Multi-Agents Interaction Approach based on Graph Network and Reinforcement Learning
Lian Liu, Zimeng Wang · 2022 9th International Conference on Dependable Systems and Their Applications (DSA) · 2022
Due to the development of artificial intelligence technology, the research of unmanned equipment has gained great progress. However, how to make multi-agents in unmanned clusters to be reasonable for interaction is the key problem to be solved at present. And the characteristics of unmanned clusters can be derived from a variety of work modes. At the same time these mode are inseparable from the interaction and collaboration mechanism. Graph reinforcement learning is a new algorithm that combines graph neural network and reinforcement learning. This paper is based on graph reinforcement learning to solve the problem of interaction between various nodes in unmanned clusters. And the approach provides the technical guidance for the future development of unmanned cluster technology.