Mobile robot navigation method and device, computer equipment and storage medium

Abstract

The invention relates to a mobile robot navigation method and device, computer equipment and a storage medium. The method comprises the following steps of: extracting features of a target point image and a scene image through a feature extraction module to obtain state features of a current state; resolving a preset expert track through an inverse reinforcement learning module to obtain a reward function; outputting a predicted execution action of a robot through a strategy network in an A3C reinforcement learning network, obtaining a predicted value function through a value network, and after the execution action obtains a next state, calculating a TD error according to the current state, the next state and the execution action to obtain a first loss function; obtaining an expert reward value according to the state features and a weight parameter, and obtaining a second loss function according to the network reward value and the expert reward value; and training the A3C reinforcement learning network and a reward network to obtain a trained mobile robot navigation model for navigation. According to the invention, the accuracy and efficiency of indoor navigation of the robot can be improved, and the generalization ability is high.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to a mobile robot navigation method, device, computer equipment and storage medium based on reward network and inverse reinforcement learning. Background technique [0002] Mobile robot navigation is one of the most basic tasks of robots. All kinds of robots must have perfect navigation capabilities before they can be put into practical applications. Robots under traditional navigation tasks will be equipped with RGB cameras or depth cameras, radar, GPS and other sensors, and some even rely on high-precision prior maps, etc., which require a large cost of manpower and material resources. In recent years, it has been realized through reinforcement learning. Robot navigation only needs visual sensors to achieve better results and reduces the dependence on other sensors, which has become a hot research topic. However, the complexity and uncertainty of the environment, the ...

AI Technical Summary

Problems solved by technology

However, the complexity and uncertainty of the environment, the low utilization of data and difficult generalization of reinforcement learning itself, and the diversity of tasks all pose challenges to the robot navigation of reinforcement learning.

[0003] The traditional robot navigation technology is mainly based on the understanding of the entire environment scene. It needs to use external auxiliary equipment and its own sensors to achieve accurate pose solution. However, it generally needs to rely on equipment and is complex and costly. Difficulty adapting to changes in the environment

[0004] The robot navigation technology using reinforcement learning has some obstacles of reinforcement learning itself, the training efficiency is low, the utilization rate of data is relatively low, and it needs a lot of interaction with the environment to get a better model; it is difficult for the unknown environment The reward function is well set, and the generalization performance for different scenarios is weak

Images