Google’s Self-Driving Cars Are Learning to Recognize Cyclists’ Hand Signals
Cyclists, meet the nicest car you’ll ever share the road with
/https://tf-cmsv2-smithsonianmag-media.s3.amazonaws.com/filer/f4/4d/f44d3b51-6d2e-4612-bb73-57d684f30406/bicycle_courier_552.jpeg)
The saga of cyclists sharing the road with cars is long and fraught with tension and accusations of unsafe behavior from both sides of the argument. While many drivers don't give cyclists enough space, some cyclists will dangerously weave in and out of traffic.
But if self-driving cars ever dominate the roads, this could be a problem of the past. According to Google’ latest report on the state of its self-driving car, the vehicle can recognize and predict cyclists’ behavior, as well as understand their hand signals, Johana Bhuiyan reports for Recode.
“Our sensors can detect a cyclist's’ hand signals as an indication of an intention to make a turn or shift over,” according to Google's June 2016 report. “Cyclists often make hand signals far in advance of a turn, and our software is designed to remember previous signals from a rider so it can better anticipate a rider's turn down the road.”
This isn’t the first time that a car manufacturer has “taught” its vehicles to detect and understand cyclists’ hand signals. According to Bhuiyan, a recent Mercedes concept car also demonstrated the ability to recognize hand signals and gestures. But as more self-driving cars hit the road, the ability to interpret and communicate with pedestrians and people riding on all manner of bikes is important to make autonomous vehicles safe.
By using machine learning, Google reports its cars can not only recognize cyclists and their hand signals, but they can also learn from how the the encountered cyclists ride. Because the self-driving cars are equipped with 360-degree sensors, they will be able to detect cyclists moving around them, even in the dark, Angela Chen writes for Gizmodo. The Google cars are also learning how to detect and handle difficult situations, such as if a cyclist ahead is passing by a car whose driver is stepping out into the street.
“For example, when our sensors detect a parallel-parked car with an open door near a cyclist, our car is programmed to slow down or nudge over to give the rider enough space to move towards the center of the lane and avoid the door,” according to Google's report. “We also aim to give cyclists ample buffer room when we pass, and our cars won’t squeeze by when cyclists take the center of the lane, even if there’s technically enough space.”
Google's cars have come far in how they handle cyclists, but they aren't yet error free. In one incident last fall, a cyclist in Austin, Texas reported that he managed to thoroughly confuse one of Google's prototypes when he performed a track stand ahead of it at a stoplight. The maneuver, which involves the cyclist standing on the pedals and rocking back and forth, confused the car so thoroughly that it reportedly froze in place in the middle of an intersection for several minutes.
Though there are still some kinks to work out, Google's self-driving cars could be a breath of fresh air for both cyclists and drivers.