Design of an Intelligent Assistive Exoskeleton for Load-Carrying Based on FMG Signals

To address the issue of shoulder musculoskeletal disorders (MSDs) in material handling tasks, this paper proposes an intelligent assistive exoskeleton based on FMG (Force Myography) signals. The system features flexible switching between active and passive modes, combining the lightweight advantages of passive exoskeletons with the intelligent control capabilities of active exoskeletons. In active mode, the exoskeleton utilizes an FMG sensor armband to continuously capture muscle deformation signals in real time and employs deep learning models (CNN+LSTM) to identify load mass, thereby dynamically adjusting the assistive output. The system adopts a Model Predictive Control (MPC) + Sliding Mode Control (SMC) strategy to optimize assistive performance, where MPC ensures precise torque regulation, while SMC enhances disturbance rejection, improving the system’s adaptability and stability. Simulation experiments were conducted to validate the effectiveness of the system, and the results demonstrate that the exoskeleton can accurately distribute shoulder joint load while maintaining stable assistance under sudden disturbances.