Participation sought for survey on perception of Volkswagen brand in US and Germany
Broad-Ocean Motor in strategic collaboration with Ballard; $28.3M investment, 9.9% ownership; initial order for 10,000 fuel cell vehicles

NTSB issues preliminary report for investigation into Tesla Autopilot fatal crash

The US National Transportation Safety Board issued its preliminary report for the investigation of the fatal 7 May 2016, highway crash in Florida involving the Tesla Model S and Autopilot. The preliminary report does not contain any analysis of data and does not state probable cause for the crash.

The preliminary report details the collision involving a 53-foot semitrailer in combination with a 2014 Freightliner Cascadia truck tractor and the 2015 Tesla Model S. According to system performance data downloaded from the car, the indicated vehicle speed was 74 mph (119 km/h) just prior to impact; the posted speed limit was 65 mph (105 km/h).

The tractor-trailer was making a left turn from westbound US Highway 27A (US-27A) across the two eastbound travel lanes. The 2015 Tesla Model S, traveling eastbound on US Highway 27A (US-27A), struck and passed beneath the 53-foot semitrailer.

As a result of the initial impact, the battery disengaged from the electric motors powering the car. The rear hatch frame separated and folded back over the crushed rear window.



After exiting from underneath the semitrailer, the car coasted at a shallow angle off the right side of the roadway, traveled approximately 297 feet, and then collided with a utility pole. The car broke the pole and traveled an additional 50 feet, during which it rotated counterclockwise and came to rest perpendicular to the highway in the front yard of a private residence. The 40-year-old male driver and sole occupant of the Tesla died as a result of the crash.

Tesla and Mobileye part ways
According to Tesla’s statement after the accident, the Autopilot technology apparently did not detect the side of the trailer against the brightly lit sky; nor did the driver. As a result, no brake was applied. (Earlier post.)
In October 2014, Tesla started equipping Model S with hardware to allow for the incremental introduction of self-driving technology: a forward radar; a forward-looking camera; 12 long-range ultrasonic sensors positioned to sense 16 feet around the car in every direction at all speeds; and a high-precision digitally-controlled electric assist braking system. Tesla has since relied upon software updates to increase Autopilot functionality.
Tesla currently uses MobileEye’s EyeQ3 processor to provide the image analysis intelligence that enables Tesla’s semi-autonomous technology. During its quarterly earnings call on Tuesday, Mobileye announced that its work with Tesla will not extend beyond EyeQ3.
Mobileye will continue to work with Tesla on improving the performance of EyeQ3 in use in Tesla cars. A new Mobileye system—the EyeQ4—comes out in 2018.

Other than the roof structure—and the front bumper components that engaged the pole—the main body of the car was generally intact.

The car’s system performance data also revealed the driver was using the advanced driver assistance features Traffic-Aware Cruise Control and Autosteer lane keeping assistance. The car was also equipped with automatic emergency braking that is designed to automatically apply the brakes to reduce the severity of or assist in avoiding frontal collisions.

A team of five NTSB investigators traveled to the site of the accident to conduct the on-scene phase of the investigation. The team used three-dimensional laser scanning technology to document the crash location, the damaged trailer and the damaged car. NTSB investigators continue to collect and analyze performance data from the car’s multiple electronic systems. This data along with other information collected during the on-scene phase of the investigation will be used to evaluate the crash events.

A key component of the discourse following the fatal Tesla crash has been the persistence of questions as to why the vehicle’s automatic braking system did not activate in the presence of an obstacle when the driver failed to brake. The Model S has, among other sensors, a forward-facing camera and a front radar sensor, and one clue may lie in a 14 July tweet from Elon Musk, more than a month after the incident, in which he announced:

This, combined with earlier claims that the truck’s white trailer against a “bright sky” had influenced the vehicle’s decision not to brake, suggests that the Model S braking system may be using radar combined with imagery to determine when the vehicle should initiate automatic braking. If so, such an approach would make it possible to develop a system in which information from the camera image could override radar information. The exact operation of Tesla’s automatic braking system is, however, unknown.

Such a system would differ somewhat from many existing automatic braking systems offered by competing manufacturers, which often use radar alone to determine necessary automatic braking events, and employ a forward-facing camera to assist with other duties such as lane centering. Radar is typically unaffected by contrast issues (light and dark), as it uses reflected radio waves at about 76 to 77 GHz to identify and classify objects.

A typical forward-facing automotive radar sensor has an approximate range of up to 250 meters (around 800 feet), which at 74 mph takes roughly 7.5 seconds to cover. Many automotive radar sensors take less than 100 milliseconds to signal the presence of an obstacle and calculate its relative speed and distance. Whether or not that signal is accurately processed by the automatic braking system is influenced to a large extent by the software that determines whether or not an automatic braking event is required. It is not known whether or not the Tesla’s radar sensor recognized the presence of the tractor-trailer prior to impact, nor how soon before impact the tractor-trailer crossed in front of the Tesla.

Caption: Automotive radar sensors must be calibrated along the (a) centerline of the vehicle, so that the (b) horizontal and (c) vertical ranges are correctly calibrated to the position of the vehicle itself. Horizontal and vertical range specifications are often application-specific. Source: Toyota Click to enlarge.

Musk tweeted on 17 July that Tesla had scheduled:

The next release of Autopilot (8.0), promised to be Tesla’s “biggest release since [Autopilot] v1”, is said to be running in select vehicles, with public release some time this summer.

All aspects of the crash remain under investigation. While no timeline has been established, final reports are generally published 12 months after the release of a preliminary report. The Florida Highway Patrol and Tesla Motors are parties to the ongoing investigation.

(Jack Rosebro contributed the section on automatic braking and sensors.—Ed.)


Account Deleted

I think the most interesting news is that Tesla and Mobileye is parting. I admit I did not see that coming. It could have been really bad news with regard to Tesla’s ability to launch Model 3 on schedule. However, the following statement from Musk has convinced me it is not. I quote “This was expected and will not have any material effect on our plans. MobilEye’s ability to evolve its technology is unfortunately negatively affected by having to support hundreds of models from legacy auto companies, resulting in a very high engineering drag coefficient. Tesla is laser-focused on achieving full self-driving capability on one integrated platform with an order of magnitude greater safety than the average manually driven car.”

Other news I have read have said that Tesla has hired top talent for chip design from AMD among others. Apparently Tesla is designing some of its own sensors and processors for the next generation autopilot hardware that will go into Model S and X first and then Model 3 at launch. Another, thing I read at electric is that the Giga factory will use a unique new cell format. Instead of the standard 18650 they will use a new cylindrical form factor and unique chemistries as well.

What all this tells me is that Musk is unhappy with the speed of how traditional suppliers work. Musk wants the innovation to go much faster and using standard components will just slow down the development of Tesla’s products. So Tesla does it alone for the key technologies that will enable the transition to a zero carbon economy on a global scale. If others will not do it Tesla will do it and they are not waiting for anyone to get on board.

Also I read that Tesla’s autopilot version 8 should enable Tesla cars to automatically exit highways along with major interface improvements. Tesla goal with autopilot is to make it 10 times safer than manual driving in all situations and achieve that goal as fast as possible. Tesla believes they are currently twice as safe (It is a belief because the data are not yet extensive enough to prove it with regard to the death statistics. We need at least 50 deaths to prove it). The beta designation for the auto pilot will not be removed until the autopilot is fully autonomous and conclusively documented to be 10 times safer than the average human driver. That would translate into 1 death per 900 million miles versus 1 death per 90 million miles as it currently is in the US.


The comments to this entry are closed.