addition, an integrated sensor module is designed to enable analysis of various contact-type interactions of users. The interaction with the robot for the language/emotion model defined by the designed complex sensor/communication-based interface and the degree of freedom of control is discussed

Keywords: Robot Mechanism and Control
Abstract: In recent years, factory automation is growing in sectors like agriculture and fisheries, where grippers are used for various foods. Particularly, flexible foods present issue due to their non-uniformity and flexibility. To address these challenges, we propose an Enclosing-Scooping Gripper designed to handle flexible foods. This gripper consists of a Scooping Plate and a Silicone net. The Scooping Plate addresses the difficulty of gripping flexible foods from the ground due to their pliability and is designed in a scooping style to safety grasp them from the base. Additionally, the Silicone net structure is used to prevent escape flexible food after gripping, due to the flexibility of the foods. Finally, to validate the performance of the proposed Scooping-Enclosing Gripper, we selected four types of flexible foods from the food groups studied in existing food gripper research and conducted five trials each. We confirmed that the gripper could safely grasp all types of flexible foods in all trials.

Keywords: Robot Mechanism and Control, Robotic Applications
Abstract: This paper introduces a novel modular anthropomorphic robotic hand, leveraging the freedom and constraint topologies (FACT) framework to create advanced compliant mechanisms that replicate human joints characteristics. The design features a modular underactuated monolithic compliant finger with 7 degrees of freedom and a thumb with 5 degrees of freedom, as well skin-like attributes, alongside a monolithic compliant wrist with 2 degrees of freedom, resulting in a lightweight, versatile, and practical hand that mimics human dexterity for a broad range of robotic applications. Our design prioritizes user repairability and modularity, facilitating easy attachment and detachment of cable-driven fingers without specialized tools. Finger design includes a flexible skin module for adaptive grasping and embedded touch transmission mechanism incorporating a Force Sensing Resistor (FSR) for tactile feedback, enabling precise gestures and robust grip through a cable-driven actuation with a differential system to achieve 2 types of motion using one motor, offering decoupled modes of grasping: complete and partial flexion-extension only of the distal phalanx using three cables, as well is able to performer abduction-adduction motion by using a second motor with a single pair of cables. Additionally, the hand incorporates a monolithic compliant wrist joint, offering 2 degrees of freedom to provide seamless, natural movement that closely replicates the human wrist's flexion and extension, as well as radial and ulnar deviation, controlled by one motor and a pair of cables each. The hand has been designed to be manufactured using additive methods allowing the entire hand to be produced in a single print session using low-cost filaments for desktop 3D printer such as PETG and PLA, ensuring economic feasibility and wide reproducibility for various robotic systems, this development represents a significant advancement in robotics, providing a sophisticated yet accessible solution for robotic manipulators tools.

Keywords: Sensors and Signal Processing, Robot Mechanism and Control
Abstract: This paper presents the theory and numerical calculation results to improve the tip position accuracy of the Flexible Sensor Tube (FST) by integrating rotation angle sensors and Inertial Measurement Units (IMUs). The FST consists of multiple links connected by 1-degree-of-freedom passive rotational joints, each equipped with a rotation angle sensor to measure the joint's rotation angle. Traditionally, FST reconstructs the entire shape by solving forward kinematics using the measured joint angles to obtain the positions of all link tips. However, measurement errors in the rotation angle sensors lead to cumulative errors in the link's attitude, affecting the tip position accuracy. This paper proposes a method to enhance tip position accuracy by estimating joint angles using a combination of rotation angle sensors and IMUs. IMUs measure attitude relative to gravity, allowing attitude acquisition without accumulating measurement errors. Numerical simulations demonstrate that the proposed method achieves higher precision in estimating the FST tip position compared to using a single sensor. This approach addresses the cumulative error issue, providing a more accurate and reliable estimation method for FST applications in various extreme environments.

Keywords: Sensors and Signal Processing, Robot Mechanism and Control, Robotic Applications
Abstract: This paper presents a real-time attitude control system designed for a fish robot. The system is developed under the assumption that approximately 80% of the fish robot's body is submerged during testing, rather than being completely underwater. The system consists of a power unit that drives the fish robot and a control unit that manages its movements. The tail-flapping operation of the fish robot is driven by a Brushless DC (BLDC) motor with a rated voltage of 24 V. Servo motors are employed to control the movements of each fin. To ensure sufficient power supply to both the BLDC motor and the servo motors, six 3.7 V Lithium-Polymer (LiPo) batteries are connected in series using a Battery Management System (BMS). The control unit for attitude control consists of an MBED LPC1768 Microcontroller Unit (MCU) and an MTi-3 Inertial Measurement Unit (IMU). The data collection interval and attitude control cycle are set at 100 Hz, allowing the system to collect 9-axis attitude data, estimate the attitude, and perform control actions. The collected 9-axis attitude data and control unit data are recorded in the EEPROM at 40ms intervals. Additionally, a Radio Frequency (RF) communication module is installed for remote operation of the robot. The external remote controller generates coupled movements to produce the fish robot's surge, roll, pitch, and yaw motions. It also allows for PD gain adjustment and mode changes.

Keywords: Control Theory and Applications
Abstract: This paper proposes an improved stability analysis of the system with time-varying delays using the uniformly delay partitioning method and the generalized integral inequality based on free matrices. For the stability analysis of the system with time-varying delays, we use the Lyapunov-Krasovskii method which is widely used. The delay partitioning method enables to find the less conservative upper bound by establishing the Lyapunov-Krasovskii functionals according to each delay subintervals. Based on the delay partitioning method, a Lyapunov-Krasovskii functional contains the information of subintervals is constructed. Also, the generalized integral inequality based on free matrices can obtain more enhanced results, since this inequality uses both the single integral term and the higher-order multiple integral terms. Using these methods and the reciprocally convex approach, sufficient stability conditions are derived in the linear matrix inequalities form. Finally, one numerical example shows the effectiveness of the proposed stability analysis.

Keywords: Control Theory and Applications
Abstract: This study aims to achieve an enhanced admissibility criterion for singular systems with time-varying delays. Through state decomposition approach, we improve the computational complexity of admissibility analysis by separating the singular system into a differential subsystem and an algebraic subsystem. We obtain a less conservative admissibility criterion by further including the relationship between the state variables of the subsystems and by constructing Lyapunov-Krasovskii Functionals(LKFs) containing the state vector and its derivative in the double integral term. The admissibility criterion is represented in terms of linear matrix inequalities utilizing the integral inequality that estimates the upper limit of the integral term containing the state vector and its derivative. The comparison with prior works using a numerical example confirms the superiority of the proposed admissibility criterion.

Keywords: Control Theory and Applications, Artificial Intelligent Systems, Industrial Applications of Control
Abstract: This study uses a recurrent meta-cognitive fuzzy neural network (RMCFNN) to present an adaptive fractional order (FO) terminal sliding mode control (TSMC) method for the robust current management of active power filter (APF). By taking into account the fact that the external disturbances and parametric perturbations of the APF are bounded, a fractional order terminal sliding mode control is created. Due to an additional degree of freedom, the suggested scheme with a FO sliding surface can provide improved finite-time high-precision tracking performance as compared to the traditional TSMC approach. Next, in order to obtain an absorbing model-free feature resulting from RMCFNN, a novel observer-based FOTSMC is constructed. The construction of specialized online updating systems for the parameters and structure of RMCFNN aims to enhance the capacity to manage uncertainties. Meanwhile, Lyapunov theory can be used to obtain finite-time convergence characteristic and closed-loop stability. Ultimately, the findings of modeling and experimentation show that the suggested observer-based FOTSMC has better control performance than other current schemes and is simple to build using a microcontroller.

Keywords: Sensors and Signal Processing, Process Control Systems, Control Theory and Applications
Abstract: In the present work, a new real representation of reduced biquaternion matrix and the special properties of Vec(Phi_{EOF}) are used to solve the minimal norm Hermitian solution, pure imaginary Hermitian solution and pure real Hermitian solution of reduced biquaternion matrix equation, and the necessary and sufficient conditions are provided. Moreover, the corresponding numerical algorithms are given, and the real representation method and complex representation method are compared in numerical example. And it is shown that the method of this paper produces less error and shorter CPU time than the complex representation method, which is demonstrated the effectiveness and reasonableness of these algorithms.

Keywords: Control Theory and Applications
Abstract: The investigation of the oscillation theory in solving differential equations holds significant practical value and finds extensive applications in domains, such as radio engineering, communication engineering, computer networks, control systems and biology. This paper is concerned with the oscillatory and asymptotic behavior of third-order neutral differential equations with damping and distributed deviating arguments. By employing generalised Riccati transforms and integral averaging techniques, we establish some new theorems to ensure that all solutions of this equation oscillate or converge to zero, and the obtained results extend and improve some known results in the literature. Finally, we provide illustrative examples to demonstrate the significance of our main results.

Keywords: Artificial Intelligent Systems, Control Theory and Applications, Process Control Systems
Abstract: The saddle-point problems with nonlinear coupling operators frequently arise in various control systems. However, traditional primal-dual methods are constrained by fixed regularization factors. In this paper, a novel generalized primal-dual correction method is proposed to adjust the values of regularization factors dynamically. Numerical results show the proposed method outperforms other benchmark methods.

Keywords: Control Theory and Applications, Industrial Applications of Control, Autonomous Vehicle Systems
Abstract: Automated clutch is an important module in automated manual transmission. Clutch engagement control plays a crucial role, since different and conflicting goals have to be satisfied: preservation of driver comfort, fast engagement and small friction losses. In this paper, a proposed optimal control strategy to improve the clutch starting up quality of automated manual transmission vehicles during starting-up phase, based on LQR combined with Pontryagin’s maximum principle is developed. Based on the friction clutch dynamics, a plant model is designed with engine speed and clutch slip speed as state variables and the rate of change clutch clamping force as control variable. Matlab/Simulink platform has been used to simulate the controller performance for different operating conditions. Simulation results revealed that the control strategy improves the starting up quality in achieving fast engagement, reducing clutch frictional power loss and minimizing the jerk.

Keywords: Control Theory and Applications
Abstract: This paper investigates the existence of positive solutions to boundary value problems for nonlinear fractional differential equations with emph{varphi}-Laplacian operator, which is fundamental to the study of control theory for such problems. Based on fixed point theorems, several new existence results for positive solutions, in terms of the value of parameters h_{1},h_{2},h_{3} are obtained. Finally, two examples are given to illustrate the main results.

Keywords: Control Theory and Applications
Abstract: The existence and uniqueness of solutions of nonlinear fractional differential equations is the key point for stability theory and the design of feedback stabilization controllers. In this paper, we investigate the existence of solutions for initial problems of fractional hybrid differential equations of variable-order 0

Keywords: Control Theory and Applications
Abstract: This paper proposes a distributed algorithm for estimating the network size, which is the total number of agents in a network. Our approach is based on an optimization problem, where the solution corresponds to the network size and the objective function can be decomposed into individual agents' objectives. This enables the use of distributed methods such as the primal-dual gradient method. We focus on a continuous-time primal-dual gradient method and adapt it using an implicit-explicit scheme to run in discrete time. This approach eliminates the need for small step sizes and ensures rapid convergence. Unlike existing methods that require specific initial values, our method can provide the network size regardless of the initial values, making it robust to network changes.

Keywords: Control Theory and Applications
Abstract: The existence of solutions to partial differential equations is an important problem in control theory, as it ensures the feasibility of the associated mathematical models in practical applications. In this paper, we obtain the existence of at least one negative energy solution for a class of cooperative elliptic systems with perturbed terms. The innovations are as follows: (1) the nonlinearities are sub-linear or asymptotically-linear at infinity, and super-linear or asymptotically-linear at zero. (2) the nonlinearities allow to be sign-changing. (3) many examples can be applied to our results.

Keywords: Control Theory and Applications, Robot Mechanism and Control, Robotic Applications
Abstract: EffiDynaMix is a novel non-parametric dynamics modeling method that combines mathematical dynamics models with the Gaussian Mixture Model (GMM) for data-driven modeling. This gray-box approach simplifies dataset creation and enhances model generalization. EffiDynaMix outperforms traditional methods like conventional GMM, Gaussian Processes (GP), and Long Short-Term Memory (LSTM) networks, particularly in training efficiency and accuracy with new data. It focuses on dynamic equations to boost learning efficiency and adaptability to new scenarios, offering significant advancements in the precision and computational efficiency of robotic systems.

Keywords: Navigation, Guidance and Control, Autonomous Vehicle Systems
Abstract: Nonlinear model predictive control is a reliable and effective approach for optimization-based motion planning. In safety-critical control systems, controllers must solve inequality-constrained optimization problems. However, scene understanding, which consists of perceiving the driving environment and designing safety constraints, may complicate the optimization problem and is a resource-intensive process. In this paper, we propose a unified scene understanding method that uses occupancy grid maps (OGMs) to design a single unified constraint. We also propose a novel method for designing OGMs that method accounts for noise and uncertainties. We use this OGM approach for scene understanding to design a single constraint that ensures that only cells with occupancy probability values less than a predefined threshold can be traversed. We embed this constraint into the optimization problem as a single unified discrete barrier state. In the experiments, we compare the performance of the proposed method with that of an augmented Lagrangian method. The motion planning results in a pop-up obstacle avoidance scenario using an unmanned mobile vehicle demonstrate the advantages of the proposed method, such as reduced time costs and improved safety

Keywords: Control Devices and Instruments, Robot Mechanism and Control
Abstract: : This paper presents balancing control of a sphere by gyroscope actuators. A scissored pair of gyroscope actuators is designed and developed. Two scissored pairs of gyroscope actuators are implemented to control two angles, roll and pitch. Each scissored pair controls roll angle and pitch angle, separately. Induced torques are measured by empirical studies. Balance of the gyroscopic actuators on the sphere is controlled.

Keywords: Autonomous Vehicle Systems, Civil and Urban Control Systems
Abstract: Cross-view geo-localization (CVGL) is the problem of determining the location of a ground-level camera with respect to an extensive database of geo-tagged aerial images. While GNSS can provide location data, using images alone for localization can be especially beneficial in scenarios where GNSS signals are obstructed or unreliable. Although CVGL is treated as a retrieval-based visual place recognition (VPR) task, it is more challenging due to the critical viewpoint domain gap. To reduce this gap, contrastive learning methods have demonstrated superior performance, especially by applying hard negative sampling. However, the issue with previous hard negative sampling is its high computational cost, making training time-consuming when performed at every epoch. In this study, we propose a more effective hard negative sampling method. We introduce sampling strategies that remove unnecessary similarity map calculations, effectively computing only for the queries. We utilize the proposed effective hard negative sampling method, conducting contrastive learning with the information noise-contrastive estimation (InfoNCE) loss function. Our work demonstrated more efficient training and superior performance compared with the existing approaches on CVUSA and VIGOR, common cross-view datasets.

Keywords: Autonomous Vehicle Systems, Navigation, Guidance and Control
Abstract: The performance of autonomous backward driving control is subject to not only internal but also external conditions. In this paper, the way of improving the performance of autonomous reverse steering control of a vehicle is suggested by considering external road conditions such as the slope of the road. The control method is desired to overcome the problem caused by the changes on the road by a learning technique. Firstly, we learn external environmental variables and physical values using neural networks and then use them in the target trajectory of the vehicle. The feasible study of the intelligent autonomous reverse steering control is provided.

Keywords: Human-Robot Interaction, Robotic Applications, Artificial Intelligent Systems
Abstract: In this paper, the contact estimation problem (contact point localization and force identification) is tackled in a data-driven manner. We consider a collaborative robot equipped with proprioceptive sensors, specifically joint torque sensors and a base force/torque sensor. In the proposed method, a generative model, specifically a diffusion model, is utilized. A noise prediction network, conditioned on the proprioceptive sensor measurements, eliminates the noise from the Gaussian white noise to generate samples representing a contact point. Moreover, to identify the contact force associated with the samples, a properly designed quadratic programming problem is solved. Simulation experiments show the effectiveness of the proposed method.

Keywords: Robotic Applications, Control Theory and Applications, Artificial Intelligent Systems
Abstract: Learning from Demonstration is attracting attention due to its ability to solve complex tasks with a few demonstrations. However, its effectiveness in novel environments and responsiveness to local perturbations remain questionable. To address this problem, we propose an approach that defines and trains a dynamical system over latent space of variational auto-encoder, endowed with a Riemannian metric which defines inner product or distance between points, and also can make curved geodesic curve. The latent space of VAE is treated as a Riemannian manifold with a metric that includes meaningful information about the demonstration data distribution. Dynamic systems are formulated as Riemannian Motion Policies over the latent space, allowing them to be composed with other policies to generate complex motion that ensure reactivity. Our approach is validated on a 2D point-mass system goal-reaching task, demonstrating better robustness under disturbances and improved generalization to out-of-distribution initial points.

Keywords: Control Theory and Applications, Robotic Applications
Abstract: In this paper, we present a design approach for the optimal friction observer for flexible joint robots (FJRs) within the robust internal-loop compensator (RIC) framework. In disturbance observer (DOB) and RIC-based control structures, designing a low-pass filter (Q-filter) is a traditional but important issue. However, in conventional friction observer schemes for FJRs, systematically designing the Q-filter to meet any design specifications is challenging. Therefore, we propose an optimal friction observer that enables the systematic design of the sensitivity function, complementary sensitivity function, and Q-filter using the H ∞ mixed sensitivity optimization scheme. The optimization problem minimizes the H ∞ norm of the transfer function from disturbance inputs (e.g., friction and sensor noise) to the system outputs. Therefore, we can achieve an optimal friction observer that better suppresses both sensor noise and friction in robot joints than conventional observers. Furthermore, a simple design guideline for the prefilter is suggested to suppress joint torque sensor noise in a specific frequency range. The practical effectiveness of the proposed friction observer will be verified through simulation and experiment.

Keywords: Human-Robot Interaction, Artificial Intelligent Systems
Abstract: Disturbance Observer (DOB) is a well-known type of observer used to identify the residuals between the robot model and the real robot. This paper presents a long-term historical velocity memory for estimating disturbances in free motion for the learning-based method. However, estimating disturbances in free motion using the learning-based method is challenging when the joint angular velocity is zero. Since free motion disturbance can be easily estimated using the current joint velocity when a robot is moving, but when stationary, it relies not on the velocity but on a past motion history, which lacks sufficient data to describe the free motion disturbance accurately. Therefore, we propose the multiple queue-based motion discriminator (MQ-MD) and Motion sign discriminator (MSD), which use historical motion data as input for the learning-based method. The main results show that our approach reduces the noise in inference values compared to a baseline and decreases the number of thresholds, thus reducing the number of hyperparameters to tune. This study confirms that considering past joint velocities is essential for accurately estimating disturbances in free motion.

Keywords: Artificial Intelligent Systems, Robot Vision, Robotic Applications
Abstract: In this paper, we explore the impact of structure pruning on model compression. Structured pruning targets specific structures within the model for removal, such as entire neurons, channels, or filters in convolutional neural networks. This differs from weight pruning, which removes individual weights regardless of their location in the model. On top of our previous publications, the focus of this work is to reduce mobile stems in CNN-transformer architectures. Here, the mobile stems often make transformer architectures more efficient for deployment on mobile devices or other resource-constrained environments. The majority of pruning methods for the mobile stems take a sequential process consisting of training, pruning, and fine-tuning stages. In contrast, our automatic selection of magnitude or similarity-based filter pruning criteria from a specified pool of criteria and the specific pruning layer at each pruning iteration is guided by the network’s overall loss on a small subset of training data. To mitigate the abrupt accuracy drop due to pruning, the network is retrained briefly after each reduction of a predefined number of floating-point operations (FLOPs). The optimal pruning rates for each layer in the mobile stems such as VGGNet, ResNet, and MobileNet are automatically determined. Experiments on the VGGNet, ResNet, and MobileNet models on the CIFAR-10 and ImageNet benchmark datasets demonstrate the effectiveness of the proposed method. We discuss our works in progress and remaining tasks in the future.

Keywords: Artificial Intelligent Systems
Abstract: SAR (Synthetic Aperture Radar) uses radio waves to generate images, allowing for image collection regardless of weather and time. It can quickly scan large areas and produce high-resolution images. However, developing and operating SAR systems is costly, and specialized data processing techniques are required, making data collection limited. To address these issues, research is being conducted to train on synthetic SAR images to identify real SAR images.

Keywords: Exoskeletal Robot, Biomedical Instruments and Systems, Robotic Applications
Abstract: As the result of previous origami-based vacuum pneumatic artificial muscles (OV-PAM) contributing to the stable force output throughout the actuation, we tried to make use of the stable force output and apply it to a wearable device. To make use of that, we began with designing a new polygonal actuator using a new manufacturing method. With the working area of 325mm2, the actuator managed to produce approximately 70N when 70kPA negative pressure was applied. The output force showed a decrease of about 20% to 50kpa when the actuator was actuated to half of its total actuation length. This force output to actuator ratio was taken into consideration when designing the wearable device making sure that the area with only stable force output will be applied to wearer. The initial design of the wearable device proposed an investigation of the application of the wearable device during squat lifting. To make sure that the actuator did not obstruct the movement of wearer, a tendon mechanism was designed to transmit the force produced by the actuator.

Keywords: Industrial Applications of Control, Robotic Applications, Process Control Systems
Abstract: Recently, the number of smart manufacturing factories utilizing robot-based logistics systems is increasing. In order to automate the robot-based transportation system, supply, and recovery of logistics (parts, finished products, etc.) within a factory, various types of robots that can drive and process logistics are needed. Robots equipped with self-driving algorithms, like mobile manipulators, play a role in handling logistics lifting and unloading and transportation, allowing the process to be changed flexibly, and unlike existing logistics, they are capable of producing a variety of products in small quantities. Therefore, in smart manufacturing plants, a method of cooperating between mobile manipulators suitable for the process and robots that only perform transportation is needed for production efficiency. In this paper, we propose a cooperation method between mobile manipulators and transportation robots for logistics picking and placement. It is expected that smart factory logistics automation and flexible production will be possible using the proposed method.

Keywords: Industrial Applications of Control, Artificial Intelligent Systems
Abstract: This paper presents a predictive framework for real-time monitoring of 2.5 MW wind turbines using a hybrid deep learning architecture on a cloud platform. The core methodology integrates a Gaussian noise-enhanced ResNet with LSTM networks to capture both spatial and temporal dependencies, enabling accurate predictions of gearbox temperature and power output. Deployed on Amazon SageMaker, the model leverages cloud scalability for efficient real-time data processing, advancing traditional predictive models. Empirical evaluations demonstrate the model’s superior performance, highlighting its potential to enhance the sustainability and reliability of wind energy systems.

Keywords: Information and Networking, Control Theory and Applications
Abstract: In this paper, we investigate the existence of solutions for nonlinear fractional Langevin differential inclusion involving two fractional orders p, q ∈ (0, 1] with Dirichlet boundary conditions. By utilizing the fixed point theorem of α-admissible for multifunction, new existence result of solutions is obtained. Finally, an example is given to illustrate our main result.

Keywords: Biomedical Instruments and Systems, Artificial Intelligent Systems
Abstract: With the changes in modern lifestyle, the incidence of chronic venous insufficiency (CVI) is increasing, but the diagnosis of CVI remains challenging due to the difficulty in quantifying the progression and its reliance on the skills and experience of the physician. To address this issue, we propose a robotic ultrasound examination framework using a manipulator. This framework performs automated ultrasound examinations through the manipulator while automatically correcting errors based on ultrasound image feedback. The manipulator maintains a constant contact force during the measurement and evaluates the quality of the ultrasound images in real-time through a deep learning network. Experimental results demonstrate that the proposed system maintains a consistent contact force and correctly assesses ultrasound image quality. This suggests that the proposed framework enables the automation and quantification of CVI examinations.

Keywords: Artificial Intelligent Systems, Navigation, Guidance and Control, Autonomous Vehicle Systems
Abstract: To delivery robots, high accuracy of elevator button detection can satisfy requirements for delivery robots moving floor-by-floor well in building. However, previous button detection networks can't exclude mirror button well. In this paper, we propose a new pipeline to solve this problem. We made a new pipeline by unifying elevator button detection network, mirror segmentation network and our algorithm. First, object detection network YOLOv8 detects all elevator buttons in image by bounding box and return all box information including positions and classes. At the same time, segmentation network U-NET segments mirror region and return mirror segmentation mask. Lastly, our algorithm excludes mirror button information from YOLOv8's button information based on U-NET's masks. A new pipeline exclude mirror button stably using Our button grouping mechanism in our algorithm. So, we can validate performance of our algorithm quantitatively and qualitatively in our own datasets including elevator button image with mirror.

Keywords: Robot Vision, Artificial Intelligent Systems, Robotic Applications
Abstract: The occurrence of wildfires poses major challenges to disaster management and environmental protection, so early fire monitoring and management is crucial. With advances in deep learning, early wildfire detection algorithms from a drone perspective offer a promising solution. However, the complex fire environment results in low detection accuracy and poor real-time performance. Therefore, a lightweight real-time flame core detection algorithm based on UAV infrared images to solve the above problems is introduced. This paper proposes many innovative techniques that are different from existing algorithms. Firstly, this paper introduces a novel drone infrared imagery fire core dataset, utilizing infrared images for detection to circumvent obstruction issues. Secondly, an efficient processing (EP) module and a lightweight feature-sharing decoupled detection head (Fast head) are introduced to reduce network complexity. Finally, this work proposes Adaptive Sample Attention Loss (ASA loss) and simultaneously adopts Normalized Wasserstein Distance loss (NWD loss), which improves the network's detection accuracy for small and long-distance fires. Experimental verification shows that the model size of our algorithm is only 4.0M, which is 36.5% smaller than the original YOLOv8n network, and the number of parameters is reduced by 38%. The computational requirement was only 5 GFLOPs (-38.3%) and an average precision (mAP) of 77.5% (@50-95 IOU) was achieved, a 1% improvement.

Keywords: Robotic Applications, Control Theory and Applications, Sensors and Signal Processing
Abstract: In this paper, we introduce an underground parking lot SLAM (Simultaneous Localization and Mapping) method through sensor fusion of a stereo camera and 2D LiDAR without multiple cameras, which is necessary for an electric vehicle (EV) charging robot to drive to its destination. A ZED camera, which is mounted on the robot, enables the RTAB-Map (Real-Time Appearance-Based Mapping) algorithm to create a 3D map. Considering the structural characteristics of the underground parking lot, a 2D LiDAR was additionally used to complement the mapping in environments with insufficient features. During the RTAB-Map SLAM process, the robot repeatedly revisits the same paths to detect loop closures and update the map.

Keywords: Robotic Applications, Control Theory and Applications, Autonomous Vehicle Systems
Abstract: Path planning is one of the crucial elements in autonomous vehicle navigation, alongside simultaneous localization and mapping and path-following control. Effective path planning must be ensured even under adverse conditions, such as narrow passages or environments with complex obstacles. However, algorithms based on random sampling often fail in path planning or take a considerable amount of time to generate a path due to the nature of random sampling. Although many modified algorithms have been developed to address this issue, limitations still exist. This paper proposes an enhanced algorithm that incorporates intermediate vertex sampling between nodes and path pruning techniques to improve the efficiency of path planning. Compared to existing algorithms, our proposed algorithm shows a 2.32 % reduction in path length, despite an increase in path planning time, and it is capable of planning new paths that existing algorithms cannot find.

Keywords: Artificial Intelligent Systems, Autonomous Vehicle Systems, Navigation, Guidance and Control
Abstract: This paper presents an enhanced method for multi-object detection and tracking using the YOLOv10 algorithm and DeepSORT tracking algorithm augmented with a particle filter (PF) utilizing Kullback-Leibler distance (KLD) sampling. YOLOv10, known for its balance between computational efficiency and detection performance, is employed for object detection. DeepSORT, integrated with PF and KLD sampling, handles object tracking. The proposed method aims to improve tracking accuracy and robustness, especially in dynamic and complex environments. Simulations conducted on the DanceTrack dataset demonstrate the effectiveness of the proposed approach. Performance is evaluated based on metrics such as HOTA, DetA, and MOTA, showing significant improvements in tracking accuracy.

Keywords: Rehabilitation Robot, Human-Robot Interaction, Control Theory and Applications
Abstract: This paper introduces RAPHAR: Rehabilitation Assist Platform with Haptic Augmented reality Robot, a home-based upper limb rehabilitation robot capable of 3D motion. The system combines an omni-wheel based mobile base that can move in 2D with a spherical manipulator and a linear motor, enabling 5-DOF 3D motion. Additionally, it features two 6-axis F/T sensors that independently generate force feedback for hand and arm movements. Using an admittance-based control method, the system calculates and provides assist as needed for each movement. By integrating haptic feedback and AR technology, this system creates an immersive interactive rehabilitation environment, aiming to enhance patient outcomes at home through highly engaging adaptive exercises.

Keywords: Process Control Systems, Control Theory and Applications, Navigation, Guidance and Control
Abstract: The objective of this doctoral thesis is to design a distributed formation control system for swarms of unmanned aerial vehicles which addresses the challenges of scalability, collision avoidance, failure recovery, energy efficiency, and control performance. The swarms are arranged in tightly/loosely coupled architectures, which are based on homogeneous nodes in a distributed network of leader-follower/leaderless structures. The model of each node in the swarm formation is based on the nonlinear/linear dynamic model of a quadcopter, i.e. an unmanned aerial vehicle. The goal is to design the formation control of swarms of unmanned aerial vehicles, which is divided into high- and low-level control. From the high-level control perspective, the main contribution is to propose continuous path planning which can quickly react to events. Setpoints are generated for the swarms of unmanned aerial vehicles considering the complex movement of a hierarchical formation, soft landing, and failure recovery. The hierarchical formation and soft landing are executed using a fixed formation. Reconfiguration of the formation after node failures is implemented using a shortest path algorithm, combinatorial algorithms, and a thin plate spline. Besides this, from the low-level control perspective, the main contribution is to manoeuvre the nodes smoothly. The tracking of setpoints and stabilisation of each node is handled by a nonlinear sliding mode control with proportional derivative control and a linear quadratic regulator with integral action. The proposed strategies are evaluated using simulations, and the obtained results are compared and analysed both qualitatively and quantitatively using different scenario-relevant metrics. In addition, this doctoral thesis (Anam Tahir. Formation Control of Swarms of Unmanned Aerial Vehicles. Doctoral Dissertation, University of Turku, Turku, Finland, September 2023, ISBN: 978-951-29-9411-3. Available: https://urn.fi/URN:ISBN:978-951-29-9411-3) includes the five publications: (I) doi: 10.1016/j.jii.2019.100106. (II) doi: 10.1109/ACCESS.2020.2988773. (III) doi: 10.7148/2020-0168. (IV) doi: 10.1109/ELMAR49956.2020.9219027. (V) doi: 10.1109/ACCESS.2022.3181244.

Keywords: Navigation, Guidance and Control, Control Theory and Applications, Biomedical Instruments and Systems
Abstract: Untethered navigation of magnetic catheter is a technique that utilizes an external magnetic field to guide movement of catheter tip through a complex environment. Significant progress has been made in the development of catheter steering in clinical setup. However, it is still confined to the majority of treatment procedures due to its cost-effectiveness. In this paper, a compact bed-typed electromagnetic actuator (EMA) system is developed to steer the tip of the catheter by controlling the external magnetic field in the region of interest (ROI) surrounding the human heart. Coil configuration and coil structure were designed and optimized to generate the maximum magnetic field, regarding the size of the human body. The independent magnetic field control method is used for magnetic field generation to orient the catheter tip in the desired direction. The EMA system performances were verified and simulated using finite element method.

Keywords: Navigation, Guidance and Control, Control Theory and Applications, Biomedical Instruments and Systems
Abstract: Determining the position microrobot or initial position of the medical tools is a crucial task that significantly affects the accuracy of controlling the microrobot to the target position. This paper proposed magnetic induction-based three-dimensional localization method to determine the position and orientation of a miniature robot or tool at an arbitrary point in 3D. This developed method employed three receiving (Rx) coils mutually perpendicular to each other along the x, y, and z axes attached on the robot/tool, and one transmitting coil (Tx) fixed at the end-effector (EEF) of a 6-DOF robotic arm. The magnetic field generated by Tx coil was computed by Point-Dipole model, the Faraday’s law was applied to compute the electromotive force (emf) and the emf values measured at the Rx coils were applied to solve the position and orientation of the object. To evaluate the proposed method, some experiments were carried out and the Root-Mean-Square Error (RMSE) was using to calculate the error of the estimated position and orientation.

Keywords: Navigation, Guidance and Control, Sensors and Signal Processing, Biomedical Instruments and Systems
Abstract: Capsule endoscopes are widely used in medical diagnostics for internal examinations and monitoring. Modern capsules are often maneuvered using magnetic actuators, which require precise control to navigate through the gastrointestinal tract effectively. Accurate knowledge of the capsule’s orientation is essential for this control, and it can be achieved using an Inertial Measurement Unit (IMU). This paper presents the design and implementation of a millimeter-scale IMU suitable for integration into a capsule endoscope. The IMU uses the MPU-6050 processor equipped with a Digital Motion Processor (DMP) to obtain the orientation in terms of quaternions by fusing gyroscope and accelerometer measurements. Roll, pitch, and yaw are obtained by converting quaternions to angles, avoiding gimbal lock. Experimental results demonstrate the successful integration of the IMU within the capsule, yielding precise orientation estimations. The performance of the orientation estimation is measured by Root Mean Square Error (RMSE), with an RMSE of 0.96 degrees for roll, 2.10 degrees for yaw, and 0.64 degrees for pitch.

Keywords: Navigation, Guidance and Control
Abstract: In traditional Multi-Agent Path Finding (MAPF) problems, environments are typically represented as grid graphs. However, due to the high computational cost in continuous time domains, recent studies have employed roadmap graphs. This study assesses how different methods for generating roadmap graphs affect path planning. Through simulations in static environments such as factories and warehouses, we observed a balance in performance based on the attributes of each method. Notably, overlooking rotational costs in graph searches can lead to significantly higher path costs. Incorporating these insights, we devised graph generation techniques that markedly enhance path planning efficiency.

Keywords: Navigation, Guidance and Control, Industrial Applications of Control
Abstract: Multi-Agent Path Finding (MAPF) is a problem that involves planning collision-free paths from start to goal locations for individual agents. Numerous studies have been conducted to solve MAPF. However, there are many challenges that should be addressed to apply the planned paths in real-world applications. This paper explores these issues, focusing on path planning and execution for multi-robot operations. Briefly, during the planning phase, both the characteristics of the robots and the workspace should be considered for direct application in robot control. In the execution phase, the system needs to adapt to unexpected environmental changes.

Keywords: Navigation, Guidance and Control, Robot Vision, Autonomous Vehicle Systems
Abstract: Localization using onboard sensors installed in vehicles is essential for technologies such as autonomous driving and HD map creation. Localization with systems composed of low-cost sensors is receiving attention. Offline computation-based localization can be used in applications such as SLAM for HD map creation and environmental change detection. These applications require accurate localization and precise feature detection. However, low-cost sensors produce data with high uncertainty. In the case of GNSS, buildings around urban roads can cause significant errors. In this paper, we propose offline pose correction for vehicles driving on urban roads, taking into account the inaccuracy of the data. The proposed pose correction overcomes data uncertainty by reflecting the structural features of road intersections. Validation was conducted using GNSS, odometry, and traffic light observations collected from drives on urban roads.

Keywords: Process Control Systems, Industrial Applications of Control
Abstract: The power plant operation simulator is a system that can simulate a power plant for various operating conditions. Therefore, the fidelity to the power plant is an important factor. But in general, fidelity tends to decrease with longer operation period. One of the reasons is that it is difficult to reflect performance changes due to deterioration or aging of power plant facilities. In order to solve these issues, we describe the simulation system with a process model parameter tuning function based on power plant operation data. We analyze four types of optimization algorithms by applying parameter tuning, and confirmed the error rate of around 1% compared to the actual data through the method. The developed simulator is installed and has been operating in the power plant site. In addition, in connection with other simulator, more detailed and analytical simulation is possible as needed.

Keywords: Rehabilitation Robot, Human-Robot Interaction, Robot Mechanism and Control
Abstract: There is a challenge in providing adequate rehabilitation training for patients who require effective gait training due to a shortage of physical therapists and limited training time. In this paper, we discuss the design of a pelvic assistance arm to be equipped on a wheeled gait training device intended for continuous rehabilitation of patients.As a reference for the design, we measured the pelvic displacement during walker-assisted gait and the pelvic assistance force provided by PT (physical therapists) in previous studies. Additionally, we simulated the torque of each joint of the assistance arm expected during gait assistance based on the measurement results of the assistance force. Using these reference data, we designed the pelvic assistance arm and proposed a gait training device equipped with the arm.

Keywords: Rehabilitation Robot
Abstract: This paper presents the importance of the physical realizability of position perturbations and the consideration of the robot dynamics for the reliable and accurate estimation of the 3-dimensional (3-D) mechanical impedance transfer function matrix (MITFM) with a nonparametric stochastic estimation method.

Keywords: Rehabilitation Robot, Human-Robot Interaction, Sensors and Signal Processing
Abstract: Recently, wearable user movement assistance devices have emerged as one of the ways to solve the problem of aging. In this study, a system for effective assistive force support was built by predicting the user's lower extremity joint movements and classifying actions such as walking and squatting. In order to support flexion and extension of the hip joint and extension of the trunk, a suitable wearable harness and an external 6DoF assistive force generator were developed and tested.

Keywords: Rehabilitation Robot, Robot Mechanism and Control, Robotic Applications
Abstract: A lower extremity constraint-induced movement therapy (LE CIMT) using a spasticity reduction device was used to treat four stroke patients suffering from hemiplegia. Upon intervention, both walking speed and stability improved, reliance on assistance from an accompanying individual or support device decreased, and both improvements to standing position during daily activities and walking activity were seen.

Keywords: Rehabilitation Robot, Sensors and Signal Processing, Robot Vision
Abstract: A study related to life after stroke found that balance and walking difficulties were the most requested area of research, especially among the elderly. In a study of stroke patients, they exhibited slower walking speeds, asymmetrical gait, unstable balance, and lower functional performance. To measure quantitative gait parameters, various systems have been developed to measure and analyze gait information. However, commercial gait analysis systems are expensive and not completely non-wearable. Therefore, we developed a system that combines the Tactile sensor and the RGB camera to perform accurate gait analysis while considering non-wearability and the low cost of the system. We performed a pilot test to analyze the temporal and spatial parameters of the walking-in-place with the proposed gait analysis system. The pilot test was able to obtain quantitative gait parameters. For future work, we plan to measure ground contact detection of abnormal gait in stroke patients as well as spatial and temporal gait parameters by extending tactile sensors longer according to the direction of walking.

Keywords: Robot Mechanism and Control, Robotic Applications
Abstract: Several actuators using pneumatic pressure have been developed to create light and flexible movements in various soft robot fields. However, various systems such as pneumatic pumps and regulators are required to use pneumatic pressure. To simplify this pneumatic system, this study introduces an actuator that increases pneumatic pressure through heating of the air and simultaneously operates a shape memory alloy to create linear movement through a driving method that integrates the two forces. To confirm that it was strengthened by SMA, the displacement and pressure were measured and compared at each driving force (Heat expansion & SMA, SMA, Heat expansion), and the increased driving force was confirmed. Additionally, to verify high performance, the increased chamber with 80 SMA springs fabricated, and the displacement control at a weight of 20 kg was performed. For displacement control, a PID controller was applied to the PWM control method for the power applied to the nichrome wire.

Keywords: Robot Mechanism and Control, Robotic Applications
Abstract: To ensure stable walking of a quadruped robot in complex terrains, it is necessary to generate command trajectories adapted with the terrain. This paper proposes a general control technique that allows stable walking in various terrains by adapting the robot's center of mass (CoM) pitch to the slope of the terrain and imposing constraints on the model predictive control (MPC) to ensure foot placement in safe positions.

Keywords: Robot Mechanism and Control
Abstract: Recent advancements in Remote Actuation Systems (RAS) have led to the development of the innovative X-ray Transparent Remote Actuation System (XTRAS). This paper explores various force transmission methods, such as pneumatic, hydraulic, and cable-based systems, highlighting their adaptability in achieving mechanical motion. However, challenges arise in converting hydraulic power into rotational motion using cables. Despite efforts to overcome these challenges, including experiments with different components and configurations, gear-driven mechanisms emerge as a resilient solution. Through interdisciplinary collaboration and innovative design approaches, the study showcases the potential of XTRAS in applications requiring X-ray transparency, particularly in medical imaging. Experimental results demonstrate the feasibility of utilizing hydraulic pressure to drive rotational motion, paving the way for innovative applications in medical imaging and beyond.

Keywords: Robot Mechanism and Control
Abstract: Cable-Driven Parallel Robots(CDPR) is a type of parallel robot driven by flexible cables, characterized by low inertia, high load capacity, and a wide workspace. CDPR with an incompletely restrained positioning mechanism (IRPM) maintain stability under external forces such as gravity. In contrast, CDPR with a completely restrained positioning mechanism (CRPM) have their end-effector positions fully determined by the cables, making it easier to maintain stability. However, if the end-effector moves outside the base platform area where the cables are anchored, it cannot maintain the constraints. Therefore, this paper designs a CDPR with an expandable end-effector to perform tasks outside the base platform area while using a fully constrained CDPR. By calculating the cable lengths based on the tool's position through coordinate transformation, the end-effector's position can be estimated. The workspace analysis based on the model showed that the area below the base platform is 200mm * 100mm * 100mm. Simulations revealed that the tool could use a larger area than the actual workspace through rotation alone.

Keywords: Robot Mechanism and Control, Human-Robot Interaction, Control Theory and Applications
Abstract: A bilateral control system aims to share force and position information bidirectionally between similar manipulators. The manipulator has unknown nonlinear plant perturbations caused by friction in joints. This makes it difficult for the bilateral control system to achieve stability and good transparency performance. In this paper, an adaptive control system is developed for single-axis DD-motor which is used as the joints of the manipulator to handle the aforementioned issue. The adaptive control makes the DD-motor linear time-invariant by conforming an output of the DD-motor to an output of the reference model. In addition, the bilateral position and force controllers are designed for the reference models. If the outputs of DD-motor converge to the outputs of the reference model, stability of the proposed control system is guaranteed by passivity. Some simulation results show that the proposed control system performs great transparency of both position and force tracking when the plants have time-varying nonlinear friction.

Keywords: Robot Mechanism and Control, Robotic Applications, Sensors and Signal Processing
Abstract: Dexterous robot hand fingertips are essential for performing precise, delicate, and complex tasks. These fingertips allow robots to handle fragile and intricate objects safely and efficiently. In this study, we introduce a novel fingertip design integrated with a 6-axis force/torque (F/T) sensor, which enables accurate calculation of contact location. Although our goal is to achieve precise contact location estimation in three-dimensional space (x, y, z), the current experimental validation focused on specific axes within different regions due to deformation observed during manipulation. This approach allowed us to obtain reliable data. Given these limitations, the experimental results showed an average error of less than 1㎜, indicating that the designed fingertip can achieve a high level of precision in manipulation tasks. Future work will extend this analysis to full three-dimensional contact location estimation and further refine the fingertip design to reduce deformation effects.

Keywords: Robot Mechanism and Control, Industrial Applications of Control, Process Control Systems
Abstract: The operation of multi robots has been a much-debated topic due to its potential to be applied in many different business sectors. Incheon International Airport is not an exception and we have been introducing a variety of robots to assist in airport operations and passenger experience. However, every new robot we bring to the airport leads us to implement a whole new system that monitors and manages the robot. To solve this issue, we have constructed a functional model for a multi-robot control system. Based on the model, the entire process in which multi robots collaboratively synchronize their maps, analyze the context data, and finally generate and allocate collaborative tasks is specified. Incheon Airport aims to further build upon the proposed model and eventually develop a centralized multi-robot control system that can essentially unify the data format, functions and task specifications for robots from multi-vendors.

Keywords: Robot Mechanism and Control, Control Theory and Applications
Abstract: In this paper, adaptive sliding mode control is introduced for underactuated mechanical systems with significant parametric uncertainty using a fuzzy logic approach. The TORA system, often tested as a benchmark for underactuated systems control, serves as the primary focus of this study. To approximate the nonlinear dynamics of the TORA system, a fuzzy basis function is employed, enabling the design of a function-based control system. Additionally, the integration of an adaptive control method enhances the robustness against approximation errors of the fuzzy system. The stability of the proposed control system based on the function approximation was validated using the Lyapunov second theorem, confirming the need for adaptive control. Simulation results demonstrated the effectiveness of the control system under inaccurate models, revealing significant robustness against mismatched errors. Further performance improvements through advanced development of fuzzy systems are also suggested as potential areas for future research.

Keywords: Robotic Applications, Artificial Intelligent Systems, Information and Networking
Abstract: As the routine operations are starting to become highly automated, it is crucial to develop autonomous solutions that are infrastructure independent. Achieving this is challenging due to the ever-changing landscape of mines, which complicates infrastructure development. In response, this paper introduces a robust framework employing drones to gather data from hard-to-access areas in mines and deliver the data back to the base station for routine monitoring purposes. These tasks include gathering data from operational vehicles (mine trucks, loaders etc.), as well as various sensors (e.g. monitoring rock bolts) and relaying the data to the mine's base station for monitoring purposes. The proposed framework is based on autonomous navigation using a known point cloud map of the mine, proximity detection via Ultra WideBand~(UWB) radios and the data transfer is accomplished through the IEEE~802.15.4 communication standard, operating in the 868~MHz ISM band, with the aim to guarantee long range operation. On the mission level, the drones act as data mules capable of autonomously extracting data from operating vehicles, storing the data onboard and eventually delivering the data to the base station, which is enabled through a Point and Click (PAC) autonomy framework based on global planning, reactive navigation, communication link and behavior management. The efficacy of this framework has been demonstrated through real-world experiments conducted at a test mine in Sweden, validating the overall architecture of the proposed solution.

Keywords: Robot Vision, Robotic Applications, Autonomous Vehicle Systems
Abstract: Visual servoing plays a crucial role in robotics, spanning across a great spectrum of applications from autonomous cars to aerial manipulation. This article proposes TFMarker, a novel tangible fiducial pattern for enabling camera-assisted guided landing of UAVs by using the visual features from color markers as the main source of information. TFMarker is structured around a 4-point fiducial marker, allowing for accurate, precise, and consistent pose estimation in different environments and lighting conditions, while also offering resilience to motion blur. The presented detection framework is based on a three-step architecture, where the first step uses Gaussian and color filtering in addition to morphological operation in order to generate a robust detection of the markers. The second step uses the Gift Wrapping Algorithm, to organize the same-color markers based on their relative positioning with respect to the off-color marker. Finally, the Perspective-n-Point optimization problem is solved in order to extract the pose (i.e. position and orientation) of the proposed pattern with respect to the vision sensor. The efficacy of the proposed scheme has been extensively validated in indoor and SubT environments for the task of autonomous landing using a custom-made UAV. The experimental results showcase the performance of the proposed method, which presents a better detection rate in both environments while retaining similar accuracy and precision to the baseline approach. For the video of the experimental evaluation please refer to the following link: https://youtu.be/Zh13OObp15Q