Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
How Soon Will Self-driving Cars Be Everywhere?
As part of an excellent presentation and recent podcast, analyst Frank Chen of VC firm Andreessen Horowitz takes a crack at answering the question above. First, this from an audio presentation with slide deck, “16 questions about self-driving cars”:
So when will this beautiful world happen? … NuTonomy says 2018 they’re live in Singapore, top ten cities by 2020. Delphi and Mobileye say they’ll have self-driving systems available to the car manufacturers by 2019. … 2020 GM says that’s when it will have its cars. Ford says 2021 they’ll have Level 5 cars available to fleet makers. BMW ships the iNEXT in 2021. Tesla who’s arguably out ahead of this right now says 2023. Uber says that its entire fleet will be autonomous by 2030. And IEEE says that by 2040, 40% of all care on the road are autonomous. So you see quite a range of predictions on when this glorious future happens, and then once it starts we don’t know what the demand curve will be…. But look, it’s going to happen in our lifetime, which is probably not something I would have predicted ten years ago.
As to that last point: How fast will driverless cars spread once they are first on the road? Here on the podcast is Chen, Carl Pope (former executive director and chairman of the Sierra Club and author — with former NYC mayor Mike Bloomberg — of the upcoming book Climate of Hope: How Cities Businesses and Citizens Can Save the Planet), and a16z’s Sonal Chokshi:
Sonal Chokshi: How fast do you guys think that will happen? What’s the rough timeline?
Frank Chen: Well, if you look at public estimates given by the auto manufacturers. I’m talking about incumbents now. You’ll see a range, anywhere from 2019 to 2024 being the year where they first introduce their autonomous vehicle. It might happen even earlier for fleet sales. The most aggressive auto manufacturers are saying 2019 is when this revolution starts and then we have a question of what’s the adoption curve. Will it be S-shaped like it was for the iPhones? Will it be straight up to the right because it’s just so awesome? And obviously that has implications of which companies are going to win, but it’s going to start soon.
Carl Pope: I think one of the things that’s important is if we go to shared vehicles … every shared vehicle replaces roughly eight other vehicles. Even today, with today’s technology and today’s ownership patterns, if you only have an eighth as many cars on the road but people are still driving as much, more or less, that means every car drives eight times as far a year. Now we turn over the automobile fleet every thirteen years and most of the projections people are making about how fast this happens assumes that thirteen-year turnover rate. … If you only have an eighth as many cars and they’re driven eight times as far and they’re retired, let’s say, five times as quickly, the whole automotive fleet is turned over in three years. So it might start a little later than the optimists think, might start in 2022, but it might be over by 2025.
Chockshi: Why might it be over?
Pope: Because in three years the whole fleet might have rolled over and once you start selling autonomous cars in large numbers people will not buy cars that have to be driven. Being able to drive your own car is going to become a very expensive option.
Hopefully I correctly matched all the voices to the right person. If you have an interest in the subject, I urge you to listen to the entire episode. And you might want to also check out this congressional hearing from today on driverless car regulation.
Published in Culture, Technology
My argument is that driverless cars will make the economics of Uber accessible and affordable all the time. Most people can’t afford to just take a taxi everywhere. Now they could. Imagine the cost of a bus ride but with the comfort of a luxury car? For going anywhere! It would be amazing.
Think how often humans get into fatal accidents for lack of carelessness? Even if it is no more safe than human driving if it is cheaper and more convenient it will be a big improvement.
Fair points, but there are people for whom this will mean they don’t buy a car. Even if it is only 10% of the population that is a significant decrease in car density. The other possibility is not that car density goes down, but that street usage becomes more efficient. If your car can go park itself why would you need lots of street parking which takes up a lot of space. You can have a few large parking structures that cars go to and come out. Also driverless cars might be able to use up less space on a street because of faster reaction time and better coordination. Imagine if you could safely drive a car with less space between cars at higher speeds? For highways this will make a difference. Plus you won’t have rubber necking, or inefficient lane usage. How much traffic is snarled up because some idiot doesn’t know what lane to be in to get to the right exit? It may be small stuff but it adds up. Also peoples behaviors can change.
Not to be too much of a Luddite, but self-driving cars provide too much of an opportunity for progressives and big-government types. There will inevitably be a move to use government to control where and when people can “drive” to different locations – all in the name of safety and efficiency. Only x number of people should be allowed to visit the Grand Canyon any given day; therefore, sorry, your self-driving car will not be authorized to go to the Grand Canyon. A pro-life march is being planned for Washington DC – oh, sorry, the number of vehicles allowed into DC on that day is being limited. The number of vehicles on the road need to be reduced – well, why should any one person have a vehicle that they take somewhere and then park; why not have all vehicles “shared” and have a computerized driving plan worked out so the vehicle can drive many people around all day (and then pick up people it dropped off at work and return them to their homes). And on it will go, into the brave new future! [I can’t do the link, but I’m reminded of the taxi cabs in the Arnold Schwarzenegger version of Total Recall.]
Anyone who hates the idea of driverless cars should drive around Houston for two weeks and then tell me what you think of the idea. 2998 people died in car crashes in Texas in 2010. The number is higher for 2016, but I can’t find it at the moment. I understand that there are challenges, but I simply do not buy that the costs outweigh the benefits.
What’s going to be neat is when your car drops you off for work, it will then pick up passengers and make money for you while you’re at the office. That’s my dream anyway.
meh. Let’s keep it that way. Actually, there is a point at which – call me old-fashioned if you will – increases in technology become more harmful than helpful. I’m not saying it’s that way with driverless cars, though I might suspect that it is… but look at the way Twitter and Facebook have changed our political discourse for the worse. Maybe it’s just growing pains, but there is also a bit of the “tower of babel” element to things that I cannot shake off simply because progress is inevitably a good thing. Looking at history, much of progress is indeed a good thing. But I wouldn’t say that all of it is. So as for all those benefits of driverless cars: Thanks, but No Thanks.
All supervised by a human occupant who can take control. And not in the rain or snow, I would bet. And not when there’s construction, nor police directing traffic around a sporting event or concert.
IOW not autonomous.
He misses the most important social question: what is going to happen to all those people who drive for a living?
Everything else about self-driving cars is wonderful, but this one aspect is a potential nightmare.
Cheaper, how is it going to be cheaper? More convenient, like that airplane you board only to wait for some mechanic? Who is going to liable for injury and death, somebody with deep pockets? Are you going to put your family in a self driving car knowing that the chance of injury and death is any percent of the current rate? Will anybody trust sharing the roads with self driving trunks which might fail to proceed safely at any time?
There is no way with the complexity and reliability problems of a self driving vehicle they can be delivered to the masses at any acceptable cost.
Has any reader every driven to Toronto, Canada, at night in the winter? The snow builds up along the curbs so that the roads become narrower as the winter goes on. The cyclists still are out. Then there are the pedestrians. It is a hair raising combination which share the changing lanes and winter conditions of Toronto streets very uneasily.
What is the default mode of a self driving vehicle? It turns the controls of the vehicle back to the human driver instantly if it cannot compute. I hope the human driver is not texting, sleeping, reading their kindle, making out, eating, watching TV, or sightseeing.
A good design would be fail-function, fail-safe. That means after experiencing a failure, it would revert to a backup system that still performs. A second failure would revert to a safe mode that doesn’t necessarily perform the desired function, but shuts down safely.
An example using a human driver would be driving in heavy rain. Once your windshield fills up with water and you can’t see, you have a visibility failure and can no longer drive safely. But you revert to another mode, which is to turn on the windshield wipers, allowing you to continue driving. Now if your windshield wipers fail, you wouldn’t just keep driving. You would fail-safe, which means pull over and stop until you can safely see again.
Any reasonable autonomous vehicle design would use a similar logic structure. Suddenly surrendering control to an unsuspecting human driver is not an option; it is unsafe and an example of failing badly.
I’m an out of state person who has driven around Texas and Houston and I have to say that a number of the drivers you have are a bit nuts and reckless. Plus the parallel roads that feed the interstates and highways can be hazardous. I’ve seen near collisions where a guy exits the highway at 75 and then starts changing lanes to the right while somebody is just pulling out of a plaza parking lot onto the parallel road. I’m not going to get deep into other things, but yes it can be a real pain and one needs to keep an eye on the side view mirrors for the guy speeding Z patterns through traffic like he’s Rickey Bobby.
That’s because Pittsburgh is a testing place for autonomous vehicles and robotics in general. Carnegie Mellon is known for science and engineering with a history of robotics. I think that’s part of the reason why Uber moved a testing and research facility to Pittsburgh to take advantage of the Universities an existing technical expertise. Plus Pgh has a history of innovation. And a good place to test vehicles on disordered roads. If a self driving car can work in Pgh, then it’ll work most places.
The human just sits there but you aren’t allowed to just read a book or take a nap. You have to be paying attention. Rain or snow is irrelevant and we get plenty of both. We also have tons of construction, again, irrelevant. Police directing traffic isn’t an issue.
Can you elaborate on “irrelevant” and “isn’t an issue”?
Everything I have seen has indicated that the market is predicted to move away from individually owned vehicles. You order a car like a pizza and it shows up at your door on demand. When you need a ride to work, an individual “car” shows up. When you go on vacation with the family, you order are car for your family size and luggage needs etc. No insurance. No gassing up, checking the tire pressure, oil etc. Cars will mostly be much smaller and lighter and more efficient. Even freight carrying models won’t be that big because of how efficient they are.
Areas that are currently road right of ways will be able to contain separate right of ways for vehicles as well as pedestrians and bikes because cart widths will be so much smaller. Arteries for specified travel would be slightly larger with limited pedestrian access and many roads would be completely cut off and no longer used by vehicles.
Like I said, I used to be very skeptical myself and all the discussions I see and hear are still HIGHLY preliminary but this could actually be worth trying and it could work.
Yes. That is what I meant. There are sensors of every type that allows the vehicle to obey several levels of commands. The only issue they have now have to deal with hacking. If someone jumps in front of a car, they will be struck but frankly, this is the case for vehicles now and the reaction time and performance of these cars is vastly superior. One other issue is that evasive maneuvers can be very uncomfortable for passengers in emergency situations but again, no different than regular vehicles.
I have yet to ride in one of these vehicles. I live in a neighborhood where you see them driving down the street every other minute. From all accounts they handle like normal vehicles. These UBER cars are normal cars however with sensors attached all over the vehicle. These current cars are like Model T’s compared to what such vehicles will be like in 15-20 years.
Sure that’s a wonderful theory. How much will it cost? How reliable will it be? Can the average car mechanic repair that system? (my mechanics have trouble enough with the present level of sensors and smart systems in cars) The cost of automobiles today is high enough to put them out of reach of many families. How much will all this redundant smart system add to that cost? Self driving vehicles are a fantasy that will run up against realville and join flying cars among the oddities of futurism.
Honestly not trying to be condescending here: Is this your opinion as an engineer? That they have literally solved all problems related to poor visibility in precipitation, poor road conditions, and irregular traffic routing conditions? Color me very skeptical.
All of this is true. I especially agree with the bolded sentence. The requirements for a “car mechanic” will move up from a tech certification to a four year degree in engineering. Or there will have to be a multi-tier support system like with computers. Second level support and troubleshooting will be very expensive.
Yeah, I don’t see it. Do you have kids? When the kids leave school, and have to go to band practice, or Tae Kwon Do, or swimming, what do they do with their gear if they can’t leave it in the car? Schlep it around with them?
Ever travel with a baby? You want to take all the stuff that comes with a baby and carry it to/from the car (and with you) every time you get out?
If you don’t have to drive, you’re going to be carrying that much more entertainment equipment with you – tablets, video players, books, or whatever, along with the charging cables. Can’t leave ’em in the car anymore…
What will gradually come into focus over the next few years is that there are two fundamentally incompatible types of self-driving cars, those that can coexist with human driven cars, and those that can’t. The latter, it will become apparent, will work vastly better than the former, because all the other similarly programmed vehicles on the road will be able to cooperate in ways (and at speeds) that human drivers can’t. This will exert enormous pressure to ban all human driving on all roads.
But this will not be market pressure, because markets don’t produce all-or-nothing outcomes, let alone sudden changes from one all-or-nothing situation to its opposite. This will be a purely political policy that will initiate a new epoch of transportation, imposed by force of law on the 49% who still prefer to drive.
Here, here! I enjoy driving most of the time as well.
Assuming I can still drive my own car on a track I suppose I’d be cool with letting the car drive me to work. But I love driving.
I wonder what @danhanson has to say about this.
I’m just summarizing various news reports I’ve seen (this is big local news because PGH is a test case) and supplementing with presentations that I have heard from the folks that are involved with the local effort.
My gut tells me there is a lot of hype and forward-looking analysis, or shall we say rosy predictions.
These seem like silly objections to me but I don’t have kids.
You clearly don’t have kids! There’s a lot more “stuff” involved in transporting children than the appropriate car seat.
As to the musical instrument/duffel bag question: Kid goes straight from school to Tae Kwon Do. We leave his gear in the car so he doesn’t have to haul it around all day. He’s got his backpack for school. He’s got the gear for TKD (uniform, sparring gear/helmet, etc). On days when he has to bring his French Horn home to practice, he’s supposed to haul all of that with him all day?
You don’t ever leave personal items in your car? I’d personally find it very inconvenient to take the equivalent of a taxi everywhere.
I don’t expect that bolded statement could be supported with evidence. You might not have heard about them in the news, but it doesn’t sound like you’ve had access to the engineering data and were able to see what problems they’ve detected and are trying to solve.
Here’s a recent example of one of the major outstanding issues, and what I believe are rosy predictions about how it will be solved.