Autonomous vehicles issuing "high-content" communications to pedestrians were rated as being more trustworthy and reassuring by a focus group set of participants with visual disabilities.

The study, which evaluated multiple intersection simulations using virtual reality (VR) technology, found that repeated messages of long duration were the most effective at reassuring pedestrians.

Date Posted
05/20/2020
Identifier
2020-B01458
TwitterLinkedInFacebook

Towards Inclusive External Communication of Autonomous Vehicles for Pedestrians with Vision Impairments

Summary Information

Autonomous Vehicles (AVs) are widely anticipated to reduce accidents due to their advanced sensors and collision-detection systems. However, it remains necessary to have a robust system for vehicle-pedestrian communication (VPC), given the lack of a human operator in an AV, to allow AVs to communicate easily and clearly with vulnerable road users. Much of the research in this area has focused on communications concepts for a non-disabled population. A recent study by researchers from Ulm University in Germany reframes the work to focus on inclusive communication methods.



The researchers first conducted a focus group to discuss issues in road traffic and proposed concepts, then performed a study using virtual reality (VR) to compare the effectiveness of the various VPC implementations. The VR simulation involved participants attempting to navigate sidewalks and intersections while receiving varying levels of VPC feedback from the simulated traffic. The baseline scenario involved no communication, the "low-content" message was simply "cross," and the "high-content" message was "I’m stopping, you can cross." Intersections also had one or two vehicles approaching.

FINDINGS



The researchers identified several major trends in the data:

  • The study found that persons with visual disabilities reported substantially higher feelings of control over the situation when the VPC messages were high in content.
  • Participants ranked the scenario in which two vehicles approached the intersection and both issued high-content VPC messages as being the most preferred, followed by, in descending order of preference, two vehicles and low-content messages, one vehicle and high-content messages, one vehicle and low-content messages, and one or two vehicles with no messages. This indicates that having messages be validated through repetition by multiple vehicles is reassuring to pedestrians.
  • Twenty out of the twenty-nine participants who gave feedback noted that explicit verbal communication enhanced their trust of the vehicle’s behavior.
  • Participants noted that the duration of the auditory message was important to their understanding of it, as the low-content messages were short and could be more easily missed.
  • Workshop participants noted that they would be unlikely to trust an "omniscient narrator" message in which two or more connected vehicles coordinate their messages so that only one VPC needs to be played, expressing a lack of confidence in the other vehicles’ behavior. The researchers note that the people consider each vehicle in traffic to be an independent vehicle, and that the concept of traffic acting as a unified system through AV technologies will require adaptation.
  • Seeing participants ranked visual cues as being "significantly favored" over audible warnings, though the researchers note that a combination of both would be most effective in communicating to the general population.
Deployment Locations