A Decentralized Multiagent Based Algorithm Design for Multiple Unmanned Vehicle Chasing Problem

Multiple unmanned vehicles chasing problem is a prominent protype in multiagent research due to its nature of the environmental dynamics and situation awareness complexity. Unmanned vehicles have to be highly intelligent and flexible to carry out their joint action so as to minimize the feasible moving range of the given target until being captured. However existing automatic control or multiagent based decision algorithms are either incapable of or based on some unrealistic assumptions that no feasible algorithms is able to handle all the challenges of the problem. In this paper, we make an initial effort and proposed a decentralized multiagent based algorithm design by introducing a social decision sequential model for each vehicle. With this model, vehicles can independently compute their expected utilities one by one but the joint action utilities of the team can be calculated by just once. Our design significantly decreased the computational cost of vehicles’ recursively searching for the best joint actions but our designed experimental results manifest that our designed vehicles can build fast and efficient decisions even they are not always the best.