Planning strategy for intruder agent based on game theory and artificial potential field

Jiahong Xu,Xiaofeng Liu,Xu Zhou,Huan Wang,Qiguang Lin
DOI: https://doi.org/10.1109/RCAR52367.2021.9517080
2021-01-01
Abstract:In our work, we propose a planner based on game theory and artificial potential field (APF) that enables intruder to confront defender in 2-D environment. The intruder would compute the mixed strategy approximate Nash equilibrium in the space of agent APF trajectory to maximally approach the target region while evading the pursuit of defender. First, according to the improved APF, the finite pure strategy matrix of agents is obtained by considering the opponent's response to the agents' actions iteratively. Then, Particle Swarm optimization (PSO) is taken to compute the mixed strategy approximate Nash equilibrium based on the expected profit composed of pure strategy matrix. Final, extensive simulation tests on agents are conducted to measure the performance of the planner we proposed. The simulation results show that our method not only has great advantages in the face of greedy defenders, but also is superior to the traditional APF planning algorithm.
What problem does this paper attempt to address?