Design method for unmanned surface vehicle tracking controller

Abstract

The invention discloses a design method for an unmanned surface vehicle tracking controller, and relates to the technical field of an unmanned surface vehicle tracking control technology. The method specifically comprises the following steps that: S1: according to unmanned surface vehicle tracking task requirements, firstly, designing a decision network, training to enable the decision network toobtain a decision capability, and designing an exploration function to explore an unmanned surface vehicle path tracking state; S2: designing a reward function, and obtaining the motion state of the unmanned surface vehicle through a state observation method. By use of the design method for the unmanned surface vehicle tracking controller, a deep reinforcement learning algorithm is adopted to train a deep convolutional neural network to serve as the unmanned surface vehicle tracking controller, the design process of the unmanned surface vehicle tracking controller is simplified, a neural network approaching form is favorably used for replacing mathematical derivation, in addition, the controller is automatically trained through a program, manual intervention is not required, calculation issimple, and transportability is high.

Description

technical field [0001] The invention relates to the technical field of unmanned boat path tracking, in particular to a design method for a surface unmanned boat path tracking controller. Background technique [0002] With the deepening of the intelligence of surface unmanned vehicles, the path tracking controller of surface unmanned vehicles is becoming more and more important. The traditional path tracking controller design method is based on mathematical calculation and analysis, which can prove the effectiveness of the algorithm in theory. , and efficient controller parameters can also be deduced. However, mathematical algorithms also have their inherent limitations, such as complex calculations, poor portability, and severe environmental impact. At the same time, reinforcement learning technology has been widely researched and applied in the fields of artificial intelligence, machine learning, and automatic control. , and is considered to be one of the core technologies...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the deficiencies in the prior art, the present invention provides a design method for the path tracking controller of the surface unmanned boat, which solves the limitations of the mathematical algorithm of the existing surface unmanned boat path tracking controller, complex calculation, poor portability and Seriously affected by the environment, it is impossible to realize the deep reinforcement learning algorithm to train the deep convolutional neural network as the path tracking controller of the surface unmanned vehicle, and the purpose of simplifying the design process of the path tracking controller of the surface unmanned vehicle cannot be achieved, and manual intervention is required question

Images