Motion Planning of Autonomous Personal Transporter using Model Predictive Control for Minimizing Non-minimum Phase Behavior

A novel autonomous personal transporter is introduced. When the autonomous personal transporter is in self-driving conditions, there is various disturbance acceleration for the person riding on it. This disturbance acceleration should be compensated for a safe drive. Motion planning methods using LQR and discrete model predictive control are suggested for a cart inverted pendulum model. When the cart inverted pendulum model is controlled using LQR, there is non-minimum phase behavior causing undesired undershoot response. For minimizing the non-minimum phase behavior, we suggest discrete model predictive control using pre-planned motion trajectory. For experiments, we developed an autonomous driving robot called a ball-plate mecanum robot. On the ball-plate mecanum robot, we put an iron ball with which it is possible to measure position using a resistive touch sensor. The ball is basically stabilized by PID control. However, the ball position is out of center when there is a disturbance. By measuring the position of the ball, we investigate how much the disturbance acceleration can be compensated for using motion planning methods. LQR motion planning and MPC motion planning are compared through simulations and experiments, and we consider which motion planning method is more suitable for the autonomous personal transporter.