Tesla to stop killing drivers: Software update beamed to leccy cars
'Autopilot' adds radar and driver prods
Tesla is changing how its "Autopilot" super-cruise-control works in response to the death of one of its customers.
The over-the-air software update will be automatically applied to the electric cars this month and will expand the use of radar sensors to decide whether a crash is likely to occur.
In a blog post on the Tesla website, the company explains that the radar sensors added to cars in October 2014 were initially intended as a "supplementary sensor to the primary camera and image processing system." But they will now be used to determine car actions without requiring the camera to confirm the existence of an object.
Radar is not a perfect tool to see what is going on around a car, Tesla's techies explain. People appear partially translucent and wood and painted plastic are effectively invisible. In addition, metal objects can appear much larger than they really are from one radar sensor's perspective due to metal reflecting the signal. That could result in a car slamming on the brakes to avoid nothing larger than a soda can, the company warned.
Tesla's solution is to combine the input from all radar sensors every tenth of a second to build a more reliable 3D picture of the world outside the car.
Why the change? Because it was the car's failure to distinguish between a white trailer and the bright sky behind it that led to 40-year-old Joshua Brown being killed in his 2015 Tesla Model S while driving in Florida in May.
The car was using the optional autopilot system, and both he and the computer failed to spot the 18-wheel tractor trailer. The car went under the trailer, slicing the top of the car off and killing Brown.
A preliminary report from the US National Transportation Safety Board (NTSB) in July said Brown was traveling at 74mph when he hit the trailer – the speed limit on the road is 65mph.
There was another factor initially mentioned by Tesla and covered in the new update – but it was not picked up as much as the failure to distinguish the sky from the trailer was.
Back in June, Tesla said: "The high ride height of the trailer combined with its positioning across the road and the extremely rare circumstances of the impact caused the Model S to pass under the trailer, with the bottom of the trailer impacting the windshield."
In its post, the company alludes to this when it says: "When the car is approaching an overhead highway road sign positioned on a rise in the road or a bridge where the road dips underneath, this often looks like a collision course. The navigation data and height accuracy of the GPS are not enough to know whether the car will pass under the object or not. By the time the car is close and the road pitch changes, it is too late to brake."
The implication being that perhaps Brown's car did notice something ahead of it but calculated that impact was unlikely. (Of course, if the US passed a law requiring "Mansfield bars" (underride guard hanging from rear of trailer) on the sides of trucks as well as the back, the death could most likely have been avoided altogether.)
Tesla's answer to the height issue is to use its network of cars to build up a database of road signs, bridges and other similar objects using radar to build up a blueprint. The car can then compare that blueprint with the real world as a car travels along the road. If it sees something out of the ordinary, it is far more likely to be a possible obstruction.
"If several cars drive safely past a given radar object, whether Autopilot is turned on or off, then that object is added to the geocoded whitelist," the company notes.
"The net effect of this, combined with the fact that radar sees through most visual obscuration, is that the car should almost always hit the brakes correctly even if a UFO were to land on the freeway in zero visibility conditions," the post smirks.
Tesla even claims that it will be able to see objects in front of the car ahead of you by bouncing radar off the underside of the car and recognizing them, thanks to "the radar pulse signature and photon time of flight." The idea is that your car will brake to avoid something even if the car in front doesn't (such as in heavy fog).
Why doesn't Telsa use the "lidar" system that other self-driving competitors like Google use? Lidar is basically the same idea as radar but using lasers rather than radio waves. Musk claims it's because lidar doesn't offer the same capabilities. Although it may also be because all recent Teslas have radio-wave emitters and receivers, and introducing lasers would require a significant re-engineering effort.