ColorUAV: A genetic algorithm for dynamic drone path planning under multiple constraints
Simon Pichot, Hicham Khalifé, Vania Conan, Gentian Jakllari · 2022 International Wireless Communications and Mobile Computing (IWCMC) · 2022
We propose ColorUAV, a genetic algorithm for UAV path planning under multiple constraints. Our solution enables drones to travel towards a destination point while avoiding obstacles and navigating as close as possible to base stations in order to maximize communication opportunities. ColorUAV learning phase creates UAVs able to interact with their environment and compute paths whose weight is within 17% of the optimal. More interestingly, our trained UAVs are capable of adapting to moderate changes in the environment without re-training and while remaining efficient in terms of the path quality.