Author ORCID Identifier

0000-0001-6088-781X

Document Type

Dissertation

Date of Award

8-31-2025

Degree Name

Doctor of Philosophy in Biomedical Engineering - (Ph.D.)

Department

Biomedical Engineering

First Advisor

Xianlian Alex Zhou

Second Advisor

Sergei Adamovich

Third Advisor

Jongsang Son

Fourth Advisor

Karen J. Nolan

Fifth Advisor

Jean-Francois Daneault

Abstract

As the global population ages, the demand for wearable assistive technologies continues to rise, driven by their potential to enhance mobility and independence in older adults. Effectively designed controllers for lower-limb exoskeletons to assist sit-to-stand (STS) and walking are crucial for delivering efficient, safe, and comfortable assistance during daily activities. Traditionally, controller optimization involves biomechanical modeling and user-specific customization. Musculoskeletal simulations play a central role in this process by providing insights into human-exoskeleton interaction dynamics, thereby informing and refining control strategies.

This work presents a simulation-driven approach for developing exoskeleton controllers for walking and STS using two distinct methods: optimal control and deep reinforcement learning (DRL). For walking, predictive gait simulations with assistance are generated using OpenSim Moco (direct collocation) and SCONE (single shooting), applied across generic and subject-specific musculoskeletal models. Assistive torque profiles are co-optimized within these simulations. Among single joint assistance conditions, hip assistance achieves the greatest cost of transport (COT) savings of up to —31.8% at 50 Nm, consistently outperforming knee and ankle assistance. Combining hip-knee-ankle assistance yields the highest savings, reaching —48.5% at 50 Nm. For STS, a DRL framework employing the proximal policy optimization (PPO) algorithm is developed to generate muscle-driven movements while co-optimizing hip and knee torque assistance. The combined hip-knee assistance at 50 Nm limit leads to substantial reductions in muscle activation, by 68.6% in gluteus maximus and 73.2% in the vastus muscle group, compared to the unassisted baseline.

Finally, a human subject experimental evaluation is conducted using a newly designed hip-knee exoskeleton device to validate the feasibility and effectiveness of transferring simulation-learned neural network controllers to real hardware. Metabolic measurements reveal moderate reductions in energy expenditure during assisted fast gait and STS compared to zero-torque baselines, with peak assistance torques below 10 Nm. However, exoskeleton fit, user familiarity, comfort, and latency issues may have impacted performance, highlighting the need for hardware improvements. These findings nonetheless support the practical viability of transferring computationally optimized controllers to physical devices.

In summary, this work demonstrates the utility of simulation-based approaches for developing effective and transferable controllers for lower-limb exoskeletons. These methods provide a foundation for streamlining exoskeleton controller development and advancing practical, intelligent assistive technologies.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.