Exploring Human Compliance Toward a Package Delivery Robot

Human-robot interaction (HRI) studies have found people overtrust robots in domestic settings, even when the robot exhibits faulty behavior. Cognitive dissonance and selective attention explain these results. To test these theories, a novel HRI study was performed in a university library where participants were recruited to follow a package delivery robot. Participants then faced a dilemma to deliver a package in a private common room that might be off-limits. Then, they faced another dilemma when the robot stopped in front of an Emergency Exit door, and they had to trust the robot whether to open it or not Results showed individuals did not overtrust the robot and open the Emergency Exit door. Interestingly, most individuals demurred from entering the private common room when packages were not labeled, whereas groups of friends were more likely to enter the room. Then, selective attention was demonstrated by stopping participants in front of a similar Emergency Exit door and assessing whether they noticed it In one condition, only half of participants noticed it, and when the robot became more engaging no one noticed it. Additionally, a malfunctioning robot is exhibited, showing what kind of negative outcome was required to reduce trust.