Google talks up its self-driving cars’ cyclist-detection algorithms

Today might be conspicuous in the annals of autonomous vehicles for a more serious reason, but another month is over and Google has issued its self-driving car report for June. In it are some interesting details about the system’s ability to detect and avoid cyclists.

“Cyclists are fast and agile — sometimes moving as quickly as cars — but that also means that it’s hard for others to anticipate their movements,” reads the report. “Our cars recognize cyclists as unique users of the road, and are taught to drive conservatively around them.”

The LIDAR and other sensors on the vehicles can detect bikers in any direction, or even every direction at once, as this illustration shows.

google_bike_vis2

That’s somewhere over a hundred Googlers riding their bikes around a stationary self-driving car. This is what they do all day in Mountain View, apparently. Each bike was tracked individually and its likely path predicted, probably giving the car’s AI an anxiety attack.

When cyclists are detected, they’re given extra room, and the cars won’t attempt to pass if the bike is taking up the whole lane (rarely convenient for drivers, but often necessary, cyclists will surely agree).

Hand signals are also seen, understood and taken into account when predicting a cyclist’s path later on, after the signalling hand has returned to its grip.

Two fender-benders occurred in June, neither the Google car’s fault and neither causing more than scraped bumpers.