> The companies say driverless vehicles are safer than human-driven ones when it comes to passenger safety. They maintain none of the incidents cited by the fire and police departments have resulted in passenger injury.
Passenger safety is the same myopic view that allows car companies to build vehicles that are increasingly dangerous-for-pedestrians. It seems blindingly obvious to me that vehicular safety should include its effects on non-passengers.
What a disingenuous way to frame the issue. Driverless car companies should be bending over backwards to work with public safety departments, not this “move fast and break things” shit.
> Passenger safety is the same myopic view that allows car companies to build vehicles that are increasingly dangerous-for-pedestrians.
Car companies do this by making bigger cars, so passenger safety and pedestrian safety are, to some degree, a zero sum game. Autonomous car safety doesn't seem to be the same.
> It seems blindingly obvious to me that vehicular safety should include its effects on non-passengers.
Seems fair.
> What a disingenuous way to frame the issue. Driverless car companies should be bending over backwards to work with public safety departments, not this “move fast and break things” shit.
This is only a disingenuous framing if you assume that increased passenger safety is coming at the expense of pedestrian safety, which seems like a stretch. It's just as easy to assume that, in the absence of a zero-sum element, increased passenger safety implies increased pedestrian safety. Is there any data that suggest otherwise?
> Driverless car companies should be bending over backwards to work with public safety departments, not this “move fast and break things” shit.
Postponing safer travel because of tiny edge cases might feel like the moral, cautious path, but it may well end up costing many, many lives.
> This is only a disingenuous framing if you assume that increased passenger safety is coming at the expense of pedestrian safety,
drewbug01's point was "vehicular safety should include its effects on non-passengers". Pedestrian safety is only one of many such effects, so your re-framing to consider only {passenger + pedestrian} safety seems to miss the point by quite a bit.
We know pedestrian safety isn't the only other issue because we can read the list of incidents involving autonomous vehicles from the San Francisco Fire Department.
If a car stops on the hose from a fire truck, so it can't be used to put out a fire, the issue isn't one of pedestrian safety. If it drives over the hose and causes thousands of dollars of damage, it still isn't an issue of pedestrian safety.
If a car gets in the way of a fire truck with lights flashing, responding to an alarm, than that's not an issue of pedestrian safety. Nor is it an issue of pedestrian safety if it blocks a fire truck from being able to return to its station.
> Is there any data that suggest otherwise?
There is enough data to know the relevant issues are more than just passenger and pedestrian safety, yes. This article points to it.
> Car companies do this by making bigger cars, so passenger safety and pedestrian safety are, to some degree, a zero sum game.
Nit: passenger safety and pedestrian safety aren't at odds with each other because in the event of a collision, no pedestrian is heavy enough to pose a risk to driver safety even in tiny 50s-era cars. What's actually happening is that there's a Red Queen's race between car manufacturers: if your car is heavier than your competitors' then your customers will be safer, so every generation of cars gets bigger than the last.
A building fire is only a "tiny edge case" when the firefighters can start fighting it in a timely manner, before it rapidly grows into a much bigger edge case.
Exactly! They just focus on the one thing that they see as important and screw everything else.
One solution would be to grant free reign without liability for fire & police take any measures they see fit at the moment to solve the problem. If it trashes the driverless vehicle, too bad (and oh yes, the cleanup cost is billed to the driverless car company).
After a few 100K write-offs, maybe the companies will pay attention and actually look at the problem.
I assure you that there are entire teams of engineers at both of these companies burning the midnight oil to address these problems already. They're definitely not being ignored.
> One solution would be to grant free reign without liability for fire & police take any measures they see fit at the moment to solve the problem.
In my part of the US, they absolutely have this. If you park your car in front of a fire hydrant and firefighters need to hook a hose up to it, they can and will do whatever is needed to make that happen, regardless of damage to the offending vehicle.
Yes, this (very proper) treatment of parked vehicles is exactly what I was considering.
However, since we have not already heard of an incident like a fire truck forcefully pushing out of the way an errant driverless car, it seems that it also needs to be extended to such non-parked/non-diretly-human-controlled cars?
I wonder how much emergency responders have negatively impacted the economy over the years by insisting on safety protocols that are above and beyond what are required. Two examples I can think of are the opposition to battery storage in homes in California, and the expansion of code requiring fire sprinklers in most new construction. These were done to improve emergency conditions, but may have slowed battery storage installation and new construction.
The fastest way to speed up new construction is to drop all safety protocols whatsoever—that's how you get rapid build up in developing countries, but that's also how you lose thousands of people to an earthquake or to a citywide fire. One of the characteristics of a developed nation is that we place a higher priority on not losing the development we already have than we do on developing further. Stability over growth. The rewards are lower, but so is the risk of a disaster.
There are valid tradeoffs, though. If you have a little more fire safety but as a result a significant increase in housing costs that pushes some people at the margins into homelessness, that might not be worthwhile.
Is the requirement to have two staircases in apartment buildings worth all of the extra lost space? Maybe, but also maybe not.
Ultimately we should be optimizing our policies for human well being, not safety. If we lock everybody in padded rooms, they'll be much safer than if we allow them to go out and do dangerous things like get in cars, but that's obviously a bad tradeoff.
Impounding is what they would do for regular joe's car. So why not for these "full" self driving ones?
Is someone simply telling them to be "lenient?"
First responders already deal with a lot of bullshit. Especially in San Francisco and Lo Angeles right now. Adding more to the least is peachy...
> Surely there are already existing laws for first responders and if the robo-taxis aren't following them they can just be ? for violating them.
The concern is probably that the consequences designed as an incentive for human drivers to be attentive to the rules (and which are not, otherwise, a compensation for the harms of the rules not being followed) are not a solution if robotaxis are irreducibly unable to follow the rules.
> Straight up impound the car and they'll start not doing this.
Seize, not impound. Your driverless car interferes with emergency services, it is to be given up and sold off to pay for the delays your driverless car caused. Impound implies they should be given a chance to pay to get it back -- I say no. One of the few applications of civil asset forfeiture I would be okay with.
Is it hard to put a do not stop zone in front of firehouses in their maps?
I mean they’re operating a handful of cities, there’s not that many firehouses. like it could even be manually programmed into the map as a second order rule that kicks in after stopping that activates a remote driver and then is a harsh penalty in the training functions.
The issue isn't just the immediate area in front of firehouses or mapping them (the data is publicly available). One issue is routing quality degradation when you close off smaller streets, particularly for the lower speed cruise vehicles that can't use higher speed thoroughfares. Permanently closing 50 low speed streets would probably have a significant impact on their already questionable routing. A second issue is that there have also been incidents near active emergency scenes, which are not publicly available data.
Why isn't anyone discussing the obvious solution to the emergency responder issue?
Driverless cars should be accessible to be navigated by emergency services. Instances of this usage should be publicly documented.
If there's no driver to make human decisions, and the device is incapable of adapting to emergent situations, this seems like the easiest and most-logical way to handle it.
My Roomba robot vacuum came with little beacon towers that mark keep-away zones. Maybe a solution is for first responder vehicles to have something like that they can turn on where self-driving cars do everything they can within safety limits to stay away from those.
And should pedestrians carry those as well? I guess I'd better strap one to my kids' ankles just to be safe. Oh, and my dog. And...
A self-driving car that can't react appropriately to the world as it is is not a self-driving car, it's a fantasy. The world cannot adapt to the needs of current self-driving car tech, there are just too many moving parts that would need to change. It's up to the self-driving cars to adapt to the world as it is, and if they can't do that then they're not ready.
When I got my first Roomba I quickly realized I had traded the chore of vacuuming for the chore of robot maintenance (our apartment had old, long carpet that would frequently wrap around brushes, wheels, etc. that never happened with a normal vacuum). It eventually got to the point where maintaining Roomba was more of a hassle than just vacuuming. I spent hours a month covered in dust trying to clean and debug Roomba failures.
When I realized how stupid this was, we got rid of the Roomba and used the vacuum again.
Several houses later, several generations of Roomba later, and hard floors, Roomba mostly works well. I infrequently need to tear it down to cut out hair and other stuck debris.
We should not expect first responders to adapt to and accept shortcomings of autonomous vehicles. If they endanger lives in novel ways that were never an issue before, they need to be deployed in areas where they won’t encounter those issues and/or be updated to recognize existing signals (tap, lights, traffic signal preemption devices, people telling them to move or go a different way, etc.).
If I can’t opt out of having a driverless car on my street I want to be absolutely positive it poses no higher risk to my family and neighbors than to other drivers without needing to remember to perform robot maintenance first.
Yeah that what I want to see when I'm in a crash, responders wasting time with tech support on some tech device.
And let me be super straightforward here: if a radio technician needs to be on emergency vehicle at all times, it better be that his salary comes out of self driving licenses fees and not my taxes.
Emergency responders use all sorts of radios already. They pretty much pioneered the ubiquitous use of radio in all sorts of situations. If there is one thing that first responders know, besides health and safety, it's radio. Licensed amateur radio operators tout their ability to respond in emergency situations as well, and whether that is true or not, many emergency responders are also Hams.
If you have not looked at a fire engine or ambulance lately, you would be a little surprised at how much technology they bristle with. The last time I called 9-1-1, they showed up with an EKG, and all the other machines to monitor vital signs. In fact, there are services now that advertise they will bring a fully-functional Emergency Department to the comfort of your own home.
Yeah, who wants to mess with ancillary tech that isn't part of the emergency. Nobody, but a transponder is fairly dumb. You press a button on it and it starts. Transponders don't have a mic, they don't have a frequency dial, they are just gonna send out your ID code until you shut it down again. It will be the least of worries for the first responders, especially if they will be safe from errant SDCs.
Emergency vehicles already have traffic signal preemption devices on them. Is that not recognized autonomous vehicles?
The devices you listed were chosen by emergency services to make them more efficient at their primary purpose. Beacons have nothing to do with improving their ability to do their job and are just an additional burden.
The problem discussed here isn’t these vehicles failing to see people and running them over. These aren’t Teslas. The issue that sometimes when driving the rules of the road break down. A police office stopped to direct traffic or some sticks used to designate that people should merge left. These are situations a human can handle, even if they might get flustered and make more mistakes, but a robot will find very tough. Instead you might see them trying to move slowly by deliberately through a crime scene ignoring the gestures made to try to make them stop
Great. My elderly parents will need a way to move around. I think robo-taxis will likely become the best way for them once they can't drive anymore.
US cities are not walkable. Bicycles aren't safe. Ubers are expensive and interacting with the driver is not something I want my parents to do/understand. Public transportation is pitiful.
The only hope for my elderly parents are either their kids driving them around or robo-taxis that hopefully will become cheaper than Ubers when scaled.
I don't know if "great" is the right word since robotaxis not only don't fix the problems you list, they make them unlikely to ever get fixed.
It would be better for your elderly parents (and everyone else) if instead of putting our energy into adapting to robotaxis, we made cities more walkable and invested in public transportation.
>I don't know if "great" is the right word since robotaxis not only don't fix the problems you list, they make them unlikely to ever get fixed.
Not worried about that. I need to worry about how my elderly parents can get around now/very soon and not how the US can re-engineer its cities to fix the problems I listed.
I don't see why we can't have both? Neither Waymo/Cruise nor the CPUC have any control over the urban design process. The CPUC does have some level of control over BART and other transit agencies, but they're still not the right agency to be dictating service levels. Investing in building better cities is the government's job. Sabotaging commercial services so that things get bad enough that the government is forced to act is cutting off your nose to spite your face.
Because you still need roads for robotaxis. Urban centers should be pedestrian only zones augmented with light rail. We need to stop looking at any form of car as a solution, they only exacerbate the problem since they require roads.
Even Dutch cities like Amsterdam and Utrecht don't ban taxis and other commercial vehicles from the city center because they serve a useful purpose in a tourist city, so I'm not sure why SF would have to. There's a huge amount of possibility-space between "everything is designed exclusively for cars" and "replace all roads with light rail".
"The incidents include running through yellow emergency tape, blocking firehouse driveways and refusing to move for first responders."
I don't really think they need to be any clearer if these are the types of issues first responders are facing. The first time a driverless car ran through emergency tape, that company should have lost all rights to continue running tests.
Yeah, I think that emergency vehicles and responders should receive subsidized equipment that will interface with SDCs on a technological level.
I don't think it's a good idea to map out every incident; that would reduce OpSec for the responders and make them more vulnerable, not less. But it seems that short-range transponders are a good idea. Put one in every fire truck, ambulance and police vehicle. Portable beacons could be set up to establish a perimeter.
Unfortuntately, SDCs are not able to recognize conventional symbols that first responders use. First responders have been caught without corresponding tech, and this is negligence on the part of the developers and manufacturers of SDCs.
If they expect to deploy SDCs everywhere, then start upgrading the tech infrastructure everywhere to accomodate them, because their inferior sensors and poor judgement is what's causing them to fail in this natural world of human-oriented signs and symbols.
"The cars can be updated to make way for emergency vehicles too."
It's an interesting problem. Here's a special case where the system can ignore traffic laws and expect other cars to ignore traffic laws too. It should also be sure it's a legitimate emergency vehicle, not a mall security car, or just a bad actor with flashing lights.
It's not the job of emergency services to spend time cataloging their specific complaints and propose solutions so these companies can operate. These companies get in the way of emergency services and so they complain. It's now the company's job to identify and fix the problem if they want to keep operating.
I think the request and reasoning is pretty clear in the article. Autonomous vehicles can't be given instructions by emergency personnel. They can't follow lawful orders given by a police officer – the kind of instructions that in critical situations can take the form of "do an action or you are under arrest."
It should be pointed out that it is a crime to knowingly impede emergency services.
I would say that the request is that these vehicles not be put onto streets without having some mechanism to prevent the examples in the article from happening, or maybe not approve their opration at all.
At the risk of sounding like I don't want technological progress, it should really be pointed out that robotic vehicles aren't necessary in any way. There isn't a massive problem looking to be solved by them, they're just a capitalist optimization undertaken by rent-seeking mega-corps like GM and Google. Hiring a human driver for passenger vehicles is already one of the cheapest services you can imagine, these corporations just want to take the minimum wage salary of an Uber driver for themselves.
With that in mind I think it's a fair ask of these corporations to do more to make their autonomous vehicles able to handle emergency situations where a human driver would be expected to take the instruction of a police officer or firefighter.
Passenger safety is the same myopic view that allows car companies to build vehicles that are increasingly dangerous-for-pedestrians. It seems blindingly obvious to me that vehicular safety should include its effects on non-passengers.
What a disingenuous way to frame the issue. Driverless car companies should be bending over backwards to work with public safety departments, not this “move fast and break things” shit.