Distributed PSO for Rescue Task Allocation with Multi-Type Multi-Task

Currently, task allocation evolutionary algorithm that use a centralized type for communication is a popular issue, while such scenario is very rare in practical. Given that, this paper presented a PSO-based distributed method to solve the problem of rescue task allocation with multi-task by multi-type robots. Existing distributed algorithms, such as, CBBA, PI, PI-softmax, PI-e-greedy are employed to generate solutions for solving the problems, however, there still exist room to improve. Due to task allocation problem is a NP-hard problem, and PSO is employed and modified so as to solve the above problem in a distributed manner. Firstly, the problem is descripted and the mathematical model is established, the average mission time for each task is set as the objective function, in addition, the priority sequence computation method proposed in [21] is also employed in this paper; secondly, PSO is modified to be a distributed algorithm, and to better the result, multiple sub-swarms is illustrated corresponding to multiple types of tasks or robots, and some new operations, including: the initial phase, the local and global best solution updates outside the sub swarms and among the sub swarms, the velocity and position update method and so on, are presented. The existing scenarios and other complex scenarios are solved to verify the proposed method, and then, the corresponding results illustrate that the proposed distributed PSO algorithm not only can handle the problem but also can gain the result within an acceptable time in such a rescue scenario.