2.5-dimensional angle potential field algorithm for the real-time autonomous navigation of outdoor mobile robots

Abstract

A new real-time algorithm for the autonomous navigation of mobile robots equipped with laser scanners is proposed in this paper. Different from the existing algorithms designed for 2-dimensional navigation problems, the new algorithm introduces the height information of the obstacles into the guidance process and behaves as a 2.5-dimensional angle potential field algorithm (2.5D-APF) to fulfill the navigation requirements under complex outdoor terrain conditions. First, one laser scan is partitioned into two kinds of function sectors: guidance sector and inspecting sector. Then, the guidance sector and the inspecting sectors are reconstructed to form a virtual guidance scan, where the 2.5D information is taken into account. Finally, the conventional APF is improved to analyze the virtual guidance scan and generate the navigation orders. The new algorithm is tested on a tracked mobile robot, and the experimental results validate the proposed algorithm.