Virtual range scan for avoiding 3D obstacles using 2D tools

Abstract

This paper presents a new method for seamless fusion of arbitrary range sensors for mobile robot obstacle avoidance. This method, named Virtual Range Scan (VRS), is able to deal with arbitrary sensor configurations (2D and 3D) and it is independent of the underlying obstacle avoidance strategy. This makes it a very flexible approach that can reuse existing 2D obstacle avoidance algorithms for 3D obstacle avoidance. Additionally the VRS is able to keep obstacles in memory in order to incorporate them even if they are no longer visible in the current real sensor data. To combine this memory effect with the presence of dynamic objects in the robot surroundings, a motion tracker is used. We demonstrate the ability of the VRS to realize 3D obstacle avoidance using two different sensor configurations for taking 3D range data and classical two dimensional obstacle avoidance methods.