Traversable Terrain Modeling and Performance Measurement of Mobile Robots

Abstract

Abstract : In this paper, we have described a technique for terrain traversability assessment modeling of mobile robots operating in natural terrain and presented a fast near-optimum algorithm for autonomous navigational path planning of mobile robots in rough terrain environments. The proposed method is based on visual sensing of terrain salient features and analysis of geo-location coordinates of the salient features. Using an algorithmic image processing technique, both free and obstacles spaces are differentiated and multiple candidate terrain paths are generated for optimization of trajectory terrain path of the robot. The algorithm uses a fuzzy logic terrain classifier to categories different salient features of the terrain. A virtual simulation is developed for terrain perception modeling and verification of generated trajectory path plans of the robot. The developed path-planning algorithm is computationally efficient, and suitable for implementation onboard autonomous robotic systems. Several different terrain conditions have been tested to validate the proposed approach.