Design and Assessment of External Displays on Autonomous Vehicles for Pedestrian Safety

Abstract

The most vulnerable group in road agents is pedestrians. In the Netherlands, from 2005 to 2009, the average lethality rate for vulnerable road users was 14 per 100 serious road injuries. Prior to autonomous vehicles (AV), conventional vehicles had a human driver that could communicate with pedestrians through signals such as eye gaze, head movements, and hand and arm gestures. With the introduction of AVs, pedestrians can no longer rely on such communication signals. In the future, when all of the control and responsibilities of the human driver gradually transfer to the autonomous driving program, the vehicle’s intent communication to pedestrians must evolve as well. The aim of this proposed research was to investigate the efficacy of different external human-machine-interface in communicating autonomous vehicle intent to pedestrians in crossing situations where negotiation between the AV and the pedestrian is required (i.e. jaywalking).

With SAE level 4 high automation enabled vehicles, what impact do external human-machine interfaces have on pedestrians’ crossing behaviour? what impact do external human-machine interfaces have on pedestrians’ general perception of AVs?

Three novel design concepts were created to fill the gap of the lack of visual experimentation with displaying the key mental model factors - external speedometer display of the vehicle, speed change indicator (i.e. decelerating/ accelerating), and gap estimation count down timer. The experiment was a within-subject design with 29 levels. The stimulus was structured into a 14 (design cases) x 2 (coloured vs. non-coloured) factorial design. A combination of iconography, text, anthropomorphic features and colour were compared and measured in perceived safety, urgency, usefulness, understandability, emotion comfort, as well as the influence on crossing decisions. A 100-person online study was conducted to understand the impact of external visual displays with high automation (SAE level 4) vehicles on pedestrians’ crossing behaviours. The novel concepts open a new discussion for the perception of warning designs where the new visual concepts (i.e. explicitly displaying and varying the symbolism of speed) had strong performance across all measures.