Hacker News Reader

Comment by danblick | Hacker News Reader

I've been thinking a little about the economics of self driving cars and it seems to make sense that they'd seek a partnership with a ride sharing service. While I guess the eventual goal could be "a self driving car in every garage", you'd probably pay for more for one if (1) you're using it all the time (high utilization), and (2) it introduces some other kind of savings (like not having to pay a driver). Consumers have neither really but taxi/trucking companies would have both. I guess it's a question of how much the hardware is really going to cost. Does anybody know of publications that talk about the economics here?

View on HN

I highly suggest checking out the a16z podcast. There's an incredible episode called "Cars and Cities, The Autonomy Edition." The podcast is pretty interesting generally, but this one reshaped my entire understanding of autonomous vehicles and their societal impact.

https://a16z.com/2017/02/08/transportation-as-service-autono...

Edit: The podcast is a conversation with Taggart Matthiesen, director of product at Lyft.

As of now the goal of companies leading in this area is not be a "self driving car in every garage", but ride sharing in metropolitan cities. There's no rush to provide L5 level capabilities to a farm owner in Idaho

Yet that farm owner will probably have a 'crew' of self-driving tractors.

Actually you already have mostly self driving tractors. However wheb there's that much money at stake, hiring a driver to hit "stop" if something goes wrong is a very cheap option.

Not sure why you were downvoted -- autosteer and more advanced farm vehicle automation technology has been available and in wide use for years. Sub-cm level accuracy is widely commercially available.

Planting/havesting/etc regular rows of crops on private property with computer positioning control is easy compared to high-level auto-driving on public roadways.

Similar position-automation technology is widely used in mining, construction and more.

I think the end game is more like the end of individual car ownership in cities. Company-owned or ideally(!!!) municiple fleets would be much more efficient in that idle time and parking would fade away.

I agree that's the end-game but recently I've revised my expected timetable.

I fully expected that my daughters, now 12 and 13, might never need to own cars. My wife even thinks they may never need to learn to drive. I now think that's unrealistic. It's true that for 95%+ of most journeys, self driving cars would be fine, but that last 5% or less really matters even if for only a minority of journeys. Finding a parking space can be very hard and our current streets, curbs and driveways aren't designed for self-driving vehicles. Rural settings can also be highly challenging.

I think the big changes initially will be road haulage because companies can re-engineer pickup and drop-off points to be highly autonomous friendly and automated. Next will be companies like Uber and Lyft providing driverless cabs from and to designated points. Then a gradual demographic trend to less car ownership, but driverless cars for the masses are going to take at least a decade to come IMHO. The last 10m is really tough in too many common scenarios.

I agree that the last 10 meters is going to be one of the harder things to solve, but i think you'll see people building calibrated parking spots into their driveways (some way of teaching the car how to get there from the road) and perhaps even companies doing the same with their parking lots and then publishing the details online. As self driving cars become more ubiquitous there will be companies that spring up to precision map places and feed that data into whatever databases the cars use.

Not sure about that. Sure it makes sense, taking public transport in Europe also makes sense and many people use it most of the time. But there are times when you want your own car. And with a family that happens often. Also, many people love cars. I don't think many people (esp those with family) will give up their car. Can be more expensive and less convenient and most will still do it.

If cars were really just to get around, we'd see different kinds of cars on the road. Cars are a lifestyle symbol and for many people a place where they spend a lot of time. You don't want to rely on municiple fleets for that.

Seems like part of uber's moat is the driver network -- having drivers available everywhere.

But at least that part of the moat disappears when you no longer need drivers.

Behind all the drama with the Waymo lawsuit and the tornado that was Anthony Levandowski blowing through, Uber's advanced technology group has been quietly and doggedly plugging away at the self driving problem.

They announced today that they've logged their 1 millionth test mile, and completed over 30,000 passenger rides in Pittsburgh and Tempe.

I have no idea how good their cars are compared to their competitors, but, with 200 test cars, their operation is supassed in scale only by Waymo.

They're hard at work digging their moat. Clearly Lyft has some tricks up it's sleeve as well.

Right, that's a different moat and a new playing field though. It seems like it could be less of a network effect, possibly more of an economy of scale effect. Or maybe those two usually end up coinciding.

Or maybe not, maybe they'll be some point where the software matures enough that people end up being sort of indifferent between who developed it as long as it passes some "good enough" threshold, and some of the set of developers with good enough software will license it to whoever wants it to undermine the dominant player. For example I'm not sure which of the Maps/directions apps out there right now is actually the best, but they all seem good enough to use and I can't actually tell which is superior except for rare edge cases that I won't see often enough to be able to quantify.

While 1 million miles sounds impressive, it's probably not more than any of the big OEMs has done so far. Every car manufacturer is working on self driving cars, plus some others (Waymo, Uber, Continental). The difference is that manufacturers don't talk that openly about it. They are used to keep advances secret until you can buy it. I'm sure Daimler, VW, GM and Ford test with a similar fleet.

If they come to market at the same time as others they'd need to find a manufacturer that doesn't already offer it. If they want to be first they need to beat manufacturers that have recognised the risk and can spend several billions a year without making a loss..

You're not going to be able to provide any sources for your claims because there are none. You are bullshitting.

That is interesting and odd as Uber in general has become significantly worse in directing drivers to pickup locations. In my experience over the last month, most drivers ignore the app and use common sense with a far higher degree of accuracy vs Uber's broken GPS

Reply on news.ycombinator.com