Obstacle Avoidance for a Heterogeneous Formation Using the Null Space-based Behavioral Approach and Potential Fields

Abstract

This paper proposes an obstacle avoidance approach based on null space behavioral control to autonomously guide a formation of two mobile robots, a differential-drive wheeled platform and an unmanned aerial vehicle, to overtake an obstacle modeled as a potential field. The controller is designed considering the paradigm of virtual structure, in this case the tridimensional straight line linking the two robots. Using such a paradigm the agents in the formation are controlled in such a way that the formation reaches and then preserve a desired shape as well as move according the task being accomplished. When the point of interest for control gets close to a potential higher than a predefined threshold the controller, which is based on the null space behavioral control approach works on the control signals sent to the robots, modifying them to assure that the formation deviate from the obstacle. The use of the null space technique allows giving higher priority to the subtask of avoiding the obstacle, leaving the subtask of moving the formation and preserve its shape with a lower priority, thus generating a way to move around an obstacle following a path of low potential. Note that if the priority were changed the formation would continue moving ahead, what certainly would cause a collision. In this article just fixed obstacles, with well known positions, and a family of Gaussian bell potential curves associated to them, are considered.