Waymo is rolling out a software update to enhance its robotaxis’ ability to handle traffic lights that are out of service during power outages “more decisively,” as outlined in a blog post released by the company on Tuesday. This update follows an incident where its self-driving vehicles experienced delays at intersections during a blackout in San Francisco over the weekend.
According to Waymo, the self-driving system in its robotaxis interprets non-functional traffic lights as four-way stops—just like human drivers are trained to do. This should have allowed the vehicles to carry on as usual despite the widespread outage.
However, many of the vehicles ended up requesting a “confirmation check” from Waymo’s fleet response team to ensure they were making the right decisions. All Waymo robotaxis can issue these confirmation requests, but with the extensive outage on Saturday, there was a “concentrated spike” in these requests, which contributed to the congestion that was captured on video.
Waymo acknowledged that they developed the confirmation request system “out of an abundance of caution during our early deployment,” but are now tweaking it to better align with their current operational scale.
“While this strategy worked well for smaller outages, we are rolling out fleet-wide updates to give the [self-driving software] specific context regarding power outages, allowing for more decisive navigation,” the company explained.
The software update will incorporate “even more context about regional outages” into the self-driving software. Additionally, Waymo plans to enhance its emergency response protocols by “incorporating lessons from this event.”
While the focus has primarily been on situations where Waymo’s robotaxis got stalled during the blackout, the company noted that its vehicles “successfully navigated over 7,000 dark signals on Saturday.”
“Navigating such a significant event posed a unique challenge for autonomous technology,” the company added.
The recent situation highlights that Waymo is still identifying unexpected challenges with its software and approach to developing a dependable fleet of self-driving vehicles. The company has previously had to implement multiple software updates to ensure that its robotaxis stop for stationary school buses, which led to a National Highway Traffic Safety Administration investigation and even a recall.
