Presentation
Development of an Imitation Learning-based Human-back Exoskeleton Interaction Simulation Framework
SessionThursday Poster Session
DescriptionWork-related low back pain remains one of the most common musculoskeletal disorders associated with repetitive lifting and manual material handling tasks. Lower-back exoskeletons have shown promise in reducing spinal loading during lifting. However, understanding how humans physically interact with these devices remains a major challenge. This work presents a simulation framework for studying human-back exoskeleton interaction using physics-based musculoskeletal modeling and imitation learning. More specifically, the framework integrates a musculoskeletal human model and a wearable lower-back exoskeleton within the MuJoCo simulation environment to reproduce lifting motions and interaction dynamics. The motion capture data are used to train an imitation learning policy capable of generating human-like lifting behavior under both assisted and unassisted conditions. The simulation allows detailed analysis of lumbar loading, movement adaptation, and interaction forces between the user and the exoskeleton. Moreover, the experimental data collected from human subjects wearing the same exoskeleton configuration will be used for validation. The proposed framework provides a scalable and flexible platform for evaluating exoskeleton assistance strategies, improving biomechanical understanding, and supporting future development of human-centered exoskeleton control systems.
Contributors
Alternate Presenter
Event Type
Poster
TimeThursday, October 22nd5:30pm - 6:30pm PDT
Location
Similar Presentations