Guidance and Control of a Mobile Robot Using Neural Network Correction Based on a Remotely Located Sensor

Abstract

This paper presents the guidance and control of a mobile robot using a neural network. Location of a mobile robot is determined by the global laser sensor remotely located from the robot. A cascaded controller is used as a primary controller for position control of the robot, and a deviated position error is corrected by a neural network using the reference compensation algorithm. A car-like robot has been built for the test. Several control algorithms are investigated and tested. Among them, the cascaded controller with compensation by a neural network performs best