VIBROTACTILE HAPTIC CUEING FOR HUMAN SUPERVISION OF A LEADER–FOLLOWER UAV SWARM IN A SCALED SLALOM TASK

Abstract

This thesis investigates whether vibrotactile haptic feedback can improve human supervision ofa leader–follower unmanned aerial vehicle (UAV) swarm during a scaled slalom task. In a leader– follower architecture, the operator directly pilots only the leader vehicle, while follower vehicles maintain prescribed relative positions within the formation. Although this reduces direct manual burden, the operator must still monitor follower motion, formation distortion, and obstacle relationships, which can increase workload and reduce situational awareness, especially in degraded visual environments (DVEs). To support the operator, a vibrotactile cueing strategy was developed to communicate formation deformation and gate-avoidance information. The primary evaluation was conducted through a simulation- based human-subject experiment using a three-vehicle leader–follower swarm in a scaled ADS-33- inspired slalom course. The simulation framework combined custom quadcopter dynamics, follower formation-control logic, a virtual reality piloting environment, and wearable bHaptics devices for directional cue delivery. Testing was performed in both good visual environment (GVE) and degraded visual environment (DVE) conditions, with and without haptic feedback. Objective metrics included formation-keeping error, missed-gate percentage, speed compliance, and completion time, while subjective evaluation used the Bedford Workload Scale and China Lake Situational Awareness Rating Scale. Simulation results showed that haptic cueing improved formation-keeping performance, with the strongest gains observed in longitudinal deformation and in lateral deformation under degraded visual conditions. Improvements in missed-gate performance were more modest, and obstacle-navigation success remained strongly influenced by visual condition. Subjective ratings indicated that haptic feedback improved situational awareness without substantially increasing workload. Participant feedback also suggested that haptics were most effective for communicating follower motion and overall swarm geometry. The work was further extended toward a physical swarm demonstration using custom-built quadcopters, PX4-based control, and a centralized ground-station architecture. Although a full physical experiment was not completed, preliminary testing demonstrated the feasibility of the communication, control, and haptic-integration framework. Overall, the results support vibrotactile cueing as a promising tool for improving human supervision of UAV swarms