Tesla fatal car crash prompts NTSB investigation
techcrunch.com
techcrunch.com
The big take-away from their research is that particular exit-only left-lane (which is one of two side by side carpool lanes) is poorly marked, has been consistently gotten worse as changes have been made to the road over the last several years, does not have proper lane markings (missing chevrons in the non-lane), routinely has cars cutting right across the gore point (to the point where there are dash cam videos of almost accidents at that exact point linked in the forum) and, most damning of all, the resettable safety cushion device [2] which protects the gore point had already been hit just recently and there’s evidence it had not been reset when it was hit again by the Model X, which meant the X suffered a direct impact with a narrow slice of concrete.
If anything I think NTSB should be investigating Caltrans for this one.
[1] - https://teslamotorsclub.com/tmc/threads/model-x-crash-on-us-...
NTSB should look into this in the event that the person was under autopilot at the time, or if there was some flaw in the design of the model X that should have helped or mitigated the death of the driver. I would love to see better guidelines for tesla and for emt/fire staff as the road was shut down for about 6 hours while they tried to deal with the battery. Someone from tesla eventually came down and helped the fire dept deal with the fire...
So, sure, this is a tricky and maybe even dangerous fork, and caltrans will probably put in those dividers like the ones on southbound old middlefield road, but its okay and smart to look into the driver, the model x and the subsequent 6 hours of delay and blockage to improve safety across the board.
This is explicitly what the NTSB mentioned in their tweet (included in the article) announcing the investigation:
> "2 NTSB investigators conducting Field Investigation for fatal March 23, 2018, crash of a Tesla near Mountain View, CA. Unclear if automated control system was active at time of crash. Issues examined include: post-crash fire, steps to make vehicle safe for removal from scene."
edit: correction this photo evidence from the day before shows the barrier not replaced. https://finance.yahoo.com/news/tesla-knows-fatal-model-x-030...
Does anyone know if that sort of impact would be what the IIHS calls small overlap frontal crash? IIRC when they added that test in 2012, many previously very safe cars were found to do poorly in that specific test because so much energy is concentrated into a very small area.
https://www.youtube.com/watch?v=YSYLE55iYj0
or the more recent:
https://www.youtube.com/watch?v=nnRIwQn9SA8
the latter is imo even more amazing
I'm not referring to this particular case, but similar cases in general: There are a lot of people merging right on that place in the last minute, just like there are a lot of people merge left on an exit only lane on any other highway exits. It's not because they are poorly marked, it's because the exit only lane moves faster and they want to cut other people. Even if the lanes are poorly marked (I don't think the side-by-side carpool lanes are poorly marked, there are plenty of signs indicates that the left one is exit only) and they are on the exit only lane by accident, they should still have a common sense about what's safe to do. They can always exit there and come back to the highway on the next entrance. They just decided that their time is too valuable to do that so they are just jerking around. How is that Caltrans' fault?
Informational sheet describing the benefits:
http://www.dot.ca.gov/hq/esc/oe/project_ads_addenda/06/06-0Q...
The picture shows the front side of the car completely incinerated. Not to mention, the cleanup of the crash took much longer than it should've because of the EMS had serious concerns about the melted batteries.
Tesla cars are great until you get trapped inside a battery-fueled inferno. Youtube it, those batteries literally create a meteoric fireball.
Incorrect.
“Witnesses were able to extract the driver before the flames posed any threat to him.“
Here is VIDEO of the Tesla immediately after the crash, taken by one of the guys who helped pull the driver out of the car:
The battery is only just starting to ignite at this point after suffering catastrophic damage. (You can see individual cells torn apart). I would say this is significantly safer than a dozen gallons of gas which surely would have exploded in the same scenario.
What IS true is that the aftermath is harder to cleanup because the damaged cells can pose a threat for reigniting.
Look at the Tesla in that video - the rear of the car is pretty much intact.
At least then you know everything burnable is already burnt, so it's not going to reignite.
Also IIRC Mythbusters did an episode on this.
You can clearly see at 30s the outline of the inferno on the tire. It completely incinerated all the shreds of car were destroyed from the crash.
Even at the beginning of the video, where it shows where the front of the vehicle should've been, it too is barely there because it was incinerated but the batteries.
It takes first responders hours to deal with, sounds like Tesla has traded gasoline for a much more lethal battery. Once again, cost externalization.
It's not so simple to stop a damaged lithium battery from continuously self-igniting. Read about the recent incident with the Rimac Concept One—it kept catching fire for five days!
As a firefighter, I think you and I likely have different ideas of what "simple" looks like in this case...
EVs definitely pose unique hazards, but there is nothing "simple" about a car fire.
But I suspect your training and experience with car fires is substantial compared to EV fires. I suppose I didn't meant to say they are simple but rather well understood.
I wonder when there are literally hundreds of unique popular models of EV on the road, whether there will be issues arising from having innumerable unique designs for battery packs, first responder lines etc.
Tesla in particular is actually quite good about providing training materials for dealing with their vehicles, including facilitating the production of some excellent videos on hazards during extrication.
https://www.youtube.com/watch?v=ntK3rvVl2Qw
As well as the standard safety manuals car manufacturers provide (though Tesla makes their especially easy to access).
With the Uber accident, I think the public is going to be very wary about over-hyped self driving claims. Tesla is a big offender in that regards where the marketing is that its collision avoidance and lane assistance technology is "Autopilot".
Unless there was a typo in the article I read
edit: Worth pointing out that in the previous Tesla fatality investigated by the NTSB for Autopilot-related issues, the victim was well known as someone who posted videos that showed off his Tesla's Autopilot capability. [0] I don't think anything like that is known about the 38-year-old victim in this week's incident. More details of that accident are here, from the San Jose Merc:
https://www.mercurynews.com/2018/03/27/ntsb-investigating-fa...
> Friday morning, a San Mateo man died when his Tesla Model X crashed at the Highway 101 and Highway 85 connector, authorities said. Wei Huang, 38, was traveling in the carpool lane of southbound 101 around 9:30 a.m. that morning when he collided with the barrier separating his lane from the Highway 85 flyover carpool lane, according to the California Highway Patrol.
[0] https://www.nytimes.com/2016/07/02/business/joshua-brown-tec...
With more and more auto-piloting cars on the road calculating the road and make decisions on what's the safest way to continue, wouldn't be it rather easy for this data to flow back? Spots that are often causing trouble for the brains of the autopilot could then be identified easily and be the road, markings, signage, etc improved much quicker. Today's method are most likely a mix of experience and looking at the fatality and accident statistics for last year.
I dont really have any commentary one way or another about autopilot or the safety of batteries, I'm just thinking there are probably other accidents that have happened in the industry in similar conditions to this.
For example, the Prius and Volt have been on the road for quite some time, and both cars carry relatively large batteries (though certainly not as large as that on a Model X). When involved in head-on collisions do they see such drastic damage/fires?
Just curious. I don't know that its even fair to compare those types of vehicles given the difference in the size of battery.
So today the NTSB probably investigates very, very few car crashes, but that should rise quite a bit.
They don’t investigate many per year though, so this is something special. I’m guessing they will be investigating these new technology crashes.
After all, EVs are becoming increasingly common; if fire departments need special training or equipment, now's the time to issue recommendations.
2030 is coming soon, and people have short memories.
In other words, yes, you will have high-tech enclaves ("works around Chandler because we already know how to drive there"); they will be minor compared to the rest of the world still living in 20th century (if even that). Try driving a SDC outside the population centers - this is literally uncharted territory. "But but but cities are hard, everything else is EASY!" Oh yeah? Once SDVs venture out in the open, we'll see another wave of deaths (beyond the one that begins now): "car drove itself off badly marked road", "car stranded itself and did something profoundly stupid" etc.
Driving with exact maps takes off one of the hard parts; having a point cloud of the entire continent sounds improbable.
I think that car that would drive themselves below 1mph would already make people happy. You can move things around near house, near stores, near office. At speed that are near danger free. And it feels a tad exciting as consumer experience.
No pedestrians to worry about, crashes easier to avoid, no mixing of autonomous and active traffic, and faster travel times.
If we're talking safety, thousands of low-flying objects in dense urban areas is not a good idea in the foreseeable future. And certainly not in public perception.
How is it more dangerous than 12 lane autonomous intersections?
Until such laws and tests are devised, the all testing of SDVs on real roads should be suspended..
Why are people not asking/demanding this?
Sad, given that we don't even know if Autopilot was on during this crash, and yet people immediately assume this is a self-driving issue.
Eventually, autonomous vehicles will come around but in a much more measured and cautious deployment, as opposed to the VC cash fueled headlong rush we are currently experiencing.
That is, do not allow autonomous mode, unless there is an old-school road-embedded autonav system and if there are other vehicles on the road, it be able to realtime comm with at least xx% of its directional and radial neighbors (to add predictability and reduce misinformation.) This in addition to all current guidance systems.
Obvs road-embedded systems do not account for debris, accidents, etc., but know the ideal path and would be backup-of last resort (incapacitated operator, other systems down). Real-time comms with neighbors surfaces good information and buries stale or inaccurate info.
What absolutely cannot work safely is any sort of system that relies on lane markings. Anyone who's driven around the Seattle area can tell you that much. State DOTs simply DGAF about compliance with road-marking practices, especially with respect to visibility in bad weather.
And then there's the question of snow. Additional fail-safe mechanisms of one type or another are definitely going to be necessary.
What I'm saying is that given that the technology has not yet proven itself in demanding scenarios, we may ask for more rigorous requirements if this is to become the default navigation mode of the future (no human intervention).
Do we have data on how many times people have had to avert an incident that wasn't reported (my autonav almost crashed me, but I took over)? If we had public data, we could make some kind of assessment but even then, if they proved good, we could make them even better yet with the additional support/failsafe systems.