Application of Model Predictive Control in Physical Human-Machine Interaction

In this paper, an application of Model Predictive Control (MPC) to physical human-machine interaction is presented. In particular, the study focuses on the development of a mechatronic device able to apply a well-controlled mechanical impulsive force on the human body, for clinical investigation of postural control. The need for high accuracy and repeatability led to the MPC design, which is able to manage the non-linearities related to the human-machine interaction. The hardware architecture design of the prototype, the development of the control system (based on motor current saturation) and its optimization are presented. The results of experimental trials carried out in the laboratory and on healthy subjects show that the MPC algorithm is able to provide the accuracy and robustness requested by the application.