Multiagent Pursuit-Evasion Problem with the Pursuers Moving at Uncertain Speeds

The multiagent pursuit-evasion problems have been widely investigated in related areas. Previous studies usually assumed that the pursuers move at certain speeds. However, in many circumstances the above assumption does not match the peculiarities of real pursuit-evasion cases in which the pursuers’ speeds may be uncertain. Therefore, this paper investigates the multiagent pursuit-evasion problem under the situation in which the pursuers move at uncertain speeds. The new problems of multiagent pursuit-evasion caused by the uncertainty of the pursuers’ speeds include: 1) many previous strategies plan pursuers’ paths based on their speeds, but the uncertainty of speeds will make the pursuers move to worthless target points; 2) previous strategies usually let each pursuer move to a scheduled location, but the uncertainty of speeds may make some pursuers fail to reach the scheduled locations punctually. Aiming at addressing these problems, we present the strategy which lets each pursuer flexibly help the slow neighboring pursuer. As the pursuers’ speeds are uncertain, the optimal decision of pursuers cannot be calculated directly. Thus, we analyze the alternative decision space of pursuers, which contains the decisions that may be optimal and does not contain the obviously bad decisions (such as moving away from the evader). Then, we compare the decisions in the alternative decision space based on simulated annealing resulting that the optimal decision may be selected after repeatedly comparing different decisions. The experimental results show that our strategy can generally outperform previous strategies when the pursuers’ speeds are uncertain.