An Assistive Strategy for Compliantly Actuated Exoskeletons Using Non-Linear Model Predictive Control Method

Assistive exoskeletons are a category of exoskeleton robots designed to provide extra powers and energies for elderlies or healthy subjects. The control algorithms of these robots, termed as assistive strategies, should control the robot in way that a portion of the user's required torques is provided by the robot actuators. This paper uses non-linear model predictive control (NMPC) method to propose a new assistive strategy for compliantly actuated exoskeletons. The user's required torques are estimated by using an identified model of the human-exoskeleton system. The NMPC method is then used to control the series elastic actuators of the robot to provide a user-selected portion of the estimated torques. The successful performance of the proposed assistive strategy is verified through multiple simulations. The force tracking performance of the NMPC method is also compared with a PID controller. The results clearly show the outperformance of the NMPC method in providing lower tracking errors and less control efforts.