Linear model predictive movement control with obstacle avoidance for an inverted pendulum robot

Abstract

This paper presents a movement control methodology with obstacle avoidance control for personal robots that can work alongside humans. The methodology attempts to secure safety for users and stability of robots. These safety and stability are very important parameters in designing personal robot, and the personal robot controller must consider these parameters. The proposed approach uses model predictive control (MPC) that is a suitable control technique to achieve safety and stability whenever the robot avoids obstacles. Furthermore, the personal robot in this study interacts with people. In this study, we adequately select a two-wheel inverted pendulum (IP) type robot. The proposed controller can continue to secure safety and stability by imposing additional constraints on the robot coping with obstacles. To achieve this function, the proposed controller has the functions to control multi variables such as distance, velocity, turn-angle, tilt-angle and so on, in same time under some constraints on controlled objects. The proposed controller has time-varying constraints in relation to velocity for ensuring safety on obstacle avoidance motion by only turn­angle control. In addition, it has also time-invariant constraints on the controller input and body tilt angle for stability. Finally, the simulation validation results from a movement control with obstacle avoidance for an IP robot.