Low time complexity collision avoidance method for autonomous mobile robots

Abstract

In this paper we present a fast and reliable method of obstacle avoidance for ground mobile robots both for outdoor and indoor navigation. The method compromises two contradictory approaches: non-complex implementation and human-like smooth steering. The method is applicable in any mobile robotic system regardless of used sensors. All calculations are done on virtual representation of surroundings, where the information about obstacles and a free space is presented in a unified way. The method results in a smooth movement, but is also equipped with a reactive calculation path that handles dangerous and damage-prone situations. The total calculation time for a desktop computer for presented method is 5[ms], resulting in real-time behavior of an algorithm.