AUV Self-adaptive Local Path Planning

Local path planning is one of the key technologies of robot technology. Compared to global path planning, it reflects more intelligence of robot. According to the motion characteristics of autonomous underwater vehicle (AUV) and the complexity of the sea environment,the thesis proposes a method to design an adaptive local path planner,which is suitable for AUV. The simulation results show that, according to the changes of sea environment, the planner can adjust planning strategies to improve the intelligence of AUV.Beacause sea current is a factor that can not be neglected in the local path planning, the thesis firstly analyses sea current influence on AUV movement considering the motion characteristics of AUV, then local path planner based on fuzzy logic is designed in condition with current and in condition without current. In order to avoid dimension disaster problem and departure from path or target point too far resulting from blindly obstacles-avoidance, the thesis adopts classification fuzzy rule base to control AUV, in order to ensure that when AUV avoids obstacles,it can reach target point at the same time and when AUV does not deviates from path very much,it does not collide obstacles. In the situation of without current or current is small, AUV can be controlled well using fuzzy rule base.In the situation of with current, because there is difference between personal experiences and practical situation,the thesis introduces Q learning to adjust rules. In order to improve the performance of Q learning, the thesis discusses and improves the problems of low speed of convergence, belief assignment, exploration and exploitation.In the thesis, Q learning does not directly adjust fuzzy rules,while it adjusts the width of membership function of fuzzy rules to achieve the goal of adjusting rules, and make AUV have the adaptability.Finally simulation experiment verifies the proposed theory method, proving the feasibility and effectiveness of this method.