Advanced myoelectric prostheses feature multiple degrees of freedom (DoFs) and sophisticated control algorithms that interpret user motor intentions as commands. While enhancing their capability to assist users in a wide range of daily activities, these control solutions still pose challenges. Among them, the need for extensive learning periods and users’ limited control proficiency. To investigate the relationship between these challenges and the limited alignment of such methods with human motor control strategies, we examine motor learning processes in two different control maps testing a synergistic myoelectric system. In particular, this work employs a DoF-wise synergies control algorithm tested in both intuitive and non-intuitive control mappings. Intuitive mapping aligns body movements with control actions to replicate natural limb control, whereas non-intuitive mapping (or non-biomimetic) lacks a direct correlation between aspects, allowing one body movement to influence multiple DoFs. The latter offers increased design flexibility through redundancy, which can be especially advantageous for individuals with motor disabilities. The study evaluates the effectiveness and learning process of both control mappings with 10 able-bodied participants. The results revealed distinct patterns observed while testing the two maps. Furthermore, muscle synergies exhibited greater stability and distinction by the end of the experiment, indicative of varied learning processes.

We have developed an interactive system comprising a soft wearable robot hand and a wireless task board, facilitating the interaction between the hand and regular daily objects for task-oriented training in stroke rehabilitation. A ring-reinforced soft actuator (RSA) to accommodate different hand sizes and enable flexion and extension movements was introduced in this paper. Individually controlled finger actuators assist stroke patients during various grasping tasks. A wireless task board was developed to support the training, allowing for the placement of training objects and seamless interaction with the soft robotic hand. Evaluation with seven stroke subjects shows significant improvements in upper limb functions (FMA), hand-motor abilities (ARAT, BBT), and maximum grip strengths after 20 sessions of this task-oriented training. These improvements were observed to persist for at least 3 months post-training. The results demonstrate its potential to enhance stroke rehabilitation and promote hand-motor recovery. This lightweight, user-friendly interactive system facilitates frequent hand practice and easily integrates into regular rehabilitation therapy routines.

Research in lower limb wearable robotic control has largely focused on reducing the metabolic cost of walking or compensating for a portion of the biological joint torque, for example, by applying support proportional to estimated biological joint torques. However, due to different musculotendon unit (MTU) contractile speed properties, less attention has been given to the development of wearable robotic controllers that can steer MTU dynamics directly. Therefore, closed-loop control of MTU dynamics needs to be robust across fiber phenotypes, that is ranging from slow type I to fast type IIx in humans. The ability to perform closed-loop control the in-vivo dynamics of MTUs could lead to a new class of wearable robots that can provide precise support to targeted MTUs for preventing onset of injury or providing precision rehabilitation to selected damaged tissues. In this paper, we introduce a novel closed-loop control framework that utilizes nonlinear model predictive control to keep the peak Achilles tendon force within predetermined boundaries during diverse range of cyclic force production simulations in the human ankle plantarflexors. This control framework employs a computationally efficient model comprising a modified Hill-type MTU contraction dynamics component and a model of the ankle joint with parallel actuation. Results indicate that the closed-form muscle-actuation model’s computational time is in the order of microseconds and is robust to different muscle contraction velocity properties. Furthermore, the controller achieves tendon force control within a time frame below $ 18\mathrm{ms} $, aligning with the physiological electromechanical delay of the MTU and facilitating its potential for future real-world applications.

Eye gaze tracking is increasingly popular due to improved technology and availability. In the domain of assistive device control, however, eye gaze tracking is often used in discrete ways (e.g., activating buttons on a screen), and does not harness the full potential of the gaze signal. In this article, we present a method for collecting both reactionary and controlled eye gaze signals, via screen-based tasks designed to isolate various types of eye movements. The resulting data allows us to build an individualized characterization for eye gaze interface use. Results from a study conducted with participants with motor impairments are presented, offering insights into maximizing the potential of eye gaze for assistive device control. Importantly, we demonstrate the potential for incorporating direct continuous eye gaze inputs into gaze-based interface designs; generally seen as intractable due to the ‘Midas touch’ problem of differentiating between gaze movements for perception versus for interface operation. Our key insight is to make use of an individualized measure of smooth pursuit characteristics to differentiate between gaze for control and gaze for environment scanning. We also present results relating to gaze-based metrics for mental workload and show the potential for the concurrent use of eye gaze for control input as well as assessing a user’s mental workload both offline and in real-time. These findings might inform the development of continuous control paradigms using eye gaze, as well as the use of eye tracking as the sole input modality to systems that share control between human-generated and autonomy-generated inputs.

This work introduces a real-time intention decoding algorithm grounded in muscle synergies (Syn-ID). The algorithm detects the electromyographic (EMG) onset and infers the direction of the movement during reaching tasks to control a powered shoulder–elbow exoskeleton. Features related to muscle synergies are used in a Gaussian Mixture Model and probability accumulation-based logic to infer the user’s movement direction. The performance of the algorithm was verified by a feasibility study including eight healthy participants. The experiments comprised a transparent session, during which the exoskeleton did not provide any assistance, and an assistive session in which the Syn-ID strategy was employed. Participants were asked to reach eight targets equally spaced on a circumference of 25 cm radius (adjusted chance level: 18.1%). The results showed an average accuracy of 48.7% after 0.6 s from the EMG onset. Most of the confusion of the estimate was found along directions adjacent to the actual one (type 1 error: 33.4%). Effects of the assistance were observed in a statistically significant reduction in the activation of Posterior Deltoid and Triceps Brachii. The final positions of the movements during the assistive session were on average 1.42 cm far from the expected ones, both when the directions were estimated correctly and when type 1 errors occurred. Therefore, combining accurate estimates with type 1 errors, we computed a modified accuracy of 82.10±6.34%. Results were benchmarked with respect to a purely kinematics-based approach. The Syn-ID showed better performance in the first portion of the movement (0.14 s after EMG onset).

Advancements in wearable robots aim to improve user motion, motor control, and overall experience by minimizing energetic cost (EC). However, EC is challenging to measure and it is typically indirectly estimated through respiratory gas analysis. This study introduces a novel EMG-based objective function that captures individuals’ natural energetic expenditure during walking. The objective function combines information from electromyography (EMG) variables such as intensity and muscle synergies. First, we demonstrate the similarity of the proposed objective function, calculated offline, to the EC during walking. Second, we minimize and validate the EMG-based objective function using an online Bayesian optimization algorithm. The walking step frequency is chosen as the parameter to optimize in both offline and online approaches in order to simplify experiments and facilitate comparisons with related research. Compared to existing studies that use EC as the objective function, results demonstrated that the optimization of the presented objective function reduced the number of iterations and, when compared with gradient descent optimization strategies, also reduced convergence time. Moreover, the algorithm effectively converges toward an optimal step frequency near the user’s preferred frequency, positively influencing EC reduction. The good correlation between the estimated objective function and measured EC highlights its consistency and reliability. Thus, the proposed objective function could potentially optimize lower limb exoskeleton assistance and improve user performance and human–robot interaction without the need for challenging respiratory gas measurements.

The muscular restructuring and loss of function that occurs during a transfemoral amputation surgery has a great impact on the gait and mobility of the individual. The hip of the residual limb adopts a number of functional roles that would previously be controlled by lower joints. In the absence of active plantar flexors, swing initiation must be achieved through an increased hip flexion moment. The high activity of the residual limb is a major contributor to the discomfort and fatigue experienced by individuals with transfemoral amputations during walking. In other patient populations, both passive and active hip exosuits have been shown to positively affect gait mechanics. We believe an exosuit configured to aid with hip flexion could be well applied to individuals with transfemoral amputation. In this article, we model the effects of such a device during whole-body, subject-specific kinematic simulations of level ground walking. The device is simulated for 18 individuals of K2 and K3 Medicare functional classification levels. A user-specific device profile is generated via a three-axis moment-matching optimization using an interior-point algorithm. We employ two related cost functions that reflect an active and passive form of the device. We hypothesized that the optimal device configuration would be highly variable across subjects but that variance within mobility groups would be lower. From the results, we partially accept this hypothesis, as some parameters had high variance across subjects. However, variance did not consistently trend down when dividing into mobility groups, highlighting the need for user-specific design.

Pathological tremors can often be debilitating to activities of daily living and significantly affect the quality of life. Such tremulous movements are commonly observed in wrist flexion-extension (FE). To suppress this tremor we present a wearable robot (WR) with a customized mechanical metamaterial (MM) as the physical human-robot interface (pHRI). The MM is optimized to conform to the user’s wrist posture and follow the hand’s Cartesian trajectory. This is done to minimize the shear between the pHRI and the user’s skin and consequently improve wearability. This WR is then used to effect a viscous tremor suppression using the velocity of the user’s wrist FE. We present a model for the interaction between the WR and the user with which we develop the viscous damping approach for tremor. This is then evaluated in simulation and using a dedicated test bed. This tremor suppression approach demonstrates an attenuation of 20–30 dB at various tremulous frequencies resulting in significantly lower tremor amplitudes due to the viscous damping.

Neuromuscular controllers (NMCs) offer a promising approach to adaptive and task-invariant control of exoskeletons for walking assistance, leveraging the bioinspired models based on the peripheral nervous system. This article expands on our previous development of a novel structure for NMCs with modifications to the virtual muscle model and reflex modulation strategy. The modifications consist firstly of simplifications to the Hill-type virtual muscle model, resulting in a more straightforward formulation and reduced number of parameters; and second, using a finer division of gait subphases in the reflex modulation state machine, allowing for a higher degree of control over the shape of the assistive profile. Based on the proposed general structure, we present two controller variants for hip exoskeletons, with four- and five-state reflex modulations (NMC-4 and NMC-5). We used an iterative data-driven approach with two tuning stages (i.e., muscle parameters and reflex gains) to determine the controller parameters. Biological joint torque profiles and optimal torque profiles for metabolic cost reduction were used as references for the final tuning outcome. Experimental testing under various walking conditions demonstrated the capability of both variants for adapting to the locomotion task with minimal parameter adjustments, mostly in terms of timing. Furthermore, NMC-5 exhibited better alignment with biological and optimised torque profiles in terms of timing characteristics and relative magnitudes, resulting in less negative mechanical work. These findings firstly validate the adequacy of the simplified muscle model for assistive controllers, and demonstrate the utility of a more nuanced reflex modulation in improving the assistance quality.

The human hand is an intricate anatomical structure essential for daily activities, yet replicating its full functionality in upper-limb prostheses remains a significant challenge. Despite advances in mechanical design leading to more sophisticated and dexterous artificial hands, difficulties persist in effectively controlling these prostheses due to the limitations posed by the muscle conditions of their users. These constraints result in a limited number of control inputs and a lack of sensory feedback. To address these issues, various semi-autonomous control strategies have been proposed, which integrate sensing technologies to complement traditional myoelectric control. Inspired by human grasping physiology, we propose a shared control strategy that divides grasp control into two levels: a high-level controller, operated by the user to initiate the grasp action, and a low-level controller, which ensures stability throughout the task. This work focuses specifically on slip detection methods, introducing improvements to the low-level controller to enable more autonomous grasping behavior during object holding. The proposed slip module uses distributed 3D force sensors across the artificial hand and integrates a friction cone strategy to ensure an appropriate shear-to-normal force ratio with bandpass filtering for establishing an initial stable grasp model without prior knowledge. Experimental evaluations consist of the comparison of this novel controller with conventional state-of-the-art approaches. Results demonstrate its efficacy in preventing slippage while requiring less grasping force than previous methods. Additionally, a qualitative validation was conducted to assess its responsiveness compared to human grasping reactions to unexpected weight changes, yielding positive outcomes.