Obstacle Avoidance for Unmanned Ground Vehicles in Unstructured Environments
Lorenzo Pollini, Manuele Cellini, Roberto Mati, Mario Innocenti · AIAA Guidance, Navigation, and Control Conference and Exhibit · 2007
[Abstract] The paper presents an obstacle avoidance algorithm to be used for Unmanned Ground Vehicles applications. The algorithm improves, and extends the recently developed Null Space Based Behavioral Control technique. The method divides the problem into various tasks with increasing priority. Activities with lower priority do not interfere with those having higher priority. The scenario is assumed only partially known, and the complete environment is reconstructed during the mission, with the aid of stereoscopic vision sensors. The validity of the method is verified first via computer simulations, and then by performing field experimentation. BSTACLE avoidance is one of the more complex problems to be addressed within the context of autonomous vehicles guidance design. Difficulties increase if the initial knowledge of the scenario is limited, and the outside environment must be reconstructed online, during the motion of the vehicle, and without a priori information and/or cues. The literature offers a large number of methods for the solution of the obstacle avoidance problem, and several of them use modifications of the potential algorithm adapted to represent vehicle trajectories and paths. Potentialbased techniques have the advantage of being straightforward and of easy implementation. One of the limitations encountered by these methods is the presence of local minima, which can be addressed in several ways, for instance by using harmonic functions 1 . In addition, the complexity of the scenarios is limited under the application of this