Myoelectric control paradigms have the potential to enable continuous volitional control of bionic limbs in various movement conditions. Although individuals with below knee amputations and an agonist-antagonist muscle interface (AMI) were proven to display a greater degree of continuous volitional control in bionic ankle-foot systems with respect to conventional socket-suspended prosthetic users, it remains unclear how myoelectric interfaces could translate to non-AMI prosthetic users with bone-anchored prostheses (BAP). This preliminary study proposes a human-machine interface (HMI) based on a neuromechanical model to enable volitional, continuous myoelectric control of a bionic leg in AMI and BAP users, walking across various speeds and ground inclinations. Differently from state of the art solutions, the proposed HMI is based on a digital twin of the intact leg, synthesizing the user's phantom limb musculoskeletal function as controlled by muscle activations measured from the residuum. When embedded in a real-time framework, it enabled the participants to achieve volitional modulation of prosthesis peak plantar-dorsiflexion torques timing and amplitude during overground walking at three speeds (between 1.6 and 3.96 km/h), with case studies provided during calf-raises (30, 45, and 60 bpm) and ramp ascent walking (3 and 5% incline). Before prosthesis control tests, the participants underwent a 2-day gait training session. Results showed that all three subjects learned how to alter initial muscle activation patterns so that an average of 87% of peak activation timing fell within target ranges. The proposed neuromechanical modeling technology opens new avenues toward generalizable HMIs for the volitional control of active prostheses beyond set conditions and amputation types.

The goal of this paper is to design spatial (3D) robust reference trajectories for lower limb exoskeletons through optimization. Robustness is defined as the magnitude of the smallest force applied at the Center of Mass that cannot be rejected without violating joint torque or ground reaction force bounds. A larger robustness margin in the reference trajectory increases the capability of the controller to reject perturbations. A trajectory optimization problem was augmented by adding the maximization of this robustness metric to it. The augmented trajectory optimization is then employed to design dynamic trajectories for three different tasks. We demonstrated that our method increases the robustness metric value by comparing the results from the proposed method with a nominal trajectory optimization that does not take robustness into account. Moreover, the designed trajectories were implemented on a lower limb exoskeleton with a paraplegic user to show the feasibility of the proposed method. The user managed successfully to walk on level ground, perform side-stepping, and ascend stairs.

Exoskeleton balance support can improve the ability to counteract perturbations. The process of human adaptation to this support, however, remains unclear. Here, we assessed how able-bodied individuals adapted to balance support provided by an ankle exoskeleton during walking, specifically when counteracting forward-directed pushes at the pelvis. Activation of the balance support led to immediate and clear reductions in both Center Of Mass (COM) displacement and soleus EMG activity. Further adaptations were observed across the first 35 perturbations for COM displacement and only across the first 5 perturbations for EMG activity before reaching a stable value. These findings demonstrate that adaptation to balance support is a rapid process. These results indicate that minimal training time is required for an individual to effectively utilize exoskeleton balance support.

Background/Objectives: “Hermes” is an ankle–foot orthosis (AFO) with negative stiffness designed to mechanically compensate the symptomatic increase in plantarflexion (PF) torque (i.e., ankle joint torque resistance to dorsiflexion, DF) in patients with spastic paresis. Methods: The effectiveness of “Hermes” was evaluated in twelve patients with chronic unilateral spastic paresis after stroke. Using a robotic ankle manipulator, stiffness at the ankle joint was assessed across three conditions: ankle without Hermes (A), ankle with Hermes applying no torque compensation (A+H0%), and ankle with Hermes tuned to compensate 100% of the patients’ ankle joint stiffness (A+H100%). Results: A significant reduction in PF torque was found with Hermes applying compensation (A+H100%) compared to the conditions without Hermes (A) and with Hermes applying no compensation (A+H0%). Furthermore, a significant reduction in positive dorsiflexion work was found with Hermes applying compensation (A+H100%) compared to the condition with Hermes applying no compensation (A+H0%). Hermes did not significantly contribute to additional PF torque or positive work when applying no compensation (A+H0%). Conclusions: The reductions in PF torque achieved with Hermes are comparable to those seen with repeated ankle stretching programs and ankle robot training. Thus, Hermes is expected to assist voluntary dorsiflexion and improve walking in patients with spastic paresis.

A small-scale electro-hydrostatic actuator, termed miniHydrA, was developed based on biomechanical requirements for gait and integrated into an ankle exoskeleton. The key advantage of this actuator concept lies in its compact size and the low mass of its output stage, combined with the ability to deliver high support torques, sufficient for full human assistance. During development, hydraulic cylinder leakage and friction were identified as key challenges. To address control requirements, a dedicated control strategy was proposed and implemented. The prototype exoskeleton was evaluated for joint torque tracking performance across a range of torques (0–120 Nm), both in benchtop tests and during treadmill walking trials. In benchtop experiments, zero-torque tracking was achieved with a mean absolute error ranging from 0.03 to 2.26 Nm across frequencies from 0 to 5 Hz. During treadmill walking, torque tracking errors ranged from 0.70 to 0.95 Nm, with no observable deviations in ankle joint kinematics among the three test subjects. These results show the feasibility of the miniHydrA for remote actuation. Compared to Bowden cables, commonly used in exoskeletons and exosuits, the proposed actuator concept offers two key advantages: it is better suited for high-torque applications, and its friction characteristics can be more accurately predicted and modeled, enabling more effective feedforward control.

Therapeutic and assistive exoskeletons and exosuits show promise in both clinical and real-world settings. Improving their autonomy can enhance usability, effectiveness, and cost efficiency. This Review presents a generic control framework for autonomous operation of upper and lower limb devices and reviews current advancements and future directions. We highlight how data-driven machine learning aids in intention recognition, synchronization, patient assessment, and task-agnostic control. In addition, we discuss how reinforcement learning optimizes control policies through digital human twins and how generative AI supports therapy planning and patient engagement. Richer patient-specific data and more accurate digital twins are needed for clinical validation and widespread deployment.

An active dual-joint back-support exoskeleton with motors at both lumbar and hip level was designed to reduce spinal musculoskeletal loading and preserve lumbar flexibility during lifting. A subject-specific controller estimated the moment actively generated by back muscles to counteract gravitational forces on the upper body, minimising a counter-productive abdominal muscle contraction. Eight subjects lifted a 15 kg load using free technique with four assistance levels, i.e. 0%, 30%, 50%, and 70% of the active moment. Time-averaged L5S1 compressive force and back muscle active moment estimated by an EMG-driven biomechanical model, decreased by 5.5-9.3% and 14.9-28.6%, respectively, with non-zero assistance. Higher assistance did not yield larger L5S1 compression reduction but did gain further reduction in the time-averaged back muscles active moment. No significant changes in abdominal muscle activity and minor changes in lumbar flexion were observed suggesting the controller and dual-joint design achieved their objectives.

Characterization of lower limb joint impedance during locomotion is crucial for advancing our understanding of walking biomechanics, while also guiding the development of assistive technologies and rehabilitation strategies. Identification of impedance parameters is dependent on perturbation strategies which elicit measurable responses at the joint of interest. However current perturbation methodologies applied during locomotion either work in the swing phase or employ exoskeleton based setups which can add complexity and weight to the leg. This study introduces and evaluates a perturbation methodology using a custom-designed pusher system to elicit multi-joint-level torque and angular responses during the stance phase of gait for different conditions. Controlled perturbations applied behind the tibia revealed measurable dynamic response at the knee and ankle joint, while preserving the gait cycle's overall pattern. Torque-angle and torque-velocity graphs reveal insights into energy and power changes between perturbed and unperturbed gait cycles. The method can be adapted to various walking speeds, perturbation intensities, and durations, providing the first step for future estimation of joint impedance during locomotion.

Identifying kinematic constraints between a robot and its environment can improve autonomous task execution, for example, in Learning from Demonstration. Constraint identification methods in the literature often require specific prior constraint models, geometry or noise estimates, or force measurements. Because such specific prior information or measurements are not always available, we propose a versatile kinematics-only method. We identify constraints using constraint reference frames, which are attached to a robot or ground body and may have zero-velocity constraints along their axes. Given measured kinematics, constraint frames are identified by minimizing a norm on the Cartesian components of the velocities expressed in that frame. Thereby, a minimal representation of the velocities is found, which represent the zero-velocity constraints we aim to find. In simulation experiments, we identified the geometry (position and orientation) of twelve different constraints including articulated contacts, polyhedral contacts, and contour following contacts. Accuracy was found to decrease linearly with sensor noise. In robot experiments, we identified constraint frames in various tasks and used them for task reproduction. Reproduction performance was similar when using our constraint identification method compared to methods from the literature. Our method can be applied to a large variety of robots in environments without prior constraint information, such as in everyday robot settings.

Additive manufacturing (AM) is a promising solution for handling the complexity of fabricating soft robots. However, the AM of hyperelastic materials is still challenging with a limited material range. Within this work, pellet-based 3D printing of very soft thermoplastic elastomers (TPEs) was explored (down to Shore Hardness 00-30). Our results show that TPEs can have similar engineering stress and maximum elongation as Ecoflex 00-10. In addition, we 3D-printed airtight thin TPE membranes (0.2-1.2 mm), which could inflate up to a stretch of 1320%. Combining the membrane's large expansion and softness with the 3D printing of hollow structures simplified the design of a bending actuator that can bend 180 degrees and reach a blocked force of 238 times its weight. In addition, by 3D printing TPE pellets and rigid filaments, the soft membrane could grasp objects by enveloping an object or as a sensorized sucker, which relied on the TPE's softness to conform to the object or act as a seal. In addition, the sucker's membrane acted as a tactile sensor to detect an object before adhesion. These results suggest the feasibility of AM of soft robots using soft TPEs and membranes as a promising material type and sensorized actuators, respectively.

State-of-the-art controllers for active back exosuits rely on body kinematics and state machines. These controllers do not continuously target the lumbosacral compression forces or adapt to unknown external loads. The use of additional contact or load detection could make such controllers more adaptive; however, it can be impractical for daily use. Here, we developed a novel neuro-mechanical model-based controller (NMBC) that uses a personalized electromyography (EMG)-driven musculoskeletal (MSK) model to estimate lumbosacral joint loading. NMBC provided adaptive, subject- and load-specific assistive forces proportional to estimates of the active part of biological joint moments through a soft back support exosuit. Without a priori information, the maximum assistive forces of the cable were modulated across weights. Simultaneously, we applied a non-adaptive, kinematic-dependent, trunk inclination-based controller (TIBC). Both NMBC and TIBC reduced the mean and peak biomechanical metrics, although not all reductions were significant. TIBC did not modulate assistance across weights. NMBC showed larger reductions of mean than peak values, significant reductions during the erect stance and the cumulative compressive loads by 21% over multiple cycles in a cohort of 10 participants. Overall, NMBC targeted mean lumbosacral compressive forces during lifting without a priori information of the load being carried. This may facilitate the adoption of non-hindering wearable robotics in real-life scenarios. As NMBC is informed by an EMG-driven MSK model, it is possible to tune the timing of NMBC-generated torque commands to the exosuit (delaying or anticipating commands with respect to biological torques) to target further reduction of peak or mean compressive forces and muscle fatigue.

Sensorized insoles provide a tool for gait studies and health monitoring during daily life. For users to accept such insoles, they need to be comfortable and lightweight. Previous research has demonstrated that sensorized insoles can estimate ground reaction forces (GRFs). However, these insoles often assemble commercial components restricting design freedom and customization. Within this work, we incorporated four 3D-printed soft foam-like sensors to sensorize an insole. To test the insoles, we had nine participants walk on an instrumented treadmill. The four sensors behaved in line with the expected change in pressure distribution during the gait cycle. A subset of this data was used to identify personalized Hammerstein–Wiener (HW) models to estimate the 3D GRFs while the others were used for validation. In addition, the identified HW models showed the best estimation performance (on average root mean squared (RMS) error 9.3%, $ {R}^2 $ =0.85 and mean absolute error (MAE) 7%) of the vertical, mediolateral, and anteroposterior GRFs, thereby showing that these sensors can estimate the resulting 3D force reasonably well. These results were comparable to or outperformed other works that used commercial force-sensing resistors with machine learning. Four participants participated in three trials over a week, which showed a decrease in estimation performance over time but stayed on average 11.35% RMS and 8.6% MAE after a week with the performance seeming consistent between days two and seven. These results show promise for using 3D-printed soft piezoresistive foam-like sensors with system identification regarding the viability for applications that require softness, lightweight, and customization such as wearable (force) sensors.

Lower limb exoskeletons (LLEs) are wearable devices that can restore the movement autonomy of paraplegic users. LLEs can restore the users’ ability to stand upright and walk. However, most of the commercially available and clinically used LLEs rely on the user maintaining balance through the use of crutches. Recent improvements in the design and control of LLEs and other legged robots allow for autonomous balance control. In this work, we implement and evaluate a momentum-based standing balance controller in the Symbitron LLE, consisting of eight active (torque-controlled) and two passive joints. We first investigate how gain tuning of the center of mass tracking control law, part of a multi-objective optimal controller, affects balancing performance. We apply pushes on different device locations while in parallel-stance, compare the response for different gains, and derive heuristic guidelines for controller tuning given the control architecture, high-level goals, and hardware limitations. Next, we show how this controller successfully prescribes joint torques to the LLE to maintain balance with a paraplegic user. The LLE can autonomously balance the user and reject mediolateral and anteroposterior pushes in the order of 60 N at hip height (and 40 N at shoulder height) while standing in parallel-stance, staggered-stance with both feet at the same height, and staggered-stance with a height difference of 0.05 m between the feet. This work presents a viable control strategy for torque-controlled light-weight under-actuated LLEs to keep the balance of paraplegic users during stance, which is a necessary starting point towards autonomous balance control during gait.

Every year, millions of people suffer from stroke, spinal cord injury, traumatic brain injury or other neurological or traumatic conditions that affect the peroneal nerve's motor pathway and lead to the weakness of the Tibialis Anterior and the inability to dorsiflex the foot during swing, a condition called Drop Foot. Nowadays, some Functional Electrical Stimulation devices are available commercially to cope with this condition but rely on an open-loop architecture. Closed-loop systems have only been presented in the academia with a time resolution of one step and some robustness issues. Our goal was then to tackle some robustness problems in part by increasing this time resolution to one third of the step. To do that the presented controller follows an Iterative Learning Control strategy that uses the ankle angle to correct the stimulation profile, and uses a three-part division of the swing phase to time-scale and correct this profile, avoiding some instability issues that arise from variations in the gait pattern. The algorithm was implemented and tested in a computational model of the human neuromusculoskeletal system and achieved Root Mean Square Error values of around 2 degrees being able to outperform the previously presented controller in speed of convergence and robustness.

Sensorized insoles provide a tool for gait studies and health monitoring during daily life. For users to accept such insoles they need to be comfortable and lightweight. Previous work has already demonstrated that estimation of ground reaction forces (GRFs) is possible with insoles. However, these are often assemblies of commercial components restricting design freedom and customization. Within this work, we investigate using four 3D-printed soft foam-like sensors to sensorize an insole. These sensors were combined with system identification of Hammerstein-Wiener models to estimate the 3D GRFs, which were compared to values from an instrumented treadmill as the golden standard. It was observed that the four sensors behaved in line with the expected change in pressure distribution during the gait cycle. In addition, the identified (personalized) Hammerstein-Wiener models showed the best estimation performance (on average RMS error 9.3%, R^2=0.85 and mean absolute error (MAE) 7%) of the vertical, mediolateral, and anteroposterior GRFs. Thereby showing that these sensors can estimate the resulting 3D force reasonably well. These results for nine participants were comparable to or outperformed other works that used commercial FSRs with machine learning. The identified models did decrease in estimation performance over time but stayed on average 11.35% RMS and 8.6% MAE after a week with the Hammerstein-Wiener model seeming consistent between days two and seven. These results show promise for using 3D-printed soft piezoresistive foam-like sensors with system identification to be a viable approach for applications that require softness, lightweight, and customization such as wearable (force) sensors.

While ankle exoskeleton research has mainly focused on reducing metabolic energy consumption, recent studies also aimed to support walking balance which is also important for individuals with mobility impairments that typically have an increased fall incidence. Balance supporting controllers have been designed by mimicking the balance recovery action of healthy subjects. In a previous study we designed and evaluated an exoskeleton controller that mimics the complete ankle function during balance recovery, including the contribution of muscle mechanical impedance, local feedback (e.g. stretch reflex) and global balance recovery responses based on sensory integration. While this controller successfully assisted walking balance, it is unclear if assisting muscle mechanical impedance and local feedback with an exoskeleton is beneficial, as exoskeleton assistance influences muscle activity and hence the mechanical impedance of muscle fibers. Here we developed a biomimetic exoskeleton controller that aims to assists only the slow global balance recovery response without counteracting the contribution of intrinsic mechanical impedance of muscle fibers and local feedback. Across 10 subjects, the neuromuscular controller improved balance support after perturbation compared to a minimal impedance controller. Muscle activity is reduced in response to push and pull perturbation by respectively 7.5% (p=0.06) and 19% (p<0.01) with respect to the minimal impedance controller.

BACKGROUND:Shoulder orthoses reduce the gravitational pull on the shoulder by providing an upward force to the arm, which can decrease shoulder pain caused by stress on the glenohumeral structures. OBJECTIVE:In this interventional study, the clinical effects of a recently developed dynamic shoulder orthosis were assessed in 10 patients with chronic shoulder pain. The shoulder orthosis provides an upward force to the arm with 2 elastic bands. These bands are arranged to statically balance the arm, such that the supportive force is always directed toward the glenohumeral joint and shoulder movements are not impeded. STUDY DESIGN:Clinical effect study. METHODS:The study population was provided with a dynamic shoulder orthosis for 2 weeks. In the week before the orthosis fitting, the participants had no intervention. The primary outcome measures were the mean shoulder pain scores before and during the intervention, and the distance between the humeral head and the acromion without and with orthosis. RESULTS:Ultrasound evaluation showed that the shoulder orthosis resulted in a reduction of the distance between the acromion and humeral head at different levels of arm support. In addition, it was demonstrated that the mean shoulder pain scores (range 0-10) decreased from 3.6 to 3 (in rest) and from 5.3 to 4.2 (during activities) after 2 weeks of orthosis use. In general, patients were satisfied with the weight, safety, ease in adjusting, and effectiveness of the orthosis. CONCLUSIONS:The results of this study show that the orthosis has the potential to reduce shoulder complaints in patients with chronic shoulder pain.

Background:People with Duchenne muscular dystrophy (DMD) cope with progressive muscular weakness and consequential upper extremity function loss. They benefit from arm supports, or arm exoskeletons, to assist arm function. Especially for severe muscle weakness (DMD ≥ Brooke Scale 4), the design of such arm support is challenging. This study aims to structurally develop functional and technical design requirements of arm supports for people with DMD Brooke Scale 4. Methods:An overview of clinical characteristics and a classification of clinically meaningful activities were derived from data from the Dutch Dystrophinopathy Database and available literature. Based on these, functional and technical design requirements of arm supports were developed and matched to the achievable needs of the user. Results:First, the clinical characteristics of the target population, such as strength, range of motion, and functional ability, are given. Next, clinically relevant activities of daily living are translated to functional requirements categorised in a 'must,' 'should,' and 'could' category. Last, the technical requirements to realise these functional goals are presented. Conclusions:The recommendations following from the functional user needs, technical requirements, and safety considerations can be used to make the development of assistive arm supports for people with DMD Brooke Scale 4 more user-centred.

Low-back pain is a common occupational hazard for industrial workers. Several studies show the advantages of using rigid and soft back-support passive exoskeletons and exosuits (exos) to reduce the low-back loading and risk of injury. However, benefits of using these exos have been shown to be task-specific. Therefore, in this study, we developed a benchmarking approach to assess exos for an industrial workplace at Hankamp Gears B.V. We assessed two rigid (Laevo Flex, Paexo back) and two soft (Auxivo Liftsuit 1.0, and Darwing Hakobelude) exos for tasks resembling the workplace. We measured the assistive moment provided by each exo and their respective influence on muscle activity as well as the user’s perception of comfort and exertion. Ten participants performed four lifting tasks (Static hold, Asymmetric, Squat, and Stoop), while their electromyography and subjective measures were collected. The two rigid exos provided the largest assistance during the Dynamic tasks. Reductions in erector spinae activity were seen to be task-specific, with larger reductions for the two rigid exos. Overall, Laevo Flex offered a good balance between assistive moments, reductions in muscle activity, as well as user comfort and reductions in perceived exertion. Thus, we recommend benchmarking exos for intended use in the industrial workplace. This will hopefully result in a better adoption of the back-support exoskeletons in the workplace and help reduce low-back pain.

Integration of both actuation and proprioception into the robot body leads to a single integrated system that can deform and sense. Within this work, liquid rope coiling is used to 3D‐print soft graded porous actuators. By fabricating these actuators from a conductive thermoplastic elastomer, piezoresistive sensing is directly integrated. These sensor‐integrated actuators exhibit nonlinearities and hysteresis in their resistance change. To overcome this challenge, a novel approach that uses identified Wiener–Hammerstein (WH) models is proposed to estimate the strain based on the resistance change. Three actuator types were investigated, namely, a bending actuator, a contractor, and a three degrees of freedom bending segment. By using the design freedom of additive manufacturing to set the porosity, the actuation and sensing behavior of a contracting actuator can be programmed. Furthermore, the WH models can provide strain estimation with on average high fits (83%) and low root mean square (RMS) errors (6%) for all three actuators, which outperformed linear models significantly (76.2/9.4% fit/RMS error). In these results, it is indicated that combining 3D‐printed graded porous structures and system identification can realize sensor‐integrated actuators that can estimate their strain but also tailor their behavior through the porosity.