Capturing Dynamic Trust Metrics during Shared Space Human Robot Collaboration: An eye-tracking approach
Event Type
Virtual Program Session
Human AI Robot Teaming (HART)
TimeWednesday, October 12th12:00pm - 12:15pm EDT
DescriptionIn this study, we investigated how workers’ trust dynamics changed throughout 4 phases of a human-robot collaboration experiment involving 2 different reliability levels (100% and 76% reliable). We found that eye-tracking measures captured changes in gaze behaviors under different trust conditions; specifically, trust level change reported by subjective trust survey throughout all stages negatively corresponded to fixation counts on the collaborative robot, indicating reduced monitoring activity. Gender differences were also present within trusting gaze behaviors; with increased subjective trust level, males had reduced fixation durations (less monitoring), while females demonstrated reduced stationary gaze entropy (close distribution of gaze) and increased gaze transition entropy (exploratory visual attention). In summary, eye-tracking measures can potentially become a reliable objective trust measure in place of the subjective measures and capturing additional behavior changes.