The temperature prediction model and fire control method in RoboCup Rescue

RoboCup Rescue constructs a virtual agent-rescue disaster environment with buildings burning, roads blocked, and civilians in injure. Due to the tricky tasks of building inferno control in this simulation platform, this paper presents a temperature and fire spot prediction model based on the grey system theory. Rather than the previous way that just extinguishing burning buildings, a pre-extinguish method which is much more efficient in the process of controlling fire spreading by handling not-burning buildings is applied here according to the prediction model, in which the building temperature of the next cycle can be estimated so that whether the building is about to catch fire will be foreknown and advanced decisions will be exploited by agents. Experimental results show that this method can achieve an enhanced effect during fire control process of the fire brigade agent in RoboCup Rescue competitions.