To Swarm or Not to Swarm: Human-Drone Swarm Control Approaches in Maritime Search-And-Rescue

Abstract

Drone swarms are promising tools for search-and-rescue operations but there is currently little guidance regarding control input methods and interface design. We report a controlled within-subjects human-swarm interaction experiment comparing three approaches for supervising/controlling a 20-drone swarm in a simulated maritime search-and-rescue task: per-drone waypoint/route control; swarm-level task-area control using configurable templates; and a combined hybrid interface. Outcomes included self-rated mental workload and situation awareness, and log-derived performance metrics (objects found, time-to-first-detect, area covered/overlap, drone utilization, and command counts). Swarm-level task-area control yielded the best overall results: substantially lower workload, higher situation awareness, more objects found and area covered, higher fleet utilization, and far fewer user inputs than per-drone control. Hybrid control reduced workload relative to per-drone control but introduced additional “meta-control” demands and underperformed pure swarm-level control on detection. Findings highlight the advantages of macro-level tasking and the need to better support mode switching in hybrid control designs.