This work presents a laboratory-scale track-based robotic colonoscope incorporating a force-differential steering mechanism. A syringe-driven hydraulic system selectively retracts individual tracks to generate asymmetric track-wall contact forces for steering. A theoretical model was developed to describe the relationships between hydraulic input, normal-force modulation, and steering torque and to provide a first-order framework for the mechanism. The prototype, was tested with force sensing and an IMU, was evaluated for straight propulsion and steering in single- and multi-bend flexible pipes. Hydraulic actuation tests showed a near-linear increase in track-wall normal force with increasing syringe input, although the measured force was substantially lower than the theoretical prediction. Steering experiments showed that motion without active steering produced approximately 35° of yaw before traction loss, whereas force-differential steering increased the yaw angle to approximately 45° and enabled negotiation of a 120° bend. Tests under different friction conditions further showed that reduced wall friction impaired steering performance. In a continuous S-shaped pipe, two of three independent trials successfully completed both bends, while one trial exhibited incomplete yaw recovery, indicating trial-to-trial variability and sensitivity to traction conditions. Overall, these results provide proof-of-concept evidence for force-differential steering in a tracked robotic colonoscope and demonstrate that straight propulsion and active planar steering can be achieved within a common tracked locomotion architecture. Further miniaturisation, improved hydraulic control, and evaluation in more physiologically representative environments are required before clinical applicability can be assessed.
Human–robot collaboration offers new opportunities to increase productivity, flexibility and ergonomics in work processes, however, it also introduces specific risks associated with a shared workspace between humans and robots, for which traditional risk assessment techniques may not be sufficiently effective. This article summarizes selected risk assessment techniques designed for the field of Human-Robot Collaboration (HRC) and evaluates their strengths and weaknesses with respect to their applicability in a collaborative environment. Based on the identified limitations, a new hybrid technique is presented that combines Hierarchical Task Analysis (HTA), Hazard and Operability Study (HAZOP), Process Failure Mode and Effects Analysis (HRC–PFMEA) and Risk matrix. The proposed approach integrates qualitative and quantitative risk assessments, enables risk identification at an early stage of system design, and increases the transparency of work processes. The applicability and logical workflow of the proposed method are demonstrated using a simulation scenario of an assisted assembly workstation created in a digital environment.
Unilateral Spatial Neglect (USN) significantly impacts functional outcomes post-stroke, yet traditional assessment methods, relying predominantly on two-dimensional (2D) paper-and-pencil tasks, fail to capture the disorder’s complexity across three-dimensional (3D) space and, in particular, the role of depth. This paper introduces NeglectARm3D, a platform for the objective and repeatable 3D assessment of USN that integrates the precise kinematic and force measurement of a 7-DOF torque-controlled robot (Franka Research 3) with the immersive 3D stimulus presentation of a commercial head-mounted display (Meta Quest 3). A Cartesian variable impedance controller lets the participant reach freely toward randomized 3D targets (Phase 1, subject in charge), enabling the reconstruction of a volumetric exploration map, and delivers smooth haptic guidance toward neglected targets (Phase 2, robot in charge). As a first step toward clinical deployment, we evaluated the technical behavior, safety, and tolerability of the platform in twelve healthy adults performing thirty reaching trajectories (twenty in Phase 1, ten in Phase 2). The system preserved natural movement velocity (0.10±0.02 vs. 0.10±0.01 m/s), achieved large position error in free reaching (32.46±14.71 mm) versus precise robot-led convergence (3.01±0.43 mm) with reduced interaction force (4.96±0.84 vs. 2.59±0.32 N), and maintained physiological movement smoothness (SPARC −1.87±0.29 vs. −1.63±0.32, benchmark −1.40). Usability was high (System Usability Scale 85.60±8.10), simulator sickness was low across all subscales, and all participants reported no pain and no perception of danger. These results support the technical feasibility and short-term tolerability of the integrated platform in healthy adults and justify subsequent clinical and psychometric evaluation in stroke survivors with USN.
Mental health is a significant healthcare challenge, and cognitive behavioral therapy (CBT) is a widely used therapeutic method for treating anxiety and depression. However, traditional CBT often requires access to trained clinicians and can be cost-prohibitive or logistically difficult for many individuals. To address these barriers, we developed a low-cost socially assistive robot (SAR) that uses a large language model (LLM) to guide the user through interactive at-home CBT exercises. In this exploratory study, 38 university students completed CBT exercises across a 15-day period using one of three modalities: with a robot (using an LLM for dialogue), a chatbot (using the same LLM for dialogue), or traditional CBT worksheets. We measured weekly therapeutic outcomes, changes in pre-/post-session anxiety measures, and adherence to completing CBT exercises. Our findings indicate that self-reported general psychological distress significantly decreased over the study period in the robot and worksheet conditions but not in the chatbot condition. Additionally, the SAR enabled significant single-session improvements on more days than the other two conditions combined. Mixed-effects modeling further An analysis with a mixed-effects model also suggested that the robot and chatbot conditions better reduced post-session anxiety for those with elevated levels of anxiety. Our findings suggest that SAR-guided, LLM-powered CBT may be an effective method for supporting therapeutic progress and decreasing user anxiety immediately after completing the CBT exercise. The findings underscore the potential for combining AI-driven personalization with socially assistive robotics to create accessible, scalable, and engaging mental health interventions.
IntroductionHuman-drone interaction is an emerging field investigating ways to extend the conventionally passive role of drones with autonomous, socially interactive functionalities. Drones offer unique opportunities for applications, such as exercise support, due to their mobility and flexibility. While previous research has studied drones for, e.g., pacing, gait analysis, and accessibility in running contexts, we propose that an equally crucial part of such applications is the socially interactive role of the drone. MethodsWe present a study in two phases. In the first phase, we conducted a preliminary study that informed the design of a co-design workshop through eight semi-structured interviews exploring drone support for runners. In the second phase, we designed and conducted a co-design workshop with five experienced recreational runners to reflect on the practical and ethical challenges of using drones and to envision ways in which drones could support and interact with runners in future scenarios. The collected materials were analyzed using reflexive thematic analysis and annotated visual analysis.ResultsThe results present five low-fidelity social drone prototypes, including participants’ reflections tied to each prototype. Furthermore, we present analytical themes that demonstrate the necessity of extending and re-framing approaches to designing social drones.DiscussionThe paper makes three key contributions. First, we offer an approach to co-designing social drones that focus on eliciting the experiences of participants and manifesting those in concrete, embodied design concepts. Second, we validate the existing Design Space for Social Drones framework through a concrete application domain. Our validation shows that the dimensions can be further detailed, and that non-user perspectives are a potentially crucial extension. Third, we present three recommendations that encapsulate concerns for further designs of social drones.
Concentrating on various types of grasped objects, this work presents a distributed coordinated control scheme for a dual-arm reconfigurable manipulator (DRM) via adaptive dynamic programming (ADP). Kinematic and force analyses are performed to formulate the dynamics of a single-arm manipulator and the object using joint torque feedback technology and the Newton–Euler formulation. To address uncertainties associated with the grasped object, a gradient model-based adaptive algorithm is developed for online estimation of the grasp matrix relating the object to the DRM. By employing an adaptive observer to identify unknown dynamic terms, an enhanced optimal performance index is constructed that incorporates both object tracking performance and internal force effects within the manipulator. The performance index function is approximated by only a critic neural network, and the optimal control policy is obtained by policy iteration. Lyapunov stability analysis demonstrates that internal force error, position tracking error, and critic network weight approximation error are all uniformly ultimately bounded. Experimental results confirm the effectiveness of the developed coordinated control approach.
IntroductionLong-horizon action recognition is increasingly deployed in streaming and safety-critical settings, where short but severe corruptions, such as motion blur or adversarial perturbations, can overwrite temporal memory and cause errors to persist after unreliable segments end. Selective State Space Models (SSMs) enable linear-time inference over long contexts, but deterministic discretization can amplify this memory-contamination failure mode.MethodsWe propose BayesMamba, a reliability-aware selective SSM that integrates Bayesian uncertainty into SSM discretization. BayesMamba uses Bayesian Selective Scan (B-Scan) to estimate per-frame aleatoric and epistemic uncertainty, convert uncertainty into a reliability signal, and modulate both the effective time step and input injection so that state updates are suppressed when observations are unreliable. We evaluate BayesMamba on Kinetics-400 (K400) and Something-Something V2 (SSv2).ResultsUnder clean evaluations, BayesMamba achieves comparable or slightly improved performance, with 82.4% top-1 accuracy on K400 and 68.7% on SSv2, compared with 82.1% and 68.2% for a deterministic selective SSM baseline, while retaining similar computational cost. On SSv2 under bursty motion blur, BayesMamba improves robust accuracy from 51.4% to 54.2%; under projected gradient descent (PGD, ε = 4/255, ε = 8/255, and ε = 16/255), it improves robust accuracy from 18.2% to 19.8%. BayesMamba also reduces post-corruption persistence, decreasing Persist(64) from 0.41 to 0.19 and Persist(128) from 0.38 to 0.16. Under distribution shift, it improves calibration, reducing corrupted-set ECE from 0.158 to 0.097 and corrupted-set NLL from 2.34 to 1.66.DiscussionThese results indicate that reliability-aware discretization mitigates memory contamination and improves long-horizon robustness with modest overhead while preserving clean action recognition accuracy.
Introduction:Tendon-driven exoskeletons are often limited by mechanical bulk, complex cable routing, and large externally mounted actuators. Similarly, conventional linear actuators used in wearable exoskeleton systems are often limited by low stroke-to-body-length ratios that restrict the range of motion. This study presents the mechanical design of a novel Rail-guided Inline DirEct dRive (RIDER) with a high stroke-to-length ratio for wearable exoskeletons. Methods:The proposed system creates a linearized RIDER system using a "motor-driven cart" layout that travels along a toothed twin-rail structure. The prototype system was developed using 3D-printed materials and open-source electronics, and the resultant force-velocity performance was experimentally evaluated under vertical loading. Vertical displacement was measured using an ultrasonic sensor (HC SR-04), and the corresponding velocity was computed via a finite-difference approximation. Results:The actuator demonstrated a maximum free-load velocity of 7.27 cm/s and a maximum force output of 103.7 N prior to slipping. When integrated into a tendon-driven system, the duo configuration produced a 6.22 Nm torque output with a range of motion of 120° or a 10.37 Nm torque output with a range of motion of 72°. The actuator also demonstrates a 190% (i.e., 0.58 vs 0.2) higher stroke-to-length ratio than a representative telescopic actuator of similar geometry. Conclusion:In this study, we show that the RIDER system demonstrates a wearable, rail-based actuation strategy capable of adjustable torque amplification and improved spatial-efficiency integration for human-robot interaction in wearable robotics. The current study focuses on the benchtop mechanical validation of the RIDER system as a proof-of-concept for future integration into wearable exoskeletons. This research is part of a larger study to develop adaptive exoskeletons that enable scalable performance improvements for assistive robotics in the workplace and neurorehabilitation.
An insole-type active assist device has been developed as a robotic system to dynamically correct ankle alignment at heel contact in patients with medial knee osteoarthritis. Although our previous feasibility study demonstrated that the device could be safely used during an on-the-spot stepping task, its effects on loading behavior remain unclear. This study aimed to investigate whether dynamic ankle alignment correction using the device alters horizontal ground reaction force variability and center-of-pressure sway during stepping. Force-plate data obtained from six ambulatory patients with medial knee osteoarthritis were analyzed as a secondary biomechanical analysis. Each participant performed repeated stepping trials under two conditions: a non-control condition, in which the device was worn without motor control, and a control condition, in which heel-eversion assistance was provided. Stance phases were extracted using vertical ground reaction force, and horizontal ground reaction force components, center-of-pressure sway measures, peak vertical ground reaction force, and vertical impulse were calculated. Compared with the non-control condition, the control condition reduced the ranges of anterior–posterior and medial–lateral ground reaction force components in all participants. Center-of-pressure rectangular sway area also decreased consistently, and trajectory length tended to decrease, whereas peak vertical ground reaction force and vertical impulse showed little change. These findings suggest that dynamic ankle alignment correction using a robotic insole-type device may reduce horizontal loading variability and center-of-pressure sway during stepping without substantially altering vertical loading. Because this study was based on a small sample and an exploratory secondary analysis, the findings should be interpreted as preliminary biomechanical evidence rather than evidence of clinical effectiveness.
Introduction:Autistic children are increasingly engaging with social robots in educational and support contexts, but limited research has compared the perceptual and design preferences of autistic and neurotypical children. Methods:This mixed-methods study examined the robot design preferences of 43 children aged 3-15 years, including 21 autistic and 22 neurotypical children, and included semi-structured interviews with 11 adult stakeholders, comprising six experts and five non-experts. Quantitative analyses explored relationships between robot preference and perceptual features, including size, color, movement, voice, and anthropomorphism, while qualitative thematic and co-occurrence analyses examined perceived benefits, ethical issues, accessibility challenges, and ideal design characteristics. Results:Perceptual characteristics, particularly movement intensity and color multiplicity, influenced robot preference. Across groups, children generally preferred humanoid robots with soft voices and eye movement. Gender-related differences appeared more consistent than diagnostic differences in preferences for size and movement speed. Autistic children highlighted playful and social interaction functions and showed greater variability in preferred personality-related traits. Qualitative findings indicated strong interest in robot-supported interactions, while also raising concerns about accessibility and possible emotional dependency. Discussion:These findings suggest that inclusive social robot design should prioritize modificable sensory features, moderated anthropomorphism, predictable interaction patterns, and personalization mechanisms that account for individual, gender-related, and diagnostic differences.
Robot-assisted physiotherapy has attracted increasing attention for its potential to provide repeatable, stable, and controllable physical interaction during rehabilitation-oriented therapy. However, contact-rich physiotherapy tasks remain challenging because the robot must reproduce therapist-demonstrated massage skills while adapting to non-planar and deformable body surfaces, suppressing impact during contact transition, and maintaining stable force regulation. This paper proposes a contact-aware robot-assisted physiotherapy framework that integrates task-space skill generalization, contact state estimation, dynamic velocity adjustment, force-error compensation, and bounded variable impedance control. Therapist-guided demonstrations are encoded using Dynamic Movement Primitives (DMPs) to construct a physiotherapy skill library, enabling typical massage skills, including kneading, patting, pushing, and pressing, to be generalized to new start and goal points in the robot Cartesian task space. During execution, force/torque feedback is used to estimate the contact point and surface normal, update the local task frame, regulate the approach velocity, and compensate for force-tracking errors. Experimental validation was conducted on a silicone abdominal model and in a preliminary healthy-volunteer back physiotherapy test. The results show that the proposed dynamic contact strategy suppresses excessive impact during contact transition, avoiding the 165.86 N peak impact observed under high-speed contact. The force compensation strategy reduces the force-tracking RMSE from 0.835 N to 0.395 N, corresponding to a 52.69% improvement. In addition, the bounded variable impedance strategy improves motion-force coordination compared with fixed-stiffness control. These results demonstrate that the proposed framework improves contact transition safety, force regulation accuracy, and task adaptability at the control-performance level, providing a feasible basis for further development of robotic physiotherapy systems.
Hydraulic booms used in material handling are typically commanded in cylindrical task-space coordinates, while dynamic models used for control are commonly formulated in joint coordinates. This mismatch leads to unnecessarily high-dimensional system representations when predictive control is applied to underactuated suspended loads. This paper addresses this modeling inconsistency by introducing a reduced-order state-space reformulation of grapple sway dynamics directly in cylindrical task-space velocity coordinates. The proposed input-consistent state transformation eliminates the dependence of the dynamics on input accelerations by embedding actuator–sway coupling within modified velocity states, yielding a nonlinear model driven solely by cylindrical velocity commands. Based on this representation, a nonlinear model predictive control (NMPC) framework is developed for simultaneous goal reaching, sway suppression, and obstacle avoidance. The NMPC cost function is designed to capture the practical trade-off between rapid boom motion and suppression of suspended load oscillations. In addition, approximate hydraulic actuator dynamics identified from a high-fidelity AMESim simulator model of a forwarder crane are incorporated to better reflect realistic boom behavior. Simulation studies in representative boom operation scenarios demonstrate that the proposed formulation enables effective sway regulation and accurate task-space motion control using a reduced set of states without requiring full joint-space models or payload sensing. The results indicate that the proposed cylindrical state-space reformulation simplifies predictive control design for underactuated hydraulic booms while preserving the essential dynamics of the suspended load.
Learning from demonstration (LfD) has become a popular approach with the emergence of modern transformer-based algorithms. However, the performance of these policies is limited by the quality of the demonstrations. Combining imitation and exploration promises to train policies that perform better and are more reliable. However, this requires a robotic system that can explore safely without damaging itself or the environment, especially in contact-rich tasks during which the robot must exert force on its environment to solve the task. In this study, we investigate the combination of a state-of-the-art reinforcement learning (RL) algorithm with human demonstrations to learn how to open a door with minimal task-specific engineering on an articulated soft robot arm. We found that learning from both exploration and demonstration data stored in separate buffers makes the algorithm not only more sample-efficient and robust but also allows the policy to reach a higher performance level than the provided expert demonstrations. We also show that using an articulated soft robotic arm allows us to perform RL on a real robotic system without any pretraining and with a simple safety system that does not require any additional sensors, such as force-torque sensors. Additionally, we can implicitly learn the nonlinearities stemming from the soft materials in the actuator. Our findings show that combining LfD with RL results in both better performance and more robust behaviors and indicate that articulated soft robots allow for learning contact-rich tasks safely on a real system.
Heterogeneous multi-robot teams require systems that can interpret natural-language goals, allocate tasks, and adapt to unexpected events. We developed CoMuRoS (Collaborative Multi-Robot System), a generalizable hierarchical architecture combining a centralized task-manager LLM with decentralized robot-level LLMs for executable Python generation from primitive ROS2 skills. The task manager uses static planning rules and dynamic context, including task history, robot/task status, and detected events, while onboard perception using VLM/image processing classifies events as relevant or irrelevant and triggers replanning. Hardware experiments demonstrated recovery from disruptive events, filtering of irrelevant distractions, and coordinated transport with emergent human-robot cooperation, achieving success rates of 9/10 for collaborative object recovery, 8/8 for coordinated transport, and 5/5 for human-assisted recovery. Simulation studies demonstrated intention-aware replanning. A curated benchmark of 22 scenarios, 54 tasks, and around 20 robots evaluated task allocation, classification, IoU, executability, and correctness across multiple LLMs, with correctness up to 0.91 ± 0.053; a 20-scenario replanning benchmark achieved Correctness = 0.948 ± 0.034 using Grok 3. CoMuRoS enables runtime, event-driven replanning on physical robots and supports flexible multi-robot and human-robot collaboration across diverse scenarios.
Industry 5.0 cobotics calls for collaborative robots that adapt to the operator’s physical and cognitive state in real time. Most current industrial deployments remain reactive, responding only after explicit commands and ignoring the operator’s psychophysiological condition. The three-module Human-Centric Digital Twin (HCDT) framework establishes the upstream perception-and-reasoning architecture for such systems using Vision-Language Models, and identifies closed-loop feedback and physical-assistance mechanisms as the natural next layer in the framework’s development. Building on that foundation, this paper presents a complementary, real-time instantiation specialised for the resource-constrained, vision-plus-biosignal setting typical of ergonomics-and-safety HRC, deployed on a standard collaborative robot (Universal Robots UR3 with a Robotiq 2F-85 gripper) and extended with closed-loop operator-state adaptation. The Perception module pairs MediaPipe hand-landmark tracking and a Random Forest gesture classifier with Tobii Pro Glasses 3 pupillometry. The Reasoning module combines a ten-state finite-state machine, a weighted composite fatigue score (blink rate, blink duration, PERCLOS, hand-jerk) banded into FRESH, MILD, MODERATE and SEVERE, a saccade-gated commitment fixation, and a Task-Evoked Pupillary Response (TEPR) detector for acute-stress soft-emergency-stop. The Motion module provides a closed-form analytical inverse-kinematics solver, workspace clamping and a slew-rate-limited joint-velocity controller. Velocity-based linear extrapolation of hand position over a 0.5 s horizon enables anticipatory robot pre-positioning. The system is presented as a single-operator feasibility study rather than a generalisation claim. On a leak-free split-then-augment evaluation (520 raw test samples not seen by the training fold in any orientation), the Random Forest classifier reaches 99.62% accuracy with both errors falling in the Okay-Neutral confusion pair that the geometric override layer is designed to catch; a comparison against MLP and KNN baselines on the same split shows all three classifiers within the real-time control budget. Saccade-gated early-commit is designed to recover the modelled 150–300 ms gaze-to-motion lead, and the TEPR Soft E-Stop operates on an independent 300–500 ms channel. End-to-end gesture-to-motion latency has a conservative worst-case ceiling of approximately 830 ms, formed by a camera-buffer term, the 10-frame debounce and one control tick, inside the 1-s interactive response-time limit adopted as the design target. Owing to a Tobii Pro Glasses 3 hardware fault, the physiological pathways (fatigue banding, TEPR soft-stop and gaze commitment) were verified through their downstream control responses to simulated triggers rather than validated on organic pupil data; organic, multi-participant validation is identified as future work.
