A novel target search framework for unmanned aerial vehicles in unknown environments

This study presents a simulation framework for multi-drone swarm cooperative search and detection of dynamic targets in an unknown environment. The framework initializes a target existence probability (TEP) map from prior target location estimates, evolves the map with the Fokker-Planck equation to model the diffusion and advection of uncertainty due to movement of targets. The TEP map is updated periodically with drone sensor readings with specific target detection and false alarm probabilities. A novel modification of the Manta Ray Foraging Optimization (MRFO) algorithm called the Entropy-Weighted Cyclone Foraging Manta Ray Foraging Optimization (EWCF-MRFO) algorithm is proposed to tackle the optimization problem. The proposed algorithm’s performance and robustness are tested in four different scenarios with different sensor detection and false alarm probabilities, as well as target numbers and formations. The TEP map entropy reduction and target detection performance of the proposed algorithm are compared with some other state-of-the-art novel literature optimizers over the scenarios. The results obtained by the proposed algorithm proved superior entropy reduction performance compared to the selected optimizers. The EWCF-MRFO algorithm can find the targets’ exact locations while approaching them and detect them in a short amount of time.