A Multi-agent Hunting-Foraging Behavior Control Method

Abstract

The invention relates to a multi-Agent stalking-foraging behavior control method comprising the following steps: 1) dividing a state space formed by multiple agents and the current position of prey; 2) designing a reward function; 3) performing reinforced learning according to the state space divided in the step 1) and the reward function in the step 2), controlling the Agents to perform corresponding atomic actions to stalk the prey, and stopping when a stalking success condition is satisfied so as to achieve a foraging effect. Compared with the prior art, the method has high stalking efficiency.

Description

technical field [0001] The present invention relates to Agent round-up and foraging technology, in particular to a multi-Agent round-up and foraging behavior control method. Background technique [0002] As a very important branch of distributed artificial intelligence, multi-robot system has the characteristics of fault tolerance, strong robustness, and distributed coordination. It has become a hot spot that people pay close attention to in recent years. The main issues of multi-robot system research include group structure, task allocation, communication methods, collaborative learning, etc. In order to make the research more meaningful in actual scenarios, the researchers focused on some multi-robot tasks, including formation cooperation, search, round-up, etc. [0003] Multi-robot cooperative round-up is one of the effective methods to test the work efficiency of multi-robots. The multi-robot round-up process is to use three or more wheeled robots to first cooperate to...

AI Technical Summary

Problems solved by technology

The grid method divides the state space of the robot according to the position. Although the division is simple and easy to understand, its accuracy is not high due to its rough discretization method. On the other hand, if the discretization accuracy is increased, " "curse of dimensionality"

Images