Human-robot interaction involves unavoidable collision risks, especially in uncontrolled impacts where forces rise rapidly. Conventional single-drive actuators (SDAs) exhibit high reflected inertia, limiting their intrinsic safety and motivating actuator designs that can passively reduce impact severity. This paper investigates Dual Motor Actuators (DMAs) as a promising solution for passive impact mitigation. We develop a high-fidelity collision model and a complementary analytical formulation that together capture the dynamics of robot-human collisions, including the influence of actuator inertia distribution. The models are validated through simulations and experiments using high-frequency force measurements at 2 kHz. Experimental and simulation results show that DMAs can passively reduce peak impact forces by up to 34% compared to SDAs, despite identical output torque capability. The analytical model accurately predicts peak forces and time-to-peak across a wide range of conditions, showing that collisions reach their maximum force within 17-79 ms. This narrow response window aligns with the experimentally observed 9 ms detection latency at a 1 kHz sampling rate, underscoring the importance of intrinsic mechanical safety. Together, these results demonstrate that DMAs offer a robust actuator architecture for passive impact mitigation and provide actionable design insights for safer collaborative and humanoid robots.

Even though many occupational exoskeletons exist, their adoption is still low. It is important to study which factors encourage adoption, such that appropriate measures can be taken, either by improving the design of occupational exoskeletons or by taking organizational measures such as setting company specific policies regarding the use of these devices. Although various questionnaires have been used to study the acceptance of occupational exoskeletons, these questionnaires either have a limited scope or are very extensive. The goal of this research was to develop a pragmatic and holistic instrument to assess the acceptance of occupational exoskeletons. The instrument is based on empirical data, survey data of 155 respondents, includes concepts from theory, and it is based on an existing framework of exoskeleton acceptance. Using a systematic approach based on exploratory factor analysis with survey data, 33 items were selected for the instrument representing 24 concepts. The ExoAccept instrument can be used by practitioners working with and/or in companies but also by researchers interested in evaluating subjective factors related to the acceptance of occupational exoskeletons.

E-commerce has grown rapidly, but high return rates pose economic and environmental challenges, with fulfillment errors as a key factor. While augmented reality (AR) has the potential to reduce such errors, the adoption of this technology in SMEs remains limited. With inputs from our recently conducted survey of SME employees across 17 European countries, we developed an AR-based order fulfillment system to investigate the ecological validity of this technology in the warehouse. For validation, we conducted an empirical user acceptance study comparing AR-based and paper-based approaches in a human–robot collaborative order fulfillment system. Results indicate a strong interest in AR, with significant increases in perceived enjoyment and social influence. Positively, AR does not lead to a higher level of anxiety. However, the paper-based method demonstrated higher scores for task completion time and intention to use. Findings highlight the potential of AR in the warehouse, but also the need for improved interfaces and hardware to enable broader adoption.

Monitoring pressure distribution on the physical interface of wearable robots (including the cuff of the exoskeleton, rehabilitation robot, and the socket of the prosthesis) is crucial for ensuring safe and comfortable human-robot interactions. However, the complex contour of customized physical interfaces brings challenges in integrating sensors. To address this issue, we propose a novel method for sensorizing the interfaces based on a 3D-printed Electrical Impedance Tomography (EIT) pressure sensor, which is easy to customize, regardless of the complexity of the spatial working surfaces of the interface. Considering the anisotropic conductivity of 3D-printed parts, we first investigated its properties and examined its impact on EIT imaging using simple shape 2D planar sensors. Then, an adjusted Jacobian matrix was employed in imaging to reduce the impact of anisotropic conductivity on imaging. Finally, an EIT pressure sensor embedded physical interface was customized for the forearm using 3D printing and tested on a rehabilitation cobot. The online experimental results validated that it can effectively monitor the distribution and variation of pressure between the human body and the interface.

Robotic manipulators in contact-rich tasks must satisfy motion objectives while respecting nonlinear dynamics, input and state limits, collision constraints, and admissible interaction bounds. Existing Explicit Reference Governors (ERGs) provide lightweight constraint enforcement, but most robotic ERG formulations remain motion-centric and do not explicitly govern end-effector interaction constraints. This paper proposes a Force-Regulated Explicit Reference Governor (F-ERG) that combines a pre-stabilizing force-aware controller, a momentum-based wrench observer, and an outer governor that filters a motion–wrench reference online. The governed reference contains a task-dependent wrench channel constructed from the estimated interaction wrench for External-Force Compensation or from a desired contact-force command for Interaction-Force Regulation. A Dynamic Safety Margin propagates bounds on interaction-induced configuration and velocity deviations and uses the resulting configuration bound to tighten the joint-position and collision-clearance constraints. The remaining motion and actuation constraints are evaluated along the nominal prediction, while the interaction-wrench margin is evaluated from the current observer estimate. Simulations show that the tightened F-ERG preserves the prescribed collision-clearance reserve and achieves a median high-level runtime of 3.64 ms, compared with 8.94 ms for the considered safe MPC baseline. In real-robot experiments, the tightened formulation maintained positive clearance under a suddenly applied 24.5 N load, whereas the nominal formulation reached the collision boundary. Across valid trials per condition, an admissible 10 N command produced a quasi-static measured force of 9.21±0.27 N, while an infeasible 15 N command under a 10 N limit was restricted to 7.95±0.30 N.

Robotic grippers are increasingly deployed across industrial, collaborative, and aerial platforms, where each embodiment imposes distinct mechanical, energetic, and operational constraints. Established YCB and NIST benchmarks quantify grasp success, force, or timing on a single platform, but do not evaluate cross-embodiment transferability or energy-aware performance - essential for modern mobile and aerial manipulation. This letter introduces the Cross-Embodiment Gripper Benchmark (CEGB), a reproducible benchmarking suite extending YCB and selected NIST metrics with three additional components: a transfer-time benchmark measuring embodiment exchange effort, an energy-consumption benchmark evaluating grasping and holding efficiency, and an intent-specific ideal payload assessment. Together, these metrics characterize grasp performance and cross-platform suitability. CEGB is validated on two mechanically distinct grippers. The experimental evaluation quantifies embodiment-dependent differences in transfer time, energetic efficiency, and operational capability under unified statistical reporting. CEGB provides a reproducible foundation for cross-platform, energy-aware gripper evaluation.

Cobots are designed to work in close collaboration with humans. To ensure safety in shared workspaces, ISO/TS 15066 imposes limits on the collaborative robot speeds to limit the impact forces during collisions. These limits depend on robot design, including inertia. Flexible link manipulators combine the advantage of lightweight design and the introduction of an elastic element, which is promising in the context of collaborative robots. However, reduced stiffness affects dynamic performance. As trade-offs between link flexibility for safety and dynamic performance remain unclear, this paper investigates how undesired side effects of flexibility relate to the positive aspects of compliance and lightweight design. To investigate the effect of flexibility on both safety towards a human and controlled manipulator dynamics, a generic simulation framework was designed. Safety is assessed by a simulation study on the influence of varying geometrical parameters on impact force at collision with a human head, after which the corresponding maximum allowable end-effector velocities per ISO/TS 15066 threshold were derived. To analyze control performance, a generic controller was implemented, consisting of a PID with direct strain vibration control, with gains optimized for each link design. The resulting end-effector tracking performance during point-to-point motion is evaluated. Based on these results, a co-design methodology was proposed to optimize link geometry and control gains for the fastest possible, yet safe motion for a given task. The effectiveness of the method was experimentally validated to assess whether optimal link stiffness and controller gains yield the desired dynamic performance while respecting ISO safety thresholds.

Industrial processes such as spray painting and welding rely on the precise execution of motion trajectories to ensure consistent product quality and compliance with process constraints. Achieving such precision typically requires extensive training and experience from human operators, particularly for tasks involving continuous surface coverage and coordinated motion control. These tasks are also highly repetitive in industrial environments, making automation desirable to maintain consistency and reduce manual effort. Learning from Demonstration provides a promising way to transfer expert skills to robotic systems by allowing robots to learn directly from demonstrated examples. However, in practical industrial scenarios, only a limited number of demonstrations are usually available, and the exact structure of the desired trajectory is often only partially known. These constraints make it difficult for existing trajectory learning approaches to generalize accurately while preserving both the spatial shape and the temporal characteristics of expert motions. To address this challenge, this work proposes a sequential learning framework that combines a variational autoencoder and a recurrent neural network to learn and reproduce industrial trajectories from demonstrations. The variational autoencoder learns a compact representation of the trajectory data and captures the essential spatial characteristics of expert demonstrations. However, by itself, it does not explicitly model how the motion evolves. The recurrent neural network complements this by modeling the temporal relationships between successive points in the trajectory. When used alone, recurrent neural networks can capture temporal patterns but may struggle to learn robust representations of complex high-dimensional trajectory data when demonstrations are limited. By integrating both components, the proposed framework is able to learn meaningful latent representations while also modeling sequential dependencies in the trajectory. The proposed method is evaluated on industrial spray-painting demonstrations collected in a virtual reality environment and compared with two established trajectory learning approaches based on probabilistic modeling and neural dynamic policies. Experimental results show that the proposed method significantly improves trajectory reconstruction accuracy and stability. In particular, the generated trajectories are substantially closer to the expert demonstrations according to the dynamic time warping similarity measure and maintain very low variance during trajectory generation, indicating stable and consistent reproduction of expert motions. These results demonstrate that combining latent representation learning with sequential modeling enables more accurate and robust trajectory generalization from a limited number of demonstrations. The proposed framework, therefore, offers a promising solution for automating trajectory-based industrial tasks and can potentially be extended to other applications such as welding, surface finishing, and assembly operations that require precise and repeatable motion patterns.

While vacuum-based bending actuation offers benefits such as safety and compactness in soft robotics, it is often overlooked due to its limited actuation pressure, which restricts both bending angle and force output. This study presents a crease-free, origami-inspired vacuum bending actuator that advances both state-of-the-art vacuum bending actuators and traditional origami deformation principles by introducing orderly self-folding through optimized stiffness distribution. Achieved through finite element method (FEM), this design provides several advantages: (i) Self-folding allows for high bending angles (up to 138°) in a very compact form. (ii) The crease-free design facilitates 3D printing from a single soft material using a consumer-level fused filament fabrication (FFF) printer, specifically thermoplastic polyurethane (TPU) with a Shore hardness of 60A, potentially higher flexibility and durability. (iii) The compact configuration enables modular design, supporting reconfiguration as demonstrated in adaptable locomotion soft robots. (iv) The large bending angles allow the actuator to wrap around objects, offering extensive contact compared to other designs. This capability, c

Autonomous multi-robot systems are increasingly deployed in energy-variable environments where sustained operation depends on efficient energy harvesting. Traditional path-planning methods overlook ambient energy variability and inter-robot coordination, reducing overall efficiency. This paper introduces Hierarchically Conditioned Multi-Agent Reinforcement Learning with Meta-Differential Evolution (H-CMARL-DE) for energy harvesting-aware multi-robot path planning. The method centralizes only meta-level parameters priority order, shaping weights, and subgoal settings, while policy learning and execution remain decentralized, ensuring scalability and safety. Implemented in ROS-Gazebo, the system operates in a closed-loop observe-plan-act cycle, enabling robots to coordinate via hierarchical time-space reservations while adapting to dynamic obstacles and energy fields. Simulations demonstrate up to 240% improvement in energy-harvesting efficiency with minimal increase in path length compared to non-energy-harvesting-aware approaches, confirming H-CMARL-DE's robustness and adaptability for long-term cooperative operation in resource-constrained environments.

Wearable robots require firm attachment to transmit assistive forces, but excessive or sustained strapping pressure increases discomfort and soft-tissue injury risk. Most studies rely only on force measurements at the robot’s attachment point, missing limb–cuff interactions such as micro-slippage and tissue deformation. We present a spatiotemporal strapping-pressure framework using three actively controlled inflatable straps and a sensorized cuff with twelve miniature 3D force sensors that measure in-cuff interaction forces, especially shear, alongside comfort ratings. Twenty-seven static configurations, grouped by total bladder pressure, and one dynamic configuration that redistributed pressure over time at constant total pressure, were tested under controlled cuff-to-arm relative motion using a collaborative robot. Configurations sharing the same total pressure showed statistically equivalent end-effector stiffness and damping, whereas different totals produced distinct mechanical responses. Raw cuff-sensor quasistiffness decreased with pressure due to reduced transmitted shear; after linear scaling to the load-cell data (R² = 66.9%) to compensate for incomplete surface coverage, the corrected quasi-stiffness recovered the end-effector trend. End-effector energy dissipation changed little with pressure, while in-cuff measurements revealed up to 64% lower interface losses at higher pressures. The dynamic configuration preserved mechanical coupling relative to a matched static medium-pressure condition while achieving significantly higher comfort (24% improvement, p = 0.0005), a strategy relevant to upper-limb exoskeletons for stroke rehabilitation where prolonged wear is required.

Understanding the physical interaction with wearable robots is essential to ensure safety and comfort. However, this interaction is complex in two key aspects: (1) the motion involved, and (2) the non-linear behaviour of soft tissues. Multiple approaches have been undertaken to better understand this interaction and to improve the quantitative metrics of physical interfaces or cuffs. As these two topics are closely interrelated, finite modelling and soft tissue characterisation offer valuable insights into pressure distribution and shear stress induced by the cuff. Nevertheless, current characterisation methods typically rely on a single fitting variable along one degree of freedom, which limits their applicability, given that interactions with wearable robots often involve multiple degrees of freedom. To address this limitation, this work introduces a dual-variable characterisation method, involving normal and tangential forces, aimed at identifying reliable material parameters and evaluating the impact of single-variable fitting on force and torque responses. This method demonstrates the importance of incorporating two variables into the characterisation process by analysing the normalised mean square error (NMSE) across different scenarios and material models, providing a foundation for simulation at the closest possible level, with a focus on the cuff and the human limb involved in the physical interaction between the user and the wearable robot.

Pruning is an essential agricultural practice for orchards. Proper pruning can promote healthier growth and optimize fruit production throughout the orchard’s lifespan. Robot manipulators have been developed as an automated solution for this repetitive task, which typically requires seasonal labor with specialized skills. While previous research has primarily focused on the challenges of perception, the complexities of manipulation are often overlooked. These challenges involve planning and control in both joint and Cartesian spaces to guide the end effector through intricate, obstructive branches. Our work addresses the behavior planning challenge for a robotic pruning system, which entails a multilevel planning problem in environments with complex collisions. In this article, we formulate the planning problem for a high-dimensional robotic arm in a pruning scenario, investigate the system’s intrinsic redundancies, and propose a comprehensive pruning workflow that integrates perception, modeling, and holistic planning. In our experiments, we demonstrate that more comprehensive planning methods can significantly enhance the performance of the robotic manipulator. Finally, we implement the proposed workflow on a real-world robot. As a result, this work complements previous efforts on robotic pruning and motivates future research and development in planning for pruning applications.

Conventional industrial robots are designed to be as stiff as possible to ensure high positioning accuracy. The stiffness of a structure is, however, strongly related to its mass, leading to heavy structures. This paper aims to quantify the potential gain of reducing and eliminating the stiffness constraints, which is of lesser importance for collaborative robots, by investigating the effect of applying different optimization objectives. The resulting optimized designs are quantitatively compared using a set of performance measures and evaluated against the traditional stiffness-designed approach. It was concluded that significant improvements can be made, e.g. the robot’s mass can be reduced up to 74% compared to traditionally stiff-designed robots. The dependency of the payload as well as the structural/actuator mass distribution on the optimized results is investigated and proved to have a significant influence on the potential improvements when allowing reduced structural rigidity. The relationship between structural mass and actuator mass is investigated and compared to commercially available cobots.

Implementing efficient warning systems (e.g., visual, audio, or vibration alerts) is critical in human-robot collaboration (HRC) to enhance safety. Augmented reality (AR) is a promising technology for delivering such warnings. However, several studies have yet to fully investigate the effects of AR-based warning systems on safety and trust HRC. In this work, we investigate the effects of an AR-based warning system on both safety and trust using objective and subjective measures as well as physiological response as stress indicators. Twenty-six people participated in our within-participant experiment. Our results suggest that while the AR-based warning system improves the objective measure of safety, participants did not explicitly feel that their environment was safer with the system. Nevertheless, they demonstrated a higher level of trust in the AR system. However, this positive effect on trust was not reflected in any measure of physiological response, such as heart rate variability or electrodermal activity. Further studies should explore the long-term effects of AR-based warning systems on both safety and stress reduction.

Providing assistance in wearable robots based on exertion levels enables targeted and energy-efficient support. Traditional fatigue estimators, such as surface electromyography (sEMG) are predominantly reactive, identifying fatigue only after it has occurred. Near-Infrared Spectroscopy (NIRS) offers a complementary approach by providing early insights into muscle oxygenation, with oxygen depletion strongly correlating with muscle fatigue, thereby enabling proactive monitoring. This study integrates chip-based NIRS sensors into a physical interface to monitor muscle oxygen saturation during both static and dynamic conditions. The experimental protocol included an occlusion test, a low-intensity walking trial, and a high-intensity cycling trial to evaluate the sensor's performance under varying levels of exhaustion. Preliminary results demonstrate the sensor's sensitivity to oxygenation changes, underscoring its potential for fatigue-adaptive assistance in wearable devices.

Recent advances in dynamic and reversible polymer networks have led to self-healing soft robots that can restore their physical and electrical properties after damage. However, in most cases, human intervention remains essential for the healing process. This poses a challenge, especially in working environments with limited human access or where human involvement cand hinder efficiency. To address this gap, in this article, first, the different phases of the healing process in soft robotics are discussed and then the technologies that are or can be integrated into self-healing soft robots to allow each individual phase to be performed autonomously with minimal human involvement are reviewed. Finally, in this article, the challenges of integrating all phases into self-healing soft robots are discussed and the perspectives on achieving fully autonomous self-healing in the future are offered. These phases are classified into five: damage detection, damage cleaning, damage closure, stimulus-triggered material healing, and recovery assessment. Achieving these attributes requires employing physical intelligence at the material level through the use of stimuli-responsive materials or utilizing embodied intelligence at the system level by integrating healing-assistive subsystems or a synergistic combination of both. Consequently, self-healing soft robots can achieve self-sufficiency in their healing capabilities, rendering them a sustainable solution for broader applications.