Virginia Tech, Zoox Release Results of ‘One-of-a-Kind’ Study
Earlier this month, on September 9, Gwen Sheldon penned a piece for Virginia Tech about new research between the university and Amazon-owned and Bay Area-based autonomous vehicle company Zoox that found “custom audio tones could effectively communicate with people attempting to cross streets.”
“A novel study from Virginia Tech and Zoox, the Amazon-owned autonomous ride-hailing company, provides a promising answer: through sound. The research was published at the 28th Enhanced Safety of Vehicles Conference, which was sponsored by the National Highway Traffic Safety Administration and Transport Canada,” Sheldon reported. “Using custom tones, the research team explored how sound could be used to communicate vehicle intent in participant trials. Results show that sounds can be as effective as a traditional horn in increasing pedestrian awareness and influencing crossing decisions, particularly in jaywalking situations, even without prior training on the tones.”
The whitepaper with the research is available to read,
I connected with Dr. Charlie Klauer, a research scientist at the Virginia Tech Transportation Institute Division of Vehicle, Driver, & System Safety. In a brief interview conducted over email earlier this month, Dr. Klauer, who has spent a quarter-century studying what he called “human factors transportation research primarily utilizing naturalistic driving studies to understand driver behavior,” explained one of the Institute’s, known as VTTI, goals is “designing research studies to test automotive systems.” Zoox, he told me, approached the school and asked to look into how sound helps communicate vehicle intent to pedestrians and thus influence their decision-making when crossing the street.
“Automated vehicles are currently operating in the real world, albeit in limited areas. As we look to a future where more vehicles may be operating without a driver, we should understand how pedestrian interactions are happening, how pedestrians are making decisions and what information they need to make safe decisions,” Dr. Klauer said. “I believe understanding this information can help us influence new ways for automated vehicles to communicate with pedestrians and inform what education road users need about sharing the road with autonomous vehicles.”
According to Dr. Klauer, the problem prompting the study was exploring the ways in which pedestrians “understand what to expect when sharing the road with a vehicle that doesn’t have a driver.” Pedestrians oftentimes rely on cues from drivers when crossing, particularly at intersections where they may not be a crosswalk or stoplight. When cars approach, he said, people are more likely to look towards the driver to confirm intent before beginning to cross safely.
With an autonomous vehicle, however, there is no human driver to convey said intent to pedestrians.
“We are studying this so that automated vehicles not only improve safety for riders and drivers, but they also are going to improve safety for our pedestrians as well,” Dr. Klauer said.
When asked about the primary goal of the project, Dr. Klauer said it was “to improve safety for all vulnerable road users, specifically ensuring that automated vehicles can effectively communicate with everyone,” adding it “explored how pedestrians would react to sounds without having any prior knowledge of what the sounds were supposed to mean.”
He added: “We also wanted to evaluate the participants’ perceptions of the sounds. Traditional car horns are effective and universally known but are often perceived as obnoxious and aggressive. Participants were surveyed to rank the sounds by qualities like ‘urgency,’ ‘friendliness,’ and ‘aggression.’ Another goal was to explore how audio tones can communicate with people who have low vision or who have some vision impairment. Half of our 40 participants included people who had low or visually impaired vision. One of the potential benefits of automated vehicles is to improve transportation options for everyone, including those who may not be able to drive. We wanted to make sure autonomous vehicles can effectively communicate with people who have varied physical and emotional capabilities.”
As to the mechanics of the study, Dr. Klauer explained VTTI designed an environment on its closed-course test track wherein science could accurately tally pedestrians’ responses. The pedestrians weren’t in a simulation, though: they faced real-world roadway environments where ambient traffic sounds would be naturally present. It’s this aspect that made the research unique, according to Dr. Klauer, because most studies examining so-called external human machine interfaces have been done under simulated conditions or in augmented reality settings. For its part, Zoox developed seven bespoke tones, each signifying intent such as “vehicle is moving” or “vehicle is waiting.” The noises were tested against situations where a pedestrian may need to wait for a vehicle, like when jaywalking, or when a vehicle was stopped in the road, waiting, such as when at a crosswalk.
“In mid-block ‘jaywalking’ scenarios, the purpose-built alert tones were highly successful. The research showed that these novel sounds were just as effective as a traditional car horn at influencing a pedestrian’s decision not to cross, and participants perceived the sounds much more favorably and as less aggressive,” Dr. Klauer said of the research’s findings. “In structured crosswalk scenarios, where the vehicle played ‘waiting’ tones to encourage safe passage, the sounds influenced decisions less than 50% of the time. This suggests that pedestrians rely heavily on the physical motion of the car in marked crosswalks, and that it will require repeated exposure over time for the public to learn and trust these new ‘waiting’ audio signals.”
When asked about feedback on the research, Dr. Klauer said he believes it was special due its realism and emphasized his belief “the results are much more generalizable than other studies that have been done in this space.” He went on to tell me he thinks the study is a “launching point” for future research on the topic, describing the participants as “naive” to the sounds they were exposed to during the study. “It will be important to continue this line of research to understand how pedestrians learn to interpret these sounds with repeated exposure over time to ensure the safety benefits for pedestrians but also all road users,” Dr. Klauer said.
Looking towards the future, Dr. Klauer noted pedestrian fatalities have been on the rise over the past decade. As the transportation apparatus (slowly) transitions to autonomous vehicles, he said, it’s “critical” they be designed in a way that “[prioritizes] pedestrian safety” so as to make them safe for all. “That is what we are ultimately working toward, a future where we greatly reduce and eventually eliminate fatalities on United States roadways,” Dr. Klauer said.
