Improved DQN for Dynamic Obstacle Avoidance and Ship Path Planning

The avoidance of collisions among ships requires addressing various factors such as perception, decision-making, and control. These pose many challenges for autonomous collision avoidance. Traditional methods have encountered significant difficulties when used in They are challenged to cope with the changing environment harsh motion constraints. In actual navigation ships, it is necessary carry out decision-making control under constraints ship manipulation risk. From implementation process perspective, a typical sequential anthropomorphic problem. order solve decision problem, this paper improves DQN by setting priority sample collection adopting non-uniform sampling, applied realize intelligent ships. It also verifies performance algorithm simulation environment.