The multiuser haptic-enabled robotic system (M-Hers) facilitates shared control among human operators through task-dependent authority allocation, where interaction relationships are typically dictated by task requirements. However, some of these relationships can be nonpassive, generating excess energy that violates passivity constraints and compromises system stability. To address this, we first introduce the interaction architecture (IA) to formalize how operators influence task execution. Based on this framework, we propose a tank-based two-layer task model that ensures system passivity despite nonpassive IAs. This model comprises a virtual object (VO) layer for task rendering and a virtual system (VS) layer that passively executes nonpassive IA behaviors. The VS layer uses a global energy tank to compensate for IA-induced energy violations and modify the VO model when tank energy is depleted. This structure decouples task rendering from low-level robotic control, enabling seamless integration of an arbitrary number of robots with heterogeneous dynamics and control modes. Simulation and experimental results validate the proposed method’s scalability, flexibility, and effectiveness in preserving passivity while accurately realizing diverse IAs. This approach paves the way for scalable and easy-to-deploy control framework that supports multiuser haptic interaction.

Accurate and robust locomotion mode prediction is crucial for seamless interaction between humans and assistive devices. While multimodal sensing offers a promising avenue for enhanced accuracy, existing approaches often struggle to demonstrate clear advantages over unimodal methods, largely due to a lack of understanding regarding each modality's unique characteristics and task-specific strengths. To address this, we present a systematic comparative analysis of Inertial Measurement Unit (IMU) and surface Electromyography (sEMG) modalities for human locomotion mode prediction. Utilizing a public dataset (nine subjects, 17 gait activities), our experiments rigorously evaluated performance across cyclic and non-cyclic locomotion tasks, considering both steady-state and transition-state. We investigated the impact of deep learning architectures (CNN, LSTM, TCN) and sliding-window lengths (short vs. long) on prediction accuracy and stability. Statistical analyses reveal significant performance differences dependent on modality, window length, and gait type. Notably, our findings demonstrate that optimal classification accuracy is achieved by leveraging IMU data with short windows for non-cyclic locomotion modes prediction and sEMG data with long windows for cyclic locomotion modes prediction, with Temporal Convolutional Networks (TCNs) consistently yielding superior overall results. These findings offer concrete guidelines for effectively fusing multimodal data, leveraging modality-specific strengths to enable adaptive, interpretable, and high-performance locomotion mode prediction.

Despite extensive investigations into the multi-user haptic-enabled robotic system (M-Hers), achieving scalable control design in the presence of non-passive human operators remains a key challenge. This is primarily due to the increasing complexity of stability conditions and interaction coupling as the number of operators grows. In this study, we address this challenge in two steps. First, we introduce the individual interaction environment (IIE) to isolate the passivity violations, which facilitates the independent control design for each human-robot subsystem, thereby enhancing the scalability with respect to the number of subsystems. Second, within the IIE framework, we identify passivity-violating components caused by partners' active behaviors and propose a novel augmented tank-based controller (ATBC) to guarantee passive IIE while maintaining high rendering accuracy. Specifically, the ATBC employs an energy-related power regulation strategy to enhance interaction safety and a time-varying control gain to mitigate the negative effects of power regulation on rendering fidelity. We validated the proposed method through collaborative haptic tasks on a customized M-Hers composed of three robots in four different scenarios. Comparative studies demonstrate that our approach effectively ensures IIE passivity in the presence of active human behaviors, while ensuring high reproducibility and achieving a favorable balance between passivity and rendering accuracy.

Previous studies on time difference of arrival (TDoA)-based ultra-wideband (UWB) localization typically employed an arbitrary and fixed reference anchor during the localization process. However, this scheme fails to maintain stable and high-quality performance in complex scenarios, since the localization of mobile targets suffers from multipath interference and the change in sensor-target geometry, while the reference anchor amplifies this effect. To address this issue, we present a dynamic reference anchor selection algorithm, which determines the reference anchor by evaluating the factor that fuses geographical factor and link quality. In addition, we integrate dynamic reference anchor selection into an optimized UWB localization approach to achieve high accuracy. The position is calculated by a constrained Taylor algorithm after UWB calibration and finally fine-tuned by a Kalman filter. The proposed approach was validated using data collected in both laboratory and office scenarios. The experimental results show that the 3D dynamic localization error of the proposed approach reaches 4.9 cm on average. The dynamic reference anchor selection algorithm we proposed improves the overall localization accuracy by 18.4 %, with a dominance of 87.8 % and 100 % in the laboratory and office scenario, respectively. Furthermore, the localization stability and reliability are also validated with a reduction in maximum localization error and avoidance of extreme outliers.

Robot-assisted task-oriented training demonstrates immense potential in rehabilitation area. Parallel robots, with advantages such as low inertia and high stiffness, facilitate precise haptic feedback, yet their application in rehabilitation is limited by workspace constraints. To this end, we propose a design scheme for a haptic robot based on a reconfigurable asymmetric parallel mechanism. We first introduce a two-stage multi-objective optimization method to obtain the optimal parameter configurations. Then, to achieve precise assembling of the reconfigurable mechanism in each configuration, corresponding positioning mechanisms are designed. System performance tests validate the robot's capabilities under different configurations: workspace meets design requirements, stiffness output reaches 30 N/mm, force output is 40 N, RMS of maximum back-driven force along$x, y$, and$z$axes is 7.5 N, and RMS of maximum back-driven torque around$x$and y axes is 567.4 N. mm. Target tracking and virtual channel trajectory tracking experiments demonstrate the system's haptic rendering ability for gross motor tasks (GMTs) and fine motor tasks (FMTs), respectively. The developed 6-DOF haptic robot holds promise for versatile task-oriented rehabilitation training.

Optimizing the feature representation and decoding efficiency of the steady-state visual evoked potentials (SSVEP) is critical to enhance the performance of neural signal decoding systems. Current deep learning models usually overlook the physical topological information of EEG channels, resulting in suboptimal feature extraction and limited recognition performance. To address these challenges, this study proposes a synergistically designed SSVEP recognition framework to alleviate data insufficiency, improve the feature representation, and enhance decoding efficiency. Specifically, a slicing-and-scaling technique is adopted to improve the model generalization under limited-sample scenarios. A graph-based spatial filter leverages the topological relationships among EEG channels to suppress redundant information and enhance spatial feature quality. A lightweight convolutional neural network (CNN) with fewer parameters is developed to efficiently extract discriminative temporal–spatial features for accurate SSVEP classification. Experimental results on two public benchmark datasets and one self-collected dataset demonstrate that the proposed framework outperforms baseline deep learning models, yielding improvements of at least 6.8 %, 8.5 %, and 0.5 % in peak average classification accuracy, respectively. The maximum average information transfer rates (ITRs) achieved on the three datasets were 221.4 ,106.7 , and 133.9 , respectively. By simultaneously reducing model complexity and improving decoding performance, the proposed framework offers an effective and promising approach for efficient neural signal decoding in SSVEP recognition.

Actuation efficiency is a significant consideration for soft exosuits. It has important significance for reducing system weight and improving the effectiveness of walking assistance. Inspired by a fact that desired assistance from exosuits varies over the gait stage, this study developed a power modulation unit (PMU) to adapt to human walking characteristics, aiming to reduce the power requirement of exoskeletons. The underlying principles consist of: 1) dividing the gait phase into assistance period and idle period by different gait characteristics, where the terminal stance phase is defined as the assistance period and other phases are defined as the idle period; 2) motor power is stored in elastic elements during the idle period with mechanical advantage; 3) an amplified power bursts during the assistance period. Preliminary experiments were carried out on an ankle exosuit with human users. Experimental results indicate that the proposed PMU can amplify the output power by 3.1 times while affecting little on wearers’ normal gait, which implies the potential of the developed PMU for use in wearable actuation systems. Note to Practitioners—This work was motivated by the trade-off between multipath assistance and lightweight actuation. That is, multipath assistance typically necessitates multiple motors, consequently adding to the overall weight of the system. In this work, we developed a PMU to adapt to human walking characteristics (motor power is stored in elastic elements during the idle period and released when needed), aiming to reduce the requirement for peak actuation power. Experimental results show that the PMUs can provide an average power amplification ability of 3.1 times and does not interfere with normal human gait. The developed PMU holds great potential in walking-assisted exosuits or other devices for human movement assistance.

Decoding motor imagery (MI) from electroencephalogram (EEG) signals is a cornerstone of brain-computer interface (BCI) systems. However, existing methods often face a critical trade-off between decoding accuracy and model interpretability, limiting their applicability in real-world settings. To address this challenge, we proposed the WavTSK, an end-to-end interpretable fuzzy neural network for efficient MI-EEG decoding. The WavTSK employs a deeply integrated architecture that combines a learnable wavelet-based feature extraction module with a multi-rule Takagi-Sugeno-Kang (TSK) fuzzy classifier. The feature extractor incorporates learnable wavelet filters, statistical descriptors, and an adaptive band-weighting mechanism to capture rich multi-scale time-frequency representations directly from raw EEG. The extracted representations are then processed by a TSK fuzzy reasoning layer, enabling rule-level transparency in decoding. Experiments on the four-class BCI Competition IV-2a dataset showed that WavTSK achieved mean decoding accuracies of 71.63% in cross-validation and 70.55% in the cross-session hold-out setting, consistently outperforming state-of-the-art black-box deep learning models while maintaining strong interpretability. The results highlight the potential of WavTSK as a powerful and interpretable framework for reliable EEG decoding, advancing the development of trustworthy and clinically applicable BCI technologies.

Emerging evidence suggests that prolonged virtual reality (VR) exposure may impair human sensory systems. Most research has focused on the visual, proprioceptive, and vestibular systems, but the impact of VR on haptic perception remains unclear. In this study, we investigated alterations in human sensitivity to contact force following VR gaming. A force perception task was designed to assess changes in contact force across six difficulty levels with step sizes ranging from 0.5 to 5 N. A total of 18 participants performed the task before VR, after 10 min, and after an additional 20 min of VR. The perceptual accuracy of correctly perceiving force changes at each difficulty level was measured across three test periods. The results indicated that 66.67% of participants experienced a negative impact from VR at the 1-N change step. Perceptual accuracy significantly decreased in this group, with a 9.17% reduction after 10 min and a 17.50% reduction after an additional 20 min. In contrast, minimal effects were observed in the remaining participants. These findings suggest that even short-term VR exposure can impair force discrimination in certain users, with the effects becoming more pronounced over time.

Robot-assisted ankle rehabilitation training imitating physician's professional techniques is highly important for promoting personalized training and improving clinical outcomes. In this work, we propose a two-level kernelized movement primitives (2-level-KMP) imitation learning algorithm under the kernelized movement primitives (KMP) framework, which reproduces physician's experience and optimizes the imitation trajectory during rehabilitation, to realize physician-level performance in robot-assisted ankle rehabilitation training. First, a KMP process combined with a Bayesian optimizer is used to imitate the rehabilitation trajectory. Second, the other KMP process is used to smooth the imitation trajectory further. Then the two KMP processes combined with patient-in-the-loop optimization (PILO) realize temporal rehabilitation adaptation. Finally, the 2-level-KMP algorithm is reproduced on a parallel ankle rehabilitation robot (PARR), which enables the patient's passive rehabilitation training to be empirical and adaptive. Ten ankle dysfunction patients were involved in clinical experiments, with the results showing that the proposed algorithm can accurately reproduce physician's trajectories and modulate trajectories based on patient's feedback. After ten rehabilitation exercises, the number of modulation points calculated from patient's torque feedback decreases by 85.19% on average compared with the beginning stage. A comparison between the 2-level KMP algorithm and existing algorithms shows that the 2-level-KMP algorithm can better ensure smoothness and retain the shape of the trajectory during trajectory modulation, ensuring the safety of ankle rehabilitation and retaining the experience of the physician.

Restoring the capability to perform instrumental activities of daily living (IADLs) is an imperative step towards independent living for neurologically impaired individuals. Robot-assisted task-oriented training with haptic feedback has the potential to enhance patients’ ability to perform IADLs. However, robot-assisted haptic rendering of IADLs is extremely challenging due to their complex dynamic properties and has been rarely reported. Considering the broad impedance range characteristics from free motion to hard contact, nail hammering (NH) is chosen as a representative IADL task. This paper presents our attempts to render the NH task via a customized robot. The core technologies consist of two aspects: 1) a robot-assisted haptic modeling technique with guaranteed accuracy and computation cost (by combining practical measurement data and experience-dependent analytical functions); 2) a robot-assisted haptic rendering technique involving a haptic robot with broad impedance range and sufficient force feedback (via a low gear ratio cable transmission and redundant actuation parallel mechanism) and closed-loop impedance control with guaranteed passivity and stability. Human experiments demonstrate an accurate NH task rendering that all Pearson correlation coefficients between real and virtual tasks are larger than 0.89. The modeling sensitivity analysis showed that stiffness parameter has the greatest effect on the realism of haptic rendering, with an effect size of 0.94. This study represents an important step towards comprehensive robot-assisted task-oriented therapy with haptic feedback.Note to Practitioners—The motivation of this work is to explore the techniques of robot-assisted haptic rendering of IADL tasks. On the one hand, current task modeling approaches are hard to balance accuracy and computational efficiency. On the other hand, existing haptic platforms have difficulty in meeting the broad impedance range and large force output requirements. In this work, we firstly developed a customized haptic robot via redundant actuation (enabling high robotic stiffness and force output) and low gear ratio cable transmission (enabling low friction and high back-drivability). We then built the nail hammering (NH) task model by combining practical measurement data (for accuracy) and experience-dependent analytical functions (for computational efficiency). Finally, we achieved the haptic rendering of the NH task using closed-loop impedance control with passivity and stability analysis. The proposed robot-assisted haptic modeling and rendering techniques can be extended to the haptic display of other types of IADL tasks.

Walking-assistive devices require adaptive control methods to ensure smooth transitions between various modes of locomotion. For this purpose, detecting human locomotion modes (e.g., level walking or stair ascent) in advance is crucial for improving the intelligence and transparency of such robotic systems. This study proposes Deep-STF, a unified end-to-end deep learning model designed for integrated feature extraction in spatial, temporal, and frequency dimensions from surface electromyography (sEMG) signals. Our model enables accurate and robust continuous prediction of nine locomotion modes and 15 transitions at varying prediction time intervals, ranging from 100 to 500 ms. Experimental results showcased Deep-STP's cutting-edge prediction performance across diverse locomotion modes and transitions, relying solely on sEMG data. When forecasting 100 ms ahead, Deep-STF achieved an improved average prediction accuracy of 96.60%, outperforming seven benchmark models. Even with an extended 500ms prediction horizon, the accuracy only marginally decreased to 93.22%. The averaged stable prediction times for detecting next upcoming transitions spanned from 31.47 to 371.58 ms across the 100-500 ms time advances. Although the prediction accuracy of the trained Deep-STF initially dropped to 71.12% when tested on four new terrains, it achieved a satisfactory accuracy of 92.51% after fine-tuning with just 5 trials and further improved to 96.27% with 15 calibration trials. These results demonstrate the remarkable prediction ability and adaptability of Deep-STF, showing great potential for integration with walking-assistive devices and leading to smoother, more intuitive user interactions.

Task-oriented training with haptic rendering can boost robot-aided motor learning to tasks with similar dynamics. Although multi-DOF robots better match the rendering of real task scenarios, single-DOF haptic robots show great potential for home use with enhanced task rendering performance. This study presents our attempts to optimize and develop a single-DOF cable-driven robot with appropriate workspace and force rendering capacity. The core technologies consist of two aspects: 1) a multi-objective optimization method was adopted to obtain optimal configuration of the haptic robot; and 2) a slider-crank-mechanism-based portable cable-driven robot was developed. Performance evaluation experiments demonstrated that 1) the robot has a workspace larger than 300 mm; 2) the robot can achieve 40 N force output and 40 N. mm-1stiffness for hard contact; 3) the root mean square of the resistance during free motion is 0.93 N; 4) in the purely passive case (without motor compensation), the average resistance to back drive the motor is 2.5 N. These lead us to believe that the developed robot holds the promise to serve as a robotic rehabilitation training platform for home use on the neurological-impaired patients.

Continuous motion prediction plays a significant role in realizing seamless control of robotic exoskeletons and orthoses. Explicitly modeling the relationship between coordinated muscle activations from surface electromyography (sEMG) and human limb movements provides a new path of sEMG-based human–machine interface. Instead of the numeric features from individual channels, we propose a muscle synergy-driven adaptive network-based fuzzy inference system (ANFIS) approach to predict continuous knee joint movements, in which muscle synergy reflects the motor control information to coordinate muscle activations for performing movements. Four human subjects participated in the experiment while walking at five types of speed: 2.0 km/h, 2.5 km/h, 3.0 km/h, 3.5 km/h, and 4.0 km/h. The study finds that the acquired muscle synergies associate the muscle activations with human joint movements in a low-dimensional space and have been further utilized for predicting knee joint angles. The proposed approach outperformed commonly used numeric features from individual sEMG channels with an average correlation coefficient of 0.92$ \pm $0.05. Results suggest that the correlation between muscle activations and knee joint movements is captured by the muscle synergy-driven ANFIS model and can be utilized for the estimation of continuous joint angles.

This article presents a linear active disturbance rejection control (LADRC) scheme for two-mass systems (TMSs) based on a singular perturbation (SP) approach. In the proposed scheme, the dynamics of TMSs are separated into a quasi-steady-state system at a slow time scale and a boundary-layer system at a fast time scale via the SP method. The quasi-steady-state model is used as the nominal model to design the LADRC, where internal model control (IMC) rules are employed for prescribed tracking performance. This not only reduces the model order but also makes the gain tuning straightforward, which is crucial for practical use. Based on the boundary-layer model, the active damping method is designed for vibration suppression, which aims at achieving a higher stable bandwidth of the IMC to better attenuate the residual disturbance of linear extended state observer such that the tracking accuracy can be improved. Stability of the full system was analyzed via the extended Tikhonov's theorem. Comparative experiments were carried out to verify the effectiveness of the proposed scheme.

While many voluntary movements involve bimanual coordination, few attempts have been made to simultaneously decode the trajectory of bimanual movements from electroencephalogram (EEG) signals. In this study, we proposed a novel bimanual brain-computer interface (BCI) paradigm to reconstruct the continuous trajectory of both hands during coordinated movements from EEG. The protocol required human subjects to complete a bimanual reaching task to the left, middle, or right target while EEG data were collected. A multi-task deep learning model combining the EEGNet and long short-term memory network (LSTM) was proposed to decode bimanual trajectories, including position and velocity. Decoding performance was evaluated in terms of the correlation coefficient (CC) and normalized root mean square error (NRMSE) between decoded and real trajectories. Experimental results from 13 human subjects showed that the grand-averaged combined CC values achieved 0.54 and 0.42 for position and velocity decoding, respectively. The corresponding combined NRMSE values were 0.22 and 0.23. Both CC and NRMSE were significantly superior to the chance level (p<0.05). Comparative experiments also indicated that the proposed model significantly outperformed some other commonly-used methods in terms of CC and NRMSE for continuous trajectory decoding. These findings demonstrated the feasibility of simultaneously decoding bimanual trajectory from EEG, indicating the potential of bimanual control for coordinated tasks.

Physically-coupled bimanual tasks (activities where a force effect occurs between two human limbs) involve the coordination and cooperation of bilateral arms. Such uncertain contribution of two arms is often studied under static configuration, which is not sufficient to typify all activities of daily life (ADLs). This study aims to investigate people's bilateral force production and control in dynamic tasks. Experiments were conducted with a customized robotic system that is characterized with two handles and programmable force fields between them. Fourteen healthy right-handed human volunteers were instructed to generate force with each hand when performing predefined trajectory tracking tasks, in which the sum of forces contributed by the left and the right hand is required to equal a target force. Significant asymmetry was found in the force output between bilateral hands. With the homologous muscles activated synchronously, the contribution of the left hand was larger, while when the non-homogenous muscles were activated synchronously, the laterality was subject to the moving direction. In addition, when considering the force difference between two hands in terms of direction and magnitude, the former decreased with the increase of the target force, but the latter was more sensitive to moving directions. The results reveal the unique characteristics of non-isometric force control tasks compared with isometric ones.

Parallel mechanisms (PMs) are commonly used for developing haptic devices due to low inertia, high rigidity and precision. However, limited workspace impedes their application for task-oriented robotic therapy which generally requires large motion ranges. To solve this problem, first, a PM- based reconfigurable asymmetric 6-DOF haptic interface was presented, and then a two-stage optimization method was proposed to make the robot implement two kinds of task-specific workspaces including gross motor tasks (GMTs) and fine motor tasks (FMTs). Optimization of this robot was conducted to pursue a compact size and high accuracy. The global conditioning index (GCI) and the occupied area of the robot were selected as the evaluation indices, where the GCI was derived using a dimensionally homogeneous Jacobian matrix. A multi-objective optimization method based on the genetic algorithm (GA) was utilized. The actual design parameters were finally defined from solutions of the Pareto front. The proposed two-stage optimization method provides a feasible solution for determining task-specific robotic workspace of the reconfigurable mechanism.

Advancement in healthcare solutions is one of the major success stories of our times. While biomedical science and engineering has progressed significantly, increased human longevity has created huge demands on the healthcare system, such as the workforce that is struggling to meet the needs of patients. With the advent of healthcare automation, doctor offices, hospitals, and other medical facilities have streamlined the way they deal with patients with enhanced efficiency.

Electric motors have been widely used as the actuators of robot and automation systems. This paper aims at achieving the high-precision position control of motor drive systems. For this purpose, a robust control scheme is presented by combining the internal model principle, the sliding mode technique and the extended state observer (ESO). The PID-type controller is firstly designed by using the internal model control (IMC) rules. Since the analysis of the IMC system is performed via a sliding surface, a robust sliding mode control (SMC) law is then synthesized to enhance the control ability of the system to uncertainties. However, this robust solution should make a trade-off between the chattering attenuation and the control accuracy. To handle this drawback, a linear ESO is employed to compensate the modeling errors for a higher control accuracy. The stability analysis is provided via a Lyapunov-based method, and the superiority of the proposed approach was validated by comparative experiments on a motor drive platform.

A previously developed pneumatic muscles' (PMs) actuated gait exoskeleton (with only knee joint) has been demonstrated in achieving appropriate actuation torque, range of motion (ROM), and control bandwidth for task-specific gait training. While the adopted multi-input-multi-output (MIMO) sliding mode (SM) strategy has preliminarily implemented simultaneous control of the exoskeleton's angular trajectory and compliance, its efficacy with human users during gait cycles has not been investigated. This article presents an improved bi-joint gait rehabilitation exoskeleton (BiGREX) with integrated human hip and knee joints. The results with 12 healthy subjects demonstrated that the system's compliance can be effectively adjusted while guiding the subjects walking in predefined trajectories.