Search for targets in a risky environment using multi-objective optimisation

This study develops an online unmanned aerial vehicle (UAV) path-planning strategy for autonomous search and localisation of targets in a risky environment that simultaneously optimises two objectives: (i) search and (ii) survival. The authors formulate two rewards (objective) functions corresponding to the two objectives and compute the Pareto front, formed by taking convex combinations of these rewards. In the extreme case of pure search, using entropy reduction as the reward, the trajectory of the UAV is reminiscent of a systematic search pattern. When combined with the survival objective, the search pattern appears more random, as the UAV intelligently trades off the reward of finding targets with the risk of being destroyed.