Path planning method integrating dense convolution network and competitive architecture

Abstract

The invention discloses a path planning method integrating a dense convolution network and a competitive architecture. The method includes the following steps that: a mobile robot samples mini-batch pieces of conversion information from an experience playback memory, and selects one of two fusion path planning networks as an online network and the other as a target network according to preset rules; the maximum value of a predicted target action value function is obtained through a predicted online action value function Q(s, a; w) and a corresponding greedy action; a loss function at a currenttime step is calculated according to the maximum value of the predicted target action value function and the predicted online action value function; and an online network weight w is updated with a stochastic gradient descent method according to the loss function. According to the path planning method of the invention, the dense convolution network and the competitive architecture form a more lightweight fusion path planning network, and therefore, model parameters are simplified, training costs are reduced, planning time is shortened, and the requirements of high-speed path planning can be satisfied to a certain extent.

Description

technical field [0001] The present invention relates to the fields of deep learning and artificial intelligence. Specifically, the present invention is a path planning method that integrates dense convolutional networks and competitive architectures. Background technique [0002] The path planning of a mobile robot refers to the autonomous calculation of the robot's movement path under the given environment, robot model, and specified planning goal. In practice, people generally use traditional methods such as ant colony algorithm and genetic algorithm to solve the problem. However, with the continuous development of science and technology, the environment faced by mobile robots is becoming more and more complex and changeable. Traditional path planning methods can no longer meet the needs of mobile robots. need. [0003] In response to this situation, people proposed Deep Reinforcement Learning (DRL), which integrates deep learning and reinforcement learning, in which deep...

AI Technical Summary

Problems solved by technology

Generally, path planning is based on the DQN network in DRL and its improved algorithm. However, the DQN algorithm has overestimation of the action value, and the training of the DQN network is not fast enough to meet the needs of high-speed path planning.

Images