GAMPP: Genetic algorithm for UAV mission planning problems

Abstract

Due to the rapid development of the UAVs capabilities, these are being incorporated into many fields to perform increasingly complex tasks. Some of these tasks are becoming very important because they involve a high risk to the vehicle driver, such as detecting forest fires or rescue tasks, while using UAVs avoids risking human lives. Recent researches on artificial intelligence techniques applied to these systems provide a new degree of high-level autonomy of them. Mission planning for teams of UAVs can be defined as the planning process of locations to visit (way-points) and the vehicle actions to do (loading/dropping a load, taking videos/pictures, acquiring information), typically over a time period. Currently, UAVs are controlled remotely by human operators from ground control stations, or use rudimentary systems. This paper presents a new Genetic Algorithm for solving Mission Planning Problems (GAMPP) using a cooperative team of UAVs. The fitness function has been designed combining several measures to look for optimal solutions minimizing the fuel consumption and the mission time (or makespan). The algorithm has been experimentally tested through several missions where its complexity is incrementally modified to measure the scalability of the problem. Experimental results show that the new algorithm is able to obtain good solutions improving the runtime of a previous approach based on CSPs.