Real-time obstacle avoidance of mobile robots

Abstract

This paper describes the formulation of the obstacle avoidance method to work in three-dimensional workspaces. We assume a spherical and omni-directional robot. The obstacle information is given in the form of points, which is the usual form in which sensor data is given. Obstacle avoidance methods are based on a real-time control process. Sensors collect information of the environment that is processed by the method to compute a motion command, the robot executes the motion and then the process restarts again. We show results that illustrate how this method successfully performs obstacle avoidance, and how it inherits the benefits of the original method being able to overcome the typical challenging problems of other obstacle avoidance methods but extrapolated to three dimensions.