Enhancing Gesture-Based Human-Drone Interaction in Emergency Scenarios: An Elicitation Study

Due to the large content and length of the article, we have tried to reduce the content of the abstract as much as possible: As beyond visual line of sight (BVLOS) interactions become more prevalent in logistics, surveillance, and rescue missions, the need for intuitive, real-time control methods such as gesture-based interfaces have become critical. However, most previous studies focus on non-emergency BVLOS or rely on pre-defined expert sets, ignoring instinctive human responses under stress. To address this gap, we employed Wizard of oz (Woz) paradigm to optimize gesture-based human-drone interaction (HDI) for BVLOS in emergency scenarios. We integrated VR helmets streaming real-time first-person-view (FPV) of the drone to simulate BVLOS conditions, paired with a flexible circular board to simulate dynamic obstacles at different emergency levels. We collected and analyzed 211 valid gesture samples from 12 participants and revealed a significant effect of emergency severity on gesture characteristics: Compared to non-emergency scenarios, high-emergency scenarios elicited faster (t = 3.47, p < 0.001) and smoother gestures (t = −2.78, p < 0.001). Furthermore, we derived a user-defined gesture set and evaluated it for consistency, response time, motion smoothness, and subjective feedback.