Obstacle avoidance using fuzzy neural networks

If an underwater vehicle is to be completely autonomous, it must have the ability to avoid obstacles to safely operate. Because of the strong nonlinearity of the movement of the vehicle and the complexity of unknown oceanic environment, it is not satisfying to solve the problem using traditional control methods. However, the intelligent control techniques have the inherent superiority for solving strong nonlinear and complex problems. Therefore, a new method incorporating a fuzzy logic inference with an artificial neural network is presented in this paper. The method is used to establish a controller to control an autonomous underwater vehicle (AUV) to avoid obstacles. It not only exerts some expertise, but also endows the controller with adaptability. As a result, the AUV is provided with the ability of obstacle avoidance at the beginning, which greatly shortens the lime of network learning. On the other hand, the controller can adjust itself to the variations of oceanic environment. Results of simulation using a five degrees of freedom nonlinear maneuvering mathematical model of the vehicle show that the proposed method can be efficiently applied to obstacle avoidance of an AUV in complex and unknown oceanic environment.