Communication between vehicles and bicyclists is an essential part of road safety. As autonomous systems become more prevalent on public roads, there is a need to develop legible, acceptable, and trustworthy interactions between unmanned vehicular robots and pedestrians. This paper focuses on the investigation of communication models that communicate intent with cyclists when they encounter a driverless autonomous vehicle. Through a user study, this paper explores different models of communication modules presented on an autonomous vehicle. Scenarios for testing were simulated in an immersive virtual reality environment combined with physical systems to replicate the experience of riding a bicycle in traffic to elicit relevant responses in a likely road interaction. This study shows the need for a system that allows for legible communication with autonomous vehicles operating without human control and entirely AI driven and provides concrete suggestions for the forms such a communication system might take. The results of this study shows the preference of symbols over text, light, and road projection when encountering an autonomous vehicle in a possible collision. Future studies will expand into real world applications of specific models that are found to be preferred by prior work.

AI Institute for Transforming Education for Children with Speech and Language Processing Challenges, National Science Foundation PI: Venugopal Govindaraju, co-PI: David Feil-Seifer, Amount: $20,000,000, Jan. 15, 2023 - Dec. 31, 2027