Multiple Depot MTSP using Genetic Algorithm and Reinforcement Learning

The Multiple Travelling Salesman problem (MTSP) is a generalization of the TSP, an NP-hard problem, where ‘m’ number of salesman have to travel through ‘n’ number of cities (n>m). The goal of the multiple travelling salesman is to make sure that the salesmen travel to all the cities mentioned from a depot while ensuring that the distance travelled on the total route covering all the cities is minimum, the distance here can be equated to profit or time. This research addresses the closed path multi depot MTSP problem, solved using Genetic Algorithm. The genetic algorithm, which is based on natural selection, is a technique for resolving both constrained and unconstrained optimization problems. The Genetic algorithm is especially useful in solving Np-hard problems. Using the elitist selection operator, 2-opt mutation and OX crossover, and further optimized with the help of Reinforcement learning with to get a feasible solution.