User preference of applied torque characteristics for bilateral powered ankle exoskeletons

Robotic lower limb exoskeletons have been built for augmenting human performance, assisting with disabilities, studying human physiology, and re-training motor deficiencies. At the University of Michigan Human Neuromechanics Laboratory, we have built pneumatically-powered lower limb exoskeletons for the last two purposes. Most of our prior research has focused on ankle joint exoskeletons because of the large contribution from plantar flexors to the mechanical work performed during gait. One way we control the exoskeletons is with proportional myoelectric control, effectively increasing the strength of the wearer with a physiological mode of control. Healthy human subjects quickly adapt to walking with the robotic ankle exoskeletons, reducing their overall energy expenditure. Individuals with incomplete spinal cord injury have demonstrated rapid modification of muscle recruitment patterns with practice walking with the ankle exoskeletons. Evidence suggests that proportional myoelectric control may have distinct advantages over other types of control for robotic exoskeletons in basic science and rehabilitation.

The clinical efficacy of robotic rehabilitation interventions hinges on appropriate neuromuscular recruitment from the patient. The first purpose of this study was to evaluate the use of supervised machine learning techniques to predict neuromuscular recruitment of the ankle plantar flexors during walking with ankle exoskeleton resistance in individuals with cerebral palsy (CP). The second goal of this study was to utilize the predictive models of plantar flexor recruitment in the design of a personalized biofeedback framework intended to improve (i.e., increase) user engagement when walking with resistance. First, we developed and trained multilayer perceptrons (MLPs), a type of artificial neural network (ANN), utilizing features extracted exclusively from the exoskeleton's onboard sensors, and demonstrated 85-87% accuracy, on average, in predicting muscle recruitment from electromyography measurements. Next, our participants completed a gait training session while receiving audio-visual biofeedback of their personalized real-time planar flexor recruitment predictions from the online MLP. We found that adding biofeedback to resistance elevated plantar flexor recruitment by 24 16% compared to resistance alone. This study highlights the potential for online machine learning frameworks to improve the effectiveness and delivery of robotic rehabilitation systems in clinical populations.

Robotic exoskeletons have emerged as effective rehabilitation and ability-enhancement tools, by mimicking or supporting natural body movements. The control schemes of exoskeletons are conventionally developed based on fixed torque-ankle state relationship or various human models, which are often lack of flexibility and adaptability to accurately address personalized movement assistance needs. This paper presents an adaptive control strategy for personalized robotic ankle exoskeleton in an effort to address this limitation. The adaptation was implemented by applying the experience-based fuzzy rule interpolation approach with the support of a muscle-tendon complex model. In particular, this control system is initialized based on the most common requirements of a “standard human model,” which is then evolved during its performance by effectively using the feedback collected from the wearer to support different body shapes and assistance needs. The experimental results based on different human models with various support demands demonstrate the power of the proposed control system in improving the adaptability, and thus applicability, of robotic ankle exoskeletons.

Ankle exoskeletons have garnered considerable interest for their potential to enhance mobility, support rehabilitation, and reduce fall risks, particularly among the aging population. Their effectiveness depends on accurate, real-time prediction of user intentions from wearable sensor data, as even small delays or errors can compromise stability and safety. Here, we present a motion classification framework that integrates three Inertial Measurement Units (IMUs) with eight surface Electromyography (sEMG) sensors fabricated as towel-based textile electrodes, which improve comfort, durability, and usability for long-term deployment compared to traditional gel electrodes. The dataset comprises multichannel time-series recordings of five functional daily motions, enabling a realistic evaluation of exoskeleton use in everyday environments. Using this framework, Convolutional Neural Networks (CNNs) achieved an accuracy of 99.263 pm 0.26%, substantially surpassing previously reported results in the field. Beyond overall accuracy, we address deployment-critical requirements: transfer learning enables reliable adaptation to new users with as few as ten calibration samples per motion, while robustness testing demonstrates that the system continues to provide stable classification even when individual sensors are disrupted. Together, these results highlight the feasibility of safe, high-accuracy, and real-world-ready exoskeleton control through deep learning combined with wearable textile electrodes and IMUs.

Lower limb exoskeleton robots help with walking movements through mechanical force, by identifying the wearer’s walking intention. When the exoskeleton robot is lightweight and comfortable to wear, the stability of walking increases, and energy can be used efficiently. However, because it is difficult to implement the complex anatomical movements of the human body, most are designed simply. Due to this, misalignment between the human and robot movement causes the wearer to feel uncomfortable, and the stability of walking is reduced. In this paper, we developed a two degrees of freedom (2DoF) ankle exoskeleton robot with a subtalar joint and a talocrural joint, applying a four-bar linkage to realize the anatomical movement of a simple 1DoF structure mainly used for ankles. However, bidirectional tendon-driven actuators (BTDAs) do not consider the difference in a length change of both cables due to dorsiflexion (DF) and plantar flexion (PF) during walking, causing misalignment. To solve this problem, a BTDA was developed by considering the length change of both cables. Cable-driven actuators and exoskeleton robot systems create uncertainty. Accordingly, adaptive control was performed with a proportional-integral-differential neural network (PIDNN) controller to minimize system uncertainty.

Experiments have shown that elastic ankle exoskeletons can be used to reduce ankle joint and plantar-flexor muscle loading when hopping in place and, in turn, reduce metabolic energy consumption. However, recent experimental work has shown that such exoskeletons cause less favourable soleus (SO) muscle-tendon mechanics than is observed during normal hopping, which might limit the capacity of the exoskeleton to reduce energy consumption. To directly link plantar-flexor mechanics and energy consumption when hopping in exoskeletons, we used a musculoskeletal model of the human leg and a model of muscle energetics in simulations of muscle-tendon dynamics during hopping with and without elastic ankle exoskeletons. Simulations were driven by experimental electromyograms, joint kinematics and exoskeleton torque taken from previously published data. The data were from seven males who hopped at 2.5 Hz with and without elastic ankle exoskeletons. The energetics model showed that the total rate of metabolic energy consumption by ankle muscles was not significantly reduced by an ankle exoskeleton. This was despite large reductions in plantar-flexor force production (40-50%). The lack of larger metabolic reductions with exoskeletons was attributed to increases in plantar-flexor muscle fibre velocities and a shift to less favourable muscle fibre lengths during active force production. This limited the capacity for plantar-flexors to reduce activation and energy consumption when hopping with exoskeleton assistance.

Overground Robotic Exoskeleton Training for Patients With Stroke on Walking-Related Outcomes: A Systematic Review and Meta-analysis of Randomized Controlled Trials

Background and Objectives: Analysis of gait parameters and energy expenditure in different walking speeds in trans-tibial amputees has been less investigated. This may provide an insight into how they respond to it. The main aim of this study was to investigate the effect of walking speed changes on gait parameters and metabolic energy consumption in unilateral trans-tibial amputees using the prosthesis. Methods: Thirty trans-tibial amputees fulfilling inclusion criteria were investigated for gait parameters using a ten-meter walk test while these subjects walked with three different walking speeds (comfortable, fast, and slow). Temporal-spatial gait parameters, such as step length, stride length, velocity and cadence, and Physiological Cost Index (PCI) for metabolic energy consumption were analyzed by repeated-measures ANOVA. Results: Step length of a prosthetic side in the fast walking speed (0.67±0.10 m) was higher than the normal comfortable speed (0.56±0.13 m) and least in slow walking (0.42±0.06 m). A similar trend was observed for all gait parameters. PCI (beats/m) was least in normal comfortable speed (0.08±0.09), followed by fast walking speed (0.11±0.08). It was highest when patients walked at a slow speed (0.18±0.12). There was a significant difference in all temporal-spatial gait parameters and PCI in three walking speed conditions (p<0.05). Conclusion: Walking speed significantly affected all temporal-spatial and energy parameters in unilateral trans-tibial amputees.

The aim of this pilot study was to investigate the feasibility of high-speed gait training with an exoskeleton robot hybrid assistive limb (HAL) in patients with chronic stroke, and to examine the efficacy of eight sessions (8 weeks) of gait training with a HAL compared with conventional physical therapy. Eighteen patients with chronic stroke were included in this study (nine each in the HAL and control groups). The HAL group underwent high-speed gait training with the HAL once a week for 8 weeks (20 min/session). The control group underwent conventional physical therapy for gait disturbance. Outcome measures were walking speed, number of steps, and cadence during a 10 m walking test, a timed up and go test, a functional reach test, and the Berg Balance Scale. Assessments were performed in the absence of the HAL before training and after the fourth and eighth training sessions. All patients in the HAL group completed the high-speed gait training without adverse events. The HAL group improved significantly in walking speed (55.9% increase, P<0.001), number of steps (17.6% decrease, P<0.01), and cadence (32.8% increase, P<0.001) during the 10 m walking test. The patients also exhibited significant improvements in the timed up and go test, the functional reach test, and the Berg Balance Scale after HAL training (P<0.01 in all). No statistical time-dependent changes were observed in any parameter in the control group. For chronic stroke patients, high-speed gait training with a HAL appears to be feasible and effective in improving gait and balance dysfunction despite the limitations of this nonrandomized pilot study.

Background and Purpose- A quarter of individuals who experience a stroke are under the age of 65 years (defined as young adults), and up to 44% will be unable to return to work poststroke, predominantly because of walking difficulties. No research study has comprehensively analyzed walking performance in young adult's poststroke. The primary aim of this study is to investigate how a stroke in young adults affects walking performance (eg, walking speed and metabolic cost) compared with healthy age-matched controls. The secondary aim is to determine the predictive ability of walking performance parameters for return to employment poststroke. Methods- Forty-six individuals (18-40 years: n=6, 41-54 years: n=21, 55-65 years: n=19) who have had a stroke and 15 healthy age-matched able-bodied controls were recruited from 6 hospital sites in Wales, United Kingdom. Type, location, cause of stroke, and demographic factors (eg, employment status) were recorded. Temporal and spatial walking parameters were measured using 3-dimensional gait analysis. Metabolic energy expenditure and metabolic cost of walking were captured during 3 minutes of walking at self-selected speed from measurements of oxygen consumption. Results- Stroke participants walked slower (P<0.004) and less efficiently (P<0.002) than the controls. Only 23% of stroke participants returned to employment poststroke. Walking speed was the strongest predictor (sensitivity, 0.90; specificity, 0.82) for return to work (P=0.004) with a threshold of 0.93 m/s identified: individuals able to walk faster than 0.93 m/s were significantly more likely to return to work poststroke than those who walked slower than this threshold. Conclusions- This study is the first to capture walking performance parameters of young adults who have had a stroke and identifies slower and less efficient walking. Walking speed emerged as the strongest predictor for return to employment. It is recommended that walking speed be used as a simple but sensitive clinical indicator of functional performance to guide rehabilitation and inform readiness for return to work poststroke.

The relationship of hip strength to walking and balance performance in unilateral lower limb prosthesis users differs by amputation level.

Temporal variability in the phase durations during treadmill walking

BackgroundResearch efforts in neurorehabilitation technologies have been directed towards creating robotic exoskeletons to restore motor function in impaired individuals. However, despite advances in mechatronics and bioelectrical signal processing, current robotic exoskeletons have had only modest clinical impact. A major limitation is the inability to enable exoskeleton voluntary control in neurologically impaired individuals. This hinders the possibility of optimally inducing the activity-driven neuroplastic changes that are required for recovery.MethodsWe have developed a patient-specific computational model of the human musculoskeletal system controlled via neural surrogates, i.e., electromyography-derived neural activations to muscles. The electromyography-driven musculoskeletal model was synthesized into a human-machine interface (HMI) that enabled poststroke and incomplete spinal cord injury patients to voluntarily control multiple joints in a multifunctional robotic exoskeleton in real time.ResultsWe demonstrated patients’ control accuracy across a wide range of lower-extremity motor tasks. Remarkably, an increased level of exoskeleton assistance always resulted in a reduction in both amplitude and variability in muscle activations as well as in the mechanical moments required to perform a motor task. Since small discrepancies in onset time between human limb movement and that of the parallel exoskeleton would potentially increase human neuromuscular effort, these results demonstrate that the developed HMI precisely synchronizes the device actuation with residual voluntary muscle contraction capacity in neurologically impaired patients.ConclusionsContinuous voluntary control of robotic exoskeletons (i.e. event-free and task-independent) has never been demonstrated before in populations with paretic and spastic-like muscle activity, such as those investigated in this study. Our proposed methodology may open new avenues for harnessing residual neuromuscular function in neurologically impaired individuals via symbiotic wearable robots.

This study systematically compares two personalized control strategies for ankle exoskeletons: human-in-the-loop (HIL) optimized torque profiles and stride-wise adaptive torque control. Through biomechanical analysis of 11 healthy participants walking at 1.50 m/s, we evaluated these approaches across multiple performance metrics. Both strategies demonstrated comparable effectiveness in reducing soleus (SOL) muscle activity (36.8% versus 35.8%) and biological ankle torque (34.1% versus 36.6%), with no significant differences observed at the group level (p > 0.05). However, the optimized torque mode showed 22.4% higher energy efficiency (1.189 versus 1.793 J/kg reduction in muscle activity) and induced less compensatory tibialis anterior (TA) activation (p < 0.05). While gait symmetry remained unimpaired (p > 0.05) under both modes, individual responses varied substantially, particularly in adaptation to dynamic parameter adjustments. These findings indicate that pre-optimized fixed parameters are preferable for stable walking conditions, whereas adaptive control requires further refinement for dynamic environments, providing empirical guidance for selecting ankle exoskeleton control strategies based on application needs.

Supported Treadmill Ambulation for Amyotrophic Lateral Sclerosis: A Pilot Study