Research into Automated Vehicles (AVs) typically focuses on the impact of the technology on the efficiency of road networks. However, AVs will also have significant effects on how Vulnerable Road Users (VRUs) will interact with vehicles. A recent article includes interviews with sixteen Human Factors researchers from a wide range of international institutions to get a better sense of future best practices for implementing AVs. In particular, the interviews focused on external human-machine interfaces (eHMIs) and potential applications of augmented reality (AR) in managing VRU-AV interactions.
The suggestions offered by the researchers included the following:
- Include smart infrastructure in urban environments. Researchers noted the risk that pedestrians may become used to AVs reliably stopping for them, and thus begin to jaywalk more frequently and more dangerously. By connecting AVs to an area's infrastructure, interactions between VRUs and AVs may be improved or more effectively managed.
- Vehicles should use eHMIs to convey their state, not to give directions. For example, an AV approaching an intersection should indicate "I am stopping," not "please cross." It is important that eHMIs be consistent and standardized where possible to reduce interperative confusion.
- Avoid text in eHMIs. This increases the risk of confusion or unintelligibility, for example to those who do not speak the same language, though researchers noted that many VRUs prefer explicit verbal or textual instructions.
- Use eHMIs as a secondary cue. The researchers noted that eHMIs were most effective when they were supporting implicit communication, such as vehicle speed and trajectory, and when they were clearly tied to specific vehicles rather than to infrastructure elements.
- Consider AR for one-to-many communication elements. While AR functionalities were noted to have many potential drawbacks--related to both technological issues and accessibility concerns--it may be useful to provide individualized interpretation of signs and signals in a format that is customized to each user.