Evaluation of traversability of wheeled mobile robots on uneven terrains by a fractal terrain model

• Published in: Advanced robotics Vol. 21; no. 1-2; pp. 121 - 142
• Main Authors: Yokokohji, Yasuyoshi, Chaen, Satoshi, Yoshikawa, Tsuneo
• Format: Journal Article
• Language: English
• Published: Taylor & Francis Group 01.01.2007
• Subjects: FRACTAL THEORY; FRACTIONAL BROWNIAN MOTION; MOBILE ROBOT; PLANETARY ROVER; TERRAIN MODEL; TRAVERSABILITY
• ISSN: 0169-1864; 1568-5535
• DOI: 10.1163/156855307779293652

We propose a framework to evaluate the traversability of mobile robots on an uneven terrain by using an abstract terrain model made from fractional Brownian motions (fBms). Using fBm models, one can generate various terrain models by using only two parameters. In the proposed framework, traversability of mobile robots is evaluated in a statistical manner, i.e., by repeating computer simulation many times to get the probability that the robot can travel a certain distance on the abstract terrain modeled from a given target terrain. To validate the effectiveness of the proposed framework, we prepared two different kinds of rough terrain as the target terrains and experimentally evaluated the traversability of a mobile robot in several cases, changing the wheel size and body configuration. Then, we modeled these target terrains by using fBm and evaluated the traversability of the robot in computer simulations. Probabilities that the robot can travel a certain distance obtained from the simulation agreed well with the experimental results, indicating that simulation on abstract terrain models can predict the traversability of the robot on real terrains. The proposed framework would be useful to design a rover that should traverse robustly on a planet where precise terrain information cannot be obtained in advance. It would also be useful to check the performance of a novel rover mechanism in various terrain conditions.