Safety-triggered stochastic tracking control for a cushion robot by constraining velocity considering the estimated internal disturbance

Abstract A safety-triggered tracking controller is presented for constraining the velocity of a cushion robot, which is typically used to help elderly and disabled users perform activities of daily living. In a human-robot system, the robot can be described by using a stochastic model subject to internal disturbances due to random changes in the mass and posture of different users. The proposed algorithm uses model predictive method to adjust the velocity input to each omnidirectional wheel of the cushion robot, which is novel because the velocity constraint information obtained from a kinematic model is used to design a tracking controller based on a stochastic dynamic model for constraining the robot’s actual velocity, which is described by a stochastic system. The internal disturbance to the human-robot system is estimated by a neural network to improve the robustness of the tracking system. A safety-triggered controller that constrains the trajectory tracking error of the cushion robot was built that makes the system exponentially practically mean-square stable. Simulation results confirmed that the proposed method effectively constrains the velocity and trajectory of the cushion robot, which will allow it to support different users perform activities of daily living in safe motion states.