An Analytical Approach to Avoid Obstacles in Mobile Robot Navigation

Abstract

A nonlinear supervised globally stable controller is proposed to reactively guide a mobile robot to avoid obstacles while seeking a goal. Whenever the robot detects an object nearby, its orientation is changed to be aligned with the tangent to the border of the obstacle. Then, the robot starts following it, looking for a feasible path to its goal. The supervisor is responsible for deciding which path to take when the robot faces some particular obstacle configurations that are quite difficult to deal with. Several simulations and experiments were run to validate the proposal, some of which are discussed here. To run the experiments, the proposed controller is programmed into the onboard computer of a real unicycle mobile platform, equipped with a laser range scanner. As for the simulations, the models of the same experimental setup were used. The final conclusion is that the nonlinear supervised controller proposed to solve the problem of avoiding obstacles during goal seeking has been validated, based on the theoretical analysis, and the simulated and experimental results.