Self-driving cars are all the rage today. Companies ranging from car manufacturers like General Motors and Toyota to private-hire companies like Uber, and even Internet search giant Google, are all scrambling to be the first to bring them to market. The efforts are so intense that the University of Michigan has even established a 23-acre town to help the cause. Dubbed Mcity, it allows manufacturers to safely test their autonomous cars using human props.
However, while the mock city can be used to simulate many real-life road conditions, it cannot help test gestures drivers use to communicate their intent to fellow drivers, pedestrians or bicyclists. These include cues like waving a car into a lane, or nodding at a person walking or on a bike, to indicate they can cross the road safely.
To try to come up with a solution for these everyday scenarios that self-driving vehicles would face, US car manufacturer Ford, teamed up with researchers from Virginia Tech. The team initially considered using text as a way to communicate the car’s intention, but decided it would probably not work universally given that people would have to be able to read and understand the same language. The option of using symbols was also discarded, because research shows that a majority of people do not have a good understanding of what they mean.
After some deliberation, the researchers settled on light signals, that are understood by people worldwide. A solid white light served as a warning that there were no humans in the car. A slow blinking one indicated that the car was coming to a stop, while a rapidly flashing light cautioned passersby that the car was about to accelerate.
Then came the big challenge – testing the signals on real roads. “We need[ed] to try out this new lighting to communicate the intent of the vehicle, but if you’ve got a driver behind the seat, you still have natural communication between humans like eye-to-eye contact,” said Andy Shaudt, who led the Virginia Tech research team. “So we needed to make it look like a driverless car.”
Taking inspiration from a similar study conducted two years ago, by Stanford researcher Wendy Ju, the team designed a car seat costume that the driver would wear to over his/her face and upper body. The drivers could see very clearly through a plastic visor, which was concealed by a thin reflective fabric making it invisible to people looking in.
The researchers then outfitted a Ford Transit Connect van with high-definition cameras to capture human reactions to a light bar on the windshield, which flashed one of the three signals when appropriate. Six drivers, all keeping their hands low on the wheel so as not to be detected, took turns test driving the van through the busy streets of northern Virginia in August. They drove for 1,500 hours, covered 1,800 miles and tested the light cues at more than 1,650 locations, including intersections, parking lots, and airport roadways.
While details of the study have not been revealed, Ford states that the overall reaction to the flashing lights was very encouraging. The car manufacturer plans on sharing the information with 11 other companies and collaborating with them to design a signaling system that would be understood by all. Who knew costumes could be useful for more than Halloween, or pranking strangers?
Resources: wired.com, the verge.com