Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Has Uber killed off its self-driving trucks? (backchannel.com)
91 points by steven on May 8, 2017 | hide | past | favorite | 95 comments


"Uber Freight" was supposed to be a freight forwarder, a booking service for cargo. "Uber for semitrailers", if you will. Call for a truck and driver, and someone comes, connects to your trailer, and takes it somewhere. There are already other companies offering that service. There's no one dominant player, so Uber might be able to muscle in.

Uber registered with the US DOT as a freight forwarder, not as a trucking company. Freight forwarders are carriers, even if they don't own any trucks. They have financial responsibility for the load getting to the destination in good condition. Uber doesn't get to claim they're just a booking agent.

This service doesn't seem to have launched. But that's what previous reports said Uber Freight was.


Another startup called Cargomatic tried to do this a few years ago. They raised $15MM but ended up shutting down last year due to logistical issues and not being able to meaningful generate revenue. One of the founders even came from a family who were in the trucking business. I think it's a good case study for the various challenges any entrant in this sector will need to solve in order to "disrupt" the old way of doing business.

http://www.businessinsider.com/cargomatic-uber-for-truckers-...


> logistical issues

This is a funny excuse for a company in the logistics business. You'd think they'd be good at at least that one thing.


Funny, indeed!

They supposedly had employees entering stuff into spreadsheets by hand. I recall that this was due to there being so many edge-cases, which made it difficult to automate processes.

The bigger problem I think with freight ever moving to self-driving is the legal liabilities involved. Consequences of even minor accidents are amplified with the size+mass involved.

We might see a hybrid system with auto-pilot doing the majority of the driving similar to how we currently fly planes, but I can't imagine how that would markedly improve the economic margins.


>We might see a hybrid system with auto-pilot doing the majority of the driving similar to how we currently fly planes,

That seems a very plausible scenario. Improves safety (which is a big plus!) but doesn't really do much for the financials overall.


One manned cab in a convoy with one or more self-drivers following does make for more enticing financials. You still have someone on hand to deal with paperwork, tire inspections, breakdowns, etc. with lower labor costs overall.


Potentially. But then you have to transfer drivers to the depots where they will pick up the other cabs in the convoy. The financials may work. We'll see. FWIW, driver cost seems to be about 1/3 the operational cost of trucking today. And about 21% of trucking is over 1,000 miles. [1] So that 21% is probably your target.

[1] http://www.atri-online.org/wp-content/uploads/2014/09/ATRI-O...


Driving in a convoy doesn't really solve the legal liability issue. How would the human driver presumably avoid accidents towards the back of the convoy for example? You'd have to have multiple drivers spaced out thru the convoy.

The main issue is having good safety controls for all the various edge cases you might have out on the open road.


As a German the self driving company Otto is really confusing. It doesn't only have the same name but also almost exactly the same logo as Otto, the decades old and second largest online retailer in Germany:

https://www.otto.de/

https://en.wikipedia.org/wiki/Otto_(company)


Right, absolute minimal effort required. The company was literally created as a way for uber to acquire stolen IP. They knew they'd never be around long enough to face a lawsuit, so just grab something easy to rip that's ​somewhat relevant and won't result in an immediate lawsuit.


Just in the autonomous vehicle space there's Levandowski's Otto, Ottomatika (acquired by Delphi), and Otto Motors, which is working on autonomous vehicles for warehouses. It's confusing.

https://www.ottomotors.com/ http://www.ottomatika.com/


Is the autonomous Otto logo trying to look like the front of a truck? It seems like the retailer Otto wouldn't want a truck as their logo? For me it's the retailer's font choice that makes the logos look similar.

Very similar to what jeep does:

http://symbolsnet.com/symbols/jeep-grill-symbol.html


I feel like the founders were just making a phonic play on the word 'auto'.



I hope not. After the Budweiser / Uber test I had an idea for a film script.

Red necks with pre-electronic cars and an improvised EMP device are running around hijacking self driving beer trucks.

I was thinking it would be kind of Fast & Furious style.


Is it really easier to hijack a self driving truck than a manned one? You can just point a gun at the driver. And there probably aren't any cameras or sensors on the manned truck.

Whenever a new technology comes along, people come up with tons of ways it could be abused and so won't work. I mean look at all the speculation around drones the past few years. The reality is most of our existing technology can be abused by creative people just the same way. The only reason society doesn't fall apart is because criminals happen to be fairly rare.


> You can just point a gun at the driver.

Yeah but now you're looking at 20-30 years for armed robbery instead of being just the good ol' boys never meanin' no harm, just makin' their way the only way they know, that's just a little bit more than the law will allow.


the good ol' boys never meanin' no harm, just makin' their way the only way they know, that's just a little bit more than the law will allow.

The current political climate would never allow a car painted like that!


The General Lee? I see a few with that paint job every so often.


...on television.


You are still looking at years in prison for theft. I'm not sure if 20 years in prison is really that much more of a deterrent than 5 years.


It's been 30 years since I was in High School and I still remember being taught back then that studies showed that length of prison sentence didn't have a noticeable deterrent effect.

Most crimes are committed by people who feel they have nothing to lose.


Laws aren't static.

If that happens often then we'll amend the law to increase punishment for this particular crime, just like we made a single act of copyright infringement worth up to $150k.


Or they could amend the law to fire you out of a cannon into the sun! What will that dastardly congress do next.


It doesn't even require anything clever. Just put your car in front of the truck and it stops on the road.


Pointing a gun at a person is a different crime than computer hacking.


Red necks with pre-electronic cars and an improvised EMP device are running around hijacking self driving beer trucks.

I was thinking it would be kind of Fast & Furious style.

Fast & Furious!? Go back further! Smokey & the Bandit! That's exactly the sort of thing they would do, if someone with a TARDIS plonked the Snowman and Bandit into the 21st century!

https://en.wikipedia.org/wiki/Smokey_and_the_Bandit


Why rednecks? There's a reason the mob would steal cigarette trucks then sell them below market value: quick cash.


Also a High value : weight ratio and easy storage which is where beer fails.


I'd pay to watch a movie about the Redneck Mob.

"Hey, Vinny. Hold my beer and watch this!"


You're missing a remake of Smokey and the Bandit!


I assume this would quickly be stamped out by sending convoys of trucks with a single driver/guard supervising it. So instead of 1 driver per-truck, you have a 5-truck convoy with a single on-site monitor/driver/supervisor/guard.


Hooooh man this is some mad max shit I am so down. How do I get hired as Shotgunner?


Imagine if they all drove close together and had roof railings so you could hop from one to the next for a visual inspection?


Now you're just making a videogame. And it would exceed the height limit for an 18-wheeler, you wouldn't be able to go under many bridges. If you made the main trailer shorter to add room for hand-rails on the top, then you've lost cargo space.


But mad max


This fear is overblown.

You can do everything you mentioned already.

Just put up a barrier on a road to stop the truck, and then shoot the driver.

Police aren't going to be able to protect drivers in the middle of nowhere.

Now ask yourself, why does this not happen right now?


Because murder is much more serious than just stealing stuff. Somebody killing every trucker on the road? Cops swarming.

Somebody making off with your merch? Ask a loss prevention officer how much cops care about that.


Perhaps you should ask yourself why that doesn't happen now.

The driver is the part that makes it difficult. Even if one had no moral qualms about murder, law enforcement and the public would be much more concerned about homicide vs merely property destruction and theft. If you were to just hijack a beer truck and nobody got hurt, I bet the average member of the public would hear about that in 'funny news'. If you killed the driver, you'd be infamous and hunted down.


Is there a fear in the post you're replying to?

Given that the objective was free beer, I assumed that in this story the robbers were the protagonists. The story is only possible in a world with self-driving trucks because in any other universe the goofy beer-stealing rednecks are threatening real people and that's not very sympathetic, but in the self-driving truck universe they're just stealing from robots.


Because police are more inclined to pursue crimes with a human victim. It's like the difference between robbery and burglery being where someone is in the house or not. One gets harder look because there is a terrified victim and escalation could've meant death. The other is just property loss.


Randomly put a guy in 1/5 trucks but make the guy not visible. Loudly state on the outside of the trucks that many trucks are manned. Criminals have no way of knowing which trucks have people and which don't.


Errr because murder is a really serious crime?


Nobody is going to stay up at night having nightmares about the time they still beer from a big company. Very few people are actually capable of being murderers.


I walk past the Uber/Otto trucks every day (three or four are kept on Harrison St in San Francisco between 3rd and 4th), and while the trucks are still their with their very visible lidar/cameras/etc, in the last week they've removed all mentions of Otto from the trucks (previously they had the Otto branding on the doors).


Do you have any pictures? I'd be interested in seeing them.



Yep! Specifically: https://thelastdriverlicenseholder.files.wordpress.com/2017/... (that's them parked in the lot on Harrison) and then you'd erase all the OTTO branding :)


I thought the logic of pursuing self-driving trucks was that it was generally easier than self driving taxi services:

- Mostly highway miles (or you could arrange it to be so) - Fewer weight requirements - Better economics (a greater % of the economy is dependent upon trucking logistics than taxi services)

This isn't to say that there's a big master plan of making a trucking company, more that making a self driving trucking company is a very practical step on the path to full autonomy.


My impression is that Uber simply can't afford to not be the absolute first in driverless taxis. They're burning money to stay in the taxi business. Their entire business model relies on keeping the price somewhere in the range of what it is now and not paying a driver.

It might be that the marginal advantage of not having a driver on the highway section of truck delivery is not large enough to justify building out a trucking network. They'd have to establish relationships, buy trucks, service the trucks etc. all of that would be time and money that isn't going into their big bet.


I've said this before, but this just seems crazy to me. Nothing I see suggests that 100% autonomous door-to-door driving as a commercially offered service is just around the corner. Even if it reliably worked technically you're still years away from being able to offer it to the public. Not even Uber could flaunt the law that flagrantly.

And the thing is that we're not there. Even if you say it's a decade out (which seems pie in the sky optimistic to me), can Uber really afford to burn through cash for another 10 years?

[ADDED: I didn't make my point well. I do think legal/regulatory will take some time independent of technical readiness, but IMO it's technical readiness especially in all the corner cases that's the long pole.]


Elon Musk stated that their technology is 2 years away from full autonomous driving. Tesla is planning LA -> NY fully autonomic drive by end of this year.

Waymo has already started a real pilot of fully autonomic driving for regular people in Phoenix area. I predict that means they are less than 1 year away from ditching training wheels and launching it fully.

Some think that changing laws to make legal will be the hard part but I see the opposite: states will be competing for who's first to make it possible.

Many states, including CA, already made testing possible. You just sign up and ask "can I test my self-driving cars, on real roads, possibly destroying other cars on those roads and killing people. Pretty please?".

Laws are hard when they're controversial and self-driving cars are political goldmine:

* dramatically less deaths and other kinds of destruction due to less accidents

* cheaper transportation

* the sheer coolness of self-driving

It'll take only few states in US to allow it and other states will have to follow or the politicians will look very stupid if a taxi drive in SF costs $1 and the same is $10 in Chicago.

No-one will have to flaunt the law, it'll get changed very quickly.


Waymo's Phoenix test will include an employee in the driver's seat ready to take over. (See: https://www.nytimes.com/2017/04/25/technology/waymo-to-offer..., and note the sentence "All the Waymo cars will have a technician who can take control in an emergency.")

Elon Musk said that Tesla's driverless technology was "probably better than a person" more than a year ago. (https://www.washingtonpost.com/news/the-switch/wp/2016/01/11...). What I'm saying is, Musk is a big promoter of Tesla and taking his ideas of when the technology will be ready at 100% face value is foolish.


>All the Waymo cars will have a technician who can take control in an emergency.

This is likely more due to regulatory reasons than anything else. Waymo's cars have an exceedingly low rate of required intervention.


In Palo Alto.


And Tahoe (light snow) and Seattle (rain/fog) and a couple of other places. Sure, most testing has been in the Bay Area (although at this point I'm not actually sure if that's the case anymore).


>Tesla is planning LA -> NY fully autonomic drive by end of this year.

But that's not really a very impressive feat compared to 3-5 days of non-stop driving in NYC. I mean who cares if it can go cross-country? I want it to take me from my apartment to the office to the grocery store to Target out to dinner and then back home again.

Hundreds and hundreds (probably >1,000 even) of those cross-country miles are going to be desolate and essentially carless.


If I could get something close to an RV that could do the highway section to a national park or another major US city while letting 2+ people sleep in the back I'd be very interested in getting that and doing a very cheap vacation across the country.

As it is driving on a boring highway all day and finding a hotel at night sucks. I'd much rather leave the driving to the night while we sleep.


I think that's completely the right idea, but I don't think the difficulty in "self-driving" is the carless-ness of those highway roads. Highway roads tend to be easy because you don't have a lot of edge cases. Cars are always driving at roughy the same speed, and there aren't very many lights or intersections or bicyclists or pedestrians, etc.


> Elon Musk stated that their technology is 2 years away from full autonomous driving. Tesla is planning LA -> NY fully autonomic drive by end of this year.

While that's nice, it's kinda the low-hanging fruit.

I'll be impressed if somebody manages autonomous driving in, say, New York (or Kolkata if you're feeling ambitious), at night, during snowfall, navigating a construction site and/or traffic regulated by policemen.

This is taxing but completely possible (and entirely expected) for a human driver. Whereas the lane-following autopilot in my car is routinely confused by shadows on a pleasant day.


I actually agree that if the tech works extremely reliably, the legal/regulatory hurdles may take some time but aren't going to be a major roadblock.

We are quite far apart however on how close we are to "just works."


There's a few easier steps they could take first. Long haul point to point from a few major distribution points to others is probably doable soon especially if they're designed to accommodate the autonomous trucks. Local drivers who are paid less drop off a trailer at a UberFreight depot in Southern California, an automated truck picks it up and drives freeways over to the UberFreight depot for Eastern North Carolina, there another local short haul driver gets booked to take it to it's final destination. Uber might be burning too much money to make it to this but the technology for long haul freeway driving is largely here already it avoids a lot of the difficulties of small roads.


While true, a lot of truly "long haul" trucking has switched to freight railroads anyway with inter-modal shipping. There may be some savings associated with effectively adding another mode (autonomous medium-haul) but there's at least some cost to adding the additional level of depots associated with that. I can believe it would work but I've never seen a real analysis.


I'd short Uber if it was publicly traded.

I'm not sure how far away door to door taxis are, but I also would bet against Uber having a long enough market lock in to recoup losses.


Uber's driverless car strategy is incredibly implausible. You have to believe all of the following:

1. Uber will be first to market with driverless car technology. That is, it will beat Waymo, all the actual car companies, Apple, Tesla, and 100 other startups.

2. Given 1, Uber's driverless car technology is a durable technological advantage (ie, even if it's first to market, that doesn't matter much if Waymo follows it to market 3 months later). Note that we are at this point stipulating that this technological challenge is doable with a team that you spun up out of nothing a couple of years ago, as not-the-best-resourced team on the block.

3. Given 2, the legal issues with driverless cars are resolved well enough that Uber has a viable business.

4. Given 3, Uber can successfully raise money for an incredible capital expenditure (how many cars are we talking here? 1,000,000, very conservatively? At $20,000 ea, very conservatively? So $20B investment), and then handle that kind of capital investment.

5. Given 4, build up all the supporting infrastructure for this business (parking, local cleaning, etc.)

6. Given 5, this all happens within Uber's operational runway (like, starts within 3-5 years, finishes within 10 or so)?

Personally, I'd put the odds of this at: 10% for 1 * 10% for 2 * 90% for 3 * 80% for 4 * 70% for 5 * 20% for 6 = roughly 0.1%.

But even if you're way more bullish on Uber than I am, can anyone really argue for more than 50% for 1 * 50% for 2 * 90% for 3 * 95% for 4 * 95% for 5 * 80% for 6? That's still only 16%.


So Uber just needs to win, in a sustainable differentiated way, self-driving technology, get it incorporated into actual cars in production volumes, and roll it out as a new major national car rental company based mostly on very short-term rentals. Within, as you say, their operational runway.


I think that steps 2 and 4 are the biggest questions for me, but the details about keeping the cars running & clean are also good ones. Right now Uber can externalize the costs of a running clean car. Running their own fleet will mean that stuff like gas, maintenance, cleanliness, misc stuff that 5 star uber drivers do (free candy, water etc) will no longer be externalized and bore by their drivers.

P.S. did you mean bearish instead of bullish?


Nope. People who are more bullish about Uber than I rate their chance of being first-to-market with driverless car technology at higher than 10%, for example.


And then you have the question, if people want to be in a taxi without a driver?


Easy, they just have to find an investor with 10 times as much money to irresponsibly blow through as the Saudi sovereign wealth fund. That exists, right?


I don't get what the Uber game plan with driverless taxis is. What is their competitive advantage?


In the past UBER was supposed to lead in shared rides(i.e. UberPool) , and their VC's talked about that as the thing that will bring UBER to replace private cars.

Currently, other companies seem to have the lead.And btw, when done well, this segment has a pretty strong network effects.

In the same way,if most of UBER's advantage is awareness and installs, credit card numbers, and common trip data, in the context of real self-driving cars, that's almost nothing:

Awareness and app install - a working self driving car will the miracle of the century. Everybody will hear about the company who offers it, constantly. from there the road to installs is short. And for such value as self-driving cars offer, entering a credit card number doesn't seem like a big deal.

As for common trip data? well at the very least Apple, Google, Moovit, Cellular ISP's, and maybe others have plenty of that(in one form or the other) to start.

But most importantly:time. Self-driving cars are a risky technology. it also takes time to scale manufacturing. so they'll probably be deployed relatively slowly. That gives the companies who own them enough time to not need UBER.

But UBER still has one strong card to play: flying cars.


As I recall, Travis Kalanick's overall goal to is reduce private car ownership. Driverless taxis would get Uber closer to that realization. However, my pessimistic view tells me that self-driving cars also grant Uber more control over the private transportation industry. And perhaps they could use that power -- some call it network effect -- to negotiate large-scale contracts with auto manufacturers and oil-and-gas companies.

Overall, as a consumer, I couldn't care less who is first to market, but which company offers the best quality-to-price point.

On another note, when self-driving cars need to refuel, is that payment going to be in USD or some cryptocurrency?


Lets be real, Travis ultimate goal is to be rich and powerful.

He saw private car ownership as counter-indicator to Ubers growth and thus his own wealth.


Wouldn't driverless taxi not be a good reason to buy a car yourself? Especially, when you don't have a driver license?


I did some back of the napkin calculations and a two car suburban family would save approximately $7000 using Uber to commute compared to owning a 15-20K second card over a 10 year period.

That's the market, if uber wants to scale they have to do that driverless.


$7k a year, or $7k total? no one in america is giving up their car for $700/year


Who is giving up a car?

Who will want to own, maintain, deal with parking, insure, etc etc a car for 10 years when you can have this year's model appear anywhere, anytime by just taking out your phone?

The 7k savings assume you bought a cheap car and kept it, I also did not factor tax & insurance. Savings increase dramatically if you were to finance a 35k+ 2nd car that many Americans in the burbs seem to do.


Uber loses money at that price.

Uber doesn't "scale", there is no economies of scale that kick in anywhere in their business model.


I don't understand what you 'mean at that price'

As for scale? There very much is efficiencies to be gained in becoming a true mass transit system.


But that means Uber takes on all the things that they currently offload on the drivers (car ownership, maintenance, etc).

By 'at that price's I mean Uber have currently lost 15 billion dollars subsidising taxi fares. They cannot sustain that.


I agree they can't sustain that. Without insight into their roadmap we can only speculate.

To the point of Uber maintaining driverless cars. What's to stop them going to a franchise model. Uhaul-like hubs where the cars go to get charged, cleaned etc?


Even if they aren't paying any drivers?


The drivers in Uber handle more than just the driving - they essentially handle all the other logistics of offering a taxi service. They keep the car clean, they handle any regular maintenance and servicing, they store the car when it's not in use, they manage registration and insurance, they manage any insurance claims, they fix any damage, they mop up the vomit, etc.

Once Uber goes self driving, they take on responsibility for all of these logistical activities, and they currently have no capability and capacity to do that. These activities only scale so far.


> Once Uber goes self driving, they take on responsibility for all of these logistical activities

Why would they need to do that. They're a software company, they have explicitly demonstrated no motivation to take on this work.

It's really not hard to envision them partnering with an existing bricks and mortar to provide those services - e.g. car rental companies provision through their hubs, or uhaul steps into the game and their franchise owners become uber hubs.

I really feel this is the obvious endgame, but who is to say.


Network effects (they already have many app installs and habitual users), and possibly an early mover advantage at the rate their program had been progressing.


Probably easier from a technological standpoint, but I can imagine an autonomous freight network being just as challenging as an autonomous tax service if you account for all sorts of things: sales, mountains of other regulations, compliance, etc. I can hardly imagine it's as simple as, "let's just put some sensors on trucks and the problems will solve themselves!"


You are right, trucks and also train's are certainly and easier problem to fix. However it's about market size.

Automate a train or a truck and you save on persons salary. Automate personal transport and your market is 6 billion people.


> Automate a train or a truck and you save on persons salary. Automate personal transport and your market is 6 billion people.

It's the other way around; automste personal transport, and your market is all the trips taken by people Automate freight, and your market is all the trips taken by goods, raw and finished.

The latter is a lot more trips. Even if you just restrict it to the ones taken by ground transportation.

OTOH, Uber clearly wants to piggyback last-mile delivery of goods on its personal transport service, and that's the part of goods transport where the labor cost is probably the highest proportion, because one person carries very little compare to the long-haul routes.


Turns out nailing full autonomy at highway speeds is actually just as difficult as getting urban autonomy right.


It is both easier, AND has a lot of money in it, if you can do it.


It's not exactly news that the Otto team went straight to work on cars after the acquisition.

Nor is it news that Levandowski and Kalanick hatched this scheme while Levandowski was still with Google. (read the second from top comment): https://news.ycombinator.com/item?id=12315205


Uber is likely keeping a low profile with its autonomous initiatives until the Waymo lawsuit blows overs.

I'm sure we can expect big things from them as soon as that lawsuit ends.


Wouldn't it be disastrous to do so?

Elon Musk said that Tesla will be releasing an electric truck in the later part of this year [1].

Of course, self driving capabilities will be added afterwards.

Wouldn't it make sense for Uber to have this very same capability as well?

[1]: https://www.youtube.com/watch?v=CgMqaDlPONA


lol dozens of paragraphs of speculation, and then Uber directly denies it at the very end. Talk about a waste of a read. This pretty much summarizes most blog articles these days.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: