An Obstacle Avoidance Method for Sonar-based Robots Avoiding Shape Changeable Obstacles

In this paper, an obstacle avoidance method for undersea unmanned vehicle (UUV) is proposed. The sensors employed are sonar-based ones, as the other type of sensors, like visual-based sensors, radar-based sensors are not viable for the underwater environments. The deformation of the obstacles has been observed and learnt with the combination of Back Propagation Neural Network (BPNN), and the coordinate position of the obstacle is predicted by the robot. The navigation algorithm applied could navigate the UUV avoiding collisions with the obstacles. The simulation results which could demonstrate the validation of our proposed algorithm are also presented, which are implemented by Matlab.