Research On Environment Perception And Autonomous Collision Avoidance Of The AUV Based On Multi-Sonar Array

Since Autonomous Underwater Vehicle (AUV)AUV plays an increasing role in science research, nation defense, exploit and other underwater operations, the research of AUV technology on application are attracting more and more attention. Environmental perception and autonomic avoiding technology is one of the most important technologies of AUV, that among the increasing reseach of AUV technology. It is Environmental perception and autonomic avoiding technology that guarantees the safe of AUV and decided the level of AUV intelligent. As AUV can not adapt the problems of unknown environment based on known knownledge, the thesis do some research on the environmental perception and autonomic avoiding method in unknown environment. The method of the thesis can effectively improve the environmental perception capability of AUV towards the unknown environment, and realize the online optimization of environmental perception and autonomic avoiding behavior of AUV.Firstly, the thesis describes the integrated system frame, especially the sensors of the thesis describes the integrated system of AUV. Considering the effect of multi-beam sonar system is being limited, in the paper multi-sonar system is being using to replace the multi-beam sonar system. The multi-sonar system is the research object of the paper. The paper analyzes the differences between the performance of multi single-beam sonar system and the one of the multi-beam sonar system. What is more, the paper establishes kinematics equations in six degrees of freedom of AUV.Secondly, the paper realizes the environmental perception of the multi single-beam sonar system, which is based on the capture and understanding of sonar information. According to the performance of the single-beam sonars, the paper designs the time-sharing mode for multi single-beam sonar system, by which the rick of multi-sonar system been solved. As the data of sonars is indeterminacy, the paper also do some research on the understanding method of sonar data, which is helpful for AUV to provide dependable environment information. What is more, the paper put the establishing method of environment modeling for multi single-beam sonar system of AUV forward, which is based on improvement of the grid technology.Thirdly, the autonomic avoiding method of AUV is given in the thesis, which is based on the Q-learning arithmetic. In unknwn environment, the online learning and optimization ability of the avoiding behavior of AUV is very important for AUV to adapt the environment better. According to the practical phenomena of autonomic avoiding behavior of AUV, the thesis gives the realization method of every element of Q-learning arithmetic. The research of the thesis realize the online optimization of autonomic avoiding behavior of AUV based on the environment information which is gained by the multi single-beam sonar system.Lastly, the results indicate that the autonomous collision avoidance method for AUV based on Q-learning algorithm can achieve avoidance strategy of AUV online learning and optimization, and make AUV rapidly adapt to unknown environment and new mission requirements.