17 ms·

Based on what I hear from those in the know, the perception is that it's decades away instead of years. And an even more complicated problem isn't whether or n

Based on what I hear from those in the know, the perception is that it's decades away instead of years. And an even more complicated problem isn't whether or not they can make it work with considerable constraints (geofence, no highways, avoid several spots that are hard to handle, weather, assistance via remote operation when stuck), but how to do it profitably. The cars are expensive. The maintenance is expensive. The ongoing cleaning, charging, and prepping of vehicles is expensive. Constantly updating HD maps is expensive. Of course the engineers to keep everything running smoothly are expensive. And the small army of people you need to actually talk to and help customers - something Google has basically never done before - is expensive. I love the idea of self-driving cars and I really can't wait until they're ready. But I'd say that's going to happen about as quickly as changing the infrastructure in this country. Personally I'd rather have better public transit anyway.

maximus1983 7y ago
TBH this is the pareto principle at work. Sure you can get a car roughly driving around the city autonomously (there are basic self driving cars back in the 1950s that drove on a wire track) but getting it perfect is where the vast majority of effort is going to be expended.
joe_the_user 7y ago
Yes, The thing about self-driving you can get to 90% or 99% or even 99.9% success and then find the cost of closing part of that extra sliver of non-success is the same as all the work you've done. Especially since testing those corners cases is hard since they are unusual. There's even some part of the sliver that even is "AI complete". Very small but you don't know where it is, etc. But I'd say that's going to happen about as quickly as changing the infrastructure in this country. Key point! Automation has traditionally been much, much easier when you are engineering a situation where you control all the variables. Automated factories are common. Automated warehouses are coming. Automated dock systems exist. In these situation, there's a hard border between "inside" and "outside". Roads are generally regular but the border between insight and outside fuzzy. Google can run it's self-driving taxis in a suburb in Arizona where roads are as regular as could be imagined but expanding beyond that is much, much hard than one imagines. Note: I recommend this blog which periodically tracks progress in self-driving cars as well as making intelligent, critical comments on current AI. https://blog.piekniewski.info/ https://blog.piekniewski.info/
- wpietri 7y ago
  Good points. And I'd add that most things we've automated are things that have been routinized in a way where people conform to the system, rather than the other way around. E.g., a ton of software development has basically been, "take this paper form and make a computer do the paperwork". The hard part was already done for us when everybody got trained to be subservient to the bureaucratic process. So societally we have this limited sense of how quickly computers can take something over. Real-world things like cars and domestic robots are much harder, because those domains are ones where people come first. As you say, key elements could well be AI-complete.
- secabeen 7y ago
  Even when you control all of the variables, there are plenty of situations where the automated system just runs out of runway, and kicks the can to a human. Can't read the label on this package? Push it to the exception lane, and have a human figure it out, and relabel it. Works great, but is useless for a car, because when the car hits the exception, it's 10 miles away on a residential street, and it's not going to make it to you anymore. For me, the most important thing a car can be is reliable, and every form of unattended driving cannot hit that level.
  - pas 7y ago
    Sometimes the taxi you called never arrives. It happens regularly. Of course the real problem is economic. Competing with the taxi service that uses humans is hard, because human labor is cheap.
wpietri 7y ago
The amazing thing to me is that I don't anybody who thought it was coming in the next couple years who based their estimate on anything but hope and magic. Of course, that's nothing new. We've all been on software projects where the launch date was the earliest day nobody could prove it was impossible to finish. But I'd like to think that investors dropping zillions on companies would have had better due diligence.
- joe_the_user 7y ago
  The amazing thing to me is that I don't [know] anybody who thought it was coming in the next couple years who based their estimate on anything but hope and magic. I'd agree that a bit of the proper sort of analysis would show that this wasn't easily happening. But I think the kind of thinking that imagines this as possible is extremely easy to fall into. We've all been on software projects where the launch date was the earliest day nobody could prove it was impossible to finish. Sure and I'm pretty sure a lot of software engineers have given or agreed to completely unrealistic estimates after having had this experience. If anything, the "software crisis", the inability of software engineers to provide good estimates even after hard experience and warnings is an expression that over-optimistic thinking is natural for human beings in this situation. My theory is that in situations where a human is asked about the difficulty of a mental task, they essentially can only fall back on a basic language-expression of said task and rationally try to calculate with that - the brilliance and the limitation of we language-generating, language-using creature is the ability to correctly distill a complex and messy task into fairly simple sequence of symbols. But that correctness actually exists in a complex and dynamic context, most of which we screen out. A command like "Drive to the store" doesn't take into account all the compensations needed to do the driving just as "Create an inventory system" fails to specify all the client-specific "gotchas" involved in such a task. So going from our normal language-distillation process to a complete software system that doesn't have the continual dynamic compensation of our intelligence is an inherently difficult problem - even though we can probably only do it at all because of our "language facility".
  - wpietri 7y ago
    > the inability of software engineers to provide good estimates I don't think it's inability. I think they're embedded in a system that doesn't encourage or reward honesty or accuracy. I agree it's easy for people to say, "Gosh, how hard can that be?" Which is where we get the classic, "I could build that in a week" estimate. But I don't agree that there's something inevitable about people going from a finger-in-the-air SWAG to a "we'll have self-driving cars in 2021" business plan. Normally I'd say that it's a straight up failure of both management techniques and the professional standards of the engineers. Which is true. But I think the problem is that a lot of fundraising in the "Uber for X" era is only slightly less of a con than Theranos. So I think it's more correctly seen as a failure of founders and VCs.
BurningFrog 7y ago
That sounds like one big company doesn't know how to do it faster by keeping doing the same things they're doing now.
vivekd 7y ago
>people you need to actually talk to and help customers - something Google has basically never done before As a google advertiser this made me laugh. At the same time I don't think expense will be a setback for autonomous vehicles. The reason being that if vehicles are fully autonomous there would be no reason for any individuals to own them. It would make more sense to have them operate like a taxi service. This is because without the costs of having to pay for a driver, it would be cheaper to just order an autonomous vehicle whenever you need to go anywhere than it would be own your own vehicle. Just from an efficiency standpoint it is terribly inefficient for each person to have a car. We're not driving most of the time and the only reason we have our car sitting in the driveway is we need it to be available should we need to drive. With autonomous cars, my car could on it's own drive across town to pick up other people when I'm not using it. The reason we can't do that today is because of the additional logistics and costs involved in transporting my unused car from my driveway to the driveway of the person across town who needs it. This problem would not exist in a world with autonomous vehicles. To me, it seems likely that a world with autonomous cars would be a world where the majority of vehicles in the world are owned by services like uber.
- YeGoblynQueenne 7y ago
  >> At the same time I don't think expense will be a setback for autonomous vehicles. The reason being that if vehicles are fully autonomous there would be no reason for any individuals to own them. It would make more sense to have them operate like a taxi service. This is because without the costs of having to pay for a driver, it would be cheaper to just order an autonomous vehicle whenever you need to go anywhere than it would be own your own vehicle. I have a better idea that we can implement right now without waiting for autonomous driving to become a real technological capability. We could make big cars with many seats and have them driven around town by humans- just a single human could drive a single vehicle capable of transporting 50, maybe 100 others. We could have those big cars stopping at predetermined points around the city, so that people would know where to get on and off. And we could support that service by asking the users to pay a small fee upon boarding the big car. This would cover the costs of the driver, and the big car, and still leave some to pay for infrastructure, etc. These big cars would probably have relatively restricted areas of operation, but, for the foreseeable future, so will autonomous cars (which have to be geofenced) and, like I say, we already have the technology for those big-cars-with-many-people-sitting-in-them-and-a-single-driver that I'm talking about. And since a whole bunch of people would be carried by one or two of those big cars per day, instead of each riding a single (autonomous or not) car, we could drastically reduce the number of automobiles-per-person, and, therefore, the CO2 emissions, thereby reducing pollution and protecting the environment to boot. So? Anyone want to invest on that?
StephenMelon 7y ago
Seems like autonomous personal air travel would be a far, far simpler problem to solve than road-based driving? Far less edge cases, once you are in the air the only obstacles you really need to worry about are other vehicles, birds, leaves and the odd stray plastic bag. With VTOL the whole thing could be an order of magnitude easier than driving. By the same principle, sea travel would also have less edge cases, at least in calm conditions. Is it the case that people aren’t seeing markets for those applications of autonomy or is there some other reason why there isn’t the same hype in those areas?
- gshdg 7y ago
  Vehicle cost; fuel cost; takeoff and landing; handling of edge cases (there's a reason that commercial airlines still have pilots despite autopilot being very effective in 90% of conditions).
  - dominotw 7y ago
    how expensive would it be creating areas of road that only allows autonoums vehicles, something like carpool lanes we have now. Wouldn't that cut problem size massively.
    - frosted-flakes 7y ago
      I can see that happening on major freeways, the way we have carpool lanes now, but not on a general scale.
- cmurf 7y ago
  A: Air travel generally is a far simpler problem to solve, because there's national standardization of the air navigation system, the airport paint and signage, and very clear rules of separation with very few edge cases. And yet we do not have anything close to self-flying planes no matter how much money you throw at it. And we're not even close to getting there. Whereas with cars, none of that is true, municipalities get wedged into violating their own equivalent of human interface guidelines all the time. Worn paint and signs for years, busted cross walk signals, confusing intersections, human driven cars that consistently do not follow the rules but in inconsistent ways. B. Personal air travel has a significant regulatory burden in the transition from ground to air. The ground is city + state regulated. And immediately once airborne it's FAA regulated, but not under any kind of Air Traffic Control, as almost all airspace below 1200' above ground is uncontrolled. So ATC has nothing to say about it, and no central mechanism for negotiating conflicts. Where are you allowed to takeoff and land is easily figured out today: airports only. And that's because it takes all kinds of things into account like obstruction clearance, noise abatement, anticipating engine failures and crashes. We can't even automate commercial flights - it might seem like it's more correct to say we don't automate commercial flights, but we could. That's not really true. The amount of changes to the air traffic control system, and on-board equipment for airplanes, is presently so cost prohibitive that it is effectively a "cannot be done".
mrep 7y ago
> but how to do it profitably. The cars are expensive. The maintenance is expensive. The ongoing cleaning, charging, and prepping of vehicles is expensive. Constantly updating HD maps is expensive. Of course the engineers to keep everything running smoothly are expensive. And the small army of people you need to actually talk to and help customer I believe that only applies to the waymo model where you go with the full time robo-taxi using top of the line hardware like lidar and which pretty much requires level 5. The tesla model is to use cheaper hardware which they have proven is cheap enough that consumers will pay for. I believe they also only need it to be level 3 [0] for it to be a massive success (I know a ton of people that would pay the $6,000 for the FSD upgrade if it allowed them to watch a movie/work/text during their commute most of the time). Granted, Tesla's tech has a lot to catch up but I wouldn't count their solution to the profitability problem out since they have a lot of more cars on the road collecting data/testing their algorithms, and I'm guessing they have a solution to the imperfect tech that ends up killing people problem (Selling their own auto insurance which they have already announced plans for and which will probably have special conditions for autopilot failures, and lots of cars running their quasi level 2 tech which they can use to lobby congressmen to limit liability for non-tesla drivers suability during autopilot failures). But, as with all big tech predictions, who knows. [0]: Level 3 ("eyes off"): The driver can safely turn their attention away from the driving tasks, e.g. the driver can text or watch a movie. The vehicle will handle situations that call for an immediate response, like emergency braking. The driver must still be prepared to intervene within some limited time, specified by the manufacturer, when called upon by the vehicle to do so: https://en.wikipedia.org/wiki/Self-driving_car#Levels_of_driving_automation https://en.wikipedia.org/wiki/Self-driving_car#Levels_of_dri...
- nlowell 7y ago
  It doesn't seem like Tesla knows how to make it profitable either, to be honest. Their current plan for profitability appears to be the million robo-taxi fleet by 2020 rather than the auto manufacturing business.
  - leesec 7y ago
    I think it would seem insane for them to present this opportunity to make a ton of money (robotaxis), and not fully commit to it. I mean that they are positioning themselves as if they truly believe this is going to happen.
dominotw 7y ago
>Constantly updating HD maps is expensive. Do any of the players in this space use HD maps?
- anticensor 7y ago
  Waymo.
stale2002 7y ago
> I love the idea of self-driving cars and I really can't wait until they're ready. What if I were to tell you that you can buy a self driving car right now, and that there are 100s of thousands them in consumer hands, right now? Tesla sells them. And they aren't that expensive. If you can press a button, and take your hands off the wheel, that is a self driving car. And you can buy them today.
- tim333 7y ago
  They are still working on the do not kill the driver by crashing into stationary objects feature though it seems.
- AndrewBissell 7y ago
  Sorry, nope. If your car crashes after you press that button and take your hands off the wheel, Tesla will point to the line in their terms of service which says, "keep hands on the wheel, driver must pay attention at all times." Tesla (and Elon Musk in particular) are of course happy to encourage the misconception that they already sell a self-driving car.
  - stale2002 7y ago
    You can call it whatever you want. But at the end of the day, this definitely fits the vast majority of people's definition of "self driving". You press a button, and it stays in lane, at the right speed. Most people are happy with that feature set. That is the MVP, that provides most of the benefits to most people, even if it isn't perfect.
    - oblio 7y ago
      It kills people... it's not ready.
    - cmurf 7y ago
      >But at the end of the day, this definitely fits the vast majority of people's definition of "self driving".> Complete, total, utter, bullshit. It's such bullshit, I have to ask: are you stupid? Because it's that stupid.
Nokinside 7y ago
Thing missing from the discussion is the different safety and threat model fully autonomous cars have compared to human drivers. Mass incidents are one of them. * If one autonomous cars fails catastrophically on some road, all other cars with same model and/or software are likely to do the same. Huge pileup in fast tragic can become nightmare for the car maker/operator. * If there is a software error in the overnight update or planned sabotage, hundreds of thousands of cars can get into accidents in very short time period until the failure is noticed. * Simple loss of trust due to security incident, discovery of exploit, server configuration error, hacking in the network can lead to security groundings of millions of cars at once. If all Toyota's are grounded for a week, the economic impact around the world is huge. Even if the autonomous driving works, designing the security and safety infrastructure that keeps them working will be real challenge. I'm sure it will be worked out but it's not going to happen in few years.
asdff 7y ago
I fear the unavoidable cost is going to ruin the model. The advantages of these things is that they are able to work as a unified network, keeping space, allowing space, preventing pileups from simply driving too close, etc. What about all the 30 year old $2000 corollas you see on the road? Are these people expected to shell out $50k+ to get a self driving car? Absolutely not. There will always be people who can only afford a 10, 20, 30 year old car. What happens when they own a 30 year old self driving car, outdated for decades and based on how tech companies operate today, probably unpatched and unsupported for 20 years? Self driving would really shine as a self driving bus, in a dedicated lane. Single occupant vehicles are irresponsible in cities that are choked with too many of them, but there is no alternative if all you can afford is a $2000 corolla and rail connections to work are nonexistant. But then again, a self driving bus is much less sexy to investors than a self driving car...

maximus1983 7y ago

TBH this is the pareto principle at work. Sure you can get a car roughly driving around the city autonomously (there are basic self driving cars back in the 1950s that drove on a wire track) but getting it perfect is where the vast majority of effort is going to be expended.

joe_the_user 7y ago

Yes, The thing about self-driving you can get to 90% or 99% or even 99.9% success and then find the cost of closing part of that extra sliver of non-success is the same as all the work you've done. Especially since testing those corners cases is hard since they are unusual. There's even some part of the sliver that even is "AI complete". Very small but you don't know where it is, etc. But I'd say that's going to happen about as quickly as changing the infrastructure in this country. Key point! Automation has traditionally been much, much easier when you are engineering a situation where you control all the variables. Automated factories are common. Automated warehouses are coming. Automated dock systems exist. In these situation, there's a hard border between "inside" and "outside". Roads are generally regular but the border between insight and outside fuzzy. Google can run it's self-driving taxis in a suburb in Arizona where roads are as regular as could be imagined but expanding beyond that is much, much hard than one imagines. Note: I recommend this blog which periodically tracks progress in self-driving cars as well as making intelligent, critical comments on current AI. https://blog.piekniewski.info/ https://blog.piekniewski.info/

wpietri 7y ago
Good points. And I'd add that most things we've automated are things that have been routinized in a way where people conform to the system, rather than the other way around. E.g., a ton of software development has basically been, "take this paper form and make a computer do the paperwork". The hard part was already done for us when everybody got trained to be subservient to the bureaucratic process. So societally we have this limited sense of how quickly computers can take something over. Real-world things like cars and domestic robots are much harder, because those domains are ones where people come first. As you say, key elements could well be AI-complete.
secabeen 7y ago
Even when you control all of the variables, there are plenty of situations where the automated system just runs out of runway, and kicks the can to a human. Can't read the label on this package? Push it to the exception lane, and have a human figure it out, and relabel it. Works great, but is useless for a car, because when the car hits the exception, it's 10 miles away on a residential street, and it's not going to make it to you anymore. For me, the most important thing a car can be is reliable, and every form of unattended driving cannot hit that level.
- pas 7y ago
  Sometimes the taxi you called never arrives. It happens regularly. Of course the real problem is economic. Competing with the taxi service that uses humans is hard, because human labor is cheap.

wpietri 7y ago

The amazing thing to me is that I don't anybody who thought it was coming in the next couple years who based their estimate on anything but hope and magic. Of course, that's nothing new. We've all been on software projects where the launch date was the earliest day nobody could prove it was impossible to finish. But I'd like to think that investors dropping zillions on companies would have had better due diligence.

joe_the_user 7y ago
The amazing thing to me is that I don't [know] anybody who thought it was coming in the next couple years who based their estimate on anything but hope and magic. I'd agree that a bit of the proper sort of analysis would show that this wasn't easily happening. But I think the kind of thinking that imagines this as possible is extremely easy to fall into. We've all been on software projects where the launch date was the earliest day nobody could prove it was impossible to finish. Sure and I'm pretty sure a lot of software engineers have given or agreed to completely unrealistic estimates after having had this experience. If anything, the "software crisis", the inability of software engineers to provide good estimates even after hard experience and warnings is an expression that over-optimistic thinking is natural for human beings in this situation. My theory is that in situations where a human is asked about the difficulty of a mental task, they essentially can only fall back on a basic language-expression of said task and rationally try to calculate with that - the brilliance and the limitation of we language-generating, language-using creature is the ability to correctly distill a complex and messy task into fairly simple sequence of symbols. But that correctness actually exists in a complex and dynamic context, most of which we screen out. A command like "Drive to the store" doesn't take into account all the compensations needed to do the driving just as "Create an inventory system" fails to specify all the client-specific "gotchas" involved in such a task. So going from our normal language-distillation process to a complete software system that doesn't have the continual dynamic compensation of our intelligence is an inherently difficult problem - even though we can probably only do it at all because of our "language facility".
- wpietri 7y ago
  > the inability of software engineers to provide good estimates I don't think it's inability. I think they're embedded in a system that doesn't encourage or reward honesty or accuracy. I agree it's easy for people to say, "Gosh, how hard can that be?" Which is where we get the classic, "I could build that in a week" estimate. But I don't agree that there's something inevitable about people going from a finger-in-the-air SWAG to a "we'll have self-driving cars in 2021" business plan. Normally I'd say that it's a straight up failure of both management techniques and the professional standards of the engineers. Which is true. But I think the problem is that a lot of fundraising in the "Uber for X" era is only slightly less of a con than Theranos. So I think it's more correctly seen as a failure of founders and VCs.

BurningFrog 7y ago

That sounds like one big company doesn't know how to do it faster by keeping doing the same things they're doing now.

vivekd 7y ago

>people you need to actually talk to and help customers - something Google has basically never done before As a google advertiser this made me laugh. At the same time I don't think expense will be a setback for autonomous vehicles. The reason being that if vehicles are fully autonomous there would be no reason for any individuals to own them. It would make more sense to have them operate like a taxi service. This is because without the costs of having to pay for a driver, it would be cheaper to just order an autonomous vehicle whenever you need to go anywhere than it would be own your own vehicle. Just from an efficiency standpoint it is terribly inefficient for each person to have a car. We're not driving most of the time and the only reason we have our car sitting in the driveway is we need it to be available should we need to drive. With autonomous cars, my car could on it's own drive across town to pick up other people when I'm not using it. The reason we can't do that today is because of the additional logistics and costs involved in transporting my unused car from my driveway to the driveway of the person across town who needs it. This problem would not exist in a world with autonomous vehicles. To me, it seems likely that a world with autonomous cars would be a world where the majority of vehicles in the world are owned by services like uber.

YeGoblynQueenne 7y ago
>> At the same time I don't think expense will be a setback for autonomous vehicles. The reason being that if vehicles are fully autonomous there would be no reason for any individuals to own them. It would make more sense to have them operate like a taxi service. This is because without the costs of having to pay for a driver, it would be cheaper to just order an autonomous vehicle whenever you need to go anywhere than it would be own your own vehicle. I have a better idea that we can implement right now without waiting for autonomous driving to become a real technological capability. We could make big cars with many seats and have them driven around town by humans- just a single human could drive a single vehicle capable of transporting 50, maybe 100 others. We could have those big cars stopping at predetermined points around the city, so that people would know where to get on and off. And we could support that service by asking the users to pay a small fee upon boarding the big car. This would cover the costs of the driver, and the big car, and still leave some to pay for infrastructure, etc. These big cars would probably have relatively restricted areas of operation, but, for the foreseeable future, so will autonomous cars (which have to be geofenced) and, like I say, we already have the technology for those big-cars-with-many-people-sitting-in-them-and-a-single-driver that I'm talking about. And since a whole bunch of people would be carried by one or two of those big cars per day, instead of each riding a single (autonomous or not) car, we could drastically reduce the number of automobiles-per-person, and, therefore, the CO2 emissions, thereby reducing pollution and protecting the environment to boot. So? Anyone want to invest on that?

StephenMelon 7y ago

Seems like autonomous personal air travel would be a far, far simpler problem to solve than road-based driving? Far less edge cases, once you are in the air the only obstacles you really need to worry about are other vehicles, birds, leaves and the odd stray plastic bag. With VTOL the whole thing could be an order of magnitude easier than driving. By the same principle, sea travel would also have less edge cases, at least in calm conditions. Is it the case that people aren’t seeing markets for those applications of autonomy or is there some other reason why there isn’t the same hype in those areas?

gshdg 7y ago
Vehicle cost; fuel cost; takeoff and landing; handling of edge cases (there's a reason that commercial airlines still have pilots despite autopilot being very effective in 90% of conditions).
- dominotw 7y ago
  how expensive would it be creating areas of road that only allows autonoums vehicles, something like carpool lanes we have now. Wouldn't that cut problem size massively.
  - frosted-flakes 7y ago
    I can see that happening on major freeways, the way we have carpool lanes now, but not on a general scale.
cmurf 7y ago
A: Air travel generally is a far simpler problem to solve, because there's national standardization of the air navigation system, the airport paint and signage, and very clear rules of separation with very few edge cases. And yet we do not have anything close to self-flying planes no matter how much money you throw at it. And we're not even close to getting there. Whereas with cars, none of that is true, municipalities get wedged into violating their own equivalent of human interface guidelines all the time. Worn paint and signs for years, busted cross walk signals, confusing intersections, human driven cars that consistently do not follow the rules but in inconsistent ways. B. Personal air travel has a significant regulatory burden in the transition from ground to air. The ground is city + state regulated. And immediately once airborne it's FAA regulated, but not under any kind of Air Traffic Control, as almost all airspace below 1200' above ground is uncontrolled. So ATC has nothing to say about it, and no central mechanism for negotiating conflicts. Where are you allowed to takeoff and land is easily figured out today: airports only. And that's because it takes all kinds of things into account like obstruction clearance, noise abatement, anticipating engine failures and crashes. We can't even automate commercial flights - it might seem like it's more correct to say we don't automate commercial flights, but we could. That's not really true. The amount of changes to the air traffic control system, and on-board equipment for airplanes, is presently so cost prohibitive that it is effectively a "cannot be done".

mrep 7y ago

> but how to do it profitably. The cars are expensive. The maintenance is expensive. The ongoing cleaning, charging, and prepping of vehicles is expensive. Constantly updating HD maps is expensive. Of course the engineers to keep everything running smoothly are expensive. And the small army of people you need to actually talk to and help customer I believe that only applies to the waymo model where you go with the full time robo-taxi using top of the line hardware like lidar and which pretty much requires level 5. The tesla model is to use cheaper hardware which they have proven is cheap enough that consumers will pay for. I believe they also only need it to be level 3 [0] for it to be a massive success (I know a ton of people that would pay the $6,000 for the FSD upgrade if it allowed them to watch a movie/work/text during their commute most of the time). Granted, Tesla's tech has a lot to catch up but I wouldn't count their solution to the profitability problem out since they have a lot of more cars on the road collecting data/testing their algorithms, and I'm guessing they have a solution to the imperfect tech that ends up killing people problem (Selling their own auto insurance which they have already announced plans for and which will probably have special conditions for autopilot failures, and lots of cars running their quasi level 2 tech which they can use to lobby congressmen to limit liability for non-tesla drivers suability during autopilot failures). But, as with all big tech predictions, who knows. [0]: Level 3 ("eyes off"): The driver can safely turn their attention away from the driving tasks, e.g. the driver can text or watch a movie. The vehicle will handle situations that call for an immediate response, like emergency braking. The driver must still be prepared to intervene within some limited time, specified by the manufacturer, when called upon by the vehicle to do so: https://en.wikipedia.org/wiki/Self-driving_car#Levels_of_driving_automation https://en.wikipedia.org/wiki/Self-driving_car#Levels_of_dri...

nlowell 7y ago
It doesn't seem like Tesla knows how to make it profitable either, to be honest. Their current plan for profitability appears to be the million robo-taxi fleet by 2020 rather than the auto manufacturing business.
- leesec 7y ago
  I think it would seem insane for them to present this opportunity to make a ton of money (robotaxis), and not fully commit to it. I mean that they are positioning themselves as if they truly believe this is going to happen.

dominotw 7y ago

>Constantly updating HD maps is expensive. Do any of the players in this space use HD maps?

anticensor 7y ago
Waymo.

stale2002 7y ago

> I love the idea of self-driving cars and I really can't wait until they're ready. What if I were to tell you that you can buy a self driving car right now, and that there are 100s of thousands them in consumer hands, right now? Tesla sells them. And they aren't that expensive. If you can press a button, and take your hands off the wheel, that is a self driving car. And you can buy them today.

tim333 7y ago
They are still working on the do not kill the driver by crashing into stationary objects feature though it seems.
AndrewBissell 7y ago
Sorry, nope. If your car crashes after you press that button and take your hands off the wheel, Tesla will point to the line in their terms of service which says, "keep hands on the wheel, driver must pay attention at all times." Tesla (and Elon Musk in particular) are of course happy to encourage the misconception that they already sell a self-driving car.
- stale2002 7y ago
  You can call it whatever you want. But at the end of the day, this definitely fits the vast majority of people's definition of "self driving". You press a button, and it stays in lane, at the right speed. Most people are happy with that feature set. That is the MVP, that provides most of the benefits to most people, even if it isn't perfect.
  - oblio 7y ago
    It kills people... it's not ready.
  - cmurf 7y ago
    >But at the end of the day, this definitely fits the vast majority of people's definition of "self driving".> Complete, total, utter, bullshit. It's such bullshit, I have to ask: are you stupid? Because it's that stupid.

Nokinside 7y ago

Thing missing from the discussion is the different safety and threat model fully autonomous cars have compared to human drivers. Mass incidents are one of them. * If one autonomous cars fails catastrophically on some road, all other cars with same model and/or software are likely to do the same. Huge pileup in fast tragic can become nightmare for the car maker/operator. * If there is a software error in the overnight update or planned sabotage, hundreds of thousands of cars can get into accidents in very short time period until the failure is noticed. * Simple loss of trust due to security incident, discovery of exploit, server configuration error, hacking in the network can lead to security groundings of millions of cars at once. If all Toyota's are grounded for a week, the economic impact around the world is huge. Even if the autonomous driving works, designing the security and safety infrastructure that keeps them working will be real challenge. I'm sure it will be worked out but it's not going to happen in few years.

asdff 7y ago

I fear the unavoidable cost is going to ruin the model. The advantages of these things is that they are able to work as a unified network, keeping space, allowing space, preventing pileups from simply driving too close, etc. What about all the 30 year old $2000 corollas you see on the road? Are these people expected to shell out $50k+ to get a self driving car? Absolutely not. There will always be people who can only afford a 10, 20, 30 year old car. What happens when they own a 30 year old self driving car, outdated for decades and based on how tech companies operate today, probably unpatched and unsupported for 20 years? Self driving would really shine as a self driving bus, in a dedicated lane. Single occupant vehicles are irresponsible in cities that are choked with too many of them, but there is no alternative if all you can afford is a $2000 corolla and rail connections to work are nonexistant. But then again, a self driving bus is much less sexy to investors than a self driving car...