Ubiquitous comment, as a person within the industry: there's a big difference between cars that have a hands-free driving mode, and cars that do not have a driver. The latter will be an absolute game-changer; the former can yield some marginal gains in safety and congestion, but does not really shatter any paradigms.
This is the former. The latter will certainly happen eventually, but is orders of magnitude more challenging and a number of years away.
We don't need to jump to completely driverless right away, there is a convenient intermediate step: The passenger can be the passive driver, and the car would drive to the pickup location empty. As soon as regulation allows, that step already gets you the major benefit of not having to "pay for the other dude in the car".
> the car would drive to the pickup location empty.
It would seem full driverless with nobody in the car is the big deal here. It doesn't really matter is someone will be behind the wheel later, you already had the thing driving around with no human to intervene in case it does something crazy. Don't see how that's an intermediary step.
A car driving around with no human inside poses less risk to human life than a car driving around with a human inside who is unable to control the car.
That's a carefully parsed statement -- less risk. True.
But I don't think that we're cool with the idea that because the number of humans threatened by an accident is reduced by somewhere between 50 and 20%, it's okay for flagrantly unsafe cars to be on the road.
I'm arguing for the case where the human as NO information.
Let's say, there's a bicycle/motorbike somewhat behind your cars.
For a human driver, it can be seen in only one of the 3 rear mirrors at any point in time. Unless it's slightly behind-left or behind-right (e.g. to take over) in which case it's in the dead angle and can't be seen at all.
By turning his head and using the mirrors, a human cannot see most of the 360° around a car. There are obstructed angles, not to mention that the human can only look toward one direction at once.
The self driving cars with 360° vision can see things that a human couldn't. It has a chance to react where the human didn't.
Sensor quality shouldn't be used to justify driverless cars being safer than a regular person driving a car. It's useless unless the car can actually use it to make a better decision than a human.
I once proposed that for airport rental cars - you'd call for a rental car, and it would self-drive, slowly, from the rental yard to the terminal. Then the renter takes over driving. At return, the driver gets out at the terminal, pushes the "done with rental" button, and the car drives itself, slowly, back to the rental yard.
Major destinations with big parking lots could be on the system. Convention centers, sports stadiums, major hotels. Again, the driver gets out at a convenient unloading point and the car drives and parks itself.
All autonomous driving would be around 15-20mph. At that speed, slamming on the brakes is an adequate solution to almost all problems.
But airport security was too touchy an issue a decade ago. What if one of the self-driving cars had a bomb in it? Might be worth another look now.
Put some sensors inside the car, like in the backseat and the trunk. If they sense something is in the car when you press the "Done with rental" button, it will signal you and will not move. Solved.
Driving without anyone in the car is still completely driverless. That human-free car will still need to navigate through the environment with hitting anyone or anything with zero possibility of intervention.
You'd need to verify that users have a valid driver's license. The problem is, how do you verify that the rider's identity is the same as the user who ordered the ride? Facial recognition?
No you wouldn't. Car sharing clubs like Zipcar deal with the same issue - they hire out cars on a completely unattended basis. If a customer chooses to pick up one of their cars and drive drunk or lend it to their cousin with a suspended license, the customer is liable.
Good point, don't know why I didn't think of Zipcar when I used a similar service last year.
Looks like the bigger problem might be people suing over accidents where they were distracted and forced to make a split second decision, when they 'reasonably' expected the car to be able to handle the conditions. (Not trying to shift the goal posts, just sharing a separate concern).
How do you do this verification when someone rents a car?
I've rented a car before where the only thing they verified is that I presented a credit card with the same name as the one on my driver's license, which they never asked to see because I had already put in the number online. I guess it's possible they were able to look up my photo from DMV records and were matching my face to it, but I'm guessing not.
I'm not a lawyer but I assume any funny business there would be charged to the person using another person's account, and the person who owns the account. That's how it works with regular cars if you use someone else's ID.
If you have full autonomy to get to the passenger, then you don't need them once you've picked them up.
Never mind the fact that now you have to worry about drunks, making sure insurance is cleared and people have perfect records (and know how to drive), want to drive at all (I don't -- isn't that what I'm paying for?), etc.
Besides the car driving around without a passenger on its way to pick them up, what if the passenger does not wish to pay attention? Part of what I enjoy when I take an Uber is the ability to make comments like this on HN.
If I am engaged with my phone, I am not supervising/shadowing the car.
"As a person within the industry" has there been any talk about the paths driverless will take? New roadways? Attendant-less roadways? Change in existing designation of highways? How to accomodate 10-20 people (vehicles won't look like they do now)?
I envision future photos-videos of driver and driver less "vehicles" at a busy intersection just like horses and early cars.
These are the right questions to be asking! Yes, there has been some talk -- but not remotely enough. At present, the vast majority of work in the vehicle-automation space is being done by, well, vehicle people. The assumption is that driverless vehicles will slot into the existing infrastructure with minimal modification. There's a fair amount of consideration being given to vehicle-to-infrastructure communications (so that vehicles can notify traffic signals that they are approaching, for example), but relatively little consideration being given to how the roadways will actually _physically_ evolve.
Personally, I expect that this evolution will be roughly as significant as what was catalysed by the horseless carriage.
One big challenge there is that infrastructure is expensive; replacing it moreso. To use an analogy, it's a lot cheaper to replace trains in a rail fleet than to replace track (the per-unit cost might be lower, but the logistical cost of downing a track network to switch it out can be significant, especially in last-mile areas or bottlenecks).
Horseless carriages drive on roads that were minimally modified from what the horses used (in some parts of Pittsburgh, for example, they never swapped out the Belgian block).
Is that an assumption, or a requirement? It seems to me that autonomous vehicles are based on fitting into the existing infrastructure because that's what they'll have to actually deal with. Any technology which requires new infrastructure hits a chicken and egg problem, since nobody will want to spend on infrastructure without any cars to use it, and nobody will want to buy cars that use it if the infrastructure isn't there.
Seems like it would basically be a two-step process. First, you need autonomous vehicles which work on roads as we have them now. Once they're common, they can start adding features which can take advantage of more sophisticated infrastructure when it comes. Sort of like how early cars had to work on paths built for horse carriages, but once they were common we could start building things like freeways.
Government has to get involved. Sometime and somewhere. Public safety is one aspect
I suggested to one town councilman that our town introduces a bill (ordinance) banning driver less vehicles. Just like we can ban oversized vehicles and parking in the winter. Our government should ban then now while it is not yet happening.
He thought it would be a good idea not to ban them but to capture a licensing fee.
Hopefully this will get introduced at the next meeting
Not supposed to down/up vote to show dis/agreement, but since it is already in play have a vote for tempering new-tech zeal with real world pragmatism.
PS: I don't necessarily agree w/ bans or taxes, just don't like the up/down culture of 'popularity'. The sooner the texters & PMGo players are relegated to child seats, the better for all drivers/passengers/riders, IMO.
Take the Las Vegas Strip for example. They have a lot of pedestrian bridges for people to avoid crossing streets. Granted, this also works to funnel folks into casinos. However, it would be difficult for a pedestrian to get into the street at some intersections because of fences and barricades. I have to think that most cities will evolve similarly when self-driving cars become commonplace
Everybody wants to jump straight to 4, because 3 is actually dangerous and likely to cause deaths as people are lulled into a false sense of safety. The question is how long will it take? Some people say 2 years, some people say 20.
How much has the recent interest and investment in self-driving cars impacted this estimate? 5 years ago, I would have said that level 4 automation is 20 years away. With all these billions pouring in, I'd cut that projection in less than half.
I'm certain that I won't be driving a car ever in 2030
I think it actually doesn't make that much difference. The problem, I believe, is that people are used to rapid advances in software. Lots of small iterations can take you a long way, but sometimes you just reach a hurdle which requires a breakthrough. A fully self-driving car would need a huge breakthrough in perception (which most people categorize under AI) in order to work.
Computer vision actually started out in the 60's because how hard could it be to classify objects in images? Well, decades later and it's still a hugely unsolved problem.
Robotics can be broken into 3 main categories: perception, planning, and controls. Perception is always the weak point. Robots just cannot see and understand their environments like people can. When you put a robot in a dynamic environment, they fail spectacularly. To give some idea of this, look at videos of robots falling down at the DARPA Robotics Challenge - and keep in mind those were made by teams of top roboticists with millions of dollars in a highly constrained environment (it's hard to say that obstacle course was "real").
I really, really doubt it's going to happen in a few years. I would love to be wrong though.
Last time I checked Google's self driving car was doing pretty well already without major breakthroughs. I suppose there still are problems when the weather conditions are less than ideal, but you can't compare it to /r/shittyrobots.
The car's performance on a relatively small stretch of road in California is hardly indicative of its potential performance on millions of miles of varied roadways, terrain and climate that make up the rest of the world.
well, hasn't google driven around actual city streets for many millions of miles? I could be wrong, but driving a car doesn't actually require breakthroughs in perception beyond current technology, The issue is that it just isn't economical for individual car ownership at this stage, but it could very well make economic sense for a fleet of taxis. Note, those taxis don't have to have the ability to go just anywhere, they could have predefined routes
>>sometimes you just reach a hurdle which requires a breakthrough.
When there are billions of $'s to be made. Generally there is way.
I understand from where you are coming. But if the trend grows, and one companies sees the billions coming. Others are forced to innovate. Also a whole new ecosystem is born.
See pre iPhone and post iPhone mobile industry. Sometimes you need one company to do something insane to drive others down the same path.
And interestingly, seeing challenges ahead, I don't think we will have 100% autonomous cars before 2050, if not later. 50 years ago, we were certain that perfect image recognition is just few months of work away, or at least enough money poured into it will solve the problem. It's 2016 and our best of the best image recognition algorithms can't even tell a zebra and a sofa in a zebra print apart. Since automatic cars are trying to solve the same problem I can't believe that in 2030 anyone will make 100% autonomous car, not a chance. Super advanced cruise control, like Tesla's autopilot, but better - sure, absolutely. But a car without a steering wheel? nope.
> It's 2016 and our best of the best image recognition algorithms can't even tell a zebra and a sofa in a zebra print apart.
State of the art offline CNNs have been doing classification quite well recently, arguably better than humans. See the Inception-v3 paper: http://arxiv.org/pdf/1512.00567v3.pdf
None of that changes the fact that they're easy to fool and have some pathological cases (that gambiting alludes to) that most humans consider quite dumb.
People like you are the best because it really adds to the enjoyment of great achievements when there are a healthy amount of naysayers to point out. Thank you, naysayer.
How many billions have been poured into speech recognition? I don't anyone here would trust controlling a car with Siri or Google Voice lest it misinterpret something you said.
Have you used Google Voice in the past ~year? I have a standard american accent so I might be the best case scenario for it but I haven't seen it miss a word I've said anytime I can remember in a year or two. I mostly use it in the car to reply and send texts, works even with the radio and ventilation on. I've used an Amazon Echo and that is right 99% of the time too.
Now, interpreting the words I say is a different matter, but as far as transcription I'd say we are at basically 100% right now and have been for a while.
> Now, interpreting the words I say is a different matter, but as far as transcription I'd say we are at basically 100% right now and have been for a while.
What? Sometimes I use Google's speech to text and a majority of the time it's a waste of my time. "Food near me" is not "fool's errand" (the irony isn't lost on me) and "gyms nearby" is not "Jim Dougherty", both are search terms I used last night.
shrug I don't have these issues at all. I just tried "food near me" and it worked perfectly fine. Like I said I might be the best case scenario since I speak in a basically standard american broadcast english accent.
I feel like that is more a product of the fact that the effort put into most other languages is greatly lower at this point relative to English, rather than English is particularly easy or other languages are particularly hard. It will all come in time.
Not effort - but data. A large amount of the dictation learning was done through users using and correcting the software. If you have 40,000,000 English users but only 80,000 Polish users you can immediately see why the software may not be up to par.
Google Voice to Text, from my experiences, is bloody amazing for Chinese. My Chinese is abysmal (think "never gets the tone right") and for short phrases/sentences I've never had Google fail on me. Maybe it fails on larger sentences but it would be difficult for me to believe.
Japanese is another language it does very well in. I can only count on one hand how many times it has made a mistake - and usually repeating myself a tad slower and with less slurring and it will pick it up just fine.
Well, my language (Danish) is pretty small and I would say Google voice recognition is notably worse than English. Sometimes it's way of pronouncing names is a bit weird but it is recognized when I said them.
So it might not work for everyone in the world but at some point soon it will properly be there for most national languages.
Same view here, current solutions are nothing more than Sunday drivers to me. They work well in good to near perfect conditions only where we actually need the least amount of self driving.
Wheel less driving, first show me it drives in the pouring rain at night when the roads aren't lit. Show it having the ability to obey posted speed limits, adjust to construction, and the like.
I am still surprised at how much current systems love to tail gate let alone don't suggest obeying posted limits.
Give me a car which self-drives for 4 months of the year, and requires me to drive it the other 8 months. That's still a massive advantage over the car I currently own.
Indeed but the context here are all the futuristic predictions starting with mass-unemployment of taxi and truck drivers (indeed that's what Uber is selling their market valuation on). If drivers are still needed 60% of the time, they're going nowhere.
The hype right now is so great that even if the reality is amazing it will be disappointing.
The disruptive effects still happen even if we don't get to 100%.
For example, a train of self driving trucks can be led by 1 in the front. Yes, I know, REAL trains exist, but they don't go everywhere. A train of self driving trucks could save a lot of money.
If self driving Taxis don't work in the rain, then Uber can just NOT drive them when it is raining! For example, Maybe the penetration rate will be very low in Seattle, but very high in Texas. Disrupting 20% of the Taxi market by eliminating drivers is still a very big deal.
Uber's main selling point has been availability when there are no taxis (hence surge pricing). If you can't get an Uber once it starts raining, someone else will eat their lunch.
And for trucks, will you let supermarkets go dry once it starts raining? Otherwise you need to keep drivers on retainer. And then you might as well just have them drive for minimum wage.
There would still be human driven taxis. The whole point of uber is that it is an on demand platform.
When it is sunny, the self driving cars are out and about, and the price is low.
When it is raining, the self driving cars stop taking passengers, and the price goes up, with surge pricing, which is handled by human drivers.
You don't keep drivers on retainer. You simply publish the price, and if the drivers like the price, then they can accept the job. If not, then they don't. If there is a shortage, then you use surge pricing until the shortage no longer exists at the given price point.
For trucks and supermarkets, if it is raining then you pay human drivers. If not, then self driving. BAM, labor costs have gone down by 50%.
>Indeed but the context here are all the futuristic predictions starting with mass-unemployment of taxi and truck drivers
Uber doesn't have any interest in self driving cars succeeding. It's a PR distraction to justify the exploitation of drivers ("you shouldn't complain because you are lucky to have a job at all when self driving cars are coming any day now").
Self driving cars that don't have drivers would completely destroy Uber's actual business model which is based on undercutting current taxi companies by shifting the capital, maintenance and insurance costs onto the drivers.
> If drivers are still needed 60% of the time, they're going nowhere.
Think about what you're saying here. If manual driving is still required 60% of the time there will still be mass-unemployment of taxi and truck drivers. These guys are not paid well enough that they can afford half the amount of income that they currently have.
Most drivers earn money based on the amount of time they spend transporting people/things, correct? I don't believe there are many drivers who get a great base salary regardless of work output. Supply and demand would prevent people who tried to use driving as their primary income and the task would be taken care of by people willing to do it on the side when the demand is there
Interesting. A car that could "drone" in some contexts. Leave your parking lot and wait closer for instance. No need for high speed, complex collision avoidance. Only 10mph private neighborhood mobility. That could be a nice selling point. Something like the Tesla thing, just better.
NHTSA Level 3 ("Limited Self-Driving: Vehicles at this level of automation enable the driver to cede full control of all safety-critical functions under certain traffic or environmental conditions and in those conditions to rely heavily on the vehicle to monitor for changes in those conditions requiring transition back to driver control. The driver is expected to be available for occasional control, but with sufficiently comfortable transition time. The Google car is an example of limited self-driving automation.") is OK. That means that the self-driving system can recognize situations it can't handle (snow, potholes, crowds in streets) and come to a safe stop if the driver doesn't take over.
It's Level 2 ("adaptive cruise control in combination with lane centering") that's the problem. Therein lies the deadly valley - good enough that the driver can tune out, but not good enough to handle trouble. That's where Tesla is. Level 2 with strongly enforced hands-on-wheel seems to be OK, but don't let the driver take their hands off for long periods.
Both 2 years and 20 years are good estimates, depending on whether you're talking about a robocab that ferries people from their suburban homes to the train station when the weather is nice, or a vehicle that drive empty across the country in a snowstorm.
I think the bigger question is how you get straight to 4. Ford has resources, but I feel like they're betting against history here w/ regards to ML successes.
Practical, reliable ML stems from enormous data sets. And I can't see a way of obtaining the necessary data volume other than sitting in Level 3 for a product cycle...
What you need to understand the auto industry plans their product lines 20 years in advance. Design/testing of components takes a non-trivial amount of time.
When they say, A future model will be wheelless. What they're saying is a car in 15-25 years will be wheelless. Which is about on track with the parent poster's time frame.
Most the large Semi-Truck manufacturers seem to be progressing along a similar track. Ford's forecasts of 20 years forward actual sees a complete stop of consumer vehicles moving completely over to driver-less fleet vehicles.
>Ubiquitous comment, as a person within the industry: there's a big difference between cars that have a hands-free driving mode, and cars that do not have a driver.
Surely the pay is far less for the passive driver? If not I assume it will be soon.
First, Uber drivers are already not making much. Drivers I've talked to here in SF say they are near $15/hour (though they can more in the form of bonuses if they hit certain hours-driven targets).
Second, the job of a passive driver will actually be both more boring and harder -- you have to pay attention while doing nothing, but be ready to take over in a second. It's like being a passenger, except you can't distract yourself in anyway. I'd much rather just drive.
Your example of San Francisco is an outlier - there's too much upward pressure on wages there to let driving compensation fall too low.
Let's have the same conversation about Minneapolis, Dallas, St. Paul, Austin, Kansas City, etc. Austin is a great example of a city with a lot of unemployement and underemployment due to the massive student population. Can't you imagine a college student riding around listening to lectures on the radio (assuming they're actually trying to do their job) for $8/hr? I can. And I can see the next student saying "I'll do it for $7.50".
I don't think the federal minimum wage applies to contractors sitting around waiting for the next job. The revenue while the clock is running is way above minimum wage.
Solution: pay drivers a bonus for the number of 'accidents avoided', or the amount of times they need to grab the wheel after the car sounds the 'take-the-wheel' ping. Drivers will be incentivized to pick-up passengers in more complex areas, which then improves the data and the car learns how to drive better in complex area.
I can't imagine why. If you were going to devote hours to being behind the wheel of an Uber, would you will willing to do it for "far less" than the already far-from-princely sums that an Uber driver makes?
I literally had an Uber driver this morning who didn't speak enough English to take the route I requested. She knew "Hello", and "Have a nice day." No harm done if she's a competent driver.
Would I devote hours being behind the wheel? No. I would try to find a job UNIXing for food because that's what I know. But we're not talking about me. We're talking about folks who have a driver's license, a clean driving record and can pass a background check. If they don't have to supply the vehicle the financial barrier to entry is much lower.
So yes I honestly think there will be downward pressure on wages for what is essentially a passive job. Sure you have to pay attention to prevent accidents, but that's it. Time to get out dragon dictate and start verbally writing the next Great American Novel.
If you're paying attention to prevent collisions then it's not a passive job. And if I got into a car to find the 'driver' dividing their attention between the road and their novel then I would get right out again.
It doesn't require a great deal of attention, either, and it's still mostly passive. Unless you're suggesting we should outlaw conversations or listening to the radio we're past the idea that people need their full undivided attention to drive. The question now is just what's the minimum.
I've seen people die in the ED because they treated driving as a passive activity. I find your hyperbole about outlawing conversations unhelpful. I very much look forward to computers having better driving skills than I had when I used to work in the ambulance service.
First, I note that you've got from "far less" to "downward pressure." Are you modifying your claim? I think that there is a considerably more colorable argument that it might shave 5 to 10% off the wages of a driver than that it might shave 50% off the wages of the driver.
While shifting the costs of vehicle purchase and maintenance from the driver to Uber is a factor, it's not enough to make driver wages "far less," and Uber has already been offering various assistances to drivers towards vehicle ownership, so nothing very dramatic is happening there -- and in any case, none of that savings will make it to the passenger, who ultimately is paying for the vehicle.
Making "far less" money than a typical Uber driver means around minimum-wage job or below. Let me suggest that there is no large pool of people who are interested in a roughly minimum-wage job with no security or prospect for advancement, who can plausibly save you from dying in a car crash.
(PS: There is no way that Uber would allow its drivers to dictate a novel while a passenger is in the car).
Sadly there is no RemindMe for ycombinator, but yes I believe "far less" is still my conclusion here. I didn't quantify far less and I'm not sure exactly how much it means? Thirty percent less? Fifty percent less? Not sure but yes I expect it'll be more than a 5 to 10% haircut you cited. And more importantly - profits will be far less if Uber adopts a taxi-like model of supplying the hardware and charging drivers a use fee.
I don't see why. Driving is considered a base-level skill in many of the places where Uber operates. It's one of those things you just assume everyone employable can do, like reading or writing. Uber drivers are mostly being paid for their time and for their vehicle, not for their driving skills.
This is not the former, this a step in the R&D process that is working toward the latter (and can act as PR in the meantime).
I expect that "supervised by humans in the driver's seat" here means "the car is fully self-driving, but we're still doing development and testing and we don't trust it enough yet". You can test these things on private roads, but eventually you need to get them out into the real world and collect data, figure out where the failure points are, etc.
But for sure they're building a fully-autonomous no-driver-needed car, and it's not ready yet so someone has to be there to take over when it fails.
I guess what I mean is, this is not a finished product that's a regular car with a "hands-free" mode; it's a prototype for a fully-autonomous driver-less car that's still under active development.
I.e. the reason that this exists is not to have some mostly-autonomous driver-assistance in place right now; that's useless to Uber. Rather it's part of the process of testing and improving the completely driver-less cars that they're working on.
I'm also skeptical that we will reach the point where cars will not need a driver at all in the near future. I might be wrong but it seems to me you need something close to AGI to have a fully autonomous car.
And with current approaches to self driving cars you need lots of data, sensors, fast and reliable internet connection, accurate maps and so forth, all of which are not at the level needed to have autonomy on all roads. So even if the algorithms would be good enough you'll need much more than that to get to self driving cars.
On the other hand a human only needs one working eye and a way to steer the car to be able do drive pretty much anywhere.
Agree no paradigms shattered with a driver in the seat from an economic POV. Im not sure safety gains have to be marginal. This thing, scaled over millions of miles, paid uber driver in the seat with a system that chimes to make driver take control could be far safer than distracted uber driver alone dealing with everything driving a taxi and communicating with customers entails.
I wonder how Uber drivers feel about working for a company that is actively trying to make them obsolete.
Not passing judgement or anything, just kind of a strange thing to contemplate..
Uber kept saying they were disrupting the taxi industry by letting more people participate when really in the end they're trying to remove the people from the equation altogether.
In the same way McD and other places are replacing attendants with tablets. Or, in a similar vein, how Netflix replaced sending DVDs through the mail with the Internet
It's a job, not a career
And I suspect there's some 3 years at least before self-driving is "production ready"
Uber wants to be the leader in the field because it's a threat to their business model but I don't think your average Uber driver has much to worry about for a long time.
Most Uber drivers i've talked to about this, and i basically always talk about this when using Uber, all see themselves as just doing this for a short while, 2-3 years, so they don't see themselves being affected by this.
But that's a short view, right? As Uber moves into more and more places, that means more and more drivers that will eventually be replaced. You're absolutely correct that it will not affect people driving for Uber right now, but it certainly will affect those who start driving in three years[0].
[0] This, of course, assumes that driver-less cars will be ready for prime time in three years.
Most all of us work for companies that would like to make us obsolete. Payroll is a necessary evil in the enterprise, as far as "the business" is concerned.
The irony here is that I would say most companies are, in one way or another, driven by consumption. Our present economic system, generally speaking, relies income earned in jobs to fuel this consumption.
If everyone is obsoleted, who will have the income to consume what the robots and fancy technology produce? I personally don't see the current business / economic system being very well prepared for the possibility of a post-consumer economy.
I always talk to the drivers about their feelings and so on when taking Uber. Maybe my data is skewed but most accept it as inevitable and are not worried much as they have other side jobs. The cab drivers are the ones who must be anxious. It will be really bad if they cannot be retrained for other jobs as it will add to the existing frustration around automation. We simply cannot afford to leave people out as we move on to other things. Outsourcing moved manufacturing jobs to China, our focus on climate and global warming will leading to moving out of coal leaving coal miners behind, Amazon is automating their warehouses, Self driving will leave cab drivers behind...So, how do we create new type of jobs for these people? Humanity cannot progress by leaving people behind.
> The cab drivers are the ones who must be anxious. It will be really bad if they cannot be retrained for other jobs as it will add to the existing frustration around automation.
I used my "training" - a very expensive CS degree - to drive a cab. I developed a nerve condition in college, so really I just suffered through the program without becoming competent as a programmer. Also, I dance to the beat of my own drum, and do not take orders well. My kuro5hin.org (RIP) story "Humanity's Second-best Hope" [1] is about the seasonal job I had at Amazon, just before I started taxi driving.
Cab driving offered me freedom to be my own boss. The cab company's rules were reasonable, and easy enough to comply with. As long as I paid my lease, I could go anywhere and do anything I wanted with my time. No one in the company's administration cared if I took a short break, or a long break, or spent the night in bed.
Something more important came up, and ridesharing ruined the economics of the business (Arizona makes it easy and cheap for anyone to start their own transportation company. The upstarts couldn't be bothered with playing fair), so I don't drive around in a taxi anymore.
I think these money-losing "ride sharing" companies were founded by people with delusions of grandeur and venture capitalists' money to burn. When they run out of their investors' money, the realities of transporting people from place to place will reassert themselves.
I guess we should still be using horse and buggies then. How dare we leave buggy drivers behind.
Change is inevitable. And for many people, painful. If the automation revolution occurs too fast, it will be a pretty economically painful transition period for a large chunk of humanity. On the other hand, if the transition is really really fast, the painful period may not last too long. Most things will become so cheap and abundant that most people won't have to work, or we will transition society to a 10-hour work week or something.
> Humanity cannot progress by leaving people behind.
You're way overestimating the value our civilization places on an individual human life, doubly so if they're unexceptional and male.
And that's even more true for less developed societies. The sanctity and preciousness of human life and desire to avoid suffering and achieve equal outcomes is a relatively new concept, unique to the developed nations of post WWII era.
You could stash a helpful little Asimo in the trunk. It would be like a trunkmonkey, except without the tire iron ;) But that could also be a far, dystopian-future option: agile I-Robot style bots that leaps out of the trunk to beat down anyone who tries to jack you when you stop to charge your battery.
I have a mental image of a slow, but patient Asimo trying to winch a wheelless Jeep back onto a dirt road, in a soaking wet tropical jungle, like that scene in Jurassic Park. Meanwhile, the occupants get eaten. Asimo has no event handler for that situation, and continues working patiently to get the car back on the road. He does not smell interesting to the raptors, who leave after their meal.
I worked in an IT dept as it was migrating it's staff to India. I bailed as soon as I could, and half-assed it while I was there, because that's the American Way
I saw one of these a month ago when I was biking around the North Shore (downtown Pittsburgh). Its cameras on top were far more conspicuous than the renders in the article, though.
This article said it would pick people up downtown, but how far would they go to drop off? The golden triangle is mostly grid-based, but go elsewhere in the city and the road topology gets jacked up fast (like Boston), especially in the south and west. Its idiosyncrasies took some getting used to, after living in flatter regions with gridded streets for several years (central Ohio, Salt Lake). In those places, if you want to go somewhere, you start driving in its general direction; in Pittsburgh, it's not uncommon to go in the opposite direction for a while, since the roads don't connect that way.
The article also mentioned bridges being a problem. In Pittsburgh, that's a big problem, since it supposedly has more bridges than any other city in the world.
Knowing Pittsburgh, the amount of road and bridge construction that goes on there would also concern me, not to mention the condition of the roads.
There's also a ton of roads that are extremely steep (multiple residential streets with an over 30% grade) which can be extremely dangerous in the winter.
It's an interesting choice of a city for this for sure.
I'm sure they chose the city for CMU but I wonder if the driving challenges played a roll. Navigating downtown Pittsburgh is no fun if you aren't familiar with it. I mean it's not really fun if you are familiar either.
The roads are getting better. It seems like half the roads are, will be, or have been closed at some point just this year and you can't drive more than 30 minutes without encountering a detour or being taken on one. My 10 minute commute to work is mostly streets that have had ongoing nightly or weekend work for almost a month.
In most places, when you merge onto a highway you have a ramp of decent length with a yield sign. You are able to get up to speed and merge safely.
In this particular spot, the end of your on-ramp has a stop sign. So, you're sitting at a dead stop with 65+ mph traffic 10 ft in front of you. Things get even more complicated since you're merging into an exit ramp. You have approximately 1/4 mile of road to merge into the fast moving traffic, then immediately get over a lane otherwise you're on the exit ramp!
In Houston, on I-10, there are some on ramps with stop lights to control the amount of traffic entering but they are not always used and are turned off when freeway traffic is traveling at speeds. Additionally you still have some space to accelerate which it looks like isn't available here.
It's worth keeping in mind for the HN crowd that "communicating with other autonomous vehicles" will require a huge leap in secure networking including government control of the car's software, basically.
Not just topology, but visibility too... IMHO at least, Pittsburgh streets in the denser neighborhoods (east end mainly) have a real problem with blind intersections and driveways, especially because of street parking. If the car's sensor system handles this well then that's really impressive.
Definitely not better road markings. I'm a Pittsburgh native, having spent most of my life there, and if anything the immaculate roads here in NorCal are a part of the ongoing skewed optimism regarding self-driving vehicles.
In Pittsburgh, you have potholes, ubiquitous construction (especially on PA-28), bridges (and I'm not talking Bay Bridge esque suspension bridges here), windy single lane dirt roads without markings, to say nothing of winter (snow usually obstructs what little markings there are 6-8 months out of the year), and a completely different driving style than in NorCal. Most people in Pittsburgh drive trucks or four-wheel drive Subarus (indeed, Pittsburgh has been a good niche market for Subaru -- they focus a lot of their advertising there). In Pittsburgh, roads only have two lanes max in each direction (and that's the highways). Even downtown has its fair share of dirt roads (near Allegheny River) and brick roads (roads that are made of bricks). I was shell shocked when I first moved to California and drove on the 101 with its 4 perfectly paved lanes filled with drivers exceeding 50 MPH.
Precipitation is a nightmare for self-driving cars' LADAR sensors, rendering them completely unusable. Drought-afflicted SV has zero precipitation, and Pittsburgh has almost as much as Seattle. Driving in snow is such a drastically different task/skill than driving in perfect weather that most people in California don't even know how to drive in Tahoe on their skiing trips (and the "snow" there is maintained by humans!).
If anything, us Pittsburgh-based autonomous vehicle researchers are the ones more grounded in reality. Elon hasn't even seen "the Turnpike" -- it is sad to say that the rather unfortunate fatal Tesla autopilot accident happened here in Pittsburgh.
As aside, Sebastian Thrun was in snow-laden Boston, and he and his lab also have a more realistic time estimate for driverless vehicles hitting the road than TechCrunch.
Anytime anyone in California complains about the roads (and believe it or not, people do complain about them here a lot), I just laugh.
Infamous Pgh driving scenario: "Left lanes need to exit right, right lanes need to exit left, here's 300 ft, make it happen"
Also, there are a number of short merge on-ramps that have Stop signs instead of Yield signs where you are required to come to a complete stop before merging onto high speed traffic.
It feels like a lot of what Uber does these days is just hyping up the stock price. Of course that is a very short term strategy, if your ever more audacious announcements keep falling flat. We saw a lot of this in the last throes of the dot-com bubble, before the inevitable implosion.
As for this announcement, does Uber have any credibility in this field, which combines the most individually complex AI, robotics and regulatory work that has ever been attempted? Hell, in any non-trivial engineering feats?
Also, Volvo has very active ongoing self-driving research going on. I have to wonder how much of this is Uber's supposed crack team of field experts, and how much is simply a partnership with Volvo delivering the technology.
They have acquired their way into credibility. They controversially hired the majority of the CMU robotics laboratory in order to start ATC. They also hired a number of Google self-driving engineers, including the former head of the self-driving effort. Today Uber announced their acquisition of Otto, a self-driving truck company.
The ATC is isolated from the rest of Uber.
Volvo has been working on self-driving cars for years; before the Uber ATC effort.
Regulatory work is one thing that Uber excels at.
To be fair, Google also hired/poached most of its self-driving experts from university robotics laboratories.
Lots of hills, and will have plenty of ice and snow in a few months. It seems to me like a logical place to test the resilience of a self-driving system against some predictable corner cases.
> The company has been creating extremely detailed maps that include not just roads and lane markings, but also buildings, potholes, parked cars, fire hydrants, traffic lights, trees, and anything else on Pittsburgh's streets.
If one city requires that much work, they're not scalable. Even if we assume the LIDAR on every Uber car collects the new data, the car will trip sooner or later if it can't detect potholes.
Concerning the bridges, can't they do like the Kinect: Match the image with depth-sensing lasers, so they can detect the boundaries of the bridge, stray away from pedestrians, filter out fire hydrants and trees but slow down when encountering other objects?
This is the way Google does it as well, and in my opinion is a fundamental architectural failure. There's a reason after several years Google still has tested in so few locations.
Delphi has a more robust system (in theory), but the accuracy isn't quite as good. To the best of my knowledge Ford is also taking a different approach which doesn't depend on super-detailed maps to operate. The technology to compare the surroundings against a map is far easier than figuring out the surroundings and then navigating within that (not to trivialize this - both are extraordinarily difficult and nobody has made a fully self-driving car yet).
But hey, like they always say in YC - do things that don't scale. Uber is clearly pushing ahead faster than anyone else, and they have a hell of a team behind this project.
Regarding bridges - to oversimplify the simple answer is they're far noisier for the sensors (all of them, including cameras) and it just wreaks havoc on the sensor fusion system.
You could drive down every road in the US for under 10 million dollars. So as long as you can keep them updated I don't think highly detailed maps are a problem. Especially if this is supposed to be a backup for GPS.
The maps Google uses for their self-driving cars, as I understand it, require constant maintenance. Also just the amount of data is huge. If this weren't the case, surely they would have driven in many other places by now?
I would actually love for someone who works at Google to chime in here. It'd be great to find out their system is more robust than I think it is.
I think they could do it every week, compared to the value of completely self driving cars. Especially if they get say UPS to add sensors to their trucks. The post office does something similar every single day.
Every self-driving car on the road is automatically remapping everything. Any discrepancies, like a new pothole or a disappearing tree, can be reported to the mothership for resolution and pushing out to all the other cars in the area.
They are using it for navigation not lane following. You can supplement GPS with dead reckoning for tunnels etc, but high quality maps let's a car be aware of landmarks and then use them to figure out where in the world they are. EX: With enough precision there is plenty of information in the picture to figure out where it's located. http://socialforest.org/wp-content/uploads/2014/04/Forest-Gr...
There were 4 million miles of road in the USA in 2013. Sure you could drive them - but drive them in a special sensor-equipped vehicle? In a fleet of such vehicles? In a reasonable time? Its a big job, and $2.50 a mile seems pretty low.
2.50$ is more than double what trucking costs per mile that should cover every lane mile at normal speeds. Now building 100 cars with this equipment is not going to be free, but trucks also burn far more fuel. Anyway, you doing this is hard, for a multi billion dollar per quarter company it's pocket change.
Add in hiring, benefits, vehicle transport and repair, sanitizing the data, hiring managers to manage the folks that do all that... any large project is unexpectedly expensive.
How did you get to this number? I'm familiar with a few "map the world" efforts, and this seems an order of magnitude too small. I'd love to hear about a breakthrough.
"Fundamental architectural failure" is a very bold claim to make given that absolutely no other approaches have worked as well as Google's for urban driving.
There's a serious problem at the heart of "throw a bunch of data at it" strategy - driving has an extremely long tail that's relatively fat. You can hit the "average case" relatively quickly (you can assume that Google's done that for the SoCal environment), but being able to extend that dataset to other, less likely scenarios, like "heavy downpour in SF" or "freak snowstorm" or "sudden construction due to water main break" is really hard.
None of what you said has anything to do with my point that the pre-mapping and then LIDAR-localization approach taken by Google is the only approach shown to work.
You're making a point about end-to-end trained ML systems; if anything, that's an indictment of Tesla's approach, not Google's more traditional sensor-based-approach.
* That we know of. Tesla et al. have not publicized urban driving videos/examples, so I'm assuming they're not capable of it as well as Google, right now.
That article has 0 information about Delphi's approach. In fact, the Delphi car has LIDAR too. For all we know they could be taking Google/Uber's approach.
I used to work on this - the approach is very different. That's a pretty hollow statement on the internet though, so take it as you will.
An interesting aside - it seems like most of Google's team left this year. Chris Urmson left, Anthony Levandowski left, and many of their engineers are gone as well (many went to Otto). It'll be interesting to see what happens now that they've lost all of the original leadership (Sebastian Thrun left long ago).
At a very high level, instead of comparing everything against a map, we try to figure out what it is. For example, the way Google detects traffic lights is using geometry to infer where a traffic light will be. Their maps are so good that they know where the car is, and where the traffic light is, and just use math to figure out where in the image there should be a traffic light. Then they just look at the pixels and figure out which color is lit - red, yellow, green. The way we did it is constantly scan for traffic lights (you can do lots of sanity checks, for instance using GPS are we near an intersection?), which means figuring out whether or not a traffic light is in the image, and then once we detect one figure out what color is the light.
Google's approach is simpler but is highly dependent on the maps. The other approach is easier to scale, but is obviously harder and likely takes more work to get the same results. Something I think a lot of people gloss over is that these maps take up huge amounts of data - not something you can stream over a cell network and probably not even high speed WiFi (maybe you'd download a map for your trip the night before). And also, as I understand it, the maps require regular maintenance.
The real question is how will this scale? And honestly I don't know. My theory is the tech is so hard and the rewards are so great that infrastructure will change to make it easier. By installing smart traffic lights, having cars talk to each other, and adding various road marks specifically for self-driving cars you can drastically reduce the complexity of the problem.
It's possible that they don't rely on the minutiae of the maps, just use them for planning.
For example, "don't take 5th street, it's full of potholes" or "we can't drop off the passenger on this block, there's a fire hydrant – will have to drop them off on the next block, so plan accordingly".
On the other hand, if the cars won't drive properly without these maps, in which case I too would be extremely hesitant to trust them.
They definitely use the maps. In order to accurately localize the car, a detailed map can be used as part of the likelihood function for location to answer "how likely is it that the vehicle is at position (x,y) given the most recent measurements?" If you don't have an accurate map, it's very difficult to come up with a likelihood function that is accurate enough to keep the car in its own lane.
Keeping the car in its lane is overrated. Any human who drives in snow knows the rules. If there are other cars around, center yourself with respect to them. If there are not, lanes are irrelevant. Even many narrow side streets in snowless areas force you to improvise in a similar way to handle oncoming cars.
Keeping the car in its lane amounts to keeping it from hitting anything. In any case - lane markers or not - controlling the car on shared roadways requires localizing it in terms of its environment. A high-fidelity map of the environment improves the localization accuracy that can be achieved from noisy sensors. If the map is more accurate, you can tolerate more noise from the sensors.
Most residential streets in Pittsburgh, as in a great deal of the rest of the world, do not have lane markings, though many are two-way and have parked cars on one or both sides of the road. Pittsburgh provides an additional difficulty, that many of these roads are too narrow for what is asked of them.
One of the tricks of driving in an older city with on-street parking is weaving in and out of unused spots to make way for oncoming vehicles that have no means to move over themselves—or won't, for fear of clipping a side-view mirror, or because they can't bring themselves to care about how little space they've left for others. Sometimes you have to stop and wait for one car to make its way through a tight stretch before going through yourself, rather like pulling up to a one-lane bridge that has been generated at runtime.
The trouble is that the lanes have to be imagined with respect to the current conditions, bearing in mind what other drivers are likely to expect of you; knowledge of the road is not enough.
This is a social problem as well as a technical one, and I will be interested to see the solutions.
Lane markings are good if they're there. Often they get obliterated by construction, wear, weather, precipitation. Sometimes there are several sets, rewritten after construction and only partly erased. Humans can have a hard time following them.
> ...will trip sooner or later if it can't detect potholes.
It's not a question of detecting the potholes, but of localizing the vehicle and tracking other objects in the vicinity (like other cars in traffic). High-fidelity maps make the system more robust because it can eliminate lots of measurement noise by averaging the readings from a constellation of landmarks.
> can't they do like the Kinect
This is essentially what they are doing. The problem is more difficult when the Kinect is moving.
You have to start somewhere. In addition to bootstrapping the program in a few cities, they are probably using this data to improve their pothole detection systems. Pittsburgh's a great city for that sort of thing.
I can understand self driving cars eventually being practical on long simple roads.
But I can't even imagine how one could get to my house. My road has cars parked on both sides. Sometimes vans. Sometimes they almost block the road. Sometimes there is a car coming the other way and you have to look for a space to pull into to so they can pass, or maybe back up 1/4 of a mile. Or hope the other car does.
There are cones and roadworks quite often.
I have significant difficulty driving down there, I don't see how an automatic car could any time soon at all.
This might not be an issue if I owned the car. I could turn it off and drive down there myself but an automatic taxi couldnt do that unless there was a driver waiting in the car to do that if required. And if there is, he might as well drive it all and save thousands of [currency]
I guess that makes sense.
And they can solve issues and make improvements one at a time untill it's good enough I guess.
But I doubt that will be any time soon
> I can understand self driving cars eventually being practical on long simple roads. But I can't even imagine how one could get to my house.
I can understand planes eventually being practical for transport, but the wind and noise and short distances they fly make them useless for passenger transport.
I can understand computers eventually being practical for business-y folks, but the rooms they take up make them useless for normal folk.
I can understand this internet thing is useful for nerds, but I can't see how regular people will ever use it.
I can understand a "portable" telephone is a nice idea, but the breifcase of a battery I have to carry around makes it's way to big and heavy.
... need I go on?
It's happening, whether you can imagine it or not.
Are self-driving cars legal in the United States by law, or just due to the absence of law? Seems like vehicular traffic has always had fairly tight regulations around their operation, but I'm stunned by the speed at which cars that drive themselves are hitting the road. Sure, we've had cruise-control, and cars have been adding things like auto-braking and lane-assist. Have they just taken that feature-creep further to facilitate self-driving under existing laws? Or are they simply able to do-so because they all necessitate a human driver to supposedly be alert, awake and aware? Maybe I missed it, but to me this just seems like something that would warrant some discussion before self-driving vehicles are allowed on the roads, even if the conclusion is the same.
Wikipedia covers the legal aspects reasonably well, with a map of state laws.
"In 2016, 7 states (Nevada, California, Florida, Michigan, Hawaii, Washington, and Tennessee), along with the District of Columbia, have enacted laws for autonomous vehicles."
What is smart with Uber pushing this tech is they have incentive to test a lot with a human operator, they can earn money driving people during the test, but Google doesn't earn anything when they do, limiting their tests.
As someone that follows lots of tech "trying to see the future" - to me Uber has the greatest hope of bring driving to the mainstream due to a number of factors.
First, they have a lot of experience with litigation and working with local laws.
They need to replace drivers, but unlike truck drivers - Uber will drive anywhere.
Unlike trucks, there's the potential to allow the passenger to drive the car.
Many more key reasons why Uber's efforts are worth watching, but really do hope they're able to make serious progress in bring autonomous vehicles to the mainstream.
I'd love to sit in one of those cars and see how well the cars can drive themselves, especially during the winter. It's just a really awesome experience being in a self-driving car - it'd be amazing if you could just hail one with an app.
The acquisition of Otto is interesting. I think Anthony Levandowski is incredible and the trucking industry is perfect for automation given the state of the technology and for logistical reasons. But it seems odd that Uber wants to branch out at this point.
It's interesting that they chose Pittsburgh. Having grown up there and they having lived in many different places since, I have learned to appreciate having learned to drive on those roads, especially during winter.
Many roads in Pittsburgh, and western PA in general, can be very curvy and have tight-radius. They're also often steep.
The on-ramps are peculiar because merging traffic must yield, visibility can be poor, and since some on-ramps can have tight radius and be steep, it makes it a little more challenging to merge. However, all of that is not that difficult to learn. It's much more interesting when you add in a foot of snow and ice and blowing snow.
Also, Pittsburgh is an 'older city' so large parts of it were built before cars. Highways have been 'bolted on' in some places, especially the ones that bisect downtown. Many people unfamiliar with Pittsburgh can get lost when coming in from 376 because on entering downtown, there are several quick exits and lots of lane switching.
It makes sense since CMU is there and a lot of robotics people, but interesting nonetheless.
I imagine navigating 376 coming into Downtown out the of Fort Pitt tunnel to be insane. I can barely do it as a human. Everyone is entering and trying to cross the 4 lanes to exit on the opposite side with in about 100 yards of roadway.
Ah, borrowing from the Musk marketing playbook, just say "self-driving" often enough, and enough uninformed people who have never seen contemporary luxury assisted-driving features will be impressed. Calling out Autopilot as "souped-up cruise control" is a nice touch.
> For now, Uber’s test cars travel with safety drivers, as common sense and the law dictate. These professionally trained engineers sit with their fingertips on the wheel, ready to take control if the car encounters an unexpected obstacle. A co-pilot, in the front passenger seat, takes notes on a laptop.
That doesn't seem very scalable. How many cars are in this so-called "fleet"?
Is there anybody working on autonomous vehicle tech that believes there's any substance here, or is it just adverjournalism?
I worked on a team at NREC that was building self-driving trucks for use in surface mines. The project was a descendant of CMU's work on the DARPA Grand/Urban Challenges. These vehicles actually went into production and are being used at a few mining sites around the world. Most of that team is now at Uber.
I can tell you that there is absolutely substance here. I have no idea what's going on inside Uber, but they have a group of people there (some of the only ones in the world) with experience building actual production-worthy fully-autonomous ground vehicles.
Having a safety driver in place during development and testing is common practice, and for sure they're still very much in the development and testing phase. Full autonomy on public roads is a harder problem that full autonomy in a more controlled environment like a mine site, for reasons that have discussed here and elsewhere, and I don't know how much progress they have made in dealing with those challenges. However, I have no doubt that they are working toward the goal of a fully-autonomous no-driver vehicle, not just the type of assisted-driving features you may have already seen in production.
Last summer I saw one of these autonomous Uber cars driving on a side road when I was leaving my office in the Lawrenceville neighborhood in Pittsburgh.
Was tempted to motion to jump in front of it to see how the car would react, but didn't...
Doesn't this upend Uber's whole model of non-ownership of vehicles and rejection of liability? Unless the driver is just there to be liable for having "allowed" the AI to intentionally hit something or someone!
Also, as a NYC driver, I wonder how aggressively a self-driving car could legally be programmed to defend its lane. I'm picturing Uberbots sitting still in traffic while streams of human drivers gleefully cut them off. City driving is basically a game of chicken every 10 seconds; winning at chicken depends on convincing your opponent that you are willing to have an accident.
It does. I wonder if Uber's accounting department really thinks this is a good idea. If Uber has to pay for gas, spare tires, oil changes, cleaning, theft, vandalism, insurance, moving violations, taxes, plus maintenance of what ever gizmos and gadgets it takes to auto drive these XC90s they may find it is cheaper to just underpay regular drivers.
If it goes well, then Uber has suddenly pushed forward autonomous driving and set itself at the forefront of commercial adoption.
If it is a disaster and many accidents result, Uber will set back driverless cars by a decade, burying it under a wave of pessimism, allowing their current business model to persist well into the future.
If the adoption of driverless cars succeeds this translates into less automobiles sold by manufacturers to the general public. Won't the auto manufacturers then need to increase the price of each unit sold? Eventually wouldn't Uber's savings from no longer having pay labor costs be lost due to auto makers increase in price per unit of each(now driverless) car sold?
This seems to be a lot exposure for Uber unless they get into the manufacturing business which seems like a rabbit hole. This type of exposure is the same as say a music streaming business that has to pay for content - i.e your business model is predicated on favorable enough terms from a third party in order to make a profit. And we've seen that the way around this is by being vertically integrated such as what Netflix has done with creating its own content.
1) Uber keeps tracks of the fuel ahead of time, and goes to an Uber refuel station automatically before it gets anywhere near low fuel.
2) The computer detects that the flat tire happened, pulls off to the side, and autodials the Uber triple A repair crew.
3) This one is hard actually. Perhaps the cameras could detect the flashing lights and sound of a police car, and automatically pull over? Or maybe there is just an emergency pull over button that the passenger presses?
I wonder how this will affect crime. With a driverless car, if someone wanted to hold you up, all they have to do is walk in front of the car, which will then obediently stop.
If I'm in a shady neighborhood and a drunk or threatening guy or whatever starts meandering towards me, if I'm driving I have the power to turn/drive around him/accelerate or whatever. If I'm sitting in the backseat of a driverless car, I don't have that option.
It's in the article. They hired a big part of CMU's robotics division, which made quite the headlines a while back. They were (and maybe still are?) based in Pittsburgh. It's also imaginable that Pittsburgh has some features that make it interesting, such as cold winters, loads of bridges (difficult!), and both grid and messy road patterns.
Uber has to work on self-driving technology, but in a certain way doesn't self-driving technology seem like something that will eliminate Uber's market advantage?
Right now, Uber is the go-to choice in most areas because of network effects. They have the most drivers because they have the most users. They have the most users because they have the most drivers. In a world of self-driving vehicles, anyone with capital can buy a fleet to blanket a city. If you can finance a self-driving vehicle for $400/mo, it's not that hard to launch in a new city. Sidecar operated in my top-10 city with what seemed like 5 drivers. That was too few and it was terrible (given that a lot of the time they might not be on-duty). But if one could get 20 self-driving vehicles for $8,000/mo, it's an easy way to launch. If adding new vehicles is as easy as going to a local dealership and putting down your VC money, you could get more on the road within a day to meet demand. In the early days, you could keep a close eye on demand and over-provision a little.
Self-driving vehicles change the market from being one of network effects (where Uber has a huge advantage) to one of capital spend. Uber does have a lot of capital, but what if GM wants to get into the game? GM gets cars at cost and GM has a brand that people know/trust (they're not some no-name I'm-Uber-Too).
Worse, it's a market of variable capital expenditure. With cable TV/internet, a lot of the costs are fixed costs (actually laying all the cable and such). With wireless service, there's a limited amount of spectrum and a lot of fixed costs creating the network. With self-driving taxis, it seems like the vast majority of the costs are variable costs that go up with the number of riders. There's really nothing stoping a city from having dozens or even a hundred different self-driving taxi companies. I doubt it will be hundreds as it's hard to run a good company, but it does mean there won't be a good opportunity to push high margins onto consumers. An attempt at high margins would mean another company scooping up that business.
I guess I just don't see how Uber maintains high enough margins in what will become an incredibly price-competitive market where it's relatively cheap for new players to join. If Sidecar could have spent $8,000/mo to launch in my city with good coverage, they would have made users happy rather than frustrated that there were no drivers around. They could have been successful. $96,000 isn't a lot of capital needed to start up for your first year. A Dunkin' Donuts franchise costs more than that to start up (from a little Googling). At $5/ride, vehicles would only need to do 3 rides a day to justify a $400/mo price tag. Even at $2/ride (less than public transit), you'd only need to do 7 rides a day for that $400/mo vehicle. That doesn't deal with fuel costs, but you can see how you don't need a lot of usage to cover the cost of the vehicle. If you think about fuel costs on a Prius and a maybe a 5 mile average trip (which seems quite long), you're talking about 1/10th of a gallon or around 20 cents for fuel. A trip that might cost $15 on UberX drops to under $2.50 at-cost for a service that isn't popular (its vehicles only do 7 rides/day) and way lower cost for popular services. So, Uber is going to have to push its margins way down (and keep them down) to prevent someone from wanting to enter a market that requires such little capital.
This is the former. The latter will certainly happen eventually, but is orders of magnitude more challenging and a number of years away.