Detecting deception in HRI using minimally-invasive and noninvasive techniques

Our work focuses on detecting deception in Human-Robot Interactions (HRI) by using measurement techniques that are appropriate for such interactions. In our previous research works, we obtained interesting results by using thermal and RGB-D cameras. In this paper, we approached this aspect from a different angle and used a lab-designed armband to accurately measure the participants' heart rate and skin conductance. We also developed a deception card game scenario that entices human participants to lie either to a robot or a human game partner, allowing us to monitor and understand the correlations between human physiological manifestations and their trustworthiness. Our results show the existence of statistically significant correlations between the participants' deceptive behaviour and their heart rate, skin conductance, face position, and face orientation. These results allow us to improve robots' ability to detect deception in HRI.