Glad this calls out Cruise specifically, and not 'robotaxis' in general. Been seeing a lot of coverage lumping Waymo & Cruise together, but Waymo, while it has its own share of mistakes, is generally a large step up in terms of driving performance compared to Cruise. (Source: I've ridden in both many times, and its pretty clear)
having been stuck behind both Waymo and Cruise cars.. I don't like either.
the worst is when you're stuck in a one way street, and the car just happily stays immobile with blinkers on. one time I was stuck (along with at least half a dozen other cars that I could see) for almost 40 minutes.
i'm sure you had a good experience as a rider, but i've had a horrible experience multiple times as a city commuter.
This happens where I live without robo taxis: we have a lot of effectively one way residential streets (one way because parking is allowed on both sides, not because the street is really one way). And some delivery truck will just stop on the road, put their blinkers on, and make their deliveries. Sometimes it’s an actual moving truck, you need to reverse out of the street somehow, but if someone is behind you, your stuck.
"Cruise blamed cellphone carriers for the problem."
"As I understand it, outside lands impacted LTE cell connectivity and ability for RA advisors to route cars."
I'm sorry, in what world is stopping in the road an acceptable response to losing cell service? How the f are we expected to trust a company that thinks this is not their fault, and that designing their "self driving" cars around 100% cell service or random(), is "safe"?
If you are putting "self driving" software on the road, and it behaves badly, then just like a human driver would be, you are responsible. Until Cruise, Waymo, Tesla, etc are directly, and non-transferrably liable for all "mistakes" made by their "self driving" systems we're just going to get more of this BS.
California DMV code 227.02 defines an autonomous vehicle as one that has a remote operator, and 227.38(b)(1) demands a continuously maintained and monitored two-way link from the operating business to the vehicle, even when not being operated.
I didn't see any language demanding that the vehicle become inoperable in a cellular dead zone, but if their paramount concern is to shield themselves from liability, then having the car disengage is one way to ensure that.
For residents of San Francisco that were personally affected by this, consider writing a polite note to the DMV asking them to ensure that this incident is included in this winter's upcoming disengagement reports (227.50) from Cruise, and share your point of views — both on how you expected autonomous vehicles to behave when cellular communication is disrupted, and on whether you think that the existing guidelines are sufficiently clear about that scenario. <autonomousvehicles@dmv.ca.gov>
> I didn't see any language demanding that the vehicle become inoperable in a cellular dead zone, but if their paramount concern is to shield themselves from liability, then having the car disengage is one way to ensure that.
If I block a bunch of roads with cars I am violating CVC, at least once per car, and would get fined every time. I'd also possibly get penalties on my license.
For example, say I'm driving down the road and start feeling tired. I realize that it's unsafe for me to continue driving. Can I just stop where I am and have a nap and claim I have shielded myself from liability?
CVC violations and license penalties are not liabilities. Both you and self-driving cars should earn penalties for that behavior, and honestly I'd like to see Cruise ticketed for the CVC violations documented above.
However, by coming to a stop in the middle of the road and having a nap, you are maximally protected from any damages or harm you might otherwise incur. If someone runs into you, they will be partially at fault; and, as you are stopped, you cannot run into someone else. From a purely selfish standpoint, that would be a completely appropriate choice to make.
Fortunately, the points system used in California for individual drivers issues penalties for non-harmful, non-destructive violations so individual drivers will still be penalized for taking purely selfish actions (like parking in the middle of the street, though YMMV region-to-region). There are too many drivers and not enough workers at the DMV to converse with every individual about their year of 'disengagements' while on public roadways, for the DMV to define protection of the commons by "once a year, we ask you for a self-audit of your fuckups", as they do with autonomous driving companies today.
I don't have an opinion on whether that is the best possible outcome or not right now — but it's as close as I can describe how things are today, and why you should not expect to minimize liability like an autonomous vehicle does without penalty. I hope this is of some use to your question.
Yeah, I’ve wondered if Cruise and Waymo get tickets for this stuff. But if they do, paying them is probably not a big expense for these large companies.
I don't know, as it wasn't interesting to me; and, I don't recommend contacting the DMV for retaliatory purposes. Others are welcome to speak to this if they know more, though.
Personally, I want to see the DMV define expectations for autonomous cars with respect to interrupted cellular service, so that Cruise and others cannot simply just park in the middle of the street as the selfish-lowest-liability choice. It's possible the DMV will define an expectation of the car coming to a stop, or continuing to operate for up to five minutes, or whatever.
I also feel like Cruise's disengagement report has a chance of containing materially-interesting data around why they disengaged — for example, can their cars operate without a cellular connection at all, or are they dependent on The Cloud?
> Personally, I want to see the DMV define expectations for autonomous cars with respect to interrupted cellular service
The rules for self driving cars should be at least the same as for people. If things differ at all, it should be making them harsher - people get a heavily restricted degree of latitude as we recognize human brains suffer fatigue, etc but computers don't suffer those issues.
Similarly liability (civil and criminal) for dangerous, unsafe, or simply incorrect driving shall be with the driver. With a self driving vehicle, that is the manufacturer. The whole reason for the "driver must pay attention while self driving" is to liability shift from the manufacturer of an unsafe product to the purchaser.
I agree that "stopping in the middle of the street" should be discouraged. Certainly one solution is to remove the liability for continuing to drive. Another is to increase the liability (if insufficient) for stopping in the middle of the street.
I don’t know if Cruise cars stop if cell connection is lost or not. However the remote operator can help unstuck the car. So it could be the car got stuck for some other reasons and then ROs couldn’t help.
You’re incorrectly assuming that lack of cell service causes the cars to stop. The quote you copied says it impacted their ability to remotely get the cars unstuck. What alternative would you propose (besides adding back in safety drivers)? Should the car keep going when it is not confident to do so?
If the car is not able to safely operate without constant human involvement then it definitionally requires a person on board.
If losing cellular service causes it to stop, then in emergencies you're pretty much guaranteeing a large number of roadblocks for emergency services, on top of their existing hazards to emergency services when operating in ideal environments.
> Should the car keep going when it is not confident to do so?
See above, re: if your car can't even pull over and park safely on its own it has no business being on the street.
The idea is that remote operators help get the car unstuck, but the car should always be able to avoid collisions on its own. I think these cars are supposed to pull over as well instead of blocking traffic.
Key word here being “supposed to”, of course problems occur all the time. Also once a car gets stuck, pulling over is hard because… it’s stuck.
This is a mess, but there's some beautiful irony seeing driverless cars temporarily turn streets into pedestrian-only zones. Are driverless cars the ally that the walkable city movement never knew it needed?!
"The CPUC decided to go ahead anyway. One of the three yes votes was cast by Commissioner John Reynolds, who served as head lawyer at Cruise before appointed to the CPUC by Gov. Gavin Newsom."
Perhaps this somewhat publicly unsupported rollout will serve to highlight the as yet unsolved problems with self driving in a city environment, and the rest of the world can take note and act accordingly. Unfortunately of course it is the local residents who will have to suffer.
I remember an essay on that from a transportation guy back during the first hype wave of self driving cars. He said you can rate transport by the number of passengers you can load per hour.
Rail and light rail excel in that. And loading usually doesn't disturb other traffic. Busses are worse but better than private cars. Taxi's are worse than private cars.
He thought self driving cars will be the worst of all. Low passengers per hour, a lot of empty trips, and highly disrupting in high traffic areas.
> He said you can rate transport by the number of passengers you can load per hour.
That's not really a fair metric. Taking rail requires me to take a bus at each end and increases my journey length considerably. I would be boosting the passenger per hour of 3 vehicles despite a worse experience?
I think metrics should capture the full cost and experience of the system. Having someone take transit for an 10% longer journey can be worthwhile, having someone take transit for 300% longer journey to save traffic might not be worthwhile if it reduces their labor output to society considerably.
Exactly. Such a metric dismisses the traveller's view. It's a "pat yourself on the back" kind of metric. Reminds me of ... the often misused idea that more traffic lanes don't solve problem because they will soon enough actually be used by people who find a benefit in using them (the horror).
One would hope for metrics that take into account the benefits to the travellers. A cab or similar is used because it brings a benefit to the user. NOT typically for the pleasure of spending more money.
It's really a planning discussion not a personal one. Planners where I lived chose rail as well as streets and I get reliable transport by rail or the Schroedinger street choice.
Passengers will choose alternative options if it isn’t convenient enough. So yes it is important to take personal experience into account.
Also theoretical throughput is different than the real one. A380’s CASM (cost per available seat mile) is awesome when it can be loaded in full, except that it isn’t usually the case. The program is doomed as the result.
The marginal driver cost is what creates equilibriums in the conventional modes. Either they're scheduled to work a shift, or they're owner-operators. Both create constraints on how many vehicles can be dispatched, the form they have to take to satisfy demand, and the logistics around them(train tracks, parking spaces, automotive supply services, etc.). The large footprint of private auto and taxis lies within these constraints, as does the high-cap status of buses and trains.
The new modes are actually just the cost of the vehicle and support team, which means that the fleet can feasibly operate tiny, light, high frequency vehicles instead of or in addition to huge, consolidated, scheduled ones. You could have a self-driving three-wheeler, a self-driving minibus and so on. Loading can be standardized by policy.
That they aren't doing that right away just reflects the current cost structure and legal framework being weighted towards larger prototypes following conventional modes. But the reason why a robotaxi makes a better taxi is because it isn't a taxi in practice, it's transit. It's centrally planned, it has standardized fleet management.
There are many things to love in the potential of self-driving cabs. Fleets can also be parked in distributed smaller spaces (impractical for cabs or buses - typically). Fleet cars can be idled in place (head to a street with typically available space and just stop there.) And this is even if we never get to privately owned or leased vehicles being used as part of a fleet operation when not wanted by their private owner.
Suffer what? A traffic jam on some day in a major city? That’s a pathetic bar to set for what a city can stomach with what well running autonomous taxis can mean for the future. Massive potential savings in greenhouse emissions, 24/7 on demand and reliable transport, efficient route planning preventing traffic jams, massive reduction in deaths caused by traffic accidents etc.
Right tell that to the people who have crashed and been horribly mutilated or killed from unexpected traffic jams or ambulances or fire trucks that can’t get through because of unexpected traffic.
It’s not like regular traffic where everyone knows this part of the road is jammed up at this specific time range if the cars are just locking up in random areas.
People have been killed for decades by traffic jam negligence. Another freeloader making an entire jam is just pointing out how wrong it is to let dangerous private property have unlimited use of public streets for free.
> Massive potential savings in greenhouse emissions, 24/7 on demand and reliable transport, efficient route planning preventing traffic jams, massive reduction in deaths caused by traffic accidents etc.
It's absolutely clear that emergency personnel need some way of quickly removing stalled robotaxis when the company itself is unable -- losing communications being a perfect example.
But how to do it? It's not clear what the ideal solution is, or how to prevent the ability from being used by criminals to seize control of vehicles for nefarious purposes.
Are these proposals that have tried to work out all the kinks? What's the status of this?
Its novel, because before there still was a driver you could appeal to. In occasions where the car was abandoned locked in the middle of the street, I bet the solve time would also be big enough to upset a lot of people, but such things just dont regulary happen.
But now there are suddenly plenty of cars that all of sudden block a lane (or street) and you cant talk to noone, honking makes no sense, and when police arrives they dont k ow what to do.
It’s not hard, use a tow truck. But getting a tow truck to every incident (especially in the bay area, not know for functional towing services) is the hard part.
Since there’s no driver present and the controls are presumably locked, you cant push the car to the side of the road which is what would normally happen with a stalled car.
I think it's pretty simple: fire trucks are really big. Use the fire truck to push the disabled car out of the way. This will cause damage to the car, but who cares? That's the car owner's problem. It's just like when firefighters smash the windows of illegally parked cars and put fire hoses through them.
There's videos of fire trucks doing exactly this (one even to a police car that was blocking their way to a building on fire!).
Fire trucks are made for stuff like that.
Legally parked cars are the problem for whoever was blocking the road, not the fire truck. When there's an emergency, fire trucks have wide latitude to do whatever is necessary to get to the scene. They can't cause other crashes, but causing property damage for illegally-parked vehicles is fine.
I'm sure it is, depending on just how long this re-route takes. I should hope that it's obvious that if they can simply steer around an illegally parked car without any issue, they should just do this, for many reasons. But if the double/triple-parked cars are completely blocking the road, they can plow through, and they have every right to.
A single double-parked car in my city can result in the street being completely inaccessible for thru-traffic. I've also seen them break out the spotters and back up into their turning lane.
I assume they're also reticent to implement the nuclear option, because even a cursory google search didn't return any incidents of this happening anytime recently (though obviously that doesn't mean it doesn't).
It did, however, return a few results about fire trucks being in (non head-on) collisions on city streets and several firefighters being injured on-board.
When a fire truck pushes another vehicle out of the way, it doesn't ram into it at full speed; that would be dumb. It slowly makes contact with the vehicle, then pushes. A fire truck easily has enough power to push a parked car out of the way. Of course, this doesn't happen often.
Buy a forklift, paint it scarlet red, put a revolving light on top. Pick up stalled-out robocar, dump it on its roof on the sidewalk. Issue a $500 removal-fee ticket.
If a car cannot drive without having Internet connectivity, it should not be on the road in the first place. If I don't have Google Maps it can be difficult for me to get to my destination, but I'll never stop in the middle of a road because my phone died. At least I'll pull over at a proper place. Self driving cars should do the same.
The same question was asked by the Luddites when their jobs were taken away by automated weaving looms. I doubt the answer is much different this time (yes, it does that, but in the long term that is somehow a good thing)
