Driver assistance technologies like Tesla Autopilot are billed as ways to keep human drivers out of trouble, but it’s always impressive to see just how well they can work. Footage of a Tesla in Autopilot mode predicting a likely crash on the road ahead, and stomping on the brakes to keep out of trouble, has emerged from an incident in the Netherlands. According to the video, shared by Hans Noordsij, the Tesla Model X sounded the forward collision warning and began slowing before the human driver had time to react.
What’s particularly notable in the incident – aside from the fact that, according to Noordsij, everybody involved survived – is that Autopilot predicted the collision from data two cars ahead. The sensors used in Tesla’s system can effectively peek at cars further in front, by bouncing radar signals underneath traffic ahead, or around it. It’s a fairly new addition to Autopilot’s talents, however, coming as part of Tesla’s software 8.0 update which began installing to cars in mid-September.
It’s an example of how Tesla has progressively – and, particularly for the auto industry, rapidly – upgraded its Model S and Model X vehicles, not just patching problems as per traditional automaker recalls, but noticeably improving systems. That has brought the electric car company some headaches in the past, with criticism from some quarters about how quickly it added semi-autonomous features like Autopilot. Nonetheless, this incident also demonstrates the flip side to those arguments for caution.
Although computer drivers still can’t deal with every situation a human driver might be able to, the increasingly advanced sensors installed in vehicles do have some talents that flesh & blood lacks. Being able to see “through” the car directly in front and track traffic further down the road is one example; several car companies are exploring vehicle-to-vehicle (or car-to-car) or vehicle-to-infrastructure systems, whereby cars could share reports of incidents far beyond the visibility of either humans or individual car sensors themselves. That, though, as Audi discovered when it began connecting its car dashboards with traffic light information, will require some significant infrastructure upgrades to enable.
Likely to come sooner are advances in individual car technologies. Tesla’s latest cars, for instance, use an updated suite of sensors – known in owner parlance as “AP2” versus the original “AP1” system – which significantly increases the number, and range, or radar, sonar, and camera vision systems on the cars. Autopilot upgrades bringing feature-parity between AP1 and AP2 are expected to roll out before 2016 is through.
Eventually, Tesla has said, the AP2 technology will support fully-autonomous driving. That’s still several years out, though; before that, it’ll be used for what the automaker is calling “Enhanced Autopilot” with a new range of driver assistance features. That’ll include the ability to navigate on-ramps and off-ramps without driver intervention, as well as automatically change lanes.
Original video, authorisation from the owner. Essential, no one could predict the accident but the radar did and acted by emergency braking. pic.twitter.com/70MySRiHGR
— Hans Noordsij (@HansNoordsij) December 27, 2016