Real-time Obstacle Avoidance and Person Following Based on Adaptive Window Approach

Abstract

Autonomous mobile robots need safe navigation capabilities in order to interact with humans in a dynamic environment. As a step toward this goal, we are applying a realtime obstacle avoidance method for a robot to follow behind a person. This paper presents a laser-based people tracking and following while at the same time avoiding obstacles method. A novel feature of our approach is it considers potential risks for a robot to follow a person in an unknown environment. We implement this method in the Robot Operating System framework and verify it in both virtual and real environments.