> Advocates hope self-driving cars can help reduce U.S. road deaths, which rose 7.7 percent in 2015, the highest annual jump since 1966.
This was a news story in itself [1]. It was attributed to increased driving, encouraged by economic factors such as higher employment and lower gas prices.
Not exactly a steady decline, quote from the NPR article linked above:
In its own early estimates that were released late in 2015, NHTSA reported seeing a 3.5 percent rise in vehicle miles traveled in the U.S. But it also said the fatality rate — the number of deaths per 100 million vehicle miles traveled — was rising at a slightly higher pace of 4.4 percent.
> attributed to increased driving, encouraged by economic factors such as higher employment and lower gas prices
Interesting - I always assumed that the increase in road deaths was at least largely due to increased cellphone usage, resulting in distracted drivers.
Rather than speculate on anecdotes, perhaps we can focus on actual data[0]. 2014-2015 only saw a +280 change in fatalities caused by distractions (read: texting).
Florida and Georgia combined saw a +190 drunk driving fatalities in the same period.
UPS has 400,000 employees. Autonomous driving would pay itself off very quickly even if one of uber's cars personally murdered every single UPS and FedEx employee. Of course, the issue at play isn't joblessness, but the lack of a safety net for when people lose their jobs. Still, in the u.s., very few people actually starve to death, even when unemployed.
Perhaps lack of health care will kill them all. But was their health care that good to begin with? Perhaps an actual solution for root problems is the better route.
I make the argument that the majority of our interactions with law enforcement (for those of us who are otherwise generally well mannered) stem from some driving related instance. This is the true pathway for most of us to get tangled up in the [broken[ revolving door of the legal system. I am 100% all for self driving cars and look forward to owning one as soon as there are no controls for me to interact with at any time during travel. i.e. no steering wheel, no gas or brake pedals, nothing and therefore no reason to be pulled over. I want the car to be treated legally as an extension to my home; anything that I do / can do at home can also be done in the car. I want a cocktail? take it from the house into the car while it takes me to my destination.
That is to say, you want to limit your mobility to that which is directly and entirely under the control of legislated authority with no possibility of circumventing laws for one reason or another.
I have long been arguing among friends and family that in the future it should be ILLEGAL to manually operate your vehicle without a special license. That license, would require to obtain it, a much more vigorous examination and approval than we currently have at the DMV. So, for those determined to manually operate their vehicle, there would always be the means for them to do so. But it should be sufficiently difficult that the average joe doesn't pursue it, given the utility of the self driving car.
Exactly, "Stop and Frisk" has no equivalent for drivers because it's already understood/accepted that the police can pull you over at any moment for a variety of reasons.
Question: Does this undermine California's safety standards for self-driving cars? I was very pleased with how quickly California slapped down Uber. Will California still be able to do that?
For those who don't remember: Uber was clearly experimenting with unready software on the roads with insufficient safety processes in place. When caught on camera running a red light, Uber blamed the human operator and weasel-worded their pr statement to imply that the car was not in autonomous mode. It in fact was in autonomous mode, and the driver was in observation mode. It's unclear if the failure was due to observer attention failure or lack of training.
Federal overreach. I find it incredible that they want fully autonomous cars, but we do not yet have fully autonomous aircraft. This stuff isn't safe yet, and may not be for some time. If a state doesn't want to allow it on their roads, DC should not get to overrule them.
This entire thing sounds like they are trying to solve a problem that doesn't yet exist.
As you said, how can they effectively legislate something that is not even fully fleshed out yet? Is this really the time to centralize legislation at the federal level and block states?
The primary benefit of allowing states to do it instead is that you can have more liberal experimentation in places where it makes more sense, like less dense, deserty Nevada, with lots of freeways. While more densely populated highly urbanized parts of the country can be avoided until it's ready.
I can't believe they are selling this as 'speeding the deployment' because it would be 10x easier for tech companies to convince a state/local government than the federal one. And it seems they already convinced a few states already...
The only thing I can imagine this is "speeding up" is helping the major auto companies minimize their legal costs and allow them to focus on lobbying one level of government. As well as making it easier for these companies to guide the legislation towards their favour. This is how corporatism works in the US. The legislation is designed with big-co's in mind who have teams of lawyers and lobbyists.
"Automakers would have to submit safety assessment reports to regulators, but the bill would not require pre-market approval of advanced vehicle technologies. The measure now goes to the Senate, where a bipartisan group of lawmakers has been working on similar legislation."
You and I both know that without premarket approval, one or more companies are going to get careless and are going to kill people before the technology is ready. It's going to stain public acceptance for a quite necessary technology.
Here's a new medicine we're developing! It's gonna totally work great. The science says it's possible, we just need to work out the kinks while treating hundreds of thousands of people....
Even with government approval, there will be fatal accidents. This approach is better because it DIRECTLY holds automakers liable-- they can't claim immunity because some bureaucrat approved the vehicle. And it keeps politics and technical ignorance out of the development cycle.
Also, 3,000 people die every month on U.S. roads, your medicine analogy is wrong because there is much harm from the current system.
I dunno, a financially disinterested third party that appoints an expert panel of inspectors could stop a poor decision from being made. Most of the time, the internal processes within these companies will work correctly.
What can happen is if suddenly there's a race to market that throws process off the rails or a risk factor that was dismissed because of particulars of the culture that is producing the vehicles. That's where an outside inspector and premarket approval really helps.
Assume self-driving cars exist at some point. Assume they reduce fatalities by 90 percent. Ignore injuries. Now "only" 3,000 people are being killed by autonomous vehicles because "Hey, software isn't perfect and stuff happens." How many consumer products today kill that many people a year and the public is basically OK with it?
I'm not saying that the companies shouldn't make reasonable payments as a cost of doing business but if Ford manufacturing defects today were to kill 1,000 people a year they'd be sued out of existence.
There are plenty of medical procedures that save lots of lives but occasionally kill people. The public is ok with that because it is way better than the alternative.
Medicine generally is the closest analog. On the other hand when someone dies because of a rare drug side effect, lawsuits are common. There's a reason why the drug companies are protected from liability in the case of common vaccines [and that medical malpractice insurance is so expensive].
Yes but usually those medical procedures prove that they save more people than they kill before they're given to the public. Self-driving cars have not done that.
Would be great to see a similar national viewpoint on liability issues with SDCs. For example, who is liable (the car owner? manufacturer? algo writer?)
algo writer should not be liable, unless it's gross negligence or conscious cover up (like VW cases).
However, i can imagine a set of guideliness adopted into industry which involves slow rollouts of updates (potentially in a multivariate testing manner controlling for speed variance/distribution, response time, collision proximity (?) or collisions avoided, distance traveled, user comfort score, and so on. In other words, due diligence must be done, before rolling out new releases.
Your response makes sense, but that is usually not how the legal system works...i'm curious if all sorts of tangentially or abstractly related parties will be pulled into many lawsuits and then promptly dropped. No ultimate impact w/r/t/ judgements, but a lot of wasted time and legal defense costs.
For self driving taxis, the car owner, manufacturer and algo writer might all be the same company (e.g. Alphabet). And a company that huge can just self-insure, especially since they have more data on the risk level of their cars than anyone else.
If my self driving Subaru causes an accident, I think Subaru will pay for the damages. And Subaru doesn't need to involve an insurance company in that transaction.
Volvo have already made it clear that they will take responsibility for incidents caused by autonomous Volvos. I presume that this is at least partly a marketing tool because I would certainly prefer a clear commitment on the part of the manufacturer of any autonomous car that I might buy.
Don't worry, you'll sign a piece of paperwork that says you are solely responsible for the operation of the vehicle when you take ownership of it. You'll also sign a place where you also agree to mandatory arbitration.
I have no idea what Subaru's reaction would be, but in general you can certainly be held responsible if you allow a bug in self-driving software to cause an accident.
The manufacturer simply has to state that while the software is in control and you are sitting idle, you must continue to pay attention and re-take control if anything goes awry.
In the event of an accident, the manufacturer can then state that the accident would not have occurred if the driver had maintained control as the T&Cs demanded and wash their hands of responsibility in some PR friendly way.
Then it is not a self-driving car. There's definitely a point where cars will be nominally just assistive driving under some set of circumstances but people won't pay attention because they normally don't need to.
This is one of the things that will need to be worked out as the tech gets better over time. Liability is a big deal. If the car is billed as fully self-driving, the driver really can't be held responsible (both criminally and civilly) if their car runs over someone. (And it will.)
I really hope we just skip over that and go straight from 0 to completely self driving. It creates all sorts of messier legal tangles when it could be either the human or the car's fault, instead of making it entirely the car's fault.
Which insurance company, under whose policy? That's the question. It makes a big difference to the people involved, because it affects the policies they buy and how much they cost.
An insurance company would likely offer to cover all self driving accidents for free. This is because it would be a small cost to them, and would encourage you to use the self driving mode more often, which would cause you to get in less accidents.
Why WOULDN'T an insurance company cover all this stuff for free? It literally would save them billions.
That is not quite how it works in most US states. Insurance companies cover up to a maximum amount, which is usually enough to cover all damages. However, when the maximum amount is breached, the cost is then bourne by some party which "caused" the accident. Suppose the "driver" of the SDC isnt even driving -- who is that other party? The manufacturer? The algo writer?
As with everything in life it depends. Where did the error occur? Did the brakes fail? MFG of the car. Did the software fail to register the red light? Software MFG. I'm confident regardless of circumstances they can find who is at fault, and worst case scenario of multiple parties at fault - place an appropriate percentage of blame. You know... just like with ever other lawsuit dating back to the beginning of our justice system.
Or what if the insurance company just offers to pay for all of it for free?
It would save the insurance companies billions of dollars to do this, as it would encourage their drivers to stay in self driving mode which will be way safer.
I wonder if potential changes to minimum required liability amounts will lead to higher insurance premiums vs lower? When it's a person's actions responsible for the accident it seems easier to cap these at a lower standard amount vs. a corporation's IP's actions/fault with x-billion in assets...
Compensatory damage amounts are generally independent of the defendant's means. This is as expected. Your hospital bills and pain/suffering aren't any different if the person whose car ran you over is rich or poor.
Punitive damages are another story, but in theory they're not available in cases of ordinary negligence.
Hmm. The section on exemptions to safety standards was pretty interesting. I wonder what brought that on.
The federal pre-emptions to state regulations was expected. States cannot have any higher standards than the federal ones, except for the vehicles that the states purchase for their own needs.
how come pre-emptions are expected? i would expect states cannot have any LOWER standards than federal (think min wage, emissions, etc) and can mandate higher if they choose
It is the other way around - states may not have standards unless they implement the exact federal standard:
> No State or political subdivision of a State may maintain, enforce, prescribe, or continue in effect any law or regulation regarding the design, construction, or performance of highly automated vehicles, automated driving systems, or components of automated driving systems unless such law or regulation is identical to a standard prescribed under this chapter.
According to Wired, there are 21 states with some form of regulation of self-driving vehicles.
> Self-driving vehicles have been testing on public roads since 2010, when Google hit the streets near Mountain View, California. And in the absence of congressional oversight, states have stepped in to regulate them, creating a patchwork of at least 21 different state laws and guidelines with different purposes, definitions, and priorities. This is a serious pain for the growing self-driving industry, which aspires to build cars permitted on all public roads.
Because it would suck if you crossed into another state and all of a sudden your car acted differently to conform to that state's laws (turning off auto pilot, limiting speed, etc).
CA emissions have forced the automakers to offer basically two versions of every vehicle for the past half century or so. Nobody in the industry wants a repeat of that because it costs money.
In both cases the California standard was higher, improving the environment and saving lives.
What happened to the states being policy laboratories? I thought Republicans were for local control? Or is that only for school textbooks and baking wedding cakes.
>In both cases the California standard was higher, improving the environment and saving lives.
Not the lives of the people who would have been saved by safety features in a car they weren't driving because they couldn't afford it or because the implementation of a particular safety feature was a lower priority than the implementing a second variant of emissions equipment and calibration.
Stuff like this is fairly zero sum.
>What happened to the states being policy laboratories
The federal government could implement this as a FMVSS if they wanted.
Neither party has a monopoly on saying one thing and doing another.
Question: Is it expected that, as a result of this legislation, there will be rigorous behavioral and performance definitions and standards that self-driving cars, however that is defined, will have to meet?
IOW, similar to, or in addition to, the FMVSS, will manufacturers have a series of regulations that define the minimum requirements of objective performance criteria for cars sold as "self-driving"? If so, what would those criteria be? Will there be software source code audits to exhibit that the software wasn't written specifically for the testing (vis-a-vis VW)? Finally, will superior test results become marketing fodder for manufacturers in the way acceleration and cornering are now?
It seems like the legislation would delegate responsibility for coming up with such policies to the Secretary of transportation and gives them 24 months to come up with such rules.
‘‘§ 30129. Updated or new motor vehicle safety stand-
10 ards for highly automated vehicles
11 ‘‘(a) SAFETY ASSESSMENT CERTIFICATION.—
12 ‘‘(1) FINAL RULE.—Not later than 24 months
13 after the date of the enactment of this section, the
14 Secretary of Transportation shall issue a final rule
15 requiring the submission of safety assessment certifi-
16 cations regarding how safety is being addressed by
17 each entity developing a highly automated vehicle or
18 an automated driving system. Such rule shall in-
19 clude—
20 ‘‘(A) a specification of which entities are
21 required to submit such certifications;
22 ‘‘(B) a clear description of the relevant test
23 results, data, and other contents required to be
24 submitted by such entity, in order to dem-
25 onstrate that such entity’s vehicles are likely to
One of the very few things that I see as a massive benefit of the current Trump Whitehouse:
Now would be the time for moonshot projects!
I bet Trump would be willing to free a massive budgets if under his term there were a realistic chance on something like the moonlanding. Or something that would ensure him a positive entry in the history books. I truly hople Thiel or Musk use this rare opportunity.
And even if a budget were somehow allocated, much of it would be deferred into subsequent congressional terms. An 11- or 12-figure project that has to survive multiple presidential and congressional terms would be no easy feat.
I don't think he is interested much in the presidential terms coming after his 4 or 8 years. The most vivid example for that problem is the Iran hostage crisis which led arguably to the election loss of Carter (while he put massive effort into that issue) but only a few days after his loss and into Reagan's presidency, the hostages were released.
Not sure. It seems that most have other priorities. But its sometimes not about the budget but can be something like easing up certain laws and so on. Let's see if self driving cars are really normality until the end if his first term.
tl;dr Under pressure from big investors and a public clamoring for road bots, politicians as usual cater to the whims of the moment.
2 years later: After numerous accidental deaths, the outcry of the public, and the behest of large tech donors fighting growing competition, politicians cater to the whims of the moment and heavily increase regulation and scrutiny of self-driving cars.
To elaborate: "Current federal rules bar self-driving cars without human controls on U.S. roads. States have issued a variety of different rules in the absence of clear federal guidance, and automakers have complained that California’s rules are too restrictive."
25k is less than 1% of most major auto manufacturers yearly production numbers. No article I've read on this news has put that into context and instead praises the legislation as liberating car manufacturers to bring FSD cars to the masses. Most people don't read past the headline let alone look into the facts of the article. Also nothing good EVER gets passed legislation unanimously. Last time the House passed legislation this quickly and unanimously was when they passed SOAPA. Something feels fishy about this. I haven't read the legislation directly but I bet if a journalist went through it fully, somethings would surface that people wouldn't like.
Given that this is from a base of zero, and the cap is set to rise by 4x to 100k by 2020 (which is when Waymo plans to start introducing driverless cars), I do not think that this is at all a slow pace.
GM sells sells 10M cars per year so 100k would be 1% of that. Also who's Waymo? Tesla is planning to demonstrate a CA to NY trip with zero human intervention in just a few months and Elon has said they're shooting for 2019 for L5 FSD for their entire fleet made past October 2016 (they've already put in the hardware necessary for FSD so when the software is ready a simple OTA software update is all that's necessary for you're car to have full L5 autonomy)
> they've already put in the hardware necessary for FSD so when the software is ready a simple OTA software update is all that's necessary for you're car to have full L5 autonomy
That's bull. Every other company working on fully autonomous driving is using Lidar for obstacle detection, but Tesla's cars have none (only radar and cameras). I see no reason to think their software is any better than anyone else's (in fact I would say Waymo almost certainly ahead of them there).
Why should I have any confidence that they will be able to get better results than anybody else using cheaper hardware?
Waymo can collect full video and sensor data from all the miles its cars drive. Tesla can collect some high-level logs. It's just not the same thing at all.
If, by L5 you mean, I can get in my car, at a random, drivable over legal roads location, and instruct the vehicle to take me to another (within reasonable distance) location that is drivable over legal roads - I.E. Google Maps has a route. Basically, no steering wheel required.
Benedict Evans, at A16Z, says that some of the smartest people in the automated driving field say that L5 autonomy is 15-20 years out at the earliest.
Let's just say that Elon Musk has been known to be a little "enthusiastic" when setting goals and objectives for his team.
Was being sarcastic. Googles been working on their self driving cars for nearly a decade and they've only collected a few million miles of data whereas Tesla has BILLIONS of miles of data and they've only started working on their tech a couple years ago. Tesla's tech is way ahead of Google's. Waymo is overrated
Miles of data is not the best metric. If you've ever tried to build a machine learning system you'd recognize some of the issues here. First off, where are the labels? How are you to interpret all that data?
Say Tesla has recorded video of some dude driving down his street in LA and their system detects a car door opening in his data stream. How can Tesla find out whether or not there actually was a car door opening? Human labelers (for billions of miles of data)? Multiple instruments?
If I were designing a system, I'd want a car with extra instruments on it that would help me validate what the primary instruments detected. The idea of just recording a bunch of live traffic seems pretty lame. (You have no way to validate any of the conclusions you draw from the data and hence no way to improve your ability to draw conclusions from the data.)
Did you see the article about Waymos test environment at Castle? That, along with a boatload of simulation, is exactly the kind of testing you need to do to develop this kind of a system.
Has Tesla ever demo'ed their simulation environment or other tests?
Not sure why you think simulation environment is more valuable than real world data. Waymo doesn't even have as much data in simulated environment as Tesla does in real world data, which is hilarious.
Waymo has claimed they do 8 million miles a day, and it's in simulation that they've made the lion's share of their progress the past few years. In simulation you can model thousands of variations of any particular corner case, the AI doesn't know the difference between real and virtual environments.
Dedicated test vehicles are collecting many terabytes a day. Google mentioned years ago that they're gathering over a gig per second of data from their test vehicles, and it's likely more than that now. Tesla is gathering a gig or 2 of data at a time from it's hw2 equipped cars here and there. I'm optimistic that Tesla can, by the end of the year, get to about where Google was at in 2011.
In a simulation environment you know ground truth. (Where objects actually were vs. what you saw.) You can play the same scenario from slightly different angles. Or different conditions (rain, snow, low light, bright light). You can look at counterfactuals (what would I have seen if I had done X). You can test the behavior of your system without actually setting a computer loose in the real world.
From "real world" recorded data you have none of these things.
This is the difference between observational and experimental data and it's a huge difference.
Agreed with everything you are saying -but my understanding is that the Waymo simulation system doesn't simulate Input, it simulates the objects. The whole idea being that responding to objects is a separate problem from perceiving the in the first place.
Well, I think that's because Waymo hasn't shipped a self-driving car to the general public yet. My best guess is that they want to achieve level 4 before they make the tech available on a larger scale
This account has been violating the guidelines with a lot of unsubstantive comments. Please slow down, review the guidelines, and post more thoughtfully.
On the one side - i find this quite extraordinary- a government ready to push new technologys, instead of inhibiting it- on the other (slightly paranoid) side- what is in it for the senators?
Huge portions of traffic automated, because they really love new tech? Nah.
They could have forwarded hyperloops and public traffic ages ago.
Do they get a bonus point for the next vote? Certainly not, a large part of the population earns its "good" livelihood with what this law is about to destroy.
So one wonders- can you push and incentivize the trading of still green bananas to the market- and the watch the populous lynch the producers/cripple with regulation, after the expected large failure happens?
You would be seen as tech loving - and you could prevent social turmoil + reap large lobbyist gift basket.
Of course that would be very cynical and presume foresight on the part of the politician. Still.
Similarly, the anti net neutrality congresspeople can be found simply by examining the amount of campaign funds raised from the fiber/cable/wireless providers.
The folks who voted against allowing importation of equivalent drugs into the USA from Canada - just have to look at pharma contributions to congresspeople.
The idea is to take a new technology, that is bound to create horrible accidents, and push it faster forward then is its own pace - thus creating indirectly a surge of horrible accidents in the publics eyes, that would make the public rise against said technology.
A sort of - over-evaluation to kill a company- strategy.
If driving from one state to another, especially for business purposes, doesn't count as interstate commerce I don't know what does.
See the United States Constitution (Article I, Section 8, Clause 3): The United States Congress shall have power "to regulate Commerce with foreign Nations, and among the several States, and with the Indian Tribes."
Exactly. For airplanes, the FAA is supreme and has consistently smacked down individual state/city initiatives to regulate air traffic (and rightly so, even as I'm a small federal government/states' rights advocate in general).
This was a news story in itself [1]. It was attributed to increased driving, encouraged by economic factors such as higher employment and lower gas prices.
[1] http://www.npr.org/sections/thetwo-way/2016/02/18/467230965/...