I, AV: A Ghost Driver Field Study Exploring the Application of Anthropomorphism in AV-Pedestrian Communication

Abstract

We employed the 'Ghost Driver' methodology to emulate an autonomous vehicle (AV) and explored pedestrians' (n=520) crossing behaviour in response to external human-machine interfaces (eHMIs). Three eHMI designs were created to replace absent pedestrian-driver communication; each had different anthropomorphic elements and were identified as 'explicit', 'implicit' and 'low' to reflect the conspicuity of anthropomorphism. They were displayed on an LED matrix and strip mounted to the front of a Nissan Leaf vehicle, which was driven around the University campus over 5 days. Video analysis highlighted differences in pedestrians' behaviour, with the explicit anthropomorphism eHMI extending crossing time and attracting more visual attention. Additionally, some pedestrians continued to use gestures, ostensibly to indicate their intention to cross or to thank the vehicle, despite the absence of a visible driver. While preliminary findings support the application of anthropomorphism in AV-pedestrian communications, further research will explore designs in more controlled, experimental settings. CCS CONCEPTS• Human-centered computing → Human computer interaction (HCI); Empirical studies in HCI.