Research On AUV Obstacle Avoidance Method Based On Forward-looking Sonar

The obstacle avoidance problem is a key issue in the research process of the Autonomous Underwater Vehicle(AUV).Due to the unpredictable and variable of the working environment of AUV,it is difficult for AUV to obtain all the environmental information of its working environment in advance,therefore it is impossible to use a path planning algorithm based on priori information.Along with the development of research and the improvement of technical indicators,the working environment of AUV and the tasks to be completed are becoming more and more complex,AUV is bound to become more and more intelligent.AUV is expected to have the ability to adapt to the unknown environment and the ability to make decisions independently.In this paper,the related technologies of obstacle perception and 3D reconstruction of obstacles,local path planning and AUV obstacle avoidance in unknown environment are researched.Firstly,related object models in obstacle avoidance system are established,including the model of AUV and the models of underwater dynamic obstacles.In the geodetic coordinate and the AUV's body-fixed coordinate,the force that act on the AUV is analyzed,and the AUV is kinematically modeled and kinetic modeled.Three typical motion models of underwater obstacles are established,and their kinematical equations are calbulated.Sage-Husa adaptive Kalman filtering algorithm is applied in dynamic obstacle perception,the algorithm fullfills precision of perception and real-time requirement while don't need too much computation.The motion state estimation experiment of dynamic obstacles is designed to verify the algorithm's effectiveness.Secondly,due to the unknown of the obstacle' s height in sonar data,the method of obstacle perception that using dual forward-looking sonars is proposed,and to verify the effectiveness of this method,detection experiment that using dual forward-looking sonars is designed and carried out.The sonar data is modeled by grids and disposed by mathematics morphology method,and the obstacle is reconstructed after the information is synthesized into a single three-dimensional model.Thirdly,three shortcomings that exist in the traditional Artificial Potential Field(APF)are analyzed and modified.The principle of Rapidly-exploring Random Tree(RRT)is expounded,some shortcomings of the algorithm are pointed out and have been modified,and the optimization of RRT algorithm is realized.The RRT algorithm is introduced into the APF algorithm to form the APF-RRT hybrid algorithm,which solves the local minimum value problem in the APF algorithm.Obstacle avoidance experiment in complex environment is designed to verify the effectiveness of the hybrid algorithm.Finally,the simulation platform is built,the AUV obstacle avoidance strategy is designed,and the APF-RRT hybrid obstacle avoidance algorithm is applied to the AUV obstacle avoidance.The AUV obstacle avoidance simulation experiment is designed.The results show that the hybrid algorithm can guides the AUV to avoid obstacles safely.