Daniel Janischowsky

Daniel Janischowsky

Master's Thesis

Predictive simulations of gait with ankle exoskeleton that alters energetics


Alexander Weiß (M. Sc.), Prof. Dr. Anne Koelewijn, Megan Mcallister (Queen’s University Canada), Prof. Dr. Jessica Selinger (Queen’s University Canada)


01 / 2023 – 07 / 2023


Recent advances in hardware and software have allowed for the design of exoskeletons for many applications, such as restoring bipedal locomotion for persons post-stroke or with spinal cord injuries, enhancing the locomotor capabilities of emergency and military personal during load-carriage, reducing fatigue and augment strength in unimpaired individuals, providing robot-mediated physical therapy in a clinical setting and being used as haptic devices for virtual reality applications [1].

A main objective of exoskeleton design is to reduce metabolic cost, since humans minimize metabolic cost during gait. However, this objective is not easily achieved due to the complex interactions between a user and the exoskeleton. Often lengthy design procedures are required to optimize the controller of the exoskeleton, e.g. through parameter sweeping [2], or human-in-the-loop optimization [3], which require hours of walking by the user.

Therefore, we aim to save time by including gait simulations in the design process. When doing so, it is important to include interactions between the user and the exoskeleton, which can be done using forward simulations. We have recently shown that we can use simulations to predict a user’s response to a knee worn exoskeleton under different controller conditions [4].

In this thesis, the goal is to use a forward simulation approach to predict the effect of an ankle exoskeleton which helps to answer a variety of research questions:
1: With which function can we model the exoskeleton controller in a simulation?
2: What would be the correlation between simulation and experiment for step rate, joint angles, joint moments and metabolic cost?
3: Can we predict the exoskeleton’s controller parameters that lead to the optimal gait cycle with the lowest metabolic cost?

All simulation predictions will be validated with experimental tests performed by our collaborators, with ample opportunity for cross-disciplinary exchange.

[1] P. Agarwal and A. Deshpande, “Exoskeletons: State-of-the-Art, Design Challenges, and Future Directions”, Human Performance Optimization: The Science and Ethics of Enhancing Human Capabilities, 9780190455132, 234-259, 2019.
[2] P. Malcolm, et al., “A Simple Exoskeleton That Assists Plantarflexion Can Reduce the Metabolic Cost of Human Walking,” PLoS ONE, 8.2, e56137–7, 2013.
[3] D. Gordon, et al., “Human-in-the-Loop Optimization of Exoskeleton Assistance Via Online Simulation of Metabolic Cost”, IEEE Trans. Rob., 30, 1410-1429, 2022.
[4] A. D. Koelewijn and J. C. Selinger, “Predictions of step frequency adaptations with altered energy landscapes.,” IEEE Trans. Neural Syst. Rehabil. Eng., 30, 1931-1940, 2022.