Collision Avoidance System with Model Predictive Control for Mobile Robot Navigation

Abstract

Mobile robot mission always begins with the movement of the mobile robot to a certain location where the robot performs its duties. To carry out these movements, a control method is needed to move the mobile robot’s actuator (in the form of wheels or legs) and understand the situation around the robot (perception). This research aims to realize a method that can detect obstacle and distances as well as regulate their movement. The Model Predictive Control (MPC) method is proposed to assist control as part of a low-level controller. This research proposes the use of the NN method to detect obstacles and also as part of the high-level controller on the robot. The results obtained from this method, are the smaller the horizon value with a value of 10, the time needed to reach the desired coordinate point is shorter with a result of 57 seconds.