Waymo's self-driving cars can recognize many types of shared road users, but the company on Wednesday specifically highlighted how its technology avoids cyclists.
In a blog update, Google's autonomous car subsidiary said it recognizes cyclists as a major road user, but acknowledged its software must be able to identify differences in cyclists and pedestrians. With 50,000 cyclists injured on the roads every year, it remains important for Waymo's self-driving cars to calculate the best defensive driving route.
In the video above, Waymo showed an example.
In this case, a Waymo autonomous vehicle identified a cyclist and correctly predicted the scenario that was about to unfold. The self-driving system predicted the cyclist was ready to shift out of the bike lane on the road to go around a parked trailer. The calculations were correct, and the Waymo van avoids any contact with the biker, and the self-driving car slows appropriately to give the cyclist space to pass the parked trailer on the side of the road.
The ability to identify what kind of object or road user is present is key in any situation. The company said the autonomous car cannot, for example, confuse a pedestrian for a cyclist. Importantly, Waymo's vehicle and software correctly identified schoolchildren crossing an intersection in a previous test. While a biker has a hard time suddenly changing directions, pedestrians can change direction at any time, which makes calculating a defensive driving route more difficult. The technology is able to predict objects, such as other vehicles, animals, and the examples spoken of based on their speed, trajectory, and other road factors.
Waymo continues to test its self-driving cars and operates the industry's first paid ride-sharing service in Arizona. However, the service is not open to the public. Instead, potential riders are screened and approved to hail a self-driving Waymo.