Nonlinear Model Predictive Control for Mobile Medical Robot Using Neural Optimization

Mobile medical robots have been widely used in various structured scenarios, such as hospital drug delivery, public area disinfection, and examinations. Considering the challenge of environment modeling controller design, how to achieve information from human demonstration a directly arouse our interests. Learning skills is powerful way that can reduce complexity algorithm searching space. This especially true when naturally acquiring new skills, mobile robot must learn interaction with being or limited programming effort. In this article, learning scheme nonlinear model predictive control (NMPC) proposed for path tracking. The learning-by-imitation system consists two levels hierarchy: first level, multivirtual spring-dampers presented imitation robot's trajectories; second NMPC method motion system. strategy utilizes varying-parameter one-layer projection neural network solve an online quadratic optimization via iteration over receding horizon. evaluated on emulated trajectory simulation three scenarios experiment.