A person laying on the ground in a crosswalk was likely never considered by the team to include in their training data. Those outlier situations are exactly what real world data is needed for. And the only way to properly train for most of these situations is to drive in the real world. The real world isn't perfect situations and nice lines on fresh asphalt so while base training in perfect situations is useful, it will still miss the exact same situation in a real world environment with crappy infrastructure.
Not sure what or how Cruise uses the data collected in real-time, but I can see camera visuals categorizing a person laying in the crosswalk as something like damage to painted lines, and small debris that can be ignored. Other sensors like radar and lidar might have categorized returns as something like echoes or false results that could be ignored, again because a person laying in the crosswalk is extremely unlikely. False data returns happen all the time with things like radar and lidar, millions of data points are ignored as outliers or info that can be safely ignored, and sometimes that categorization is incorrect.
A person laying on the ground in a crosswalk was likely never considered by the team to include in their training data
I didn't bother reading any further than this. The person was on the crosswalk when both cars started moving. Neither car should have been moving while anyone was still on the crosswalk.
I actually work at one of these AV companies. We definitely have training data on adults and children laying down. I'd be very very very surprised if Cruise doesn't due to all the people laying down on the sidewalks in SF. In addition, the clarity of the lidar/camera data on objects on the road is very clear. You can see the dips and potholes in the road as well as specifically see the raises of the painted lines. There's no way they weren't tracking the person.
I could see predictions on the pedestrian saying the coast is clear. Once the initial crash happens, there likely isn't enough room to stop in time even with a max break.
A different question I have is whether or not the cars have transponders or other communication devices to automatically call emergency services in case of accidents. I'm assuming not because they would probably have a lot of junk calls and I doubt the company would have spent the time to create an algorithm for when to call 911 if they didn't create an algorithm for what to do if there's a pedestrian in a crosswalk.
That's one of the big downsides of these driverless cars: if a human accidentally ran over the victim, they have the capability to get out of the car to assess the situation, call 911, and offer aid to the victim. An empty car can only ever just sit there with its hazard lights on and maybe call for emergency services.
The auto-driving company should be required to have something like an on-star operator available any time the vehicle receives an impact/shock above a certain threshold and any time physical safety measures are required. The local governments should not have to pay for the externalities created by these 'disruptive technology' jerks, especially when there are literal lives on the line.
Without knowing what type of vehicle the first car was, it's hard to say how this played out. If it was a van or truck or something else that could've easily obstructed Cruise's LIDAR system or if the other vehicle stopped ahead of the crosswalk line, the SDC would've had little to no way of knowing that there was anybody in the crosswalk.
Can't know for sure unless Cruise releases the video to the public, which they're unlikely to do until the police do their investigation.
If your vision of the crosswalk is obstructed, you don't proceed through until it's unobstructed. That's true whether it's radar, lidar, or vision. Truck in the way? Pull up as far as you can safely see, then look and proceed if clear.
There needs to be full transparency on these fleets rather than having governments bend over backwards in the name of trade secrets. We’ve gone absolutely too far in that direction with everything from vehicles on our streets to fracking chemicals in our groundwater.
We desperately need footage of this to make any conclusions.
The human driver hit her first, and knocked her into the neighboring lane...
So she was hit and flung into another lane...
...directly in front of a Cruise autonomous vehicle (AV) that was driving around by itself with no-one on board. The self-driving car then ran her over and came to a stop on top of her body, turning on its hazard lights. Her leg was pinned down by the back tire.
So it stopped Like others have mentioned, driving over someone under a car can cause more injury than not moving. Was she screaming to move forward/move back? Was she flung in a way that a human driver could have stopped in time?
If this was just a hit and run and there was no footage like what was provided by the ai car it could be that the victim or her family would be 100% on the hook for the medical bill. It will be interesting to see if the perpetrator is found and the footage surfaces with details so we can get some answers.
I'm hoping that the car has multiple cameras recording so maybe the hit-and-run driver will be caught and prosecuted. That said, unless I misunderstood the article, I think the driverless car (DC) didn't do a horrible job here. It sounded like the victim was struck and flung in front of the DC and it stopped (unfortunately on top of her). I don't know if I could have reacted better. The article wasn't clear but it read like the car contacted the police and the police instructed it to remain where it was, which is what I would have done if I were driving a car. The DC was then lifted off the woman by emergency personnel. We can't expect DCs to be magically perfect. Just like we don't expect people to be perfect. A DC is only as good as it's programming. Hopefully this incident will be studied and if a better solution is found that can be integrated into the DCs operations. I really feel bad for the woman here. I don't know but even if she shouldn't have been walking no one deserves that. Let's hope the hit-and-run driver is caught.
Yeah but the focus will be on the driverless car, I’m not saying they are blameless, just that the general heat and attention will probably not be on them now
One of my son’s coworkers was just killed in a similar incident. Woman hits a pedestrian, she freaks out and calls her boyfriend instead of emergency services, boyfriend arrives and runs over the injured pedestrian ensuring he was dead.
They are unsure which vehicle actually killed him.
Perhaps he's implying that due to the second car it may obfuscate if the first driver's actions would have been lethal or just left the person injured. He'd probably rather be tried for a hit and run resulting in injury than a hit and run resulting in a death.
If you get stuck under a car the fire dept is going to come lift it off of you. They aren't going to try and drive it off that would almost surely cause further injuries.
Reminds me of someone I knew that would keep a pint of Jack in his trunk. He drove drunk constantly and it was there for when if he ever got into an accident he was prepared to run out of the car, pop the trunk, and pound the bottle in front of all the witnesses.
Can't prove he was drunk at the time of the accident.
I know I’ve read of at least one successful case where the person fled the scene and went home, then claimed he was drinking at home. Honestly, though, there’s so many things that factor into whether an individual gets arrested or released that we’d need more examples to differentiate between just letting someone go and This One Simple Trick Judges Hate.
It's up to the police to show you were driving while intoxicated. You have witnesses corroborating that you were drinking after the accident. Any field sobriety, or blood test they give you would be worthless because it would be after that.
You might not be able to prove it, but anyone willing to chug alcohol in front of witnesses to have that kind of plausible deniability can easily be assumed to have already been drunk to start with. That just doesn't seem like it would hold up to the 'reasonable doubt' standard...