>“However, in three of these meetings, internet connectivity issues likely precluded or hampered them from seeing the Full Video clearly and fully,” the report states.
It sounds like they're having issues with the meeting platform rather than with internet on the self-driving cars themselves. Dismissing an entire company's development efforts because of an IT SNAFU makes little sense. It'd be like dismissing the efforts of OpenAI because they had issues setting up a projector at a conference.
The internet issues were when showing the video live over screen-sharing in the briefings. Cruise did send NHTSA a copy of the full video later the same day:
> Just as the NHTSA meeting was concluding, at 10:59 a.m., a NHTSA employee sent Cruise an email titled “File request: CBI Cruise Pedestrian Incident 10/2/23” with a link to upload video of the Accident. At 1:40 p.m., Cruise sent the Full Video to NHTSA.
Is it difficult to extract the video _then_ send it to the investigators? Directly? Blaming their internet connection in the meeting for the missing video is like blaming the dog for eating the homework.
I think what we need is NTSB style investigations for every incident involving a self driving car.
An independent regulatory agency should secure the data, interpret it, and provide recommendations.
If this were an aircraft crash you wouldn’t have the designer/operator/manufacturer/maintainer (the same entity in all four roles) massage videos and hold presentations. You would have a competent and independent flyaway team secure the evidence and analyse it.
This is fundamentally in the best interest of the self driving companies themselves too.
NTSB already investigates certain “commercial” incidents (usually involving railroads or busses but sometimes other transportation).
Self-driving cars should certainly be investigated, even if it’s mostly just a cursory “can we file this as user error” - as the self-driving part is a commercial operation.
As it is the NTSB already records data on various aspects of driving incidents and sometimes notices something and investigates. It’s why it’s worth reporting if you notice safety issues.
> according to a report compiled by a law firm investigating the incident. The law firm, Quinn Emanuel Urquhart & Sullivan, was hired by Cruise to determine whether its executives misled regulators
This is why lawyers are so expensive. Totally worth the money in this case to hire an "independent" auditor.
Especially this quote from the ”Summary of Principal Findings and Conclusions”: “The reasons for Cruise’s failings in this instance are numerous: poor leadership, mistakes in judgment, lack of coordination, an “us versus them” mentality with regulators, and a fundamental misapprehension of Cruise’s obligations of accountability and transparency to the government and the public. Cruise must take decisive steps to address these issues in order to restore trust and credibility.”
The unstated allegation is that they weren't hired for a warm endorsement, but to assuage the regulator and the public.
If that were true, you could expect certain things, such as casting blame towards folks who have left for company culture, or framing issues with informing the regulator as technical difficulties (eg, instead of deliberate malfeasance).
That the resulting report aligns with such expectations isn't proof - but it might raise an eyebrow here or there.
If they were playing these videos for the regulators live, as it sounds like they were, they would have known the key segments weren't shown, right?
So they'd still have a duty to point out those details, because they'd know the regulators hadn't seen them?
Edit: this reminds me of kids intentionally corrupting their .doc files when mailing to teachers, so they could get an extension. Sounds like the kind of lie that, if proper forensics could be carried out, should result in serious jail time.
> During Quinn Emanuel’s review, the employees who had met with the D.M.V. disagreed about whether the company had shown the complete video to regulators. The bigger problem was that Cruise didn’t tell regulators about what had happened, the law firm said.
> “We were lucky they didn’t pick up on the dragging,” one employee said a participant in the D.M.V. meeting had said afterward.
> > “We were lucky they didn’t pick up on the dragging,” one employee said a participant in the D.M.V. meeting had said afterward.
Yeah...that's the real problem here. Something very bad happened with a safety critical product, and anyone involveds instinct is to think that "getting lucky things weren't noticed" is a good thing.
Not to mention speaking incredibly poorly of the moral compass of whoever said that.
It’s an interesting narrative that has been crafted for Cruise, but it beggars belief that Cruise would have ever intended to send the video alone and let it “speak for itself”.
> Over 100 Cruise employees were aware of the pedestrian-dragging incident prior to the October 3rd meeting with the San Francisco mayor’s office, the National Highway Traffic Safety Administration, the DMV, and other government officials. But Cruise said nothing about the pedestrian being dragged, instead letting the video speak for itself.
Quinn Emanuel is who you hire when you need corporate deniability.
Technically, it was bad internet, not an active coverup.
Uber got off by showing video with extremely bad dynamic range "see, the killed woman wasn't visible", Cruise now suddenly - after months long of expensive lawyers work - got "bad connectivity".
I've driven on Mill Rd. plenty of times, and it actually is hard to see pedestrians who have a bad habit of walking around wearing all black at night because they don't want to use the crosswalks that may be 1/2 mile away. The glare from all the lights makes it hard to see for normal humans, too.
Part of the problem is that their safety driver was watching something on their phone and not actually watching out like they were supposed to. If they had actually been paying attention, that would've been avoidable.
Still, the automatic cars have a much better record with me than the human drivers do. For all that these few accidents were stupid and avoidable, I've seen normal humans doing things far dumber.
She was also crossing at a weird point: just beyond where Mill Ave crosses under the 202, and then starts and uphill climb on its way to the Washington intersection. Despite all this, the safety driver almost certainly would have noticed her if they cared to be doing their job instead of watching TV on their phone, as you say, because at least the area is well-lit.
The reason we average nearly 1 pedestrian death a day here is because of what you cite: unwillingness to walk further to cross at a crosswalk, people weirdly wearing all black, at night, and specifically on the arterial roads, the suicide lane meaning that you're being blinded by oncoming traffic with a person as far away from the streetlights as possible. It's crazy. (The spot where she was hit on Mill Ave, ironically, isn't any of those things; it's the end of the bridge connectors, so it's one-way there, but still.)
I didn't realize that many people died per day, but it makes sense given how many people are doing the all black at night thing and playing Frogger with traffic.
I do think people underestimate how well the lights can blind you to pedestrians, even though I also think the safety driver should've seen the victim in this case if they were paying any attention at all.
>it actually is hard to see pedestrians who have a bad habit of walking around wearing all black at night because they don't want to use the crosswalks that may be 1/2 mile away.
It's not on pedestrians to wear reflective clothing just for you. Nor is it reasonable to expect someone to make a 1/2 mile detour just to cross the road. That could add 10-15 minutes to their journey.
You are the one operating a heavy, fast-moving piece of machinery. If it's dark and there are people around, drive slowly and carefully.
> You are the one operating a heavy, fast-moving piece of machinery. If it's dark and there are people around, drive slowly and carefully.
These are 5-lane, 40 MPH roads, and no pedestrians are supposed to be there at all. We're not talking about University Dr. here, which has lots of pedestrians, but also lots of crosswalks and a couple of pedestrian bridges.
Now, I have never hit anyone, despite having encountered many people doing illegal and dangerous things, but my own caution will not change that what they are doing is inherently risky. If they roll the dice enough times, eventually they will encounter a driver who is less careful than I have been.
It does not help that some of these folks have been obviously impaired, and shouldn't have been on the streets at all.
The right type of camera at night will outperform human eyes by a lot. My aftermarket birds eye camera kit makes the road quite easily visible when I can't see it at all.
This is the sort of thing which HDR image capture should be spectacular at.
I understood that to be an unusual configuration that was unintentional, but the Swiss cheese model of failure says that accidents always happen when too many bad things align.
That doesn't apply to cars because US traffic engineers design driving to be as dangerous as possible, instead of as safe as possible, so there isn't any built-in safety margin like with airplanes.
Like, their manual sometimes doesn't allow installing a crosswalk until /multiple/ people are killed trying to cross it.
The site has lots of recommendations, but doesn't seem to give much engineering behind the recommendations or mention any testing, just a handful of random stats. They don't seem to acknowledge the tradeoffs in some goals, either, e.g. between making roadways slower and climate goals, never mind that idling cars are quite wasteful and that people need to travel.
I can believe that we can do better, and it seems to me there should be crosswalks wherever there's significant pedestrian traffic without requiring anyone to die, but a lot of their other recommendations appear about as dubious as Brazil's choice to do things like putting unmarked speed bumps in the absolute middle of nowhere.
Especially the part that's anti driverless cars. For one, most things that would help that would also help human drivers (ambiguous road markings are nobody's friends), and two, the actual vehicles, which I encounter regularly, are much better than the human drivers. In theory, someday they could enable people to get rides on demand when they need from robotic taxis and get rid of the need for massive areas devoted to parking by helping the same number of people with fewer cars. This would also be good for the climate.
The fact that they're advocating against that which undercuts their stated goals calls their reasoning into question for me.
I didn't link to it for any of those, just to show the problem existed.
> between making roadways slower and climate goals
No tradeoff there for two reasons.
1. cars use more fuel when they go faster, whereas idling cars can be optimized by turning the engine off.
2. making driving worse and more expensive is an important part of shifting travel to other modes like ebikes/mass transit. Although people don't like when you put it that way.
It's hard to turn off your car in a traffic jam, efficiency is getting cars off the road. And those transit options are often horrible for the disabled. The fight against self-driving cars is especially shortsighted.
But yeah, wanting to make things worse for everyone tends to get a lot of push-back.
> Part of the problem is that their safety driver was watching something on their phone and not actually watching out like they were supposed to. If they had actually been paying attention, that would've been avoidable
Safety drivers are a poor solution, it's too easily boring (try watching driving videos for more than 10 min straight), so it's completely Cruise fault.
You are confusing Cruise with Uber in your response. Uber had a safety driver issue. Cruise had an issue with a driverless vehicle dragging a pedestrian.
Very odd. You'd think that with an incident like this the second these connection issues popped up they'd put the car on a flatbed and tow it to an office/garage to ensure all evidence can be secured
The alleged connection issue is not between the car and the cloud.
It is connection issues with the Cruise employee presenting the video to the regulatory agencies in a video meeting.
It is also only a tiny part of the whole picture. They made written reports to the regulatory agencies which failed to disclose the fact that they dragged the pedestrian. Overall the report is quite damning.
