Autonomous underwater vehicle trajectory tracking control method for time-varying dynamics

Abstract

The invention provides an autonomous underwater vehicle trajectory tracking control method for time-varying dynamics, and belongs to the field of deep reinforcement learning and underwater vehicle intelligent control. The method comprises the following steps: firstly, establishing a trajectory tracking control problem of the autonomous underwater vehicle in a time-varying dynamic environment; then constructing a batch Markov decision process model of the trajectory tracking problem according to the time-varying dynamics of the aircraft; and then constructing a strategy network based on an attention mechanism, training the strategy network by using a meta-reinforcement learning method, and finally optimizing to obtain a control strategy to control the autonomous underwater vehicle to complete a trajectory tracking task in a time-varying dynamic environment. According to the method, the time-varying dynamic characteristics of the autonomous underwater vehicle are considered, so that the provided model-free controller is more suitable for an actual scene compared with a general reinforcement learning model, the control scheme can be better applied to practice, and meanwhile, the control method has high precision and high generalization.

Description

technical field [0001] The invention belongs to the field of deep reinforcement learning and intelligent control of underwater vehicles, and in particular relates to an autonomous underwater vehicle trajectory tracking control method for time-varying dynamics. Background technique [0002] The development of deep-sea submarine science is highly dependent on deep-sea exploration technology and equipment. Due to the complex deep-sea environment and extreme conditions, currently, deep-sea operating autonomous underwater vehicles are mainly used to replace or assist humans in deep-sea detection, observation and sampling. Over the past two decades, autonomous underwater vehicles (AUVs) have been used for a variety of underwater tasks, including seabed mapping, pipeline inspection in the oil and gas industry, missing aircraft wreckage and the location of pollution sources, etc. However, these tasks are extremely difficult and even dangerous for humans. In AUV motion control, a ch...

AI Technical Summary

Problems solved by technology

For the time-varying dynamic environment, the method divides the original tracking problem into multiple independent Markov decision-making processes, and uses the meta-reinforcement learning framework to train the policy network. This method can solve the time-varying dynamics problem; The method can not solve the defects of the time-varying dynamic model, can not follow the general reference trajectory and other problems, and realize the high-precision tracking control of the general reference trajectory in the time-varying dynamic environment

Images