How should a minimal risk maneuver of a highly automated shuttle be communicated to its users via its internal human-machine interface? Applying the media richness theory to user-centered interface design.

Abstract

In future automated public transportation, the lack of human drivers means that passengers have no one to communicate with in case of problematic situations during the transportation process. These situations originate in problems outside the system's capabilities. It will then, as a consequence, perform a minimal risk maneuver and stop, which may cause insecurities in passengers. One possible solution to address this could be the use of avatars as a means of communication. To optimize this interaction between the passengers and avatars, they need to be designed in a way that is visually appealing and not uncanny, as this can be detrimental to the experience. Considering the uncanny valley curve, that many of today's concepts have fallen into, photorealistic avatars might get ahead of the curve and out of the uncanny valley. For this purpose, we conducted an online-study comparing photorealistic MetaHumans, created using Unreal Engine 5 (UE5) and abstract avatars, abstracted by ReadyPlayerMe, in terms of the quality of their general visual appearance and the uncanny valley. We animated all avatars using motion capture with Live Link Face, making their facial expressions as realistic as possible. The results showed that the participants rated the photorealistic Meta Humans significantly better in terms of their overall appearance and perceived them to be significantly less uncanny. These findings indicate the promising potential of photorealistic avatars for use in automated on demand shuttles, where they could significantly improve the user experience and therefore meaningfully facilitate future adoption.