Slashdot Mirror
Uber Stops Self-Driving Car Pilot In San Francisco After The DMV Steps In (engadget.com)
93 Escort Wagon writes: San Francisco bicyclists can breathe a sigh of relief now that Uber has suspended testing of its autonomous fleet in the city. The company announced the decision after the California Department of Motor Vehicles suspended the registration of the vehicles involved in the testing. Uber remains "100 percent committed to California and will be redoubling our efforts to develop workable statewide rules," the company said. A spokesperson for Uber told Recode, "We are open to having the conversation about applying for a permit, but Uber does not have plans to do so."
Fuck Uber
As per this article, the DMV tried to work with Uber to get permits for the vehicles.
The DMV told Uber that if it had obtained a permit, the regulator would have given the green light to the self-driving pilot. DMV director Jean Shiomoto said in a letter sent to Uber that she would "personally help to ensure an expedited review and approval process", which she said could take less than three days.
So it's not like the city was trying to stop Uber.
Another 20 companies exploring self-driving cars — including Alphabet's Google, Tesla Motors and Ford Motor Co — have obtained California DMV permits for 130 cars.
It seems its not difficult for other companies either.
Its just that Uber doesn't want to play by the rules, even when the rules are being bent for their benefit. The Recode article in the fine summary said it would be $150 per vehicle so it's not like they were trying to kill them with fees either.
Either that, or Uber knows it's autonomous car program is not ready for prime time.
Calling someone a "hater" only means you can not rationally rebut their argument.
That first video where their car drove through a hard-red light, and with a pedestrian in the crosswalk was scary.
GM will be building self-driving cars as well, i don't know if they will be needing uber to drive them
I wonder what kind of serious problems Uber has with their self-driving cars that they suspend the testing without fuss over something as trivial as breaking the law. This must be one of the very first times.
But really fuck them for this. Taking down a taxi cartel by breaking laws designed to prop up a monopoly is one thing.
Not abiding by a permit with dangerous test vehicles on the road is quite the other.
Half expecting a snarky, "We don't accept the DMV's cancellation of our registration, and will continue to test these cars while we seek resolution with them".
The "San Francisco bicyclists can breathe a sigh of relief" comment surprised me until I saw that Uber has a problem with turning right across bike lanes. This certainly isn't a problem for all self-driving cars. In the South Bay I've seen a self-driving car do exactly the right thing: signal, merge into the lane when it turns from solid to dashed, stop at the red light, and then turn. That's a lot better than the average human at the same intersection; seeing someone signal and merge and stop would be quite unusual.
Its just that Uber doesn't want to play by the rules, even when the rules are being bent for their benefit.
I get the impression that is something of a cultural thing with Uber at this point. They're not even trying to get along. They are just acting the part of a bully and trying to do whatever they want. I think their corporate mantra is "better to beg forgiveness than ask permission" but they don't even bother with the forgiveness bit.
Either that, or Uber knows it's autonomous car program is not ready for prime time.
I think this is probably close to the mark. Uber getting involved in autonomous cars makes absolutely zero business sense. Their entire business model is based on an asset-light utilization of vehicles owned by the people that drive for them. They are basically a middle-man matching and scheduling taxi service. Actually owning the vehicles would cost a ton of money for no obvious benefit to Uber. Uber owning the vehicles (autonomous or not) would A) undermine their (absurd) argument that they aren't a taxi service and B) require a HUGE investment in assets and the attendant insurance, maintenance, upkeep, registration, and tracking. Uber has no infrastructure to do this nor do they have the capital to buy such a fleet.
Uber getting into autonomous vehicles smacks of marketing more than engineering. They want to project an image and seem hip but this is a project well outside their wheelhouse and they don't curently have the profits to support projects like this like Google or Apple or even Amazon does. I think they are trying to steal a page from the Netflix playbook and be ready for what they think will be the next industry evolution but it would make a LOT more sense for them to let someone else figure out the autonomous vehicle thing and just buy them when the time comes. They are a me-too entry into what is already a crowded field and they have no particular advantage in making autonomous cars and some very obvious disadvantages.
Nothing in the CA law says self-driving cars need permits. This sounds like a gross violation of the 5th and 18th amendments, as well as the Food and Beverage act and the Indian Lands Kerosene Act of 1842 (IANAL).
The CA DMV has this page listing the requirements for testing self driving cars. They are not onerous. Speculate as you will why Uber doesn't want to comply. It could be anything from having to report incidents on a per yearly basis to not allowing commercial operation (eg not picking up actual customers) or even just a general "fuck you, we're Uber".
Application Requirements for Autonomous Vehicle Tester Program
I am Slashdot. Are you Slashdot as well?
Uber claimed they weren't testing self driving cars, they were testing driver assist cars. Saying they don't need a permit. If they applied for the permit ($180) then its and admission its a self driving car and thus could be regulated in future as such.
Uber are little shits, now they have an in with Trump, you can expect them to track their customers for Trump, in exchange they'll get to ignore the laws like Trump co does.
Uber's system is about the same as Teslas (slightly better due to more sensors), but neither should be rolled out to customers at this point. If I can't drive without passing a driving test, why should my car be able to drive without passing a test?
"Self-Driving Car Pilot?" Man, and I thought my job was useless!
Uber will never be able to justify its valuation to investors.
They wouldn't be the first in that situation. I am inclined to agree but never is a very long time and I've seen more unlikely companies succeed. They have a lot of revenue which is actually the hardest part. Once you have that then profits are a possibility. Without revenues they are screwed. Whether the investors will make their money back kind of depends on how much they dumped in and what sort of margins Uber can make. Are they selling $2 bills for $1 or do they simply have to get to a certain scale and/or cost targets? Hard to say. In any case their profits will have to be rather substantial for the investors to not take a haircut.
They were aware of the "problem" and did not pull the fleet out of the streets immediately == unethical fucks.
In a traditional business sense, of course it makes no sense. But that doesn't apply here.
Sure it does. The laws of economics are not suspended for the benefit of Uber.
Uber is in a competitive industry with few barriers to entry, and very little profit, yet they have a market cap of $50B. How can they possibly justify that valuation to their investors?
Uber is not a publicly traded company so they do not have a market cap in the conventional sense. Their valuation is based on an extrapolation from the percentage of the company purchased by investors and the price of that investment. If you buy 10% of a company for an investment of $5B then the company is valued at $50B. That doesn't mean that it actually would be worth that amount on the open market and it certainly doesn't mean they are actually worth $50B. It means the most recent investor felt it was worth that but it is dangerous to extrapolate that too far.
You think there are few barriers to entry in their industry? I disagree. There are some pretty substantial network effects in play here. There are only so many drivers and cars to go around and they are going to tend to gravitate towards the company which is most likely to have the biggest user base. Sure, users can in theory switch easily but what good is switching to a taxi service that doesn't have any drivers? Scale will matter here which is why Uber is trying to grow as fast as possible. Uber has something like the same advantage eBay has. The buyers and sellers will tend to gravitate to the largest platform. Frankly the self driving car thing is nothing but a pointless distraction from building their network as far as I can tell.
From a publicity point of view, their defiance of the California DMV was pure genius.
Perhaps but I think you'd have a hard time showing a causal relationship between that and Uber's bottom line.
They need to be perceived as a gutsy company aggressively pursuing new tech.
No they do not. They WANT to be perceived that way (for reasons that aren't entirely clear) but wants and needs are different things.
If you or I drove down a California street without a valid driver's license, the penalty would be anything from a ticket (first time offense, no injuries to others) to an arrest (multiple offender, injuring people) and possible incarceration.
If the standard penalties were applied to whichever Uber executive authorized the law-breaking, then Uber would come into compliance rather quickly.
I don't agree with Uber's business model, but they have explicitly stated from day 1 that regulations don't apply to them, "this time it's different", we're disruptors, etc. A move like refusing to follow the DMV's process will be seen by the company's true fans as a big middle finger to those "corrupt Luddite government organizations." It's the equivalent of taking your ball and going home.
What will probably happen next is Uber will wallpaper smaller non-CA state/city officals' offices with money until they agree to let them operate. They'll turn around and say how progressive these cities are, and look at Luddite California...oh by the way we're moving to Texas/North Carolina/Georgia to spite them...
I'd be fine with Uber if they paid their drivers as employees, followed the local taxi licensing procedures like all the other cab companies do, and incurred the same expenses a traditional cab company did to get into business. They make money because they just declare the rules don't apply to them. I see why too - so many people I meet in the tech sector are classic Ayn Rand devotees who basically want the government dismantled anyway. Most techies are obsessed with automation and productivity and have no idea that basically every business in the world has to follow rules, pay corrupt officials, and has startup costs.
if they can pull this driverless thing off it will be better for so many people. Taxi drivers can retrain and get better jobs, and the clueless volunteers who drive for Uber can stop giving away the equity in their car for 90% return in fare payment - after expenses.
What better jobs? They've all been automated away.
San Francisco bicyclists can breathe a sigh of relief now
Yes, much better not to have cars that will automatically try to avoid hitting bicycles. People are much more reliable than machines... well except for the hundreds of thousands of times every year when people driven cars they hit people, other cars, and other stuff.
Sarcasm... yes. As a society we should be mandating adoption of this technology and improving it, not treating it like a safety threat to be slowed down. In the next 5 years there should be a prohibition against allowing any car on the market that doesn't have autonomous collision avoidance features. Once we have cars that won't allow drivers to hit stuff unless the driver specifically disables that feature (like choosing to disable an air bag or not buckling your seat belt) then fully autonomous is a natural progression.
If the Uber cars are really capable of avoiding cyclists, why does Uber not just say so? Also if their cars are so teachable why don't they just go through a couple proper turns with a pilot and resolve that issue forever more? If anything, regulation is highlighting some real glaring gaps in their implementation here and that's a good thing.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
I don't care if there's new technology in the car or not. The guy operating it is still a Driver not a Pilot.
Maybe there should be mandatory automated safety features on bikes as well. And pedestrians! A $40-per-shoe automated collision prevention device should be mandatory for any pedestrian who wants to cross a road.
If the Uber cars are really capable of avoiding cyclists, why does Uber not just say so? Also if their cars are so teachable why don't they just go through a couple proper turns with a pilot and resolve that issue forever more? If anything, regulation is highlighting some real glaring gaps in their implementation here and that's a good thing.
Uber is saying that these are not autonomous cars as defined under the California regulations. For the purposes of the regulations these are cars with driver assist technology being operated by licensed drivers. For regulatory purposes this should be regulated the same way as all the other car companies that are offering some sort of driver assist technology. It would be a mismatch if you make Uber go through some licensing/certification process intended for fully autonomous vehicles if they are not fully autonomous vehicles. Sure maybe Uber is trying to live in some gray area of regulation, like they like to do, but it also seems that the regulators are trying to apply some standards that may not be applicable.
As long as Uber is putting these cars on the road intending for them to one day be autonomous and in the name of training them to be autonomous, then they should be technically ready to be autonomous. If they are not technically ready to be autonomous then they need to go back to the drawing board and get better hardware.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Autonomous vehicles would be a whole new business model for Uber,
That might be fine if they had figured out a way to make their current business model profitable. Perhaps they will in due time but given the size of their losses so far that's not a certainty. It's rarely a good idea to try a second business model before you make the first one work.
Any other company trying to launch a robot driven taxi fleet without all of these pieces in-place will be at a significant disadvantage to Uber. Anybody who thinks that robot taxis are "the way of the future" would be wise to invest in or partner with Uber to make it happen.
I think that is far from a given. First off for Uber to have a sustainable competitive advantage they will have to find a path to profitability. Robot taxis aren't going to be a thing in even the most optimistic scenario for a number of years. If Uber can become the largest taxi service AND do so profitable AND not get sued into oblivion AND not run afoul of regulations then MAYBE they would be worth partnering with.
As long as Uber is putting these cars on the road intending for them to one day be autonomous and in the name of training them to be autonomous, then they should be technically ready to be autonomous. If they are not technically ready to be autonomous then they need to go back to the drawing board and get better hardware.
I've never heard a better description of a catch 22.
So, should we take every car off the road that has collision avoidance technology just because car manufacturers are collecting data with the intent of someday making the cars fully autonomous? This is technology meant to save lives. Seat belts and airbags also had their problems and failed to save lives under particular circumstances and with defects, but if we had used your criteria of provable perfection then a lot of people would be dead that didn't need to be.
Seat belts and airbags are not an apt comparison because they were always intended to be a cheap and basic solution. Autonomy is the exact opposite of that. It is hard to see if it can ever be adopted to the extent that will save a significant amount of lives. Therefore it shouldn't even enter into consideration at this point.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Having carnal relations with a corporation is just gross. Yes, I know the courts consider them "legal persons" but it's still unnatural. In a word, ewww.
And don't even get me started on having such relations with an automobile. The tailpipe is NOT to be abused in that manner. But if you try and you get your wee-wee burned off, well, tough.
Seat belts and airbags are not an apt comparison because they were always intended to be a cheap and basic solution. Autonomy is the exact opposite of that. It is hard to see if it can ever be adopted to the extent that will save a significant amount of lives. Therefore it shouldn't even enter into consideration at this point.
In what way is putting an explosive charge to inflate a bag in the steering wheel seem like a basic solution? That is some very precise engineering.
I don't see a few cameras, maybe a bit of lidar, with some coding to not hit stuff as too complicated. If you code it the right way then worst case it should stop when it sees something in its path. Cars are very complex systems already.
saying "actually owning the vehicles would cost a ton of money for no obvious benefit to Uber" is not true. Simple math shows they would make a killing. Even if you estimate that they average a measly $15/hr with their autonomous cars, working an average of 20 hours per day, 50 weeks/yr, that's $105,000 per year. I think that's what they call profitable.
I'm a certified cost accountant. Lot of problems with your analysis.
First off there are no autonomous vehicles to buy and we don't know when there will be. It is unlikely they will be cheap especially at first. Even once they become available there will be regulations and legal issues to sort out before they can become a viable business. It's going to be a while.
Second you did not consider costs at all. We don't know what the cost of an autonomous vehicle will be. You did not consider costs at all. You didn't consider the cost of fuel, the cost of insurance, the cost of engineering, the cost of administration, legal costs, the cost of maintenance, the cost of financing, the cost of regulations, and a host of other costs. Profits are revenue minus costs so until you properly consider costs your analysis is farcical. I can generate huge revenues by selling $2 bills for $1. I'll also be bankrupt in no time at all.
Third, demand will not be constant and average utilization is unlikely to be anywhere close to your rosy estimates. A substantial portion of the time the car is going to be driving unoccupied to pick up passengers. It's going to have down time to refuel. It's going to have down time for maintenance. It's going to break down now and then. It's going to get vandalized. There are going to be substantial periods where there are no fares to get. The list goes on and on. You are assuming an extremely naive best case scenario that is unlikely to exist in the real world.
$105K sounds like a lot but consider that even if we take your overly rosy assumptions about utilization and assume the car averages a modest 20mph during that time period it will cover something like 140,000 miles during that year. So you will have to replace the vehicle in just over a year most likely even assuming it performs exceptionally well. Few cars last much beyond 200K miles. We have no idea how reliable the navigation equipment on an autonomous car will prove to be. My guess is that there will be all sorts of problems with sensors going bad, getting covered with gunk and snow, etc. An autonomous vehicle will almost by definition be less reliable than an identical vehicle without the extra equipment.
Finally, I stand by my statement that there is no competitive advantage to Uber in owning their taxis. It adds a lot of cost and reduces their margins. What's to prevent GM from simply making and operating their own taxi service? Not like Uber is going to get self driving cars cheaper than GM will. Uber's advantage right now is that they don't have to buy any cars. There are only so many drivers and cars out there to utilize and Uber is the biggest fish in that pond. This gets them network effect advantages and keeps their costs (relatively) low. If they have to buy the cars then the network effect advantages of tying up all those cars as Uber drivers (rather than Lyft) goes away. Then they are just another taxi service with some fancy cars. It takes a huge amount of money to buy a gigantic fleet of cars which then has to compete against what they already had. It makes no sense.
Well they haven't made it work yet with "a bit of lidar", so there is no proof of concept for that. They can't even figure out whether to place a lidar sensor on the vehicle so that it sees everything from a low hanging bridge to a ground squirrel. On the contrary, they had a working prototype of both a seatbelt and the 'super complicated' air bag device before putting them into consumer vehicles.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Everything in the future is far from a given.
Not true at all. I can tell you with with a reasonably narrow set of error bars what the income and profitability of Coca-Cola Corporation will be for the next few years. Their business is very stable, predictable, and proven. I could probably make a pretty good guess what their balance sheet will look like 10 years from now.
Uber? No idea. Anyone who tells you they can tell you what Uber's profitability will look like is full of crap. Especially when we are talking about them building a business around a technology that doesn't even exist yet outside of the laboratory. Any investor in Uber is basically taking a huge chance on a very speculative venture. Nothing wrong with that but it's anything but predictable.
The question is: can Uber attract investment in their plan?
Just because they can attract investment does NOT mean the plan will work. Ask any venture capitalist an they will tell you that maybe 1 of ever 10 companies they invest in turns into a good business. One or two more maybe do better than break even. The rest go down in flames. Investment dollars do not in any way ensure success.
The other difference is that neither a seatbelt nor an airbag pose *any* risk of harming people *outside* the car. Autonomous vehicles pose quite a substantial risk of this type.
The Uber tests have self-driving cars but all with drivers at the wheel. California wanted to force Uber to get permits for all their cars under the concept that they're self-driving. Uber is saying that the car with a driver is not self-driving. CA is using onerous regulation to line their pockets. Meanwhile, Uber can still drive in Pittsburgh.
Only Uber could lose a PR fight with the DMV.
Uber is lying.
Under the CA regulations, any vehicle which has the capability of operating autonomously is considered an autonomous vehicle.
The fact that you have someone in the drivers seat ready to take over does not change that.
Even if you were doing the driving, and the computer was just recording and analysing your actions at the time does not change that.
-The fact that it has the capability of driving autonomously means it requires an autonomous vehicle testing permit.
So getting drivers for a new service is not a real barrier to entry, the only issue is convincing customers to request rides using it.
That's the case now because a dominant player hasn't emerged just yet. But let's say hypothetically that Uber successfully gains enough of an edge that they are tying up a quorum of available drivers at a given time. This means that customers looking for a ride are going to increasingly not going to bother trying the smaller taxi companies because they will be less likely to get a ride when they need it. This means that the drivers are going to increasingly be forced to work with Uber and they've got a cycle going that basically reduces other ride share firms to niche roles. It's kind of like with eBay. The buyers and sellers increasingly are driven to use the same platform whether they prefer it or not.
Now it might not go this way. It's possible that the taxi market will remain fragmented with multiple players kind of like it is today. But if that's the case then Uber probably won't remain a going concern as a business. They've kind of taken a shoot for the moon approach which if it works they will make a killing but if it doesn't they will lose a ton of money.
I don't think Uber is looking to make the capital investment into a fleet of self-driving cars but rather that they're looking at self-driving technology as a means to secure more drivers once self-driving vehicles become mainstream.
If that is the case then it makes even less sense. Why not just wait for someone to develop the technology (plenty of others working on it) and then just buy it when it is ready if the only purpose is to supplement their driver pool. Uber doesn't bring any special capabilities to the table here. The ONLY reason to do the research on self driving cars is if you plan to either produce products or license the technology to others. Is Uber planning to get into the car manufacturing business? Are they planning to supply technology to the car industry? None of this has ANYTHING to do with their current business model. The only thing that might make sense is if they are working on some sort of dispatch technology so that they can direct self driving cars to paying customers once the technology is ready. But compared to developing the self driving cars themselves that's a trivial endeavor.
If a driver has a self driving vehicle and has a choice between Uber, which allows him to send out his self-driving vehicle to earn money for him when he doesn't need to use it, or Lyft, which will only pay him when he's driving, I would believe the driver would select to use the Uber platform.
You are conflating two issues. Why Uber would need to be doing R&D on self driving vehicles is a completely separate issue from how self driving vehicles would be used within a taxi service. There is no obvious reason for Uber to be doing R&D on self driving vehicles. They have no special technical expertise in automotive R&D. Unless they are planning to develop some saleable technology out of the R&D then they are simply wasting their investors money on what amounts to a marketing boondoggle. Uber hasn't even figured out how to make their taxi business profitable and I'm supposed to believe they are a serious player in developing self driving cars? Makes zero sense.