Soccer robot cooperation method based on reinforcement learning

Abstract

The invention discloses a soccer robot cooperation method based on reinforcement learning. The soccer robot cooperation method comprises the steps that S1, a reinforcement learning basic model of soccer robots is constructed based on a Sarsa (lambda) algorithm with communication added, and a reward and punishment mechanism r of the reinforcement learning basic model is set; S2, a specified numberof state variables are defined based on the distance and angle between the soccer robots; and S3, operational action sets of the soccer robots are set, and the soccer robots select the next action based on the reward and punishment mechanism r, the state variables and the communication between the soccer robots. According to the soccer robot cooperation method based on the reinforcement learning,the reward and punishment mechanism in set up in the established reinforcement learning basic model, the soccer robots choose the next action according to the current environment and the reward and punishment mechanism, learning and updating are conducted by the soccer robots communicating with one another, and the cooperation efficiency of the soccer robots is effectively improved.

Description

technical field [0001] The invention belongs to the field of soccer robots, and in particular relates to a soccer robot collaboration method based on reinforcement learning. Background technique [0002] As a typical multi-soccer robot system, the football robot competition provides a good experimental platform for the study of intelligence theory and the integrated application of various technologies. The capability requirements are getting stronger and stronger, which involves a series of research topics such as robot positioning, path planning, coordinated control, target tracking and decision-making. [0003] In recent years, many scholars and experts have made a lot of research results. For example, the Chinese patent application number 201120008202.2 discloses an intelligent robot competition device, including mechanical parts and circuit control parts. The mechanical parts include ball tables, consoles and robots. , the circuit control part includes a control module ...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the problem of low cooperation efficiency between soccer robots in the soccer robot competition in the above-mentioned prior art, the present invention proposes a soccer robot collaboration method based on reinforcement learning, which is based on adding communication-based Sarsa( The λ) algorithm constructs the basic model of reinforcement learning of soccer robots, and achieves high cooperation efficiency between each other through the reinforcement learning model and the communication between soccer robots. The specific technical scheme is as follows:

Images