Last Friday, Tesla rolled out an update for its autonomous driving software that includes “Traffic Light and Stop Sign Control” capabilities. The latest Autopilot version enables vehicles to stop independently at traffic lights. However, according to driver reports, the new functionality does not always work as intended.
Some Tesla owners have noticed their cars slowing down when approaching green lights. Many have posted video evidence on social media. One driver’s footage even showed the vehicle slowing to a 6-mph crawl while nearing a green light.
According to Tesla, the recent update is still in “beta” testing, which may explain the unanticipated consequences. The company aims to improve the underlying artificial intelligence algorithm over time as it collects more data from the field.
How the Update is Supposed to Work
When working correctly, the Autopilot upgrade displays a visual prompt on user dashboards as the car approaches an intersection. The message initially reads, “Stopping for traffic control in 300 ft. Use accelerator or cruise stalk to continue,” and appears to update the distance as the vehicle gets closer.
Still, Tesla warns drivers to pay attention. After installing the update, the in-car console displays text telling passengers to be ready at any time to take immediate action. The message also articulates the conservative nature of the feature, which makes sense, given that it is still officially in the testing phase.
“This feature will be conservative, slowdown often at first, and will not attempt to turn through intersections,” explains the console display.
Safety Still the Top Concern
Although the green light slowdowns appear relatively harmless, many are concerned about how other drivers will react. Duke professor, Missy Cummings, cautions that slowing vehicles can pose dangers to human drivers who don’t expect that to happen.
In general, she is wary of drivers using an autonomous system that isn’t 100 percent reliable. “There’s no upside to this software,” she warns. “There’s a capability being released to the general public that’s known to have significant defects.”
Professor Cummings’ concerns bring to light the questions that regulators need to answer regarding appropriate autonomous driving software testing. Other experts, like Paul Godsmark of CAVCOE, believe the new Tesla feature will lead to complacent drivers and subsequent crashes, revealing the many complex safety dimensions of self-driving cars.
The Elon Musk Way
Tesla’s iterative approach to software development is nothing new. Going back to 2016, we can find tweets from Musk about Autopilot versions that were still in “beta” after public launch.
Ultimately, the eccentric CEO believes his approach is the best path to achieving completely safe, self-driving cars. Tesla has always been the one to push the limits and break new ground, which has perhaps paved the way for other models. Tesla’s competitors, like Waymo and Cruise, rely instead on virtual simulations and test fleets that use professional drivers to gather field intelligence.
In the grand scheme of things, the autonomous driving wars are only beginning. There’s a long way to go before any single company is ready to roll out the technology at scale. Until then, we may see a Tesla or two sitting quietly at a green light.