Research on motion planning based on self-learning behavior agent for AUV

According to the shortcomings of traditional reinforcement learning method when applied to autonomous underwater vehicle (AUV) engineering, such as generalization problem, risks by trial-and-error a low learning efficiency, the neural network and case based Q learning (NCQL) is proposed. The basic principle of NCQL is making use of neural net to solve generalization problem, and case based learning to make sure the convergence of learning process, and the algorithm steps are introduced. The elements of NCQL based self-learning behavior agent are introduced. Simulation tests results show that the NCQL is well done on its convergence property, and the speed of convergence is fast. NCQL has the properties of on-line learning and self-adapt learning, and it is suitable for motion planning of AUV in an unknown environment.