The application of path planning algorithm based on deep reinforcement learning for mobile robots

Abstract

To meet the need for autonomous route planning for tour guide robots in tourist venues, this paper proposes a path planning algorithm based on deep reinforcement learning. The traditional Deep Q-learning Network (DQN) algorithm two defects - overfitting and overestimation. This paper adopts a method that discards the experience pool and treats behavioural values equally, which not only solves the shortcomings of the traditional method, but also satisfies the need for mobile robots to lead tourists on tours through autonomous learning. The paper analyses the principle and process of the method and compares it with the traditional method through experiments to verify that the method outperforms the traditional method in terms of accuracy and speed.