This study presents a novel attempt to replicate human neuromechanical experiments in a virtual environment using a digital twin of the human model. The model was enhanced using a human motion simulation platform, MyoSuite, and controlled by reinforcement learning. The study compared the movement of a human elbow and the one controlled by an RL agent. The RL agent showed higher elbow impedance, likely due to quicker reaction time and superior sensory capabilities. The findings suggest that RL-controlled digital twins could be beneficial for neuromechanical research and rehabilitation therapy design.
Publication date: 6 Feb 2024
Project Page: Not Provided
Paper: https://arxiv.org/pdf/2402.02904
