Path planning for a mobile robot in a rough terrain environment

The robust control of a mobile robot in a rough terrain environment is a challenging endeavour, since any reliance on favourable surface or environmental conditions will inevitably lead to task failure. This paper presents the preliminary development of a unified navigation system, used to control a nonholonomic mobile robot in an a priori unknown outdoor domain. Accurate, high-resolution environmental data was gathered from a scanning laser rangefinder (ladar), which constitutes the robot's exteroceptive perception system. Using this data, a 3D tessellated environmental model was created that generically captures terrain traversability. By adapting the Rapidly-exploring Random Tree (RRT) approach to the tessellated model, an efficient kinodynamic path planning algorithm was devised that enables point-to-point trajectory traversal. This path planning strategy was found to be a computationally efficient method of producing robust and versatile path plans.