Keywords: Robotics, Machine Learning
Abstract: The generalizability of visuomotor policy models is crucial for their real-world usefulness in settings such as industrial environments. This is heavily impacted by the choice of visual encoder. In this paper, we integrate the DINOv2 foundation visual encoder with Diffusion Policy by designing a spatially-aware projection head, that allows the policy to shape its visual representation while benefiting from DINOv2’s robust embeddings. We evaluate this in drastic out-of-distribution conditions. As success rate can be uninformative in these conditions, where failure rates are high, we present three evaluation criteria for goal-driven policies that remain informative despite task failure. Our result shows that our approach outperforms the baseline under color alterations and camera displacements. We observe promising emergent task-relevant feature tracking using the DINOv2 visual encoder for policy learning.

Keywords: Robotics, Machine Learning
Abstract: Despite significant advances in bipedal locomotion, enabling humanoid robots to perform general whole-body tasks through meaningful interaction with their environments remains a challenging open problem. While deep reinforcement learning (RL) has recently demonstrated impressive results in dynamic walking --- even on complex and unpredictable terrain --- real-world utility demands that humanoids go beyond locomotion to execute task-oriented behaviors.

In this work, we propose a framework for teaching humanoid robots to imitate humans doing useful tasks by training policies for tracking human motion references. Our approach leverages high-quality in-house motion capture (MoCap) data, from which we perform kinematic retargeting to project human trajectories onto a humanoid platform. Crucially, we adopt a hybrid learning paradigm: the policy is trained to track upper-body and root motions from the MoCap data, and receives additional supervision from a pre-trained omnidirectional walking expert. This expert guidance, implemented via a Behavior Cloning (BC) objective, ensures that leg motion respects dynamics and kinematic constraints of the humanoid. We train policies entirely in simulation and successfully transfer them to a real humanoid robot. We validate our method on a box loco-manipulation task, demonstrating effective sim-to-real transfer and marking a step toward more capable, task-driven humanoid behavior.

Keywords: Virtual / Augmented / Mixed reality, Robotics, Machine Learning
Abstract: The rapid advancement of Vision-Language-Action models has created an urgent need for large-scale, high-quality robot demonstration datasets. Although teleoperation is the predominant method for data collection, current approaches suffer from limited scalability, complex setup procedures, and suboptimal data quality. This paper presents XRoboToolkit, a cross-platform framework for extended reality based robot teleoperation built on the OpenXR standard. The system features low-latency stereoscopic visual feedback, optimization-based inverse kinematics, and support for diverse tracking modalities including head, controller, hand, and auxiliary motion trackers. XRoboToolkit's modular architecture enables seamless integration across robotic platforms and simulation environments, spanning precision manipulators, mobile robots, and dexterous hands. We demonstrate the framework's effectiveness through precision manipulation tasks and validate data quality by training VLA models that exhibit robust autonomous performance.

Keywords: Medical Devices, Hardware Design, Mechatronics Systems
Abstract: In quantitative diagnosis of tracheomalacia using endoscopic examination, it is necessary to perform spatiotemporal quantification of cardiogenic oscillations on the tracheal wall that interfere with the measurement of respiratory fluctuations. Measurement accuracy and characteristics by high-speed endoscopic imaging are unclear against cardiogenic oscillations observed in tracheal endoscopic images, which are considered to include arterial pulse waves. To understand the mechanism of such cardiogenic oscillations, it is also necessary to distinguish between arterial pulse waves and cardiac tissue motion. In this paper, we propose an arterial pulse wave simulator consisting of a pulsatile pump and an ultra-flexible tube. Pulse transit time is measured by laser displacement sensors placed at two points on the tube deformed by the pulse wave. By adjusting the pressure of the liquid inside the tube, it is possible to generate specific pulse wave velocities within a certain range. Using the adjusted pulse wave velocity as a reference value, we quantitatively evaluate the measurement accuracy of pulse waves obtained by high-speed endoscopic imaging. We have experimentally evaluated the quality of pulse wave velocity adjustment of the developed simulator and demonstrated the estimation of pulse wave velocity by endoscopic texture tracking.

Keywords: Machine Learning, Automation, Robotics
Abstract: Robots hold great promise for performing repetitive or hazardous tasks, but achieving human-like dexterity, especially in contact-rich and dynamic environments, remains challenging. Rigid robots, which rely on position or velocity control, often struggle with maintaining stable contact and applying consistent force in force-intensive tasks. Learning from Demonstration has emerged as a solution, but tasks requiring intricate maneuvers, such as powder grinding, present unique difficulties. This paper introduces Diffusion Policies For Compliant Manipulation (DIPCOM), a novel diffusion-based framework designed for compliant control tasks. By leveraging generative diffusion models, we develop a policy that predicts Cartesian end-effector poses and adjusts arm stiffness to maintain the necessary force. Our approach enhances force control through multimodal distribution modeling, improves the integration of diffusion policies in compliance control, and extends our previous work by demonstrating its effectiveness in real-world tasks. We present a detailed comparison between our framework and existing methods, highlighting the advantages and best practices for deploying diffusion-based compliance control.

Keywords: Machine Learning, Decision-making systems, Automation
Abstract: Generalization is essential for accelerating optimization workflow, enhancing scalability and reducing computational cost. This paper presents a novel Feature-Conditioned Reinforcement Learning (FC-RL) method for engineering optimization, which enables generalization across diverse problem scenarios by conditioning the RL policy on explicit problem features. Unlike traditional optimization methods that require hyperparameter tuning for each new task, FC-RL leverages a meta-network to condition its behavior dynamically through Feature-wise Linear Modulation (FiLM). The experiments conducted across six multi-modal benchmark functions with varying dimensions, search-space, and global optimum shifts demonstrate that FC-RL often achieves global convergence without the need of retraining or retuning, while producing consistent and reproducible results due to its deterministic policy. Although FC-RL's performance depends on sufficient training diversity and is limited to novel scenarios within the known environments, these findings indicate that the proposed method is particularly well-suited for repetitive engineering optimization tasks involving variable problem configurations across multiple systems.

Keywords: Machine Learning, Automation, Robotics
Abstract: Contact-rich industrial manipulation poses a significant challenge for reinforcement learning policies, requiring dexterous interactions with objects exhibiting complex contact dynamics. Additionally, in industrial applications, completion deadlines are equally important to task success, for integration into wider processing pipelines. Further, in the standard reinforcement learning setting, failure to account for remaining time in episodic tasks can result in state aliasing or inconsistent temporal difference errors, therefore, this research work seeks to determine the most effective integration of time limits in policy learning. We propose that the remaining time be used as both an input and a scaler for the task success reward, demonstrating the effectiveness for the dexterous unscrewing of a nut from a bolt. The resulting time-based policy completes the unscrewing task with a success rate of 90% in 10 simulated trials, the highest of all approaches considered, including a standard baseline. It takes an average completion time of 21.67 seconds across the trials, given a 35 second time limit, which, while not the fasted method considered, may indicate more stable motion resulting from awareness of the time limit. Finally, the efficacy of the learned unscrewing policy is validated on a real UR5e manipulator for the nut-bolt disassembly task.

Keywords: Control Technologies, Robotics, Machine Learning
Abstract: This work proposes a learning-based approach to dynamic non-prehensile object reorientation, enabling fast reorientation of large, grasp-infeasible objects using uni-manual manipulation. Our policy is trained in simulation via reinforcement learning, utilizing a carefully designed observation space, action space, and reward function to reorient randomly sized cuboids with varied physical properties. Given an object model and a target rotation direction, the policy plans offline trajectories suitable for both simulation and real-world deployment. Although the policy is sensitive to modeling uncertainties, accurate modeling enables successful sim-to-real transfer across different objects and rotation directions. Videos are available at https://tinyurl.com/DNPR2L.

Keywords: Integration Platforms, Software Design, Network Systems
Abstract: In this paper, we propose a software platform to realize a Software-Defined Robot (SDRobot) for a large number of mobile robots performing exploration and base construction. To develop a platform that adapts to unknown environments by utilizing the redundancy of multi-robot systems and changing functionality through software, we extended the robot middleware OpenRTM-aist with a dynamic module allocation function and an automatic recovery function. The extended functions provide the foundation for a system that adapts to the environment by changing the software configuration depending on the status of resources such as control computers and the state of communication between robots. In this paper, we show an example of constructing a system with multiple RED mobile robots and confirmed the operation of the extended functions.

Keywords: Hardware Design, Mechatronics Systems, Control Technologies
Abstract: Container cranes are core equipment in port logistics, but issues remain regarding operational efficiency and safety due to factors such as the aging of operators, labor shortages, and the sway caused by strong winds or acceleration and deceleration. In particular, in the development and evaluation of remote operation and automation systems, using container cranes involves high costs and strict safety restrictions, which reduce development efficiency. Therefore, in this study, a small-scale crane model at approximately 1/30 of the actual size was designed and manufactured, and its operation was evaluated. As initial steps toward remote operation and automation, a remote-control system and an automation control system using the small-scale crane were constructed. And the feasibility of basic operations was confirmed. Furthermore, the position of the spreader was estimated using LiDAR, and its effectiveness was verified.

Keywords: Software Design, Robotics, Hardware Design
Abstract: This paper describes the winning robot system in the WRS2025 Future Convenience Store Challenge (WRS2025 FCSC), an international competition focused on tasks related to managing convenience store merchandise stock. To ensure reliable and efficient task execution, we enhanced the XYZ stage-type display system, which automates stock and disposal tasks. We introduced a two-stage recognition method that combines object detection using YOLO and identification using ArUco markers, enabling the high-precision acquisition of product type, position, and orientation. Furthermore, the work planning software was divided into three components: product management, task planning, and motion planning, improving program readability and debugging efficiency. Evaluation of the developed system at WRS2025 FCSC yielded the following results: in the preliminary round, it achieved a perfect score of 54 points in the Stock Task and 49 points in the Stock and Disposal Task. In the final round, it scored 54 points in the Stock Task and 31 points in the Stock and Disposal Task, securing the overall championship.

Keywords: Automation, Robotics, Control Technologies
Abstract: In recent years, competitions such as the World Robot Summit (WRS) have spurred the development of robotic systems aimed at automating operations in convenience stores and other retail outlets. However, many existing methods require extensive modifications to products and shelves, and are highly dependent on the environment, making their deployment in brick-and-mortar stores problematic. In this study, we propose a mobile manipulation system that performs highly accurate product stocking, disposal, and customer detection while minimizing environmental modifications. The proposed system consists of three elements: (1) highly accurate product position recognition using an industrial camera and ArUco markers; (2) robot arm control with variable grasping strategy, coordinate correction, collision prevention, and retry functions; and (3) customer approach detection using 2D LiDAR. Testing in a competition scenario simulating a real-world environment demonstrated improved grasping success rates, detection of discarded products under obstructed visibility, and safety, confirming both practicality and competition performance.

Keywords: Medical Devices, Robotics, Automation
Abstract: Automation of healthcare workflows and devices demands safe and trustworthy robotic behavior, particularly in environments shared with patients and medical staff. For ceiling-mounted imaging robots, the key challenge lies in perceiving and monitoring the 3D workspace to plan safe, collision-free motions around people and equipment. Beyond simple obstacle avoidance, semantic understanding is essential to distinguish between object types — such as patients, walking aids, or medical tools — and to adapt motion behavior accordingly. We address this challenge with a semantic-aware obstacle tracking and avoidance pipeline that extends prior 2D semantic navigation concepts into full 3D space. The approach combines 2D semantic segmentation with depth projection to estimate object positions and dimensions in real time from RGB-D data. These detections are fused in a tracking module to build a continuous, semantic world model from which class-dependent safety margins are derived. The resulting information enables adaptive motion planning that increases distance from high-risk objects (e.g., persons) or reduces velocity when close interaction is required. Experiments on a real ceiling-mounted robot in laboratory scenarios demonstrate the system's ability to enhance safety, predictability, and contextual awareness during automated healthcare procedures.

Keywords: Robotics, Control Technologies
Abstract: During autonomous navigation, mobile robots often need to avoid obstacles in their path. To address this obstacle avoidance issue, we have proposed various motion planners based on deep neural networks. Focusing on obstacle avoidance, a mobile robot is typically required to move straight for a certain duration after avoiding an obstacle before reorienting itself toward the destination. However, in such cases, it is difficult for the robot to plan these different motions based on similar image inputs. To address this challenge, we propose a novel motion planner based on a Time-Delay CNN that utilizes visually distinct time-series image inputs. Through experiments, we demonstrate that the robot is able to plan appropriate avoidance motions as described above and navigate toward the destination in both simulation and real-world environments with multiple dynamic obstacles.

Keywords: Robotics, Automation, Control Technologies
Abstract: In this paper, we propose a method of motion planning using MPPI based on obstacle cost maps, to realize stable avoidance behavior for moving obstacles. The proposed method creates cost maps considering moving obstacles, and performs motion planning based on cost function formulation with MPPI. In cost map creation, we propose a cost shape that takes into account velocity and predicted poses of moving obstacles. In motion planning, we adopt MPPI which is one of the mainstream sampling-based methods. To apply cost maps to MPPI, we formulate cost functions of MPPI that appropriately avoid obstacles. In simulation experiments, we performed moving obstacle avoidance in an environment with 50 pedestrians, and our proposed method achieved the highest success rate (94%). Our method realized stable avoidance behavior for moving obstacles even in crowded environments, by creating cost maps considering velocity and predicted poses of moving obstacles, and performing motion planning based on these maps using MPPI.

Keywords: Human-robot Interaction / Collaboration, Robotics, Medical Devices
Abstract: Autonomous medical systems must meet stringent hygiene and safety requirements while operating reliably in dynamic clinical environments. This paper presents a collision avoidance system based on the fusion of two mm-wave radar technologies - frequency modulated continuous wave (FMCW) and pulsed coherent radar (PCR). The system is fully integrated behind sealed covers of medical devices and enables unobtrusive and hygienic use without compromising functionality. We demonstrate that both radar types provide robust detection of dynamic obstacles, even through layers of disinfectants, blood and polycarbonate materials. A failsafe system architecture based on redundant sensor paths and dual microcontrollers ensures reliable operation under fault conditions. Experimental validation on a robotic X-ray system confirms the responsiveness, accuracy and suitability of the system for clinical integration. The results show that radar fusion offers a promising path to hygienic and certified safe motion planning in healthcare robotics.

Keywords: Human-robot Interaction / Collaboration, Virtual / Augmented / Mixed reality, Automation
Abstract: In the teleoperation of skilled manual tasks such as precision polishing, haptic feedback with a high degree of fidelity is considered essential. However, matching the dynamic characteristics of the leader and follower systems is challenging, often degrading feedback quality. This degradation can impair demonstration data and reduce learning performance for both machine learning systems and human users. Conventional solutions have relied on complex hardware or control schemes, increasing cost and limiting applicability. Several studies have explored improving haptic perceptual quality via minimal-cost vibrotactile signal processing that deforms or exaggerates the original haptic signal. To examine the effects of different processing strategies, we compared four vibrotactile deformation methods: perceptual low-pass filtering, frequency-bin amplitude change emphasis, envelope-based nonlinear gain, and task-specific band-pass filtering emphasizing vibration features identified from real polishing data. Pre-recorded vibrations were processed offline and presented to participants using a custom mock polisher with two voice-coil actuators. A perceptual evaluation revealed no significant differences in switch response behavior to vibration changes, indicating contact and pressing of the metal surface. In contrast, the task-specific band-pass filtering condition yielded significantly higher clarity, naturalness, and confidence scores than the others. The main contributions of this work are the development of a low-cost, medium-fidelity metal mock polisher and a task-specific vibration frequency deformation strategy for polishing, which can improve the perceived clarity of pressing sensations compared to both the strategies drawn from prior work and all other strategies tested here.

Keywords: Virtual / Augmented / Mixed reality, Human Factors, Human-robot Interaction / Collaboration
Abstract: Tactile information is known to be used and integrated into various fields, such as robotics, virtual reality, and healthcare, to improve immersion, telepresence, performance, and substitute for different sensations. Vibrotactile stimulation is a widely used haptic modality, owing to its ease of modification and integration into existing interfaces. Many studies have focused on the characteristics of the magnitude spectrum in the frequency domain. We focus on the phase spectrum as a novel parameter to modify for a wider variety of rendered vibrotactile stimuli. In this study, we evaluated the effect of the phase spectrum on the tactile perception by generating Noise-Texture chimeras, which consists of the magnitude spectrum of colored noise and the phase spectrum of measured skin vibration for tracing certain textures. The results demonstrated that the phase spectrum is crucial to the tactile perception as well as the magnitude spectrum, indicating the possibility of rendering various vibrotactile textures with the usage of the phase spectrum and colored noise.

Keywords: Entertainment and Educational Systems, Assistive Robotics, Integration Platforms
Abstract: Recent advances in sports coaching technology enable feedback via two main approaches: (a) terminal feedback, which provides performance metrics after gameplay, and (b) continuous feedback, which delivers real-time cues through methods like vibrotactile or auditory stimuli. However, most systems focus on individual performance, making them unsuitable for team sports, where coordinated movements and strategies are essential. This paper proposes a framework for real-time, simultaneous guidance in team sports. The system tracks player positions using a single RGB camera and provides continuous vibrotactile feedback, preserving auditory and visual attention for gameplay. The key contributions are: a lightweight player tracking system using YOLOv4 and coordinate transformations, achieving position estimates at ~20 Hz, and a simultaneous vibrotactile guidance system that enables teams to form target formations by following the stimuli. We validate our system through two experiments. First, we assess the tracking accuracy in a virtual environment with four color-tagged moving models, analyzing RMSE across different speeds. Second, we evaluate the effectiveness of vibrotactile guidance by tracking and guiding two subjects through six target formations while measuring performance metrics.

Keywords: Robotics, Assistive Robotics, Human-robot Interaction / Collaboration
Abstract: This paper presents a novel handheld grip-shaped haptic device that effectively integrates vibrotactile and robotic-touch interfaces for turn-by-turn navigation. Inspired by our common expressions used daily for directions like ``turn right at the third corner,'' our system leverages vibration counts to indicate ``how many corners ahead'' and the direction change of a servo motor's horn (robotic-touch) to show ``which way to turn.'' This approach provides a clearer and more situated instruction than existing systems that give only directions and/or subjective distances to the destination or those that just present the entire routes all at once. A systematic user study with 13 participants demonstrated the effectiveness of the proposed device over existing solutions that use either vibration or robotic-touch modalities alone. Specifically, our device allows users to navigate the multi-intersection environment significantly faster than the robotic-touch-only baseline. Subjective evaluations obtained from questionnaires further indicate that our device enhances ``intuitiveness'' and ``efficiency'' over the vibration-only system, and improves ``reliability'' and ``safety'' compared with the touch-only system.

Keywords: Human Factors
Abstract: This paper aims to identify bottlenecks in learning lathe operation and find ways to reduce the workload of the identification, in order to analyze effective and efficient learning processes for lathe operation training for beginner students. This paper considers bottlenecks suitable for practical training and proposes a method for measuring bottlenecks using an eye tracker to reduce the workload required for identification. Based on this method, bottlenecks during student training are identified and discussed.

Keywords: Rehabilitation Systems, Assistive Robotics, Medical Training
Abstract: The objective of this study is to analyze the load on the musculoskeletal system of the lower limbs of small dogs walking on slippery flooring, focusing on differences in the walking patterns of small dogs. In Japan, small dogs such as Poodles, Corgis and Chihuahuas are often kept indoors on slippery flooring. Based on veterinary experience, these small dogs often suffering from lower limb joint diseases, and it is thought that this is caused by stress on the musculoskeletal system of the lower limbs when walking on slippery floors. Therefore, in this study, we investigated the gait of small dogs walking on slippery floors and estimated which muscles are stressed. Furthermore, we measured the gait of small dogs using motion capture and force plate system, and entered the measurement results into a musculoskeletal simulation model of small dogs that we developed. The simulation results identified that the muscles involved in adjusting balance during walking are heavily stressed in the gait of small dogs walking on slippery floors.

Keywords: Assistive Robotics, Medical Devices, Hardware Design
Abstract: With global aging, the incidence of motor dysfunction is increasing. Especially, the upper limb motor dysfunction requires assistance and rehabilitation to involve in daily activities such as eating and brushing one’s teeth. In recent decades, wearable robots have made remarkable progress, with a wide variety of upper limb assistive robots being developed and commercialized. However, it relies on a control system using sensors and software, and there is a possibility that a sudden collision could damage the wearable robots or the users. Therefore, to further improve safety, it is desirable to add mechanical safety devices such as torque limiters. In previous studies, we proposed a new torque limiter with a soft gear transmission mechanism and verified its feasibility. On the other hand, to apply this torque limiter to wearable robots, it is necessary to model the deformation behavior of the soft gear and establish a design methodology. In this study, we proposed a modeling method for soft gears and conducted the evaluation experiments using a finite element analysis and a prototype. The proposed model predicted slip torque values of 0.78 Nm (forward) and 1.54 Nm (reverse), with an error rate of approximately 13% from the experimental values. Wearable robots equipped with the soft gears have the potential to improve safety and be used in daily life.

Keywords: Human-robot Interaction / Collaboration, Assistive Robotics, Medical Devices
Abstract: A method was developed for increasing the accuracy of the continuous wrist joint angle estimations for myoelectric prosthetic hands.The method considers cepstral coefficients, which efficiently represent the frequency spectrum characteristics of electromyography (EMG) signals as features. The root mean square myoelectric power (RMS) is conventionally used for myoelectric prosthetic control; however, the sensitivity of the RMS to angle changes is low, especially at the fine muscle contraction level. The proposed method incorporates low-order cepstral coefficients into the feature vector as well as the RMS myoelectric power. This approach increases the angle estimation accuracy through capturing the changes in the spectral shape related to the firing patterns of the motor units. The results of an angle estimation experiment involving wrist dorsiflexion and palmar flexion movements showed that adding a few cepstral coefficients substantially increased the estimation accuracy, particularly when estimating the small flexion angle range. The cepstral coefficient can be used to effectively estimate joint angles from EMG signals, contributing to the development of smoother and more intuitive myoelectric prosthetic control.

Keywords: Robotics, Automation, Machine Learning
Abstract: Reinforcement Learning (RL) has shown great promise for efficiently learning force control policies in peg-in-hole tasks. However, robots often face difficulties due to visual occlusions by the gripper and uncertainties in the initial grasping pose of the peg. In this paper, we propose a robust tactile insertion policy that can align the tilted peg with the hole using active inference, without the need for extensive training on large datasets. Our approach employs a dual-policy architecture: one policy focuses on insertion, integrating force control and RL to guide the object into the hole, while the other policy performs active inference based on tactile feedback to align the tilted peg with the hole. In real-world experiments, our dual-policy architecture achieved 90% success rate into a hole with a clearance of less than 0.1 mm, significantly outperforming previous methods that lack tactile sensory feedback (5%). To assess the generalizability of our alignment policy, we conducted experiments with five different pegs, demonstrating its effective adaptation to multiple objects.

Keywords: Robotics, Machine Learning
Abstract: Recent advances in AI technologies have enabled the advanced design of complex control policies. In contrast, focusing on the body, many robots still employ simple bodies that can limit adaptability to environments. Studies in embodied robotics have shown that well-designed bodies can partially replace the role of control and computation with physical body–environment interactions, yet such designs still depend heavily on expert intuition. There is a need for a systematic theoretical framework for body design, as well as a method for joint optimization of the body and controller. To address this, we introduce the Embodied Perceptron, a theoretical framework that unifies neural networks and physical body systems. In this view, the body itself acts as a perceptron: mechanical parameters correspond to weights, and physical nonlinearities play the role of activation functions. By representing physical constraints as weights and nonlinear properties as activation functions, a physical body can be modeled in neural-network form. The system representation enables us to explicitly and theoretically explain that the body can substitute for part of the neural control. As an application, we co-optimize control policy and muscle configuration in a musculoskeletal robot and show that the resulting embodied intelligence can provide inherent stability, improve learning efficiency, and drastically reduce model size—even with a single-neuron controller. The results bridge the informational and physical worlds and provide a pathway toward understanding and systematic design of embodied AI systems.

Keywords: Plant Engineering, Machine Learning, Automation
Abstract: We propose spectral-to-spatial distillation, a novel denoising framework for real-time anomalous sound detection. While anomalous sound detection is critical for industrial applications, its reliability is often compromised by background noise, leading to frequent false positives. Our framework addresses this problem by distilling the knowledge of a general-purpose spectral filtering network into an environment-specific spatial filtering network. Specifically, we generate distillation targets, which are denoised audio signals, by using a pretrained foundation model and the distillation targets serve as training targets for the spatial filtering network. This distillation process leverages a single reference audio clip of the target sound to automatically generate training targets, thereby eliminating the need for any per-target fine-tuning. To train robustly with distillation targets, we also introduce the Semantic Clarity improvement, a novel quality metric that uses a foundation model to measure the semantic similarity improvement between the distillation targets and the reference audio. Experiments show that our framework achieves best denoising and anomaly detection performance while maintaining real-time processing capabilities, making it a practical solution for noisy industrial environments.

Keywords: Control Technologies, Decision-making systems
Abstract: The use of water refrigerated tooling in a laboratory scale was used to simulate conditions of warm (500°C) stamping of aluminum plates with 0.8 mm in thickness.The work presents load-displacement curves of room temperature and heat treated samples during stamping at 120 and 240 mm/min, effect of the treatment in the anisotropiccharacteristics of the rolled sheet plate and electron microscopy analysis along with element detection, as this alloy contains small amounts of silicon. The processes resembles a solution treatment and, thus, also an aging treatment at 120 C for 12 h was carried out in order to understand the application possibilities of the manufacturing process in the automotive or aero-space industry.

Keywords: Control Technologies, Robotics, Automation
Abstract: This study proposes a novel fault-tolerant control strategy for a quadcopter by integrating dynamic equilibrium analysis with nonlinear model predictive control (NMPC). First, we formulate an optimization problem based on dynamic equilibrium analysis. The feasibility of maintaining a constant altitude under a fault condition is determined by solving this problem. This analysis yields an optimal target that minimizes yaw angular velocity, while satisfying physical limitations. The target then serves as an online-updated reference for NMPC. This integrated approach enables a single controller to manage various rotor faults with a single fixed set of controller weights. Furthermore, the controller explicitly handles angular velocity constraints within the sensor limits, enhancing its reliability in a real quadcopter. The effectiveness of this strategy is verified through numerical simulations using a model based on an actual quadcopter, which enables visualization of operational bounds for various fault conditions. Simulations also demonstrate that stable flight can be achieved, while satisfying constraints even after a severe rotor fault.

Keywords: Robotics, Machine Learning, Control Technologies
Abstract: Bridging the sim-to-real gap remains a fundamental challenge in robotics, as accurate dynamic parameter estimation is essential for reliable model-based control, realistic simulation, and safe deployment of manipulators. Traditional analytical approaches often fall short when faced with complex robot structures and interactions. Data-driven methods offer a promising alternative, yet conventional neural networks such as recurrent models struggle to capture long-range dependencies critical for accurate estimation. In this study, we propose a Transformer-based approach for dynamic parameter estimation, supported by an automated pipeline that generates diverse robot models and enriched trajectory data using Jacobian- derived features. The dataset consists of 8,192 robots with varied inertial and frictional properties. Leveraging attention mechanisms, our model effectively captures both temporal and spatial dependencies. Experimental results highlight the influence of sequence length, sampling rate, and architecture, with the best configuration (sequence length 64, 64 Hz, four layers, 32 heads) achieving a validation R2 of 0.8633. Mass and inertia are estimated with near-perfect accuracy, Coulomb friction with moderate-to-high accuracy, while Damping coefficient and distal link center-of-mass remain more challenging. These results demonstrate that combining Transformers with auto- mated dataset generation and kinematic enrichment enables scalable, accurate dynamic parameter estimation, contributing to improved sim-to-real transfer in robotic systems.

Keywords: Machine Learning, Robotics, Automation
Abstract: The performance of robotic imitation learning (IL) largely depends on the quality of human demonstrations. To address this challenge, we present AIREC-Basic, a leader–follower teleoperation system equipped with a dual-arm mobile manipulator that enables efficient data collection. The system employs 8-DoF redundant arms to realize diverse task postures, but such redundancy can reduce the consistency of demonstrations. To overcome this issue, we propose a novel control strategy, Soft Homing Control (SHC), which mitigates redundancy while preserving intuitive operator control, thereby improving dataset consistency. We validate our approach on three household tasks using state-of-the-art IL algorithms (ACT, Diffusion Policy, and HSARNN). Experimental results show that SHC significantly reduces joint trajectory variance and improves task success rates, particularly in scenarios with strong trajectory constraints and frequent contacts.

Keywords: Robotics, Decision-making systems, Integration Platforms
Abstract: Causal Bayesian Networks (CBNs) in robotics are often learned in simulation due to the considerable amount of data required for training. However, discrepancies between simulation and the physical world can cause the learned causal relations to fail in real-world scenarios. Thus, the sim-to-real evaluation is a critical step to deploy a simulation-learned CBN in the real-world. The main challenges in this process are the lack of real-robot evaluation datasets that capture the complexity, noise, and variability of physical environments, which are missing in simulation. In this paper, we propose a set of task-agnostic guidelines for real-robot data collection to evaluate Causal Bayesian Networks (CBNs). The guidelines are generalizable and can be applied to collect real-robot datasets across different robot tasks and platforms. To demonstrate this, we apply them to a robotic platform performing one concrete task, e.g., the robot TIAGo performing a two-cube stacking task, and we collect the real-robot dataset from 100 trials. As a case study, we demonstrate how the dataset can be used to evaluate a simulation-trained CBN on real-robot executions, reporting 10% accuracy drop from sim-to-real transfer. We present this as a first step towards standardized and quantifiable sim-to-real evaluation for CBNs.

Keywords: Robotics, Automation, Integration Platforms
Abstract: The integration of automated systems for the manipulation of deformable objects in manufacturing remains a time-consuming process. This challenge arises primarily from the absence of standardized grasping parameters applicable to the full range of manufactured products and materials. This paper presents a method for automating the study of the lifting parameters of deformable objects using a robotic manipulator with a pneumatic grasping system. A detailed description of the design, implementation, and evaluation of a custom pneumatic gripper integrated with a robot arm is provided. Through experiments utilizing various gripping surfaces and materials, the influence of surface patterns, material properties, and pneumatic pressures on lifting performance was investigated. The results demonstrate significant correlations between material type, surface design, and supply pressure in the context of gripping porous objects. The proposed method enables rapid characterization of the interaction between materials and pneumatic grasping systems. This approach facilitates the integration of pneumatic gripping systems into fully automated manufacturing lines handling deformable objects.

Keywords: Robotics, Integration Platforms, Mechatronics Systems
Abstract: The disassembly of small household appliances poses significant challenges due to their complex and curved geometries, which render traditional rigid fixtures inadequate. In this paper, we propose a modular vacuum-based fixturing system that leverages commercially available balloon-type soft grippers to conform to arbitrarily shaped surfaces and provide stable support during screw-removal tasks. To enable a reliable deployment of the system, we develop a stability-aware planning framework that samples the bottom surface of the target object, filters candidate contact points based on geometric continuity, and evaluates support configurations using convex hull-based static stability criteria. We compare the quality of object placement under different numbers and configurations of balloon hands. In addition, real-world experiments were conducted to compare the success rates of traditional rigid fixtures with our proposed system. The results demonstrate that our method consistently achieves higher success rates and superior placement stability during screw removal tasks.

Keywords: Hardware Design, Robotics, Mechatronics Systems
Abstract: Multiple modular robots use attachment/detachment couplings to adapt to diverse tasks and environments. Sharing the same physical components across different robots and applications requires couplings that (1) have low backlash, (2) are low cost, and (3) allow easy attachment and detachment. However, the presence of multiple couplings on a robot makes it challenging to maintain reliable mechanical and electrical connections. We propose a 3D-printable screw-type electromechanical coupling (``Screw Coupling''), demonstrate its integration into heterogeneous components, and present applications on multiple robots. We also introduce a system that continuously applies loads and moments to the coupling during operation, enabling real-time evaluation and ensuring long-term reliability.

Keywords: Robotics, Software Design, Hardware Design
Abstract: This study presents a system integration approach for planning schedules, sequences, tasks, and motions for reconfigurable robots to automatically disassemble constrained structures in a non-destructive manner. Such systems must adapt their configuration and coordination to the target structure, but the large and complex search space makes them prone to local optima. To address this, we integrate multiple robot arms equipped with different types of tools, together with a rotational device, into a reconfigurable setup. This flexible system is based on a hierarchical optimization method that generates plans meeting multiple preferred conditions under mandatory requirements within a realistic timeframe. The approach employs two many-objective genetic algorithms for sequence and task planning with motion evaluations, followed by constraint programming for scheduling. Because sequence planning has a much larger search space, we introduce a chromosome initialization method tailored to constrained structures to mitigate the risk of local optima. Simulation results demonstrate that the proposed method effectively solves complex problems in reconfigurable robotic disassembly.

Keywords: Software Design, Hardware Design, Integration Platforms
Abstract: Modular reconfigurable robots offer a high degree of flexibility through dynamic hardware morphology changes, but their potential has been constrained by software limitations, particularly the static design philosophy of URDF, the de facto standard format. This paper identifies three fundamental research gaps that hinder the development of reconfigurable robots: (1) the disconnect between programmatic ``decomposition'' of models and runtime ``composition,'' (2) the lack of standard mechanisms for updating kinematic topology at runtime, and (3) the limited semantic expressiveness of the URDF format itself. To address these challenges, we propose scikit-robot, a unified framework that integrates multiple components for reconfigurable robotics. The framework establishes a design-to-deployment workflow through: (1) an integrated URDF toolchain handling decomposition, reconfiguration, and mesh optimization, (2) a dynamic root transformation algorithm enabling flexible physical connections, and (3) a comprehensive hash-based model management system providing asset identity verification and supporting simulation-to-hardware integration. Experimental validation on multiple real modular robot systems demonstrates stable operation and direct model transfer from simulation to physical hardware, confirming the effectiveness of our integrated workflow approach. scikit-robot provides researchers and developers with a practical foundation for dynamic and adaptive robotic systems.

Keywords: Human-robot Interaction / Collaboration, Robotics, Hardware Design
Abstract: Robots are increasingly being used to replace humans in performing dangerous tasks, and aerial robots are particularly popular for work at heights. Both underactuated and fully actuated multirotors have mainly been used, but the range of tasks they can perform is limited due to their low degree of operational freedom. Articulated aerial robots are attracting attention as one solution to this problem. Due to the complexity and numerous disturbances involved in high-altitude work, teleoperation by humans is still necessary, and research is ongoing. Most of these studies focus on conventional multi-rotors, and it is difficult to intuitively control articulated aerial robots. Therefore, in this study, we propose a new teleoperation framework that allows operators to intuitively control all degrees of freedom of an articulated aerial robot simultaneously. The proposed framework consists of a floating-based device that acquires operating inputs by utilizing the freedom of movement of both human hands, and a system that generates commands to the robot from those inputs. The effectiveness of the proposed framework was verified through operating experiments using a real robot and wall cleaning task.

Keywords: Virtual / Augmented / Mixed reality, Human-robot Interaction / Collaboration, Robotics
Abstract: Effective teleoperation involves immersive and responsive visual feedback to support depth perception and spatial understanding to achieve precise control. Standard camera views naturally constrain the operator’s Field of View (FoV) of the remote scene, especially in cluttered or dynamic scenarios. We present a real-time RGB-D fusion system that expands the operator’s FoV by employing immersive 3D reconstruction. Our system incorporates the Azure Kinect RGB-D sensor into Unreal Engine using the Robot Operating System (ROS) communication, rendering live sensor information onto a spherical mesh. This allows for smooth, wide-FoV rendering of the scene with greater peripheral context and depth continuity. In contrast to planar or depth-free systems, the proposed method is enhanced by live depth deformation for more interactive teleoperation, leading to better scene understanding. This architecture lays the basis for flexible, high-fidelity remote interaction for robotics applications. All our developments and implementations are publicly available at https://github.com/isri-aist/rgbd-immersion

Keywords: Robotics, Mechatronics Systems, Human-robot Interaction / Collaboration
Abstract: This paper presents the design and evaluation of a web-based multi-robot teleoperation platform that enables remote access to heterogeneous robots for education and research. In contrast to most existing remote laboratories, which are limited to single-robot control or simulation environments, the proposed system integrates real-time video streaming, bidirectional control, and user management within a modular client–server web architecture. The platform was deployed with three distinct robotic systems—an industrial manipulator, a humanoid, and a quadruped—each hosted on independent servers with public IP access. A cross-border user study with Technological University Dublin was conducted, where participants performed guided teleoperation tasks and completed structured surveys. The results demonstrated high usability, low-latency performance, and positive educational impact across diverse robot morphologies. These findings suggest that the proposed platform provides a scalable and accessible solution for remote robotics education and international collaboration, bridging the gap between simulation-based training and real-world robot interaction.

Keywords: Robotics, Human-robot Interaction / Collaboration, Virtual / Augmented / Mixed reality
Abstract: Teleoperation enables human operators to control robots in remote environments, yet its integration with physical human–robot interaction (pHRI) for handling cumbersome objects remains limited. This work introduces a mixed reality (MR) teleoperation interface using an object-based control strategy, enabling the remote operator to manipulate a virtual point on the object rather than the robot's tool center point. The approach was evaluated in collaborative manipulation of an object exceeding the robot's payload capacity and compared with conventional control in a user study. Object-based control supported accurate, differentiated rotational behaviors and was rated more favorably in usability while maintaining low workload, indicating its potential for precise, intuitive manipulation of heavy or unwieldy objects.

Keywords: Human-robot Interaction / Collaboration, Robotics, Telecommunication Systems
Abstract: As robots are deployed into new domains, teleoperation with a human-in-the-loop remains an important method for training new skills. Direct control of a remote robot over a network demands high concentration of the human teleoperator and remains challenging due to unavoidable network delays and congestion that can degrade performance. In this work, we demonstrate a shared control solution using well established virtual fixtures (VFs) as a teleoperator assistance tool and propose a novel VF generation pipeline leveraging mate constraints in available CAD data of the manipulated objects. This approach simplifies the definition of VFs by taking parameters from relationships already defined within the CAD data. Using an industry-sponsored task board and a bilateral leader-follower hand-guided robot scenario, we demonstrate how a set of VFs can be constructed and activated in a series to support teleoperators with a multi-step manipulation task including the pressing of buttons, peg-in-hole with the picking and inserting of a Multimeter Probe Plug, and a novel VF task of opening of a hinged door. We present data from a pilot user study with eight teleoperators and 67 trial attempts with two scenarios (with and without VFs enabled) across three test conditions of round-trip network delays of 0 ms, 100 ms, and 250 ms. We found teleoperators had an increased task success rate, lowered the total travel distance, and overall reduced task execution time, in the best case of 0 ms delay by 26 seconds, or 21%, when VFs were enabled versus when they were not. Performance was maintained or improved at higher network delays.

Keywords: Mechatronics Systems, Human-robot Interaction / Collaboration, Robotics
Abstract: This study presents a vision-based tactile sensor (VBTS) design inspired by the multilayered anatomy of the human fingertip, incorporating a rigid “nail” structure to enhance contact area measurement and fine texture detection. Five flexible skin materials were mechanically evaluated, and optimal candidates were selected based on tensile strength and transparency. VBTS prototypes with (w-VBTS) and without (wo-VBTS) the nail structure were fabricated and tested. Contact area experiments on flat surfaces revealed that the w-VBTS generally produced larger contact areas, particularly with Ecoflex 00-10, which exhibited skin-like conformability. Fine step detection tests using a Solaris–Slacker skin demonstrated that the w-VBTS accurately detected periodic surface patterns down to 1 mm pitch, with a maximum distance error of 0.69 mm, while the wo-VBTS failed to detect the smallest pitch under comparable loading. Power spectral density analysis confirmed that the w-VBTS could sense minute surface variations at low normal forces, indicating reduced deformation loss due to the nail structure. These results validate that integrating a fingernail-inspired rigid layer improves VBTS performance in contact characterization and high-resolution texture sensing.

The proposed biomimetic approach has potential to enhance robotic manipulation capabilities, particularly in tasks requiring stable grasping and fine surface discrimination.

Keywords: Machine Learning, Robotics, Human-robot Interaction / Collaboration
Abstract: Tactile sensors embedded in robotic systems are vulnerable to hardware-level faults such as short-circuits (saturated high-value output, e.g., 5 N) and open-circuits (zero-value output), which may arise from realistic physical failures such as broken signal lines or internal disconnections. In this work, I explicitly model such hardware-induced faults based on the physical wiring structure of tactile sensor arrays and investigate their impact on object recognition accuracy. I first perform a sensitivity analysis to identify sensor regions where abnormal values (e.g., 0 N or 5 N) have particularly strong influence on recognition performance. Building on these insights, I design a robust recognition framework using a Transformer-based recognition model, and evaluate three strategies: (1) fault-aware training with simulated failure patterns, (2) selective removal of known faulty sensor data, and (3) online correction of abnormal values by converting them into statistically neutral values with minimal impact on the learned force distribution. The experiments show that while fault injection during training provides some robustness, it is difficult to generalize across all possible failure patterns. In contrast, fault detection and correction at inference time significantly restore recognition accuracy under severe fault conditions. These results highlight the necessity of integrating both hardware fault modeling and adaptive inference mechanisms for reliable tactile perception in real-world robotic systems.

Keywords: Robotics, Human-robot Interaction / Collaboration, Assistive Robotics
Abstract: Radial pulse measurement is an important physiological assessment method in healthcare and wellness monitoring. While most existing pulse measuring devices are dedicated instruments requiring precise placement, this work explores a human-like approach using a general-purpose humanoid robot. In this work, we present a proof-of-concept study on radial pulse measurement using the AIREC humanoid robot equipped with a 3-axis 8×6 matrix uSkin tactile sensor embedded in its palm. A dual-arm motion strategy allows one hand to support the subject’s wrist while the other applies gentle pressure for stable human-robot contact. Contact stability is evaluated by both normal and shear measurements. Then, the tactile signals from the array are scanned to identify the optimal sensing location, and bandpass filtering is applied to extract the pulse waveform. Preliminary results show accurate pulse rate estimation compared with a commercial pulse oximeter, demonstrating the feasibility of a human-like bio-information measurement in humanoid robots.

Keywords: Mechatronics Systems, Human Factors
Abstract: This study proposes a wearable fingertip force estimation approach that employs a single three-axis force sensor mounted on the side of the fingernail, enabling measurements under natural contact conditions without obscuring the fingerpad. Fingertip deformation during contact is captured by the sensor, and calibration experiments are performed using a table-mounted force sensor to relate the measured lateral deformation to actual normal and tangential forces. Two modeling approaches—a 6th-order multi-input multi-output transfer function model and an 8th-order state-space model—are identified from the calibration data. The transfer function model achieves high offline accuracy (R² ≈ 0.9, RMSE ≤ 7%FS), while the state-space model offers smoother, low-latency output suitable for real-time feedback. These results clarify the relationship between fingertip deformation and applied forces, providing a scientific basis for improving skill transfer, training protocols, and haptic simulation in precision manual tasks.

Keywords: Robotics, Machine Learning
Abstract: This paper presents the application of Hypergraph Convolutional Networks (HGCNs) for tactile spatial processing in multifingered robotic hands. Building on prior work employing Graph Convolutional Networks (GCNs) for modeling irregular sensor layouts, we address the architectural complexity introduced by topological segmentation approaches through the use of hypergraphs, which naturally capture higher-order relationships among tactile sensors. We evaluate HGCNs, standard GCNs, and feedforward neural networks (FNNs) on object geometric property recognition using eight objects and multimodal input (touch states, taxel coordinates, and joint angles). Our results demonstrate that HGCNs achieve high recognition rates of 96.61% while reducing model redundancy, and that hyperedge structure and types of hypergraph adjacencies significantly influence model performance. These findings suggest HGCNs offer scalable and effective tactile data processing.

Keywords: Robotics
Abstract: Although considerable research in humanoid robotics focuses on developing general manipulation capabilities, whole-body manipulation—leveraging all body links for manipulation tasks—has received limited attention. To explore the potential of whole-body manipulation, we propose a framework enabling humanoid robots to carry multiple objects simultaneously using their entire body. We introduce two types of body holding primitives: under-arm holding and wrist-torso holding, significantly expanding humanoid robot manipulation capabilities. To achieve the task of multi-objects loco-manipulation, we leverage compliant control to effectively stabilizes object holding using body links. Furthermore, We integrate existing learning-based whole-body locomotion controllers with a dynamic feedforward PID controller to generate accurate upper-body movements for manipulation while maintaining stable locomotion.

Keywords: Robotics, Hardware Design
Abstract: Small-scale planetary exploration missions have been attracting significant attention in recent years. To maximize the scientific return from these missions, compact rovers are required. Hopping is a promising locomotion strategy for achieving both high traversability and compactness. However, hopping on soft granular terrain like regolith is challenging due to significant energy dissipation into the medium. The dynamics of sand flow during hopping are not yet well understood, especially for diagonal trajectories. This study investigates sand behavior during diagonal hopping using DEM simulations. We focused on how the added-mass effect and the effective friction coefficient vary with the hopping angle. Our results revealed a fundamental trade-off: shallower hopping angles decrease the inertial resistance from added mass, yet simultaneously increase the effective friction, which enhances propulsive force. In addition, we applied observed friction behavior to hopping simulations and confirmed the effectiveness. These insights provide a physical basis for the development of more efficient hopping mechanisms for future planetary rovers.

Keywords: Robotics, Machine Learning
Abstract: Deep Reinforcement Learning (RL) is now commonly used for controlling legged robots. Several recent studies have demonstrated impressive results in solving increasingly complex robotic tasks such as navigation in unstructured environments or loco-manipulation. However, this complexity often comes with intricate learning setups requiring tedious reward shaping and features to help convergence. In this work, we tackle these issues and achieve loco-manipulation with a humanoid robot using a RL algorithm that enforces constraints through stochastic terminations during policy learning. We keep the number of rewards low by reformulating them as constraints when they can be intuitively expressed that way. Moreover, we study the relevance of various learning features encountered in the literature and show that providing observations without noise or privileged information to the critic are two straightforward ways to boost locomotion performances on rough terrains. We also demonstrate that the proposed minimalist architecture is not limited to pure locomotion but extends to a loco-manipulation task involving upper limbs. Videos are available at https://humanoid-cat.github.io

Keywords: Robotics
Abstract: Cockpit-style exoskeleton systems are a promising approach for the whole-body teleoperation of humanoid robots, yet they face two major challenges. The first is the difficulty of intuitive torso posture control, arising from the operator's seated and constrained posture. The second is the lack of transparent feedback for kinematic singularities, which do not manifest as physical forces. This paper proposes two novel methods to address these respective challenges: 1) an indirect torso control interface that converts the operator's foot-actuated torque into the robot's torso angular velocity, and 2) a haptic feedback scheme that guides the operator's arm away from singularities based on the gradient of the arm's manipulability measure. We implemented these methods into a teleoperation system and successfully demonstrated tasks such as picking up an object from the floor and an operator's active avoidance of singular postures.

Keywords: Robotics, Hardware Design, Mechatronics Systems
Abstract: Humanoid robots are inherently unstable, making fall management critical, especially for hardware-based reinforcement learning (RL), where falls frequently occur. This paper introduces a lantern-shaped mechanical cover designed for a kid-sized humanoid robot to mitigate damage during falls and support autonomous recovery. A multimodal fall detection method integrating inertial, proprioceptive, and acoustic sensors was implemented alongside an improved stance phase detection algorithm that eliminates reliance on heuristic thresholds. Hardware experiments on the Hybrid Leg biped robot demonstrated improved walking robustness and revealed a 57.4% success rate for autonomous recovery after induced falls. Results indicated that perturbations in vertical (z) and positive forward (x) foot trajectories posed the greatest challenges to successful recovery.

Keywords: Robotics, Rehabilitation Systems, Assistive Robotics
Abstract: Due to their inherent compliance and safe interaction with their environment, soft robots play a significant role in wrist rehabilitation. However, current designs lack bending in multiple directions and have limited payload capacity. This article aims to explore various configurations of pneumatic soft fingers, specifically targeting the different bending directions of the wrist joint. A pair of parallel-connected soft actuators is mounted on the index and ring fingers to achieve extension and flexion movements by simultaneously pressurizing both actuators. Pronation and supination are achieved by activating two diagonally positioned actuators, while radial and ulnar motions are accomplished through two side-connected actuators attached to the little finger and thumb. A nonlinear static analysis based on the Yeoh model is conducted to validate the design while concerning its bending and deflection. A series of experiments are carried out to verify the bending and payload capacity of the pneumatic soft finger. The results show that it achieves a bending angle of 270◦, while carrying a payload of 200 g. Moreover, the effectiveness of soft design is validated by bending a 660 g metal frame, which is designed to mimic a human hand, in multiple directions. Finally, four soft fingers were employed to successfully perform flexion deviation on a real human hand. This paper represents the initial phase of utilizing pneumatic soft fingers to achieve multiple wrist bending movements.

Keywords: Robotics, Medical Devices, Mechatronics Systems
Abstract: Endoscopic-based double balloon represents an advanced technique for diagnosing and treating bowel cancer. However, existing designs face challenges including complexity and high cost due to the use of hybrid actuation, leading to prolonged procedure duration and patients’ pain. This article introduces a novel design of a multi-section soft robot actuated by cables. It is composed of two sections, each capable of bending in two planes. Additionally, the distal section can perform compression and extension. Four tendons separated by 90◦ are used to control the bending of each section. Each pair of parallel cables is attached to a DC motor through a double groove pulley. Moreover, four independent cables are utilized to compress and extend the distal section. Finite element analysis is conducted to evaluate the performance of the prototype concerning its bending and displacement. Experimental validation is carried out to investigate the capability of the design in terms of bending and payload capacity. The results show that the robot can bend by more than 270◦ under a payload of 50 g. This paper represents the first phase for developing a soft colonoscope and relaxing the complexity of current designs.

Keywords: Assistive Robotics, Medical Training, Human-robot Interaction / Collaboration
Abstract: The constraints of cable-driven actuation have largely limited mediolateral (ML) assistance from soft exosuits to hip abduction. To address this limitation, we developed a soft pneumatic exosuit actuated by pneumatic artificial muscles (PAMs) capable of providing bidirectional assistance to both hip abduction and adduction to guide the wearer’s step width. Given the challenges in accurately modeling PAM-driven systems, we propose a Human-in-the-Loop (HIL) approach that treats the human-exosuit system as a black box and uses Bayesian optimization to tune control parameters. In walking experiments with two healthy participants, the proposed HIL framework successfully identified participant-specific PI gains during the parameter tuning session. In the subsequent validation, the controller based on the optimized gains reduced the mean step-width RMSE by 43.4 % for Participant 1 and 26.8 % for Participant 2 compared with the condition without assistance. The optimized gains also provided an additional improvement of 3.76 % for Participant 1 and 1.07 % for Participant 2 relative to the controller with initial gains. These results indicate that the proposed approach offers a practical method for personalizing assistance in soft pneumatic exosuits and contributes to the development of devices that enhance ML balance.

Keywords: Rehabilitation Systems, Assistive Robotics, Medical Devices
Abstract: This study evaluates a soft robotic wearable device, equipped with a Hyper-extension Pneumatic Actuator (HPA), to prevent deep vein thrombosis (DVT). We investigate the hemodynamic efficacy of a novel "Combined" exercise protocol, which strategically integrates the high-load advantage of resistance exercise with the wide range of motion from induced exercise, all delivered by a single soft actuator system.The effects of four distinct exercise modalities on key hemodynamic parameters, including time-averaged maximum blood flow velocity (TAMAX), were systematically evaluated in healthy participants. Results demonstrated that all active exercise modalities showed a strong tendency to augment blood flow velocity compared to rest (effect size r>0.8). Notably, the "Combined" protocol yielded a statistically significant increase in TAMAX compared to exercise without the device (p < .05).This enhancement is attributed to the protocol's ability to elicit high-quality, high-tension muscle contractions by balancing exercise load with a sufficient range of motion, thereby optimizing the muscle pump mechanism. These findings provide engineering guidelines for developing effective DVT prevention strategies, demonstrating that the intelligent integration of a control protocol—not just the hardware alone—is key to maximizing hemodynamic efficacy.

Keywords: Assistive Robotics, Machine Learning, Medical Devices
Abstract: In this study, we propose a method for autonomously operating a surgical robot by controlling visual attention using the robot's own state in deep predictive learning. The proposed method, named TAIRNN (Task-Attentive Informed Recurrent Neural Network) uses a state-conditioned Query to retrieve visual Keys to deep predictive model. Experimental results of point-to-point movement showed that this method can reach the final target with improved accuracy and in fewer steps compared to conventional methods that rely solely on image information. The results demonstrate that an approach that incorporates a robot's self-state awareness into its visual attention mechanism is effective in suppressing task-irrelevant visual noise and improving control stability.

Keywords: Control Technologies, Integration Platforms, Mechatronics Systems
Abstract: Model Predictive Control (MPC) is increasingly being adopted across various industrial sectors due to its capability to manage constraints and nonlinearities effectively. This paper presents a novel approach for dynamically updating the MPC model and optimizing the prediction horizon during operation, ensuring both real-time performance and robustness. The method seamlessly integrates hardware and software, allowing for continuous operation without any downtime. Experimental results indicate that the proposed approach sustains optimal control performance, even in the face of changes in system characteristics caused by aging and wear.

Keywords: Control Technologies, Network Systems
Abstract: Event-triggered control is a method in which the measured values are sent from sensors to controllers only when an event-triggering condition is satisfied. In the case where the controller is given in advance, it is necessary to design event-triggering mechanisms (ETMs). In this paper, a design method of decentralized event-triggering mechanisms (DETMs) is proposed based on non-monotonic Lyapunov functions. Decentralized ETMs efficiently work for a sensor network in which multiple sensors are located in a distributed way. Using non-monotonic Lyapunov functions, it is expected that the number of times that the event occurs is decreased. The design problem of DETMs is reduced to an LMI (linear matrix inequality) feasibility problem. The effectiveness of the proposed method is presented by a numerical example.

Keywords: Robotics, Control Technologies, Software Design
Abstract: In recent years, manipulation tasks in high place by aerial robots have been realized. In particular, rotor distributed manipulators equipped with rotors on each link of articulated structure have attracted attention for their ability to perform high DoF tasks through whole-body deformation. However, in many conventional rotor distributed robots, joint control and thrust control were separated for model simplification, or the dynamics of joint motion were ignored and robot body was approximated as a rigid multirotor at each control cycle. As a result, they are mainly applied to tasks comprising static motion, and its application to tasks requiring dynamic deformation is limited. Therefore, in this work, we aim to enable dynamic deformation by a rotor distributed manipulator. We clarify the whole-body dynamics model of a rotor distributed manipulator that integrate joint torque and thrust, and develop its control system. In the proposed inverse dynamics method, we formulate a problem based on quadratic programming to minimize the exerted thrust and joint torque while achieving the desired whole-body motion. Then, we implement a control system for a rotor distributed manipulator incorporating the proposed inverse dynamics method. We verify the effectiveness of the proposed method through trajectory generation of actuator inputs and dynamic motion simulation. To the best of our knowledge, this is the first time to establish a control system for dynamic deformation by integrating joint torque and thrust for rotor distributed robots.

Keywords: Robotics, Mechatronics Systems, Hardware Design
Abstract: Deformable aerial robots with articulated structures have attracted increasing attention for their ability to perform complex tasks through in-flight morphological adaptation. However, most existing implementations rely on mechanical actuators, which increase total weight and vulnerability to external impacts. To address these limitations, we propose a deformable quadrotor platform actuated by antagonistic McKibben Pneumatic Actuators (MPAs), which offer lightweight, flexible, and robust actuation. We develop a dynamic model of the quadrotor incorporating joint angles actuated by MPAs, and characterize the relationship between internal pneumatic pressure and joint angle. Based on this model, we construct a prototype platform and evaluate its performance through a series of experiments. We first identify the optimal actuator length by analyzing joint deformation under varying pressure conditions. Next, we demonstrate stable flight during in-air morphing transitions, such as X-type, T-type, and H-type configurations, with position and attitude errors remaining within acceptable ranges. The results confirm that the proposed system enables stable flight using soft pneumatic actuation and paves the way for future aerial manipulation and morphing applications.

Keywords: Robotics, Control Technologies
Abstract: Soft grippers have attracted attention as a safe and adaptable means of grasping fragile or irregularly shaped objects. However, challenges remain in performing advanced tasks such as in-hand manipulation. This study proposes a soft-rigid hybrid two-finger hand that integrates position-controlled motors and flexible links, based on a design concept that combines the mechanical properties of rigid and flexible materials. The proposed hand directly drives the flexible links using position-controlled motors, enabling active control of the interaction between the flexible structure and the environment. Leveraging this feature, a position-based force control strategy is introduced, in which fingertip positions are adjusted to shift the equilibrium of the flexible links, indirectly modulating the contact force through their passive deformation. The effectiveness of the proposed method is validated through simulations of object grasping tasks, demonstrating stable and dexterous in-hand manipulation.

Keywords: Virtual / Augmented / Mixed reality, Human Factors
Abstract: This study proposes a new method for organizing and accessing information in MR environments by anchoring data to parts of the human body (such as the head or hands). The system utilizes Microsoft HoloLens 2 and eye-tracking technology to enable users to intuitively access virtual information through gaze-based operations. This approach introduces a relative spatial coordinate system based on the user's body. User studies were conducted to verify the effectiveness of body-based tagging and information management in MR spaces. The results confirmed that this information access method is usable without issues, and it was revealed that rules exist governing the relationship between the access time and location of body-tagged information. These results suggest that body-based information tagging has the potential to enhance the usability of MR for everyday tasks.

Keywords: Virtual / Augmented / Mixed reality, Assistive Robotics, Rehabilitation Systems
Abstract: In the context of preserving traditional performing arts and facilitating efficient motor skill acquisition, there is a need to develop posture guidance methods. Traditionally, visual imitation and verbal instruction have been the primary means of guidance; however, haptic feedback is considered particularly effective for posture instruction and form correction, as it can directly convey movement nuances that are difficult to communicate through visual or auditory channels alone. Although various haptic information transmission devices have been developed, they often present issues such as discomfort caused by reaction force support structures or restrictions on joint range of motion. To address these limitations, we focused on haptic feedback using air jets. Air jets can present forces in arbitrary directions relative to the body part on which the nozzle is mounted, and the approach can be extended to the entire body, enabling posture guidance even for joints involved in complex movements. In this study, we proposed and developed a haptic feedback device for guiding whole-body posture. We demonstrated that the device can provide effective posture guidance and clarified the influence of the frequency characteristics of the presented haptic feedback. Furthermore, we experimentally demonstrated that the system enables complex, multi-joint, and large-range posture guidance.

Keywords: Human Factors, Intelligent Transportation Systems, Integration Platforms
Abstract: The development of autonomous and respective multi-agent systems creates both technological innovations and significant societal and regulatory challenges. Of particular note is the social acceptance of these systems once they are integrated into their usage environment. This is a critical factor that is typically limited to the end product, while the development process and the underlying technologies remain largely disregarded. However, there are indirect technologies like virtual testing, which are increasingly used in the respective product development, for which their potential effects on the acceptance among stakeholders are overlooked. To address this source of indirect skepticism, this paper introduces the Indirect Technology Acceptance Framework (iTAF) to investigate the practical applicability and acceptance dimensions using the example of virtual validation methods. The framework reveals specific implications for each framework layer: the technological layer requires procedural transparency; the institutional layer necessitates harmonized standards across jurisdictions; the societal layer highlights the influence of communication strategies and safety demonstrations; and the interactive layer emphasizes continuous stakeholder engagement throughout the development process. The model successfully integrates technical, regulatory, and societal dimensions, acknowledging the unique position of indirect users as critical stakeholders in societal legitimation and integration. It also offers a foundation to align development processes and especially system validation to the stakeholders’ needs regarding the trustworthiness of the system of interest (SOI). To illustrate the iTAF, it is applied as a demonstrative example to the social acceptance of virtual validation in the presence of legal uncertainty.

Keywords: Virtual / Augmented / Mixed reality, Entertainment and Educational Systems, Human-robot Interaction / Collaboration
Abstract: In recent years, various haptic devices have been used to reproduce the senses of touch and force when interacting with virtual fluids. This study focuses on reproducing the resistance force experienced during underwater walking movements in a virtual reality (VR) space using a lower-limb force-feedback exoskeleton. To realize disaster simulations with a greater sense of fear and realism for the user, we focused on the necessity of parameter design based on human subjective evaluation when constructing a force-feedback model for the sensation of underwater motion. This study aims to clarify the influence of visual viscosity information on the user's subjective evaluation. Based on the drag model and the added mass force model proposed in previous research, we constructed a composite model in which the contribution ratio of both can be varied. It is conducted an evaluation experiment by combining two types of fluid visuals with different viscosities and four force-feedback conditions. The analysis revealed that visual viscosity information affects the criteria for judging the sensation of viscosity and the way force feedback is perceived. Furthermore, it became clear that as the viscosity changes, the correlation structure among the subjective evaluation items also changes.

Keywords: Machine Learning, Virtual / Augmented / Mixed reality, Medical Devices
Abstract: Accurately recognizing objects is crucial for ensur ing the safety of automobiles and autonomous robots. However, fast movements, such as sprinting, cause motion blur, which complicates detection tasks. In this study, we developed an algorithm to enhance the accuracy with which 3D point clouds recognize individuals in sprinting motion. Data were collected using LiDAR sensors, which are commonly employed in au tonomous driving technology. Our previous research explored methods for removing and correcting motion blur by adjusting the reference values for noise removal based on sprinting speed. Initial results showed that addressing dynamic noise signifi cantly improved recognition accuracy compared to the original data. One challenge encountered during the prior investigation, however, was the excessive correction of lateral and vertical body movements. To tackle this issue, we proposed a refined method targeting motion blur specifically caused by sprinting. This method detects inter-frame blur in three dimensions. By comparing the proposed approach with the prior investigation’s results, we confirmed further improvements in recognition accuracy.

Keywords: Hardware Design, Control Technologies, Human-robot Interaction / Collaboration
Abstract: This paper proposes a thin-film sensor that visualizes heat transfer at contact surfaces through color changes, using thermochromic pigments specifically designed for material identification. By capturing the color change as time-series image data, the spatial distribution of heat transfer across the contact surface can be obtained, enabling differentiation of contact materials. The sensor provides a visual representation of heat flow across the interface between the sensor and the object. Experimental results demonstrate the potential of the proposed sensor for material identification and spatial heat transfer analysis.

Keywords: Hardware Design, Robotics, Mechatronics Systems
Abstract: In recent years, drones have become widely used. Consequently, their potential to cause harm or damage to people, property, or infrastructure has also increased. As drones have become more prevalent, so has the need to counter them and prevent potential misuse. Unlike net or projectile-based systems, a gripper enables precise midair capture and secure handling of the intruder drone without causing uncontrolled descent or collateral damage. This paper presents our iterative development process for such a mechanical gripping system for use in civil counter-UAS applications. After a brief overview of existing gripping systems, we will describe the initial development steps and progress to increasingly complex design solutions. During the development process, three prototype designs were built and evaluated, one after another. Finally, we present our final design, which satisfies all requirements for integration into the UAS and can be used to catch the intruders in mid-flight.

Keywords: Hardware Design, Mechatronics Systems, Human-robot Interaction / Collaboration
Abstract: Timely food rewards are crucial for effective dog training, but providing immediate reinforcement is difficult when the handler cannot be near the dog. Existing systems, such as stationary feeders, restrict the dog’s movement and cannot deliver rewards flexibly. Our previous back-mounted feeder (1.2 kg) enabled successful remote food delivery, but its size and weight meant it could only be used for medium- to large-sized dogs. Remotely controlled food-dispensing devices solve this by allowing spatially unconstrained rewards without limiting the dog’s movement. However, for real-world use, such devices must be compact, lightweight, and unobtrusive to avoid disrupting natural behavior or requiring long habituation periods. Hence, in this study, we present the development and validation of a collar-mounted feeder (hereafter, “collar-feeder”) capable of remotely dispensing food multiple times. The newly developed collar-feeder weighs only 143.6 g and can store up to 20 food items. A compact internal design with fewer electronic parts reduces its size and weight. Experiments showed a 100% feeding success rate among dogs unbothered by the device, while some dogs mistook it for a toy. These findings highlight the potential applicability of the collar-feeder in dog training and behavioral discipline. With further functional improvement, the collar-feeder presented in this paper may gain widespread adoption. For instance, by integrating biometric sensors and a food-tracking system, the device could serve as a valuable tool in households and veterinary clinics for pet health monitoring and management.

Keywords: Hardware Design, Integration Platforms, Robotics
Abstract: Robotic grippers are often designed as custom solutions to meet specific handling requirements, resulting in a large number of component variants and high lifecycle costs. Optimizing grippers for individual usecases will maximize performance. When they share components, cost may be reduced, however, performance may be compromised. This work aims at minimizing the number of components while maintaining a minimum performance. Co-designing grippers that share components across multiple use cases is an expensive combinatorial optimization problem. Unlike conventional optimisation approaches that yield point-based solutions, the proposed method here identifies the intersection of permissible design variables—so-called solution spaces—to define a compact, standardised set of interchangeable modules. This way, gripper families can be identified at significantly reduced numerical cost. The approach is applied to grippers intended for 6 different cylindrical objects of varying mass, geometry, stiffness and friction coefficient. The number of component variants could be reduced from 21 for individual gripperes to 9 for the resulting gripper family.

Keywords: Rescue Systems, Hardware Design
Abstract: Auditory scene analysis using drones, a technology known as "drone audition", is a promising method for locating survivors in disaster areas. This technology requires a structure to mount measurement equipment onto the drone. A previous study proposed a versatile structure designed to be attached to the drone's landing gear and evaluated its applicability based on four criteria: mass limitation, load capacity, vibration characteristics, and ease of adjustment. As for vibration characteristics, it was confined to the theoretical evaluation and was required to be evaluated under actual flight conditions. In this research, we evaluate the vibration characteristics of the structure during flight by comparing acceleration responses at several points on the conventional and proposed structures. The results confirmed that the proposed structure's vibration characteristics are superior to the conventional one's for minimizing the impact on the measurement equipment during flight. Furthermore, it was found that the proposed structure exhibits larger vibrations than the conventional one upon landing. This is acceptable, since acoustic measurements are not performed during landing. Therefore, it became clear that the proposed structure must have sufficient strength to withstand the impact force during landing.

Keywords: Human Factors, Human-robot Interaction / Collaboration, Entertainment and Educational Systems
Abstract: This study examines the “benefit of inconvenience” in tourism through an experiment conducted in Nijo, Kyoto. Participants used two intentionally inconvenient tools: “Blur Navigation,” which gradually obscures walked paths and lacks search functions, and an “Unfriendly Camera,” which requires entering the reason and location for each photo and has a three-second delay in capturing. Emotional changes were measured before and after the tour using PANAS and POMS2, and scenery memory was assessed with a post-tour quiz. Results showed that positive emotions such as “Enthusiastic,” “Excited,” and “Proud,” as well as negative emotions like “Ashamed” and “Nervous,” were significantly correlated with higher memory scores. Questionnaire responses indicated that the obscuring of paths and lack of search functions encouraged active route selection and attention to surroundings, while the camera delay promoted intentional photography but sometimes caused missed opportunities. These findings support that moderate inconvenience can enhance engagement, emotional richness, and memory retention in tourism experiences.

Keywords: Entertainment and Educational Systems, Human Factors
Abstract: In the ever-expanding gaming market, there is constant demand to improve game quality. One of the key elements in enhancing quality is the Non-Player Character (NPC), which is a game character not controlled by the player, and is used for purposes such as balancing the game. Particularly, the quality of NPCs influences the user experience in games in which human players cooperate with NPCs. Artificial intelligence (AI) is an effective tool for improving the quality of NPCs. In recent years, many developers have turned to AI to improve NPCs, aiming to make interactions less scripted and more adaptive.

To investigate the effects of the characteristics of AI-based NPC as a partner, we developed an original game, where the human player connects to an AI-based NPC via a spring and they must work together to collect as many targets as possible. We prepared six types of AI-based NPC with different initiative and performance levels ; three levels of initiative (low, medium, and high) and two levels of performance (low and high). Eleven young people were participated in the experiment and we recorded the subjective rating of the gameplay experience and the performance of the human-NPC team. The results revealed that high-performance NPC had a positive impact on gameplay performance and satisfaction. On the other hand, it was observed that human players were unable to perform to their full potential when AI-based NPCs exercised initiative for long periods. The findings of this study are expected to provide valuable insights into the design of AI-based partner NPCs.

Keywords: Human-robot Interaction / Collaboration, Robotics, Human Factors
Abstract: In this study, we investigated how robot appearance---human-like versus non-human-like---affects personal space during direct robot approaches from eight directions and examined the relationship between personal space and perceived likability. Twenty-three Japanese male participants participated in the experiment. The participants remained stationary while each robot approached from different directions (0^{circ}--315^{circ} in 45^{circ} increments), and the distance at which discomfort was perceived was recorded as the personal space threshold. Likability was assessed via a questionnaire. Across all conditions, a consistent negative correlation was observed between likability and personal space: participants who rated the robot more favorably tended to allow it to approach them more closely. No significant differences in personal space or likability were found between the two appearance conditions. However, the Human-like condition showed an almost circular personal space shape with an approximate radius of 1.35 m, slightly extended in its rear part, whereas the Non-human-like condition tended to have a personal space shape with a larger front and smaller rear parts. These results indicate that likability is closely associated with interpersonal distance regulation and suggest that manipulating perceived likability may serve as an effective strategy for managing personal space in service and interactive robot contexts. These findings may inform robot design approaches aimed at maintaining socially comfortable distances and improving overall user acceptance.

Keywords: Human Factors, Hardware Design, Medical Devices
Abstract: The Thermal Grill Illusion (TGI), a perceptual phenomenon arising from the central integration of spatially interlaced warm and cold stimuli, provides a unique window into the psychological and physiological processes underlying pain perception and central sensitization. While previous studies have primarily explored TGI using large-area thermal arrays, the mechanisms by which TGI can be elicited within a narrow contact area and the relationship between illusory sensations and heat transfer dynamics at the skin interface remain largely unexplored. To address this gap, we developed a compact, smartphone-controlled device capable of delivering precisely regulated, localized TGI stimuli via Peltier elements and copper contact plates. The system incorporates high-accuracy thermistor calibration using the Steinhart–Hart equation and PID-based temperature control within ±1 deg, enabling reproducible manipulation of thermal gradients. Experiments with 17 healthy participants assessed two temperature ranges (20–35 deg, 18–42 deg applied to the palm and upper arm, demonstrating that greater temperature disparities yield stronger perceptions of warmth and burning, and that site-specific differences emerge under smaller disparities. By enabling controlled induction of TGI in confined regions, this device provides a versatile research tool for investigating the interplay between thermal perception, skin heat transfer, and nociceptive processing—offering a foundation for both basic psychophysiological research and future clinical applications.

Keywords: Rehabilitation Systems, Machine Learning, Human Factors
Abstract: Several methods have been developed to capture motion during real-time walking and provide feedback; however, these approaches may not always be suitable for every trainee owing to individual physical differences. In our previous study, we proposed a generative adversarial network (GAN)-based method to generate target gaits for active seniors. The generator incorporated the biomechanical relational network (BMR-Net) to extract inter-variable features of gait. However, the effectiveness of this block in generating individualized target gaits has not yet been verified. In this study, we examined whether the generator incorporating the BMR-Net (BMR-NetGAN) is effective in generating target gaits that reflect individual motion characteristics. In particular, we constructed a 2D transposed convolution GAN, which is generally effective for bidirectional feature extraction in the temporal and variable domains, and a generator without BMR-Net, and compared their results with those of BMR-NetGAN. The results demonstrated that BMR-NetGAN is an effective model for generating ideal gaits that reflect individual motion characteristics, as evidenced by adjustment patterns of lower-limb joint angles on the ZX plane that were not observed in the 2D transposed convolution GAN. Furthermore, an analysis of lower-limb joint motion indicated that BMR-NetGAN may successfully generate target gaits that account for left–right balance in individual participants.

Keywords: Robotics, Control Technologies, Human-robot Interaction / Collaboration
Abstract: This paper introduces a local control framework for mobile manipulators that enables coordinated motion between a nonholonomic base and a robotic arm without relying on any sensors mounted on the platform. The method is based on velocity vector fields defined in the mobile base frame and uses only internal joint measurements from the manipulator. A reduced kinematic model provides the translational end-effector (TEE) position, which drives all base motion decisions. By evaluating distance- and angle-based thresholds, the controller generates smooth linear and angular velocity commands that guide the base through extension, alignment, and retreat behaviors. The system supports real-time human interaction, such as kinesthetic guidance across the full reachable workspace, while respecting the nonholonomic constraints of the platform. Since the base controller is independent of the arm’s control strategy, it is compatible with autonomous planning, teleoperation, or demonstration-based teaching. The proposed control strategy is particularly suited for collaborative scenarios where intuitive human guidance is essential and reliance on global localization is unnecessary. Experimental results with physical kinesthetic guidance demonstrate the platform’s ability to reactively follow end-effector motion, preserving reachability and manipulability across the full workspace.

Keywords: Robotics, Machine Learning, Decision-making systems
Abstract: Learning effective control strategies for deformable object manipulation remains a major challenge in robotics, especially when tasks require fast, coordinated motions between multiple manipulators. In this work we address the problem of dynamic bimanual cloth manipulation, where two robotic arms must coordinate fast, fluid motions to place a cloth onto a surface. Our method uses Proximal Policy Optimization (PPO) with Dynamic Movement Primitives (DMPs) as the policy output, enabling smooth and parameterized trajectory generation. We compute rewards such as cloth height, corner alignment, and movement direction only at the end of each training episode, while the robot control continues at a higher frequency, and introduce a probing phase to obtain knowledge about cloth dynamics. We implement our approach in NVIDIA Isaac Sim with realistic cloth dynamics. Experiments show that this setup allows the robots to learn fast, coordinated bimanual cloth placement using only occasional reward feedback.

Keywords: Robotics, Automation, Control Technologies
Abstract: Low-volume, high-mix manufacturing presents unique challenges for visual inspection, where frequent product changes make automation difficult. Manual inspection remains common but is slow and error-prone, while fully automated systems are often too costly for small and medium-sized enterprises (SMEs). We present a method to optimize the sequence of image acquisitions in robotic visual inspection. The problem is formulated as a unidirectional weighted graph and solved using Travelling Salesman Problem (TSP) techniques. Unlike prior work focused on single-camera setups, we address the more complex case of two-camera inspection with larger numbers of inspection points, introducing a geometric grouping strategy that clusters inspection points by planar regions derived from object geometry. This enables efficient parallel use of cameras while maintaining low planning complexity. The proposed framework supports agile reconfiguration of inspection tasks, making it suitable for high-mix industrial environments. In simulation of a real-world scenario, our method reduces inspection cycle times by up to 2.35 times while maintaining near-optimal sequencing, demonstrating its potential to make multi-camera robotic inspection more practical for agile manufacturing.

Keywords: Robotics, Human-robot Interaction / Collaboration, Machine Learning
Abstract: Autonomous grasp planning remains a key challenge in robotic manipulation, particularly in unstructured environments where object types, poses, and arrangements vary. This work presents a data-driven grasp planning method for a robotic manipulator tasked with clearing a table containing diverse objects. The method encodes human-demonstrated grasping strategies by representing Cartesian trajectories with Dynamic Movement Primitives (DMPs), whose parameters are predicted by a neural network from grasp-specific inputs. A second neural network estimates feasible grasp poses based on the object pose estimate data, which is used as the new goal parameter for the grasp trajectory generation.

To reduce demonstration effort, synthetic datasets are generated via data augmentation of the recorded trajectories. The approach is implemented on a real system and evaluated on four objects with varying shapes and sizes. The experiments show a high success rate for grasping as well as the ability to incorporate new objects into the system through minimal additional effort.

Keywords: Robotics, Integration Platforms, Virtual / Augmented / Mixed reality
Abstract: This paper presents an integrated system that combines robotics, depth vision, and digital twin technologies to automate and enhance the fabrication of structural wooden panels and their installation on-site by human workers. The methodology for enabling better and faster panel production and on-site installation integrates AI and Robotics into BIM (Building-Information-Model), allowing the entire build process to be simulated and optimized in a digital environment before deployment. The robotic system for panel assembly and the computer vision system work hand in hand with the robots, making real-time decisions, detecting issues, and ensuring consistent quality. The interface to a digital twin enables on-site crew guidance for seamless panel integration into new homes. The system is evaluated on more than 50 real-world panels and demonstrates improvements in installation efficiency by reducing the time needed to frame the house by 70%, decreasing the cost by approximately 30%, and improving quality traceability, compared to conventional building methods. This work exemplifies cognitive collaboration between robots and humans in modern manufacturing environments.