Optimizing Exoskeleton Assistance with Neuromusculoskeletal Simulations and Reinforcement Learning

Xianlian Alex Zhou, Ph.D.
Associate Professor of Biomedical Engineering
NJIT

Abstract: Robotic exoskeletons represent a breakthrough technology for enhancing motor performance in both healthy and aging individuals, as well as improving gait in mobility-impaired patients. However, designing and evaluating these systems remains a complex, time-consuming, and expensive process. This typically involves prototyping, human testing, and iterative design refinement. A key challenge lies in delivering precisely timed assistance tailored to the wearer’s movements, which requires a deep understanding of muscle biomechanics and its adaptation to the assistance. In this talk, we will explore several case studies demonstrating the use of neuromusculoskeletal (NMSK) simulations to optimize exoskeleton assistance and analyze the resulting neuromuscular responses. By integrating human NMSK models with robotic exoskeleton control, we conduct virtual human-in-the-loop evaluations, enabling us to assess and optimize exoskeleton performance in a simulated environment. Specifically, we present two methods for predicting optimized walking assistance: one based on optimized muscle reflex control and the other using direct collocation motion optimization. We will also showcase how deep reinforcement learning (DRL), when combined with NMSK simulations, can be leveraged to develop adaptive exoskeleton controllers for various activities, without the need for human experiments. Our findings highlight the exceptional potential of NMSK simulations in optimizing different types of exoskeleton assistance, while also offering valuable insights into the co-adaptation between human muscle and exoskeleton control. Lastly, we will discuss the deployment of these optimized assistive strategies in real-world systems and share the results of their experimental evaluations. Our research underscores the transformative role of NMSK simulations and advanced control methods in shaping the future of exoskeleton technology, with far-reaching implications for both rehabilitation and human augmentation.

Bio: Dr. Zhou is an Associate Professor of Biomedical Engineering at the New Jersey Institute of Technology (NJIT) and the Director of the BioDynamics Lab. He earned his Ph.D. in Mechanical Engineering from the University of Iowa in 2007, following his B.S. and M.S. degrees from Shanghai Jiao Tong University. Before joining NJIT in 2018, Dr. Zhou was a Principal Research Scientist at CFD Research Corporation in Huntsville, Alabama, where he led the Human Performance and Biodynamics group. At CFD Research, he served as Principal Investigator (PI) or Co-PI on more than 10 Department of Defense (DoD)-sponsored projects focused on human performance and injury protection. At NJIT, Dr. Zhou’s current research focuses on computational biomechanics, wearable robotics, and personalized medicine. His recent work has been primarily supported by the U.S. Army, the DoD Defense Health Agency, and the CDC/NIOSH.