Pairing Predictive Simulations and Human Experiments in Ankle Exoskeleton Walking

Weiß A, Chen A, Janischowky DB, Shih I, Phillips G, Selinger JC, Koelewijn A (2024)

Publication Type: Conference contribution

Publication year: 2024

Publisher: IEEE

City/Town: New York City

Pages Range: 925-930

Conference Proceedings Title: 2024 10th IEEE RAS/EMBS International Conference for Biomedical Robotics and Biomechatronics (BioRob)

DOI: 10.1109/BioRob60516.2024.10719905

Abstract

Exoskeletons can enhance human gait and reduce metabolic energy expenditure, potentially enabling users to walk further and longer before experiencing fatigue. However, metabolic energy reductions often require hours of walking to tune control parameters to the user and allow the user to adapt to the device through complex human-machine interactions. Predictive simulations could expedite this process by finding near-optimal device and gait parameters ahead of human testing. Here, we investigated if we could create participant specific predictive gait simulations that replicate energetic, kine-matic, and kinetic adaptations to walking with bilateral ankle exoskeletons. We hypothesized that our predictive simulations would generate similar metabolic reductions to those achieved by our human participants' and that these reductions would be explained by similar kinematic and kinetic adaptations. Our hypothesis was partially confirmed, as the simulations predicted an average change in metabolic energy expenditure (12.6± 3.3 %) that matched the experimentally measured change (13.0 ±7.1 %). Moreover, the ankle kinematics during both base-line and exoskeleton walking were similar. However, underlying kinetic responses during exoskeleton walking differed between the simulations and experiments. Furthermore, we did not correctly predict differences in metabolic savings for individual participants. We expect that these differences are related to factors such as individual adaptation rate, muscle activity, and musculotendon dynamics. We conclude that our predictive simulations are currently not accurate enough to predict individual adaptations when walking with an exoskeleton, despite being able to well predict average group changes in metabolic cost and kinematics.

Authors with CRIS profile

Alexander Weiß Machine Learning and Data Analytics Lab Anne Koelewijn Professur für Computational Movement Science

Additional Organisation(s)

Sonderforschungsbereich 1483/1 - Empathokinästhetische Sensorik - Sensortechniken und Datenanalyseverfahren zur empathokinästhetischen Modellbildung und Zustandsbestimmung (SFB EmpkinS)

Involved external institutions

Queen's University CA Canada (CA)

How to cite

APA:

Weiß, A., Chen, A., Janischowky, D.B., Shih, I., Phillips, G., Selinger, J.C., & Koelewijn, A. (2024). Pairing Predictive Simulations and Human Experiments in Ankle Exoskeleton Walking. In 2024 10th IEEE RAS/EMBS International Conference for Biomedical Robotics and Biomechatronics (BioRob) (pp. 925-930). New York City: IEEE.

MLA:

Weiß, Alexander, et al. "Pairing Predictive Simulations and Human Experiments in Ankle Exoskeleton Walking." Proceedings of the 2024 10th IEEE RAS/EMBS International Conference for Biomedical Robotics and Biomechatronics (BioRob) New York City: IEEE, 2024. 925-930.

BibTeX: Download