4th International Conference on Intelligent and Fuzzy Systems (INFUS), Bornova, Turkey, 19 - 21 July 2022, vol.504, pp.119-127
Between 1998 and 2017, 1.3 million people died, 4.4 billion were wounded, or were displaced due to natural disasters, according to CRED (2018). This paper considers the problem of maximizing the survival probability of survivors in severe and large-scale natural disasters scenarios using multi-robots. A novel two-stage method was used to determine the locations of relief depots and robots' routes based on a survival probability function. The fuzzy c-means clustering approach was employed in the first step to find the best sites for relief depots, and an ant colony optimization algorithm was utilized to find the best paths between depots and survivors. We presented a novel objective function that maximizes survivor survival probability while decreasing a total number of deaths and total distance traveled by robots. The proposed method was compared against the k-means clustering algorithm and the most often used objective function in multi robot task allocation (MRTA) problems, total distance traveled. According to simulation results, our strategy outperforms others in terms of reducing the overall number of deaths and improving the probability of survival.