Human-in-the-loop optimization of exoskeleton assistance during walking

A biomechanics and energetics dataset of neurotypical adults walking with and without kinematic constraints

Numerous studies have explored the biomechanics and energetics of human walking, offering valuable insights into how we walk. However, prior studies focused on changing external factors (e.g., walking speed) and examined group averages and trends rather than individual adaptations in the presence of internal constraints (e.g., injury-related muscle weakness). To address this gap, this paper presents an open dataset of human walking biomechanics and energetics collected from 21 neurotypical young adults. To investigate the effects of internal constraints (reduced joint range of motion), the participants are both the control group (free walking) and the intervention group (constrained walking - left knee fully extended using a passive orthosis). Each subject walked on a dual-belt treadmill at three speeds (0.4, 0.8, and 1.1 m/s) and five step frequencies ( - 10% to 20% of their preferred frequency) for a total of 30 test conditions. The dataset includes raw and segmented data featuring ground reaction forces, joint motion, muscle activity, and metabolic data. Additionally, a sample code is provided for basic data manipulation and visualisation.

Rethinking Exoskeleton Simulation-Based Design: The Effect of Using Different Cost Functions

Designing an exoskeleton that can improve user capabilities is a challenging task, and most designs rely on experiments to achieve this goal. A different approach is to use simulation-based designs to determine optimal device parameters. Most of these simulations use full trajectory tracking limb kinematics during a natural gait as a reference. However, exoskeletons typically change the natural gait kinematics of the user. Other types of simulations assume that human gait is optimized for a cost function that combines several objectives, such as the cost of transport, injury prevention, and stabilization. In this study, we use a 2D OpenSim model consisting of 10 degrees of freedom and considering 18 muscles, together with the Moco optimization tool, to investigate the differences between these two approaches with respect to running with a passive knee exoskeleton. Utilizing this model, we test the effect of a full trajectory tracking objective with different weights (representing the importance of the objective in the optimization cost function) and show that when using weights that are typically used in the literature, there is no deviation from the experimental data. Next, we develop a multi-objective cost function with foot clearance term based on peak knee angle during swing, that achieves trajectories similar (RMSE=7.4 deg) to experimental running data. Finally, we investigate the effect of different parameters in the design of a clutch-based passive knee exoskeleton (1.5 kg at each leg) and find that a design that utilizes a 2.5 Nm/deg spring achieves an improvement of up to 8% in net metabolic energy. Our results show that tracking objectives in the cost function, even with a low weight, hinders the simulation's ability to change the gait trajectory. Thus, there is a need for other predictive simulation methods for exoskeletons.

Experiment-free exoskeleton assistance via learning in simulation

Exoskeletons have enormous potential to improve human locomotive performance1-3. However, their development and broad dissemination are limited by the requirement for lengthy human tests and handcrafted control laws2. Here we show an experiment-free method to learn a versatile control policy in simulation. Our learning-in-simulation framework leverages dynamics-aware musculoskeletal and exoskeleton models and data-driven reinforcement learning to bridge the gap between simulation and reality without human experiments. The learned controller is deployed on a custom hip exoskeleton that automatically generates assistance across different activities with reduced metabolic rates by 24.3%, 13.1% and 15.4% for walking, running and stair climbing, respectively. Our framework may offer a generalizable and scalable strategy for the rapid development and widespread adoption of a variety of assistive robots for both able-bodied and mobility-impaired individuals.

Chimeric Forecasting: An experiment to leverage human judgment to improve forecasts of infectious disease using simulated surveillance data

Forecasts of infectious agents provide public health officials advanced warning about the intensity and timing of the spread of disease. Past work has found that accuracy and calibration of forecasts is weakest when attempting to predict an epidemic peak. Forecasts from a mechanistic model would be improved if there existed accurate information about the timing and intensity of an epidemic. We presented 3000 humans with simulated surveillance data about the number of incident hospitalizations from a current and two past seasons, and asked that they predict the peak time and intensity of the underlying epidemic. We found that in comparison to two control models, a model including human judgment produced more accurate forecasts of peak time and intensity of hospitalizations during an epidemic. Chimeric models have the potential to improve our ability to predict targets of public health interest which may in turn reduce infectious disease burden.

Torque modulation mechanism of the knee joint during balance recovery

Exploring the torque modulation mechanisms of human joints is critical for analyzing the human balance control system and developing natural human-machine interactions for balance support. However, the knee joint is often overlooked in biomechanical models because of its limited range of motion during balance recovery. This poses a challenge in establishing mathematical models for the knee joint's torque modulation mechanisms using computer simulations based on the inverted pendulum model. This study aims to provide a simplified linear feedback model inspired by sensorimotor transformation theory to reveal the torque modulation mechanism of the knee joint. The model was validated using data from experiments involving support-surface translation perturbations. The goodness-of-fit metrics of the model, including R2 values and root mean square errors (RMSE), demonstrated strong explanatory power (R2 ranged from 0.77 to 0.90) and low error (RMSE ranging from 0.035 to 0.072) across different perturbation magnitudes and directions. Through pooling samples across various perturbation conditions and conducting multiple fits, this model revealed that knee torque is modulated using a direction-specific strategy with adaptable feedback gains. These results suggest that the proposed simplified linear model can be used to develop assistive systems and retrieve insights on balance recovery mechanisms.

Adaptive Functional Electrical Stimulation Delivers Stimulation Amplitudes Based on Real-Time Gait Biomechanics

Functional electrical stimulation (FES) is often used in poststroke gait rehabilitation to decrease foot drop and increase forward propulsion. However, not all stroke survivors experience clinically meaningful improvements in gait function following training with FES. The purpose of this work was to develop and validate a novel adaptive FES (AFES) system to improve dorsiflexor (DF) and plantarflexor (PF) stimulation timing and iteratively adjust the stimulation amplitude at each stride based on measured gait biomechanics. Stimulation timing was determined by a series of bilateral footswitches. Stimulation amplitude was calculated based on measured dorsiflexion angle and peak propulsive force, where increased foot drop and decreased paretic propulsion resulted in increased stimulation amplitudes. Ten individuals with chronic poststroke hemiparesis walked on an adaptive treadmill with adaptive FES for three 2-min trials. Stimulation was delivered at the correct time to the dorsiflexor muscles during 95% of strides while stimulation was delivered to the plantarflexor muscles at the correct time during 84% of strides. Stimulation amplitudes were correctly calculated and delivered for all except two strides out of nearly 3000. The adaptive FES system responds to real-time gait biomechanics as intended, and further individualization to subject-specific impairments and rehabilitation goals may lead to improved rehabilitation outcomes.

Assistance force-line of exosuit affects ankle multidimensional motion: a theoretical and experimental study

BACKGROUND

The talocrural joint and the subtalar joint are the two major joints of the ankle-joint complex. The position and direction of the exosuit force line relative to these two joint axes can influence ankle motion. We aimed to understand the effects of different force-lines on ankle multidimensional motion.

METHODS

In this article, three assistance force line schemes for ankle exosuits were proposed: perpendicular to the talocrural joint axis (PT), intersecting with the subtalar joint axis (IS), and parallel to the triceps surae (PTS). A theoretical model was proposed to calculate the exosuit's assistance moment. Seven participants completed four experimental tests of ankle plantarflexion, including three passive motions assisted by the PT, PTS and IS schemes, and one active motion without exosuit assistance (Active).

RESULTS

The simulation results demonstrated that all three exosuits were able to produce significant moments of ankle plantarflexion. Among these, the PT scheme exhibited the highest moments in all dimensions, followed by the PTS and IS schemes. The experimental findings confirmed the effectiveness of all three exosuit schemes in assisting ankle plantarflexion. Additionally, as the assistive force lines approached the subtalar joint, there was a decrease in ankle motion assisted by the exosuits in non-plantarflexion directions, along with a reduction in the average distance of ankle angle curves relative to active ankle motion. Furthermore, the linear correlation coefficients between inversion and plantarflexion, adduction and plantarflexion, and adduction and inversion gradually converged toward active ankle plantarflexion motion.

CONCLUSIONS

Our research indicates that the position of the exosuit force line to the subtalar joint has a significant impact on ankle inversion and adduction. Among all three schemes, the IS, which has the closest distance to the subtalar joint axes, has the greatest kinematic similarity to active ankle plantarflexion and might be a better choice for ankle assistance and rehabilitation.

Human-in-the-Loop Optimization of Knee Exoskeleton Assistance for Minimizing User's Metabolic and Muscular Effort

Lower limb exoskeletons have the potential to mitigate work-related musculoskeletal disorders; however, they often lack user-oriented control strategies. Human-in-the-loop (HITL) controls adapt an exoskeleton's assistance in real time, to optimize the user-exoskeleton interaction. This study presents a HITL control for a knee exoskeleton using a CMA-ES algorithm to minimize the users' physical effort, a parameter innovatively evaluated using the interaction torque with the exoskeleton (a muscular effort indicator) and metabolic cost. This work innovates by estimating the user's metabolic cost within the HITL control through a machine-learning model. The regression model estimated the metabolic cost, in real time, with a root mean squared error of 0.66 W/kg and mean absolute percentage error of 26% (n = 5), making faster (10 s) and less noisy estimations than a respirometer (K5, Cosmed). The HITL reduced the user's metabolic cost by 7.3% and 5.9% compared to the zero-torque and no-device conditions, respectively, and reduced the interaction torque by 32.3% compared to a zero-torque control (n = 1). The developed HITL control surpassed a non-exoskeleton and zero-torque condition regarding the user's physical effort, even for a task such as slow walking. Furthermore, the user-specific control had a lower metabolic cost than the non-user-specific assistance. This proof-of-concept demonstrated the potential of HITL controls in assisted walking.

Multidisciplinary collaboration on exoskeleton development adopting user-centered design: a systematic integrative review

Purpose: The world population is ageing, along with an increasing possibility of functional limitations that affect independent living. Assistive technologies such as exoskeletons for rehabilitative purposes and daily activities assistance maintaining the independence of people with disabilities, especially older adults who wish to ageing-in-place. The purpose of this systematic integrative review was threefold: to explore the development team compositions and involvement, to understand the recruitment and engagement of stakeholders, and to synthesise reported or anticipated consequences of multidisciplinary collaboration.Methods: Databases searched included PubMed, CINAHL Plus, PsycINFO, Web of Science, Scopus, and IEEE Xplore. A total of 34 studies that reported the development of exoskeleton adopting user-centered design (UCD) method in healthcare or community settings that were published in English from 2000 to July 2022 were included.Results: Three major trends emerged from the analysis of included studies. First, there is a need to redefine multidisciplinary collaboration, from within-discipline collaboration to cross-discipline collaboration. Second, the level of engagement of stakeholders during the exoskeleton development remained low. Third, there was no standardised measurement to quantify knowledge production currently.Conclusion: As suggested by the synthesised results in this review, exoskeleton development has been increasing to improve the functioning of people with disabilities. Exoskeleton development often required expertise from different disciplines and the involvement of stakeholders to increase acceptance, thus we propose the Multidisciplinary Collaboration Appraisal Tool to assess multidisciplinary collaboration using the UCD approach. Future research is required to understand the effectiveness of multidisciplinary collaboration on exoskeleton development using the UCD approach.IMPLICATIONS FOR REHABILITATIONGlobal trend of population ageing causes a higher risk of disability in older adults who require rehabilitation and assistance in daily living.Assistive technologies such as exoskeletons have the potential to contribute to rehabilitation training and daily activity assistance demand closer multidisciplinary collaboration.A Multidisciplinary Collaboration Appraisal Tool using user-centered design approach (MCAT) is proposed to understand the effectiveness as well as limitations and barriers associated with multidisciplinary collaboration in developing exoskeletons.

Design and Evaluation of a Bilateral Semi-Rigid Exoskeleton to Assist Hip Motion

This study focused on designing and evaluating a bilateral semi-rigid hip exoskeleton. The exoskeleton assisted the hip joint, capitalizing on its proximity to the body's center of mass. Unlike its rigid counterparts, the semi-rigid design permitted greater freedom of movement. A temporal force-tracking controller allowed us to prescribe torque profiles during walking. We ensured high accuracy by tuning control parameters and series elasticity. The evaluation involved experiments with ten participants across ten force profile conditions with different end-timings and peak magnitudes. Our findings revealed a trend of greater reductions in metabolic cost with assistance provided at later timings in stride and at greater magnitudes. Compared to walking with the exoskeleton powered off, the largest reduction in metabolic cost was 9.1%. This was achieved when providing assistance using an end-timing at 44.6% of the stride cycle and a peak magnitude of 0.11 Nm kg-1. None of the tested conditions reduced the metabolic cost compared to walking without the exoskeleton, highlighting the necessity for further enhancements, such as a lighter and more form-fitting design. The optimal end-timing aligns with findings from other soft hip exosuit devices, indicating a comparable interaction with this prototype to that observed in entirely soft exosuit prototypes.

Estimating human joint moments unifies exoskeleton control, reducing user effort

Robotic lower-limb exoskeletons can augment human mobility, but current systems require extensive, context-specific considerations, limiting their real-world viability. Here, we present a unified exoskeleton control framework that autonomously adapts assistance on the basis of instantaneous user joint moment estimates from a temporal convolutional network (TCN). When deployed on our hip exoskeleton, the TCN achieved an average root mean square error of 0.142 newton-meters per kilogram across 35 ambulatory conditions without any user-specific calibration. Further, the unified controller significantly reduced user metabolic cost and lower-limb positive work during level-ground and incline walking compared with walking without wearing the exoskeleton. This advancement bridges the gap between in-lab exoskeleton technology and real-world human ambulation, making exoskeleton control technology viable for a broad community.

Gait Recognition and Assistance Parameter Prediction Determination Based on Kinematic Information Measured by Inertial Measurement Units

The gait recognition of exoskeletons includes motion recognition and gait phase recognition under various road conditions. The recognition of gait phase is a prerequisite for predicting exoskeleton assistance time. The estimation of real-time assistance time is crucial for the safety and accurate control of lower-limb exoskeletons. To solve the problem of predicting exoskeleton assistance time, this paper proposes a gait recognition model based on inertial measurement units that combines the real-time motion state recognition of support vector machines and phase recognition of long short-term memory networks. A recognition validation experiment was conducted on 30 subjects to determine the reliability of the gait recognition model. The results showed that the accuracy of motion state and gait phase were 99.98% and 98.26%, respectively. Based on the proposed SVM-LSTM gait model, exoskeleton assistance time was predicted. A test was conducted on 10 subjects, and the results showed that using assistive therapy based on exercise status and gait stage can significantly improve gait movement and reduce metabolic costs by an average of more than 10%.

Evaluation of controllers for augmentative hip exoskeletons and their effects on metabolic cost of walking: explicit versus implicit synchronization

Background: Efficient gait assistance by augmentative exoskeletons depends on reliable control strategies. While numerous control methods and their effects on the metabolic cost of walking have been explored in the literature, the use of different exoskeletons and dissimilar protocols limit direct comparisons. In this article, we present and compare two controllers for hip exoskeletons with different synchronization paradigms. Methods: The implicit-synchronization-based approach, termed the Simple Reflex Controller (SRC), determines the assistance as a function of the relative loading of the feet, resulting in an emerging torque profile continuously assisting extension during stance and flexion during swing. On the other hand, the Hip-Phase-based Torque profile controller (HPT) uses explicit synchronization and estimates the gait cycle percentage based on the hip angle, applying a predefined torque profile consisting of two shorter bursts of assistance during stance and swing. We tested the controllers with 23 naïve healthy participants walking on a treadmill at 4 km ⋅ h-1, without any substantial familiarization. Results: Both controllers significantly reduced the metabolic rate compared to walking with the exoskeleton in passive mode, by 18.0% (SRC, p < 0.001) and 11.6% (HPT, p < 0.001). However, only the SRC led to a significant reduction compared to walking without the exoskeleton (8.8%, p = 0.004). The SRC also provided more mechanical power and led to bigger changes in the hip joint kinematics and walking cadence. Our analysis of mechanical powers based on a whole-body analysis suggested a reduce in ankle push-off under this controller. There was a strong correlation (Pearson's r = 0.778, p < 0.001) between the metabolic savings achieved by each participant with the two controllers. Conclusion: The extended assistance duration provided by the implicitly synchronized SRC enabled greater metabolic reductions compared to the more targeted assistance of the explicitly synchronized HPT. Despite the different assistance profiles and metabolic outcomes, the correlation between the metabolic reductions with the two controllers suggests a difference in individual responsiveness to assistance, prompting more investigations to explore the person-specific factors affecting assistance receptivity.

Enhancing human-human musical interaction through kinesthetic haptic feedback using wearable exoskeletons: theoretical foundations, validation scenarios, and limitations

In this perspective paper, we explore the use of haptic feedback to enhance human-human interaction during musical tasks. We start by providing an overview of the theoretical foundation that underpins our approach, which is rooted in the embodied music cognition framework, and by briefly presenting the concepts of action-perception loop, sensorimotor coupling and entrainment. Thereafter, we focus on the role of haptic information in music playing and we discuss the use of wearable technologies, namely lightweight exoskeletons, for the exchange of haptic information between humans. We present two experimental scenarios in which the effectiveness of this technology for enhancing musical interaction and learning might be validated. Finally, we briefly discuss some of the theoretical and pedagogical implications of the use of technologies for haptic communication in musical contexts, while also addressing the potential barriers to the widespread adoption of exoskeletons in such contexts.

Impact of Haptic Cues and an Active Ankle Exoskeleton on Gait Characteristics

OBJECTIVE

This study examined the interaction of gait-synchronized vibrotactile cues with an active ankle exoskeleton that provides plantarflexion assistance.

BACKGROUND

An exoskeleton that augments gait may support collaboration through feedback to the user about the state of the exoskeleton or characteristics of the task.

METHODS

Participants (N = 16) were provided combinations of torque assistance and vibrotactile cues at pre-specified time points in late swing and early stance while walking on a self-paced treadmill. Participants were either given explicit instructions (N = 8) or were allowed to freely interpret (N=8) how to coordinate with cues.

RESULTS

For the free interpretation group, the data support an 8% increase in stride length and 14% increase in speed with exoskeleton torque across cue timing, as well as a 5% increase in stride length and 7% increase in speed with only vibrotactile cues. When given explicit instructions, participants modulated speed according to cue timing-increasing speed by 17% at cues in late swing and decreasing speed 11% at cues in early stance compared to no cue when exoskeleton torque was off. When torque was on, participants with explicit instructions had reduced changes in speed.

CONCLUSION

These findings support that the presence of torque mitigates how cues were used and highlights the importance of explicit instructions for haptic cuing. Interpreting cues while walking with an exoskeleton may increase cognitive load, influencing overall human-exoskeleton performance for novice users.

APPLICATION

Interactions between haptic feedback and exoskeleton use during gait can inform future feedback designs to support coordination between users and exoskeletons.

Human-in-the-loop optimization of rocker shoes via different cost functions during walking

Personalised footwear could be used to enhance the function of the foot-ankle complex to a person's maximum. Human-in-the-loop optimization could be used as an effective and efficient way to find a personalised optimal rocker profile (i.e., apex position and angle). The outcome of this process likely depends on the selected optimization objective and its responsiveness to the rocker parameters being tuned. This study aims to explore whether and how human-in-the-loop optimization via different cost functions (i.e., metabolic cost, collision work as measure for external mechanical work, and step distance variability as measure for gait stability) affects the optimal apex position and angle of a rocker profile differently for individuals during walking. Ten healthy individuals walked on a treadmill with experimental rocker shoes in which apex position and angle were optimized using human-in-the-loop optimization using different cost functions. We compared the obtained optimal apex parameters for the different cost functions and how these affected the selected gait related objectives. Optimal apex parameters differed substantially between participants and optimal apex positions differed between cost functions. The responsiveness to changes in apex parameters differed between cost functions. Collision work was the only cost function that resulted in a significant improvement of its performance criteria. Improvements in metabolic cost or step distance variability were not found after optimization. This study showed that cost function selection is important when human-in-the-loop optimization is used to design personalised footwear to allow conversion to an optimum that suits the individual.

A neuromorphic device mimicking synaptic plasticity under different body fluid K+ homeostasis for artificial reflex path construction and pattern recognition

The ionic environment of body fluids influences nervous functions for maintaining homeostasis in organisms and ensures normal perceptual abilities and reflex activities. Neural reflex activities, such as limb movements, are closely associated with potassium ions (K+). In this study, we developed artificial synaptic devices based on ion concentration-adjustable gels for emulating various synaptic plasticities under different K+ concentrations in body fluids. In addition to performing essential synaptic functions, potential applications in information processing and associative learning using short- and long-term plasticity realized using ion concentration-adjustable gels are presented. Artificial synaptic devices can be used for constructing an artificial neural pathway that controls artificial muscle reflex activities and can be used for image pattern recognition. All tests show a strong relationship with ion homeostasis. These devices could be applied to neuromorphic robots and human-machine interfaces.

Shaping high-performance wearable robots for human motor and sensory reconstruction and enhancement

Most wearable robots such as exoskeletons and prostheses can operate with dexterity, while wearers do not perceive them as part of their bodies. In this perspective, we contend that integrating environmental, physiological, and physical information through multi-modal fusion, incorporating human-in-the-loop control, utilizing neuromuscular interface, employing flexible electronics, and acquiring and processing human-robot information with biomechatronic chips, should all be leveraged towards building the next generation of wearable robots. These technologies could improve the embodiment of wearable robots. With optimizations in mechanical structure and clinical training, the next generation of wearable robots should better facilitate human motor and sensory reconstruction and enhancement.

Dynamic and Static Assistive Strategies for a Tailored Occupational Back-Support Exoskeleton: Assessment on Real Tasks Carried Out by Railway Workers

This study on occupational back-support exoskeletons performs a laboratory evaluation of realistic tasks with expert workers from the railway sector. Workers performed both a static task and a dynamic task, each involving manual material handling (MMH) and manipulating loads of 20 kg, in three conditions: without an exoskeleton, with a commercially available passive exoskeleton (Laevo v2.56), and with the StreamEXO, an active back-support exoskeleton developed by our institute. Two control strategies were defined, one for dynamic tasks and one for static tasks, with the latter determining the upper body's gravity compensation through the Model-based Gravity Compensation (MB-Grav) approach. This work presents a comparative assessment of the performance of active back support exoskeletons versus passive exoskeletons when trialled in relevant and realistic tasks. After a lab characterization of the MB-Grav strategy, the experimental assessment compared two back-support exoskeletons, one active and one passive. The results showed that while both devices were able to reduce back muscle activation, the benefits of the active device were triple those of the passive system regarding back muscle activation (26% and 33% against 9% and 11%, respectively), while the passive exoskeleton hindered trunk mobility more than the active mechanism.

Continuous Locomotion Mode and Task Identification for an Assistive Exoskeleton Based on Neuromuscular-Mechanical Fusion

Human walking parameters exhibit significant variability depending on the terrain, speed, and load. Assistive exoskeletons currently focus on the recognition of locomotion terrain, ignoring the identification of locomotion tasks, which are also essential for control strategies. The aim of this study was to develop an interface for locomotion mode and task identification based on a neuromuscular-mechanical fusion algorithm. The modes of level and incline and tasks of speed and load were explored, and seven able-bodied participants were recruited. A continuous stream of assistive decisions supporting timely exoskeleton control was achieved according to the classification of locomotion. We investigated the optimal algorithm, feature set, window increment, window length, and robustness for precise identification and synchronization between exoskeleton assistive force and human limb movements (human-machine collaboration). The best recognition results were obtained when using a support vector machine, a root mean square/waveform length/acceleration feature set, a window length of 170, and a window increment of 20. The average identification accuracy reached 98.7% ± 1.3%. These results suggest that the surface electromyography-acceleration can be effectively used for locomotion mode and task identification. This study contributes to the development of locomotion mode and task recognition as well as exoskeleton control for seamless transitions.

Lower limb biomechanics of fully trained exoskeleton users reveal complex mechanisms behind the reductions in energy cost with human-in-the-loop optimization

Exoskeletons that assist in ankle plantarflexion can improve energy economy in locomotion. Characterizing the joint-level mechanisms behind these reductions in energy cost can lead to a better understanding of how people interact with these devices, as well as to improved device design and training protocols. We examined the biomechanical responses to exoskeleton assistance in exoskeleton users trained with a lengthened protocol. Kinematics at unassisted joints were generally unchanged by assistance, which has been observed in other ankle exoskeleton studies. Peak plantarflexion angle increased with plantarflexion assistance, which led to increased total and biological mechanical power despite decreases in biological joint torque and whole-body net metabolic energy cost. Ankle plantarflexor activity also decreased with assistance. Muscles that act about unassisted joints also increased activity for large levels of assistance, and this response should be investigated over long-term use to prevent overuse injuries.

IMU-Based Energy Expenditure Estimation for Various Walking Conditions Using a Hybrid CNN-LSTM Model

In ubiquitous healthcare systems, energy expenditure estimation based on wearable sensors such as inertial measurement units (IMUs) is important for monitoring the intensity of physical activity. Although several studies have reported data-driven methods to estimate energy expenditure during activities of daily living using wearable sensor signals, few have evaluated the performance while walking at various speeds and inclines. In this study, we present a hybrid model comprising a convolutional neural network (CNN) and long short-term memory (LSTM) to estimate the steady-state energy expenditure under various walking conditions based solely on IMU data. To implement and evaluate the model, we performed level/inclined walking and level running experiments on a treadmill. With regard to the model inputs, the performance of the proposed model based on fixed-size sequential data was compared with that of a method based on stride-segmented data under different conditions in terms of the sensor location, input sequence format, and neural network model. Based on the experimental results, the following conclusions were drawn: (i) the CNN-LSTM model using a two-second sequence from the IMU attached to the lower body yielded optimal performance, and (ii) although the stride-segmented data-based method showed superior performance, the performance difference between the two methods was not significant; therefore, the proposed model based on fixed-size sequential data may be considered more practical as it does not require heel-strike detection.

Optimizing exoskeleton assistance to improve walking speed and energy economy for older adults

BACKGROUND

Walking speed and energy economy tend to decline with age. Lower-limb exoskeletons have demonstrated potential to improve either measure, but primarily in studies conducted on healthy younger adults. Promising techniques like optimization of exoskeleton assistance have yet to be tested with older populations, while speed and energy consumption have yet to be simultaneously optimized for any population.

METHODS

We investigated the effectiveness of human-in-the-loop optimization of ankle exoskeletons with older adults. Ten healthy adults > 65 years of age (5 females; mean age: 72 ± 3 yrs) participated in approximately 240 min of training and optimization with tethered ankle exoskeletons on a self-paced treadmill. Multi-objective human-in-the-loop optimization was used to identify assistive ankle plantarflexion torque patterns that simultaneously improved self-selected walking speed and metabolic rate. The effects of optimized exoskeleton assistance were evaluated in separate trials.

RESULTS

Optimized exoskeleton assistance improved walking performance for older adults. Both objectives were simultaneously improved; self-selected walking speed increased by 8% (0.10 m/s; p = 0.001) and metabolic rate decreased by 19% (p = 0.007), resulting in a 25% decrease in energetic cost of transport (p = 8e-4) compared to walking with exoskeletons applying zero torque. Compared to younger participants in studies optimizing a single objective, our participants required lower exoskeleton torques, experienced smaller improvements in energy use, and required more time for motor adaptation.

CONCLUSIONS

Our results confirm that exoskeleton assistance can improve walking performance for older adults and show that multiple objectives can be simultaneously addressed through human-in-the-loop optimization.

Human-in-the-Loop Optimization of Wearable Robotic Devices to Improve Human-Robot Interaction: A Systematic Review

This article presents a systematic review on wearable robotic devices that use human-in-the-loop optimization (HILO) strategies to improve human-robot interaction. A total of 46 HILO studies were identified and divided into upper and lower limb robotic devices. The main aspects from HILO were identified, reviewed, and classified in four areas: 1) human-machine systems; 2) optimization methods; 3) control strategies; and 4) experimental protocols. A variety of objective functions (physiological, biomechanical, and subjective), optimization strategies, and optimized control parameters configurations used in different control strategies are presented and analyzed. An overview of experimental protocols is provided, including metrics, tasks, and conditions tested. Moreover, the relevance given to training or adaptation periods was explored. We outline an HILO framework that includes current wearable robots, optimization strategies, objective functions, control strategies, and experimental protocols. We conclude by highlighting current research gaps and defining future directions to improve the development of advanced HILO strategies in upper and lower limb wearable robots.

Human-in-the-Loop Adaptive Control of a Soft Exo-Suit With Actuator Dynamics and Ankle Impedance Adaptation

Soft exo-suit could facilitate walking assistance activities (such as level walking, upslope, and downslope) for unimpaired individuals. In this article, a novel human-in-the-loop adaptive control scheme is presented for a soft exo-suit, which provides ankle plantarflexion assistance with unknown human-exosuit dynamic model parameters. First, the human-exosuit coupled dynamic model is formulated to express the mathematical relationship between the exo-suit actuation system and the human ankle joint. Then, a gait detection approach, including plantarflexion assistance timing and planning, is proposed. Inspired by the control strategy that is used by the human central nervous system (CNS) to handle interaction tasks, a human-in-the-loop adaptive controller is proposed to adapt the unknown exo-suit actuator dynamics and human ankle impedance. The proposed controller can emulate human CNS behaviors which adapt feedforward force and environment impedance in interaction tasks. The resulting adaptation of actuator dynamics and ankle impedance is demonstrated with five unimpaired subjects and implemented on a developed soft exo-suit. The human-like adaptivity is performed by the exo-suit in several human walking speeds and illustrates the promising potential of the novel controller.

Joint power, joint work and lower limb muscle activity for transitions between level walking and stair ambulation at three inclinations

To enhance human mobility, training interventions and assistive lower limb wearable robotic designs must draw insights from movement tasks from daily life. This study aimed to analyze joint peak power, limb and joint work, and muscle activity of the lower limb during a series of stair ambulation conditions. We recruited 12 subjects (25.4±4.5 yrs, 180.1±4.6 cm, 74.6±7.9 kg) and studied steady gait and gait transitions between level walking, stair ascent and stair descent for three staircase inclinations (low 19°, normal 30.4°, high 39.6°). Our analysis revealed that joint peak power, limb and joint work, and muscle activity increased significantly compared to level walking and with increasing stair inclination for most of the conditions analyzed. Transition strides had no increased requirements compared to the maxima found for steady level walking and steady stair ambulation. Stair ascent required increased lower limb joint positive peak power and work, while stair descent required increased lower limb joint negative peak power and work compared to level walking. The most challenging condition was high stair inclination, which required approximately thirteen times the total lower limb joint positive and negative net work during ascent and descent, respectively. These findings suggest that training interventions and lower limb wearable robotic designs must consider the major increases in lower limb joint and muscle effort during stair ambulation, with specific attention to the demands of ascent and descent, to effectively improve human mobility.

User preference optimization for control of ankle exoskeletons using sample efficient active learning

One challenge to achieving widespread success of augmentative exoskeletons is accurately adjusting the controller to provide cooperative assistance with their wearer. Often, the controller parameters are "tuned" to optimize a physiological or biomechanical objective. However, these approaches are resource intensive, while typically only enabling optimization of a single objective. In reality, the exoskeleton user experience is likely derived from many factors, including comfort, fatigue, and stability, among others. This work introduces an approach to conveniently tune the four parameters of an exoskeleton controller to maximize user preference. Our overarching strategy is to leverage the wearer to internally balance the experiential factors of wearing the system. We used an evolutionary algorithm to recommend potential parameters, which were ranked by a neural network that was pretrained with previously collected user preference data. The controller parameters that had the highest preference ranking were provided to the exoskeleton, and the wearer responded with real-time feedback as a forced-choice comparison. Our approach was able to converge on controller parameters preferred by the wearer with an accuracy of 88% on average when compared with randomly generated parameters. User-preferred settings stabilized in 43 ± 7 queries. This work demonstrates that user preference can be leveraged to tune a partial-assist ankle exoskeleton in real time using a simple, intuitive interface, highlighting the potential for translating lower-limb wearable technologies into our daily lives.

Effect of hip abduction assistance on metabolic cost and balance during human walking

The use of wearable robots to provide walking assistance has rapidly grown over the past decade, with notable advances made in robot design and control methods toward reducing physical effort while performing an activity. The reduction in walking effort has mainly been achieved by assisting forward progression in the sagittal plane. Human gait, however, is a complex movement that combines motions in three planes, not only the sagittal but also the transverse and frontal planes. In the frontal plane, the hip joint plays a key role in gait, including balance. However, wearable robots targeting this motion have rarely been investigated. In this study, we developed a hip abduction assistance wearable robot by formulating the hypothesis that assistance that mimics the biological hip abduction moment or power could reduce the metabolic cost of walking and affect the dynamic balance. We found that hip abduction assistance with a biological moment second peak mimic profile reduced the metabolic cost of walking by 11.6% compared with the normal walking condition. The assistance also influenced balance-related parameters, including the margin of stability. Hip abduction assistance influenced the center-of-mass movement in the mediolateral direction. When the robot assistance was applied as the center of mass moved toward the opposite leg, the assistance replaced some of the efforts that would have otherwise been provided by the human. This indicates that hip abduction assistance can reduce physical effort during human walking while influencing balance.

A Data-Driven and Personalized Stance Symmetry Controller for Robotic Ankle-Foot Prostheses: A Preliminary Investigation

People with unilateral transtibial amputation generally exhibit asymmetric gait, likely due to inadequate prosthetic ankle function. This results in compensatory behavior, leading to long-term musculoskeletal impairments (e.g., osteoarthritis in the joints of the intact limb). Powered prostheses can better emulate biological ankles, however, control methods are over-reliant on non-disabled data, require extensive amounts of tuning by experts, and cannot adapt to each user's unique gait patterns. This work directly addresses all these limitations with a personalized and data-driven control strategy. Our controller uses a virtual setpoint trajectory within an impedance-inspired formula to adjust the dynamics of the robotic ankle-foot prosthesis as a function of stance phase. A single sensor measuring thigh motion is used to estimate the gait phase in real time. The virtual setpoint trajectory is modified via a data-driven iterative learning strategy aimed at optimizing ankle angle symmetry. The controller was experimentally evaluated on two people with transtibial amputation. The control scheme successfully increased ankle angle symmetry about the two limbs by 24.4% when compared to the passive condition. In addition, the symmetry controller significantly increased peak prosthetic ankle power output at push-off by 0.52 W/kg and significantly reduced biomechanical risk factors associated with osteoarthritis (i.e., knee and hip abduction moments) in the intact limb. This research demonstrates the benefits of personalized and data-driven symmetry controllers for robotic ankle-foot prostheses.

Interaction force modeling and analysis of the human-machine kinematic chain based on the human-machine deviation

A mechanical model for a human-machine interaction force based on the man-machine kinematic chain is established. This is combined with screw theory and a virtual rigid body model for the human-machine interaction force is proposed. This model interprets the basic principle model of the human-machine contact force. The deviation of the human-machine kinematic chain is calculated using the virtual model. In order to carry out the calibration simulation for the model, a 6-sps parallel mechanism is taken as an example to illustrate the construction principle of the human-machine interaction virtual rigid body model. This is calibrated by introducing finite element software. Finally, using the knee exoskeleton as an example, a numerical simulation is introduced. This illustrates the relationship between the driving force of the exoskeleton, the human-machine deviation as well as the virtual stiffness. The modeling method of this paper provides theoretical reference for controller design of human-machine interaction forces in the future.

Effects of Personalization on Gait-State Tracking Performance Using Extended Kalman Filters

Emerging partial-assistance exoskeletons can enhance able-bodied performance and aid people with pathological gait or age-related immobility. However, every person walks differently, which makes it difficult to directly compute assistance torques from joint kinematics. Gait-state estimation-based controllers use phase (normalized stride time) and task variables (e.g., stride length and ground inclination) to parameterize the joint torques. Using kinematic models that depend on the gait-state, prior work has used an Extended Kalman filter (EKF) to estimate the gait-state online. However, this EKF suffered from kinematic errors since it used a subject-independent measurement model, and it is still unknown how personalization of this measurement model would reduce gait-state tracking error. This paper quantifies how much gait-state tracking improvement a personalized measurement model can have over a subject-independent measurement model when using an EKF-based gait-state estimator. Since the EKF performance depends on the measurement model covariance matrix, we tested on multiple different tuning parameters. Across reasonable values of tuning parameters that resulted in good performance, personalization improved estimation error on average by 8.5 ± 13.8% for phase (mean ± standard deviation), 27.2 ± 8.1% for stride length, and 10.5 ± 13.5% for ground inclination. These findings support the hypothesis that personalization of the measurement model significantly improves gait-state estimation performance in EKF based gait-state tracking (P ≪ 0.05 ), which could ultimately enable reliable responses to faster human gait changes.

Tapping Into Skeletal Muscle Biomechanics for Design and Control of Lower Limb Exoskeletons: A Narrative Review

Lower limb exoskeletons and exosuits ("exos") are traditionally designed with a strong focus on mechatronics and actuation, whereas the "human side" is often disregarded or minimally modeled. Muscle biomechanics principles and skeletal muscle response to robot-delivered loads should be incorporated in design/control of exos. In this narrative review, we summarize the advances in literature with respect to the fusion of muscle biomechanics and lower limb exoskeletons. We report methods to measure muscle biomechanics directly and indirectly and summarize the studies that have incorporated muscle measures for improved design and control of intuitive lower limb exos. Finally, we delve into articles that have studied how the human-exo interaction influences muscle biomechanics during locomotion. To support neurorehabilitation and facilitate everyday use of wearable assistive technologies, we believe that future studies should investigate and predict how exoskeleton assistance strategies would structurally remodel skeletal muscle over time. Real-time mapping of the neuromechanical origin and generation of muscle force resulting in joint torques should be combined with musculoskeletal models to address time-varying parameters such as adaptation to exos and fatigue. Development of smarter predictive controllers that steer rather than assist biological components could result in a synchronized human-machine system that optimizes the biological and electromechanical performance of the combined system.

An Energetic Approach to Task-Invariant Ankle Exoskeleton Control

Robotic ankle exoskeletons have been shown to reduce human effort during walking. However, existing ankle exoskeleton control approaches are limited in their ability to apply biomimetic torque across diverse tasks outside of the controlled lab environment. Energy shaping control can provide task-invariant assistance without estimating the user's state, classifying task, or reproducing pre-defined torque trajectories. In previous work, we showed that an optimally task-invariant energy shaping controller implemented on a knee-ankle exoskeleton reduced the effort of certain muscles for a range of tasks. In this paper, we extend this approach to the sensor suite available at the ankle and present its implementation on a commercially-available, bilateral ankle exoskeleton. An experiment with three healthy subjects walking on a circuit and on a treadmill showed that the controller can approximate biomimetic profiles for varying terrains and task transitions without classifying tasks or switching control modes.

Simulating Muscle-Level Energetic Cost Savings When Humans Run with a Passive Assistive Device

Connecting the legs with a spring attached to the shoelaces, called an exotendon, can reduce the energetic cost of running, but how the exotendon reduces the energetic burden of individual muscles remains unknown. We generated muscle-driven simulations of seven individuals running with and without the exotendon to discern whether savings occurred during the stance phase or the swing phase, and to identify which muscles contributed to energy savings. We computed differences in muscle-level energy consumption, muscle activations, and changes in muscle-fiber velocity and force between running with and without the exotendon. The seven of nine participants who reduced energy cost when running with the exotendon reduced their measured energy expenditure rate by 0.9 W/kg (8.3%). Simulations predicted a 1.4 W/kg (12.0%) reduction in the average rate of energy expenditure and correctly identified that the exotendon reduced rates of energy expenditure for all seven individuals. Simulations showed most of the savings occurred during stance (1.5 W/kg), though the rate of energy expenditure was also reduced during swing (0.3 W/kg). The energetic savings were distributed across the quadriceps, hip flexor, hip abductor, hamstring, hip adductor, and hip extensor muscle groups, whereas no changes were observed in the plantarflexor or dorsiflexor muscles. Energetic savings were facilitated by reductions in the rate of mechanical work performed by muscles and their estimated rate of heat production. By modeling muscle-level energetics, this simulation framework accurately captured measured changes in whole-body energetics when using an assistive device. This is a useful first step towards using simulation to accelerate device design by predicting how humans will interact with assistive devices that have yet to be built.

Multivariate CNN Model for Human Locomotion Activity Recognition with a Wearable Exoskeleton Robot

This study introduces a novel convolutional neural network (CNN) architecture, encompassing both single and multi-head designs, developed to identify a user's locomotion activity while using a wearable lower limb robot. Our research involved 500 healthy adult participants in an activities of daily living (ADL) space, conducted from 1 September to 30 November 2022. We collected prospective data to identify five locomotion activities (level ground walking, stair ascent/descent, and ramp ascent/descent) across three terrains: flat ground, staircase, and ramp. To evaluate the predictive capabilities of the proposed CNN architectures, we compared its performance with three other models: one CNN and two hybrid models (CNN-LSTM and LSTM-CNN). Experiments were conducted using multivariate signals of various types obtained from electromyograms (EMGs) and the wearable robot. Our results reveal that the deeper CNN architecture significantly surpasses the performance of the three competing models. The proposed model, leveraging encoder data such as hip angles and velocities, along with postural signals such as roll, pitch, and yaw from the wearable lower limb robot, achieved superior performance with an inference speed of 1.14 s. Specifically, the F-measure performance of the proposed model reached 96.17%, compared to 90.68% for DDLMI, 94.41% for DeepConvLSTM, and 95.57% for LSTM-CNN, respectively.

Towards Personalized Control for Powered Knee Prostheses: Continuous Impedance Functions and PCA-Based Tuning Method

Optimizing control parameters is crucial for personalizing prosthetic devices. The current method of finite state machine impedance control (FSM-IC) allows interaction with the user but requires time-consuming manual tuning. To improve efficiency, we propose a novel approach for tuning knee prostheses using continuous impedance functions (CIFs) and Principal Component Analysis (PCA). The CIFs, which represent stiffness, damping, and equilibrium angle, are modeled as fourth-order polynomials and optimized through convex optimization. By applying PCA to the CIFs, we extract principal components (PCs) that capture common features. The weights of these PCs serve as tuning parameters, allowing us to reconstruct various impedance functions. We validated this approach using data from 10 able-bodied individuals walking. The contributions of this study include: i) generating CIFs via convex optimization; ii) introducing a new tuning space based on the obtained CIFs; and iii) evaluating the feasibility of this tuning space.

User-Centered Configuration of Soft Hip Flexion Exosuit Designs to Assist Individuals with Multiple Sclerosis Through Simulated Human-in-the-Loop Optimization

Soft exosuits hold promise as assistive technology for people with gait deficits owing to a variety of causes. A key aspect of providing useful assistance is to keep the human user at the center of all considerations made in the design, configuration, and prescribed use of an assistive device. This work details a method for informing the configuration of a soft hip flexion exosuit by 1) modeling the user's shape and movements in order to simulate the mechanical interaction of the exosuit and user, 2) incorporating the mechanical effects of the exosuit into a muscle-driven musculoskeletal gait simulation, and 3) using the results of these simulations to define a cost function that is minimized via Bayesian optimization. This process is carried out for models of four different people with multiple sclerosis, and the final optimized configurations for each subject are compared. For all users, the estimated metabolic cost of transport was reduced below baseline, no-device levels. This work represents a step toward more individualized, user-centric modeling of assistive devices, and demonstrates a system for informing the physical configuration of an exosuit on a case-by-case basis using real patient data.

Predictive Control of Peak Achilles Tendon Force in a Simulated System of the Human Ankle Joint with a Parallel Artificial Actuator During Hopping

Latest advances in wearable exoskeletons for the human lower extremity predominantly focus on minimising metabolic cost of walking. However, there currently is no robotic exoskeleton that gains control on the mechanics of biological tissues such as biological muscles or series-elastic tendons. Achieving robotic control of biological tissue mechanics would enable prevention of musculoskeletal injuries or the personalization of rehabilitation treatments following injury with levels of precisions not attained before. In this paper, we introduce a new framework that uses nonlinear model predictive control (NMPC) for the closed-loop control of peak tendon force in a simulated system of the human ankle joint with parallel exoskeletal actuation. We propose a computationally efficient NMPC's inner model consisting of explicit, closed-form equations of muscle-tendon dynamics along with those of the ankle joint with parallel actuation. The proposed formulation is tested and verified on movement data collected during dynamic ankle dorsiflexion/plantarflexion rotations executed on a dynamometer as well as during walking and running on a treadmill. The framework designed using the NMPC controller showed a promising performance in keeping the Achilles tendon force under a predefined threshold. Results indicated that our proposed model was generalizable to different muscles and gaits and suitable for real-time applications due to its low computational time.

State-Based Versus Time-Based Estimation of the Gait Phase for Hip Exoskeletons in Steady and Transient Walking

The growing demand for online gait phase (GP) estimation, driven by advancements in exoskeletons and prostheses, has prompted numerous approaches in the literature. Some approaches explicitly use time, while others rely on state variables to estimate the GP. In this article, we study two novel GP estimation methods: a State-based Method (SM) which employs the phase portrait of the hip angle (similar to previous methods), but uses a stretching transformation to reduce the nonlinearity of the estimated GP; and a Time-based Method (TM) that utilizes feature recognition on the hip angle signal to update the estimated cadence twice per gait cycle. The methods were tested across various speeds and slopes, encompassing steady and transient walking conditions. The results demonstrated the ability of both methods to estimate the GP in a range of conditions. The TM outperformed the SM, exhibiting a root-mean-squared error below 3% compared to 8.5% for the SM. However, the TM exhibited diminished performance during speed transitions, whereas the SM performed consistently in steady and transient conditions. The SM displayed a better performance in inclined walking and demonstrated higher linearity at faster speeds. Through the assessment of these methods in diverse conditions, this study lays the groundwork for further advancements in GP estimation methods and their application in assistive controllers.

Hip and Knee Joint Kinematics Predict Quadriceps Hyperreflexia in People with Post-stroke Stiff-Knee Gait

Wearable assistive technology for the lower extremities has shown great promise towards improving gait function in people with neuromuscular injuries. But common secondary impairments, such as hypersensitive stretch reflexes or hyperreflexia, have been often neglected. Incorporation of biomechanics into the control loop could improve individualization and avoid hyperreflexia. However, adding hyperreflexia prediction to the control loop would require expensive or complex measurement of muscle fiber characteristics. In this study, we explore a clinically accessible biomechanical predictor set that can accurately predict rectus femoris (RF) reaction after knee flexion assistance in pre-swing by a powered orthosis. We examined a total of 14 gait parameters based on gait kinematic, kinetic, and simulated muscle-tendon states from 8 post-stroke individuals with Stiff-Knee gait (SKG) wearing a knee exoskeleton robot. We independently performed both parametric and non-parametric variable selection approaches using machine learning regression techniques. Both models revealed the same four kinematic variables relevant to knee and hip joint motions were sufficient to effectively predict RF hyperreflexia. These results suggest that control of knee and hip kinematics may be a more practical method of incorporating quadriceps hyperreflexia into the exoskeleton control loop than the more complex acquisition of muscle fiber properties.

An EMG-Based Objective Function for Human-in-the-Loop Optimization

Advancements in wearable robots aim to improve the users' motion, performance, and comfort by optimizing, mainly, energetic cost (EC). However, EC is a noisy measurement with a physiological delayed response that requires long evaluation periods and wearing an uncomfortable mask. This study aims to estimate and minimize an EMG-based objective function that describes the natural energetic expenditure of individuals walking. This objective is assessed by combining multiple electromyography (EMG) variables from the EMG intensity and muscle synergies. To evaluate this objective function simply and repeatedly, we prescribed step frequency (SF) via a metronome and optimized this frequency to minimize muscle activity demands. Further, a linear mixed-effects model was fitted for EC, with the EMG variables as fixed-effects and a random intercept that varies by participant. After the model was fitted to the data, a cubic polynomial was used to identify the optimal SF that reduces the overall EMG-based objective function. Our analysis outlines that the proposed objective function is comparable to the EC during walking, the primary objective function used in human-in-the-loop optimization. Thus, this EMG-based objective function could be potentially used to optimize wearable robots and improve human-robot interaction.

A transition point: Assistance magnitude is a critical parameter when providing assistance during walking with an energy-removing exoskeleton or biomechanical energy harvester

Researchers and engineers have developed exoskeletons capable of reducing the energetic cost of walking by decreasing the force their users' muscles are required to produce while contracting. The metabolic effect of assisting concentric and isometric muscle contractions depends, in part, on assistance magnitude. We conducted human treadmill experiments to explore the effects of assistance magnitude on the biomechanics and energetics of walking with an energy-removing exoskeleton designed to assist eccentric muscle contractions. Our results demonstrate that the assistance magnitude of an energy-removing device significantly affects the energetics, muscle activity, and biomechanics of walking. Under the moderate assistance magnitude condition, our device reduced the metabolic cost of walking below that of normal walking by 3.4% while simultaneously producing 0.29 W of electricity. This reduction in the energetic cost of walking was also associated with an 8.9% decrease in hamstring activity. Furthermore, we determined that there is an assistance magnitude threshold that, when crossed, results in the device transitioning from assisting to hindering its user. This transition is marked by significant increases in muscle activity and the metabolic cost of walking. These results could aid in the future design of exoskeletons and biomechanical energy harvesters, as well as adaptive control systems, that identify user-specific control parameters associated with minimum energy expenditure.

Simulating the effect of ankle plantarflexion and inversion-eversion exoskeleton torques on center of mass kinematics during walking

Walking balance is central to independent mobility, and falls due to loss of balance are a leading cause of death for people 65 years of age and older. Bipedal gait is typically unstable, but healthy humans use corrective torques to counteract perturbations and stabilize gait. Exoskeleton assistance could benefit people with neuromuscular deficits by providing stabilizing torques at lower-limb joints to replace lost muscle strength and sensorimotor control. However, it is unclear how applied exoskeleton torques translate to changes in walking kinematics. This study used musculoskeletal simulation to investigate how exoskeleton torques applied to the ankle and subtalar joints alter center of mass kinematics during walking. We first created muscle-driven walking simulations using OpenSim Moco by tracking experimental kinematics and ground reaction forces recorded from five healthy adults. We then used forward integration to simulate the effect of exoskeleton torques applied to the ankle and subtalar joints while keeping muscle excitations fixed based on our previous tracking simulation results. Exoskeleton torque lasted for 15% of the gait cycle and was applied between foot-flat and toe-off during the stance phase, and changes in center of mass kinematics were recorded when the torque application ended. We found that changes in center of mass kinematics were dependent on both the type and timing of exoskeleton torques. Plantarflexion torques produced upward and backward changes in velocity of the center of mass in mid-stance and upward and smaller forward velocity changes near toe-off. Eversion and inversion torques primarily produced lateral and medial changes in velocity in mid-stance, respectively. Intrinsic muscle properties reduced kinematic changes from exoskeleton torques. Our results provide mappings between ankle plantarflexion and inversion-eversion torques and changes in center of mass kinematics which can inform designers building exoskeletons aimed at stabilizing balance during walking. Our simulations and software are freely available and allow researchers to explore the effects of applied torques on balance and gait.

Interacting effects of AFO stiffness, neutral angle and footplate stiffness on gait in case of plantarflexor weakness: A predictive simulation study

To maximize effects of dorsal leaf ankle foot orthoses (AFOs) on gait in people with bilateral plantarflexor weakness, the AFO properties should be matched to the individual. However, how AFO properties interact regarding their effect on gait function is unknown. We studied the interaction of AFO bending stiffness with neutral angle and footplate stiffness on the effect of bending stiffness on walking energy cost, gait kinematics and kinetics in people with plantarflexor weakness by employing predictive simulations. Our simulation framework consisted of a planar 11 degrees of freedom model, containing 11 muscles activated by a reflex-based neuromuscular controller. The controller was optimized by a comprehensive cost function, predominantly minimizing walking energy cost. The AFO bending and footplate stiffness were modelled as torsional springs around the ankle and metatarsal joint. The neutral angle of the AFO was defined as the angle in the sagittal plane at which the moment of the ankle torsional spring was zero. Simulations without AFO and with AFO for 9 bending stiffnesses (0-14 Nm/degree), 3 neutral angles (0-3-6 degrees dorsiflexion) and 3 footplate stiffnesses (0-0.5-2.0 Nm/degree) were performed. When changing neutral angle towards dorsiflexion, a higher AFO bending stiffness minimized energy cost of walking and normalized joint kinematics and kinetics. Footplate stiffness mainly affected MTP joint kinematics and kinetics, while no systematic and only marginal effects on energy cost were found. In conclusion, the interaction of the AFO bending stiffness and neutral angle in bilateral plantarflexor weakness, suggests that these should both be considered together when matching AFO properties to the individual patient.

A machine learning approach to quantify individual gait responses to ankle exoskeletons

Predicting an individual's response to an exoskeleton and understanding what data are needed to characterize responses remains challenging. Specifically, we lack a theoretical framework capable of quantifying heterogeneous responses to exoskeleton interventions. We leverage a neural network-based discrepancy modeling framework to quantify complex changes in gait in response to passive ankle exoskeletons in nondisabled adults. Discrepancy modeling aims to resolve dynamical inconsistencies between model predictions and real-world measurements. Neural networks identified models of (i) Nominal gait, (ii) Exoskeleton (Exo) gait, and (iii) the Discrepancy (i.e., response) between them. If an Augmented (Nominal+Discrepancy) model captured exoskeleton responses, its predictions should account for comparable amounts of variance in Exo gait data as the Exo model. Discrepancy modeling successfully quantified individuals' exoskeleton responses without requiring knowledge about physiological structure or motor control: a model of Nominal gait augmented with a Discrepancy model of response accounted for significantly more variance in Exo gait (median R2 for kinematics (0.928-0.963) and electromyography (0.665-0.788), (p<0.042)) than the Nominal model (median R2 for kinematics (0.863-0.939) and electromyography (0.516-0.664)). However, additional measurement modalities and/or improved resolution are needed to characterize Exo gait, as the discrepancy may not comprehensively capture response due to unexplained variance in Exo gait (median R2 for kinematics (0.954-0.977) and electromyography (0.724-0.815)). These techniques can be used to accelerate the discovery of individual-specific mechanisms driving exoskeleton responses, thus enabling personalized rehabilitation.

A Transformer-Based Neural Network for Gait Prediction in Lower Limb Exoskeleton Robots Using Plantar Force

Lower limb exoskeleton robots have shown significant research value due to their capabilities of providing assistance to wearers and improving physical motion functions. As a type of robotic technology, wearable robots are directly in contact with the wearer's limbs during operation, necessitating a high level of human-robot collaboration to ensure safety and efficacy. Furthermore, gait prediction for the wearer, which helps to compensate for sensor delays and provide references for controller design, is crucial for improving the the human-robot collaboration capability. For gait prediction, the plantar force intrinsically reflects crucial gait patterns regardless of individual differences. To be exact, the plantar force encompasses a doubled three-axis force, which varies over time concerning the two feet, which also reflects the gait patterns indistinctly. In this paper, we developed a transformer-based neural network (TFSformer) comprising convolution and variational mode decomposition (VMD) to predict bilateral hip and knee joint angles utilizing the plantar pressure. Given the distinct information contained in the temporal and the force-space dimensions of plantar pressure, the encoder uses 1D convolution to obtain the integrated features in the two dimensions. As for the decoder, it utilizes a multi-channel attention mechanism to simultaneously focus on both dimensions and a deep multi-channel attention structure to reduce the computational and memory consumption. Furthermore, VMD is applied to networks to better distinguish the trends and changes in data. The model is trained and tested on a self-constructed dataset that consists of data from 35 volunteers. The experimental results show that FTSformer reduces the mean absolute error (MAE) up to 10.83%, 15.04% and 8.05% and the mean squared error (MSE) by 20.40%, 29.90% and 12.60% compared to the CNN model, the transformer model and the CNN transformer model, respectively.

Using human-in-the-loop optimization for guiding manual prosthesis adjustments: a proof-of-concept study

Introduction: Human-in-the-loop optimization algorithms have proven useful in optimizing complex interactive problems, such as the interaction between humans and robotic exoskeletons. Specifically, this methodology has been proven valid for reducing metabolic cost while wearing robotic exoskeletons. However, many prostheses and orthoses still consist of passive elements that require manual adjustments of settings. Methods: In the present study, we investigated if human-in-the-loop algorithms could guide faster manual adjustments in a procedure similar to fitting a prosthesis. Eight healthy participants wore a prosthesis simulator and walked on a treadmill at 0.8 ms-1 under 16 combinations of shoe heel height and pylon height. A human-in-the-loop optimization algorithm was used to find an optimal combination for reducing the loading rate on the limb contralateral to the prosthesis simulator. To evaluate the performance of the optimization algorithm, we used a convergence criterium. We evaluated the accuracy by comparing it against the optimum from a full sweep of all combinations. Results: In five out of the eight participants, the human-in-the-loop optimization reduced the time taken to find an optimal combination; however, in three participants, the human-in-the-loop optimization either converged by the last iteration or did not converge. Discussion: Findings from this study show that the human-in-the-loop methodology could be helpful in tasks that require manually adjusting an assistive device, such as optimizing an unpowered prosthesis. However, further research is needed to achieve robust performance and evaluate applicability in persons with amputation wearing an actual prosthesis.

Advances on mechanical designs for assistive ankle-foot orthoses

Assistive ankle-foot orthoses (AAFOs) are powerful solutions to assist or rehabilitate gait on humans. Existing AAFO technologies include passive, quasi-passive, and active principles to provide assistance to the users, and their mechanical configuration and control depend on the eventual support they aim for within the gait pattern. In this research we analyze the state-of-the-art of AAFO and classify the different approaches into clusters, describing their basis and working principles. Additionally, we reviewed the purpose and experimental validation of the devices, providing the reader with a better view of the technology readiness level. Finally, the reviewed designs, limitations, and future steps in the field are summarized and discussed.