Study at a glance
Summary
Disclaimer: The following summary details independent academic research. Theia3D is intended for research, sports performance and educational use only. It is not cleared as a medical device and is not intended to be used for the diagnosis, treatment, mitigation or prevention of any disease, injury or medical condition.
Editor's note: Theia3D version not specified in the paper. This study uses Theia3D as a measurement tool to derive whole-body angular momentum from markerless video, rather than evaluating Theia3D's own kinematic accuracy. The paper's Data Collection text does not name a specific camera brand or model for the markerless system used.
Study Overview
Design
A within-subject study systematically varying the magnitude and duration of torque assistance delivered by a bilateral robotic hip exoskeleton during simulated treadmill slip perturbations, quantifying gait stability using whole-body angular momentum derived from markerless motion capture.
Participants
Eight able-bodied young adults (3 women, 5 men; mean age 22.75 years; mean height 174.8 cm; mean weight 66.4 kg).
Data Collection
Each participant wore a custom bilateral robotic hip exoskeleton while walking at a fixed speed on a force-instrumented split-belt treadmill, experiencing unexpected anteroposterior slip-like perturbations delivered through an abrupt, transient change in belt velocity. Torque assistance was applied under 28 experimental conditions, including 25 combinations of five assistance magnitudes and five assistance durations using a trapezoidal torque profile, an existing energy-optimized spline profile, a no-assistance condition with the device worn but inactive, and a no-exoskeleton control condition, each repeated four times in randomized order for a total of 112 perturbation trials per participant. Trials were recorded using markerless motion capture cameras and processed through Theia3D to generate a full-body kinematic model used to calculate whole-body angular momentum; participants also rated their perceived stability after each trial using a standardized survey instrument.
Analysis
Whole-body angular momentum range, the difference between its maximum and minimum normalized value across the perturbed and recovery strides, was calculated for each trial and condition. A linear mixed-effects model tested the effects of assistance magnitude, duration, and their interaction on whole-body angular momentum range, with subject included as a random effect. A continuous response surface across the full assistance parameter space was estimated using interpolation, and bootstrap resampling was used to establish confidence intervals around the estimated optimal assistance parameters.
Key Findings
- Whole-body angular momentum range was governed by a significant interaction between assistance magnitude and duration, rather than magnitude alone. At short assistance durations, increasing torque magnitude significantly increased (worsened) whole-body angular momentum range, while at longer durations the effect of magnitude on stability became non-significant.
- The best-performing assistance parameter combination, identified at 25% of peak biological hip torque sustained over four times the perturbation length, significantly reduced whole-body angular momentum range by 25.7% relative to walking without the exoskeleton (p = 0.005) and by 27.4% relative to an existing energy-optimized controller (p = 0.02).
- An existing controller previously optimized for energy efficiency during steady-state walking showed no significant stability benefit over not wearing the exoskeleton at all during slip perturbations, directly supporting the authors' hypothesis that energetics-focused exoskeleton control is insufficient for improving reactive stability during an actual destabilizing event.
- Repeating the same perturbation protocol four times produced no significant change in whole-body angular momentum response (p = 0.29), indicating that the stability improvements observed were attributable to the specific assistance parameters tested rather than participants' motor adaptation or learning across repeated exposure to the same perturbation.
- Optimal assistance parameters, and perceived stability ratings, varied substantially between individual participants. While the theoretical optimal assistance magnitude and duration could be estimated across the group, bootstrap confidence intervals around the optimal magnitude specifically were wide, reflecting high inter-subject variability and underscoring that a single, universal assistance profile may not work equally well for every individual.
What This Means for Movement Science Researchers and Exoskeleton Development Teams
For teams developing lower-limb exoskeletons intended to improve stability rather than just reduce energy cost, this study offers a specific, actionable finding: assistance duration, not magnitude alone, determines whether torque assistance helps or actively hinders stability recovery during a slip, and a controller tuned for walking efficiency cannot be assumed to also provide a stability benefit during a genuine perturbation. The substantial inter-subject variability observed in optimal assistance parameters is an equally important takeaway, pointing toward the need for individualized, rather than universal, assistance tuning as this technology moves toward real-world deployment, particularly as this line of research extends toward older adult populations where fall risk and the benefit of effective stability support are greatest.
The Role of Theia3D in Movement Analysis
Theia3D was used as the markerless motion capture processing software throughout this study, generating a 16-segment, 54-degree-of-freedom full-body kinematic model from synchronized video that was used to calculate whole-body angular momentum, the study's primary stability outcome measure, across 112 perturbation trials per participant. The authors note that markerless motion capture, while carrying some reduction in kinematic accuracy compared with marker-based systems, was selected specifically to avoid the substantial marker occlusion that can occur during the rapid, destabilizing limb movements of a perturbation response, where missing marker data would otherwise be difficult to reliably reconstruct.
Study Limitations and Future Directions
The authors note that this study was conducted exclusively in healthy young adults as a necessary first step, given the fatigue and safety considerations associated with extensive perturbation-parameter testing, and that further data collection with older adult populations is needed before these findings can be extended to the population ultimately most relevant to fall-related stability support, since age-related gait changes may significantly affect how individuals respond to exoskeleton assistance. Practical time constraints in the study limited testing to a discrete set of assistance parameters for only anteroposterior slip perturbations, which limits understanding of how these findings might generalize to other perturbation types or directions. Whole-body angular momentum range, as calculated in this study, does not capture longer-term temporal recovery dynamics beyond the immediate perturbed and recovery strides, limiting the scope of these findings to immediate recovery behavior rather than the full recovery process. Perturbation onset was assumed to be known in advance for experimental consistency in this study, a real-world limitation that the authors note future work integrating online perturbation detection will need to address. The authors identify developing adaptive control profiles for a broader range of perturbation types, building frameworks for individualized assistance personalization, and leveraging biomechanical simulation to reduce the burden of extensive human-participant testing as clear next steps.
Citation: Tagliaferri, M. T., & Kang, I. (2026). Systematic evaluation of hip exoskeleton assistance parameters for enhancing gait stability during ground slip perturbations. IEEE Transactions on Biomedical Engineering. https://doi.org/10.1109/TBME.2026.3709674
Link: Read the full study
Does Exoskeleton Assistance Timing Matter More Than Torque Magnitude for Stability?
This study used Theia3D-derived whole-body angular momentum to show that a hip exoskeleton's assistance duration, not torque magnitude alone, determines whether it stabilizes or destabilizes recovery from a simulated slip, with a longer-duration assistance profile outperforming both no exoskeleton and an existing energy-optimized controller. If your work involves developing or evaluating stability-focused exoskeleton control strategies and you want to quantify dynamic balance from markerless video, talk to our team about using Theia3D in your own perturbation research.