Migratory Genetic Algorithm with Pheromone Information for Autonomous Coverage Path Planning Using Uavs

Abstract

The multi-objective multiple Traveling Salesmen Problem (MOMTSP) can be used to model different applicational problems such as multi-agent path planning in robotics. This paper proposed a new hybrid algorithm with migration mechanism to solve the MOMTSP with an unknown depot. The proposed algorithm, namely, NSGACO can be divided into two parts. Part one consists of the hybrid algorithm that combines the Non-dominated sorting genetic algorithm II (NSGA-II) and the Ant colony algorithm (ACO) through a subtour library to solve the MOMTSP. This hybridization realizes the benefits of both algorithms throughout iterations, allowing the search for a better Pareto optimal front. Part two consists of the migration mechanism which evaluates the fitness of the current depot and guides the system to migrate to a more optimal depot. The performance of the proposed NSGACO is compared with four single or hybrid algorithms over the TSPLIB benchmark problems. The results show that the NSGACO outperforms other algorithms in terms of solution quality and diversity. The NSGACO is also able to effectively search for an optimal depot location.

Keywords: Multi-Objective Optimization, Multi-agent system, Hybrid Algorithm, GENETIC ALGORITHM, Ant colony optimization.