Sensor Based Path Planning in Highly Constrained Environments for Agile Autonomous Vehicles

This paper presents at a class of new methods which explore and describe an unstructured environment according to the free space as seen by a vehicle within the environment. The proposed methods use such sensor-based information to present a safe, goal-seeking path through the environment. The methods are suitable to both online, reactive pathnding and obstacle avoidance as well as online or offline global navigation and goal-seeking behaviours. Experimental results on ground vehicles and in simulation are presented to demonstrate the path-planning capabilities of the proposed methods