Neurophysiological and behavioral studies of human-swarm interaction tasks

Abstract

This paper studies human-swarm interaction performance by evaluating both the neurophysiological and behavioral characteristics of human subjects. By utilizing our unique test facility, we conduct a series of real-world-scenario-inspired tasks in which subjects are asked to guide a group of ground robots with various configurations to arrive at a sequence of randomly assigned targets. A range of neurophysiological and behavioral sensors are used to measure how cognitive states, e.g., cognitive load, behaviors, e.g., gazes, and performances, e.g., success rate, of human performs unfold in real time as the tasks evolve. Through an analysis of changes in gaze and cognitive load, we gain a wider understanding of the mechanisms of task failure; most notably the difficulty of estimating the complete state of the robotic group. The results of this study can help to inform the design of efficient interaction policies which can maximize task effectiveness between humans and robot swarms.