Intelligent ship autonomous collision avoidance and path planning method based on reinforcement learning

Abstract

The invention discloses an intelligent ship autonomous collision avoidance and path planning method based on reinforcement learning, and the method comprises the following steps: 1), obtaining the surrounding environment information of a ship and the information of the ship, and sensing an environment state space; (2) calculating collision avoidance parameters according to relevant information including the obstacle position, the navigation speed and the navigation direction in the environment state space, and judging whether a collision risk exists or not; 3) if there is no collision risk, directly performing path planning; if the collision risk exists, building an intelligent ship collision avoidance model built by fusing the LSTM and the reinforcement learning principle, searching the optimal collision avoidance strategy for avoidance, and obtaining the navigation speed and the navigation direction required by collision avoidance of the ship; and 4) after execution of the collisionavoidance strategy is finished, determining a critical position point where the collision risk disappears as a new starting point according to a set condition, and then performing path planning againby utilizing a path planning algorithm. According to the invention, the LSTM neural network is introduced, and the optimal strategy is updated by using the Bellman equation so that the continuity of the collision avoidance action is realized.

Description

technical field [0001] The invention relates to path planning technology, in particular to an intelligent ship autonomous collision avoidance and path planning method based on reinforcement learning. Background technique [0002] With the development of artificial intelligence technology, the development of ship intelligence and automation has become a mainstream trend. At present, there are many studies on collision avoidance and path planning of intelligent ships. If existing methods are used for collision avoidance and path planning, it will often result in a large amount of analysis, failure to comply with collision avoidance rules, and untimely and unintelligent path planning. It is difficult to achieve fast collision avoidance decision-making and path planning. In order to ensure the safety of intelligent ship navigation, a method that can realize autonomous collision avoidance and real-time path planning is urgently needed. Contents of the invention [0003] The t...

AI Technical Summary

Problems solved by technology

At present, there are many studies on collision avoidance and path planning of intelligent ships. If existing methods are used for collision avoidance and path planning, it will often result in a large amount of analysis, failure to comply with collision avoidance rules, and untimely and unintelligent path planning. Difficult to achieve fast collision avoidance decision-making and path planning

Images