The learning mechanisms underlying the rapid and energy-efficient adaptation of human locomotion to varying environmental and physical conditions are not well understood. Our goal is to model locomotion adaptation using a feedback controller, reinforcement learning component, and memory mechanism for curriculum learning tasks such as sequences of split-belt treadmill and exoskeleton, and temporal credit assignment. Next steps also include incorporating hip and ankle exoskeleton dynamics and control to the current model and converting existing work for open source distribution. With a more robust model of human motor learning, we can better understand locomotion learning disorders, improve rehabilitation methods, and design more supportive robots.

I am participating in SuperUROP because I want to gain meaningful, long-term, research experience in the motor control field. I’ve previously taken courses in reinforcement learning, machine learning, and robotics and can expand this knowledge towards real world applications. Additionally, Im excited to collaborate and learn from members of the lab and contribute towards open source science and neuromotor rehabilitation.