Multiple Objectives for UAVs Paths Optimization for a Disaster Area Reconnaissance

Abstract

The benefits of early reconnaissance in the event of emergency response can be invaluable for the success of the emergency teams present in the area. The use of unmanned aerial vehicles (UAVs), which are restricted only by weather conditions, introduces a path-planning optimization requirement. We model a disaster area as a set of rectangular regions, prioritized by the expected number of victims. UAVs, working as a team, provide complete coverage of all regions to gather information about conditions on the ground. This paper proposes a new set of objective functions to maximize the number of detected victims in the earliest stages of the search while minimizing completion time. A new solution representation is introduced together with problem-specific mutation operators. Experimental research using NSGA algorithms provides insight into how the model's features affect solution optimization.