A Team Theoretic Approach to Multi-Agent Cooperative Search without Inter-Agent Communication
Titas Bera, Mohit Ludhiyani, Arup Kumar Sadhu, Ranjan Dasgupta · AIAA SCITECH 2022 Forum · 2022
View Video Presentation: https://doi.org/10.2514/6.2022-0758.vid This paper addresses the problem of autonomous multi-agent cooperative search without communication in an unknown environment using a decentralized framework. The cooperative nature of the multi-agent search implies that an individual agent’s action should corroborate towards optimizing a team performance. This problem is challenging as agents neither exchange information about the environment nor broadcast actions among themselves. In the context of multi-agent search, this work proposes an integrated decision and control theoretic solution which is capable of generating feasible controls for the agents in pursuit of achieving a team objective. Specifically, a perception based algorithm (p-CS) is developed which allows an agent to estimate probable strategies of other agents and realize an integrated decision and control based on such estimates. The probabilistic completeness of the proposed algorithmic method is shown theoretically. The performance of the co-operative search algorithm is compared with greedy strategies using Monte-Carlo simulations for different example scenarios and with a known look ahead q-Step search method. The simulations also show the robustness of the solution with respect to the agents’ inaccuracies in perception.