Neural reflexive control of a mobile robot

Abstract

A control strategy for a service mobile robot in which the motion and the navigation of the vehicle are based on primitive reflexive behaviors that incorporate no a priori planning is presented. Such a scheme provides the robot with the ability to run through the whole free space of a hall with unknown obstacles while avoiding collisions, using only a limited set of primitive behaviors. It uses several proximity sensors placed around the robot, a neural-network-based controller for the locomotion system, and a simple odometer-based navigation system for point-to-point movements. The robot moves in a stripe pattern and avoids obstacles by going through already covered regions. If the robot runs into an obstacle, it circumnavigates it and returns to the initial stripe. At the end of each stripe where the workspace perimeter is reached, the robot turns and moves to the next stripe until the entire area is covered.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>