Autonomous Car Progress

[EVNow]

My kids are 17 and 19, so that is no longer a concern for me, but I don't see how you couldn't have minivans that already have 1, 2, or 3 car seats in them ready to go. In the app you could choose the vehicle of your needs. During peak times of course your wait might be a little longer, but not unreasonable.

When do you think this will happen in all top 100 cities in US ?

ps : I still don't see how I can drop off my kids to school / pre-school in time using such a service. Extremely time sensitive - given the kids get ready just in time (mostly). So, it would be quite a bit of change in the way one lives - not just the way someone uses transportation.

 

[rxlawdude]

At IAA Mobility 2021, BMW demo'ed their automated valet parking project. There is also automated charging and car wash.

When it actually is used in production vehicles, you'll have a point.

 

[daniel]

True, but the problem can be somewhat mitigated by multiple seats. For example using vans. During peak 6 passengers in the car. Non peak: one passenger.

Vans are more expensive than smaller cars. Carrying one person in a huge van is cost-inefficient. Efficiency requires that the tool be right for the job.

It would cost more than individual cars?

Having a fleet of 100 cars costs more than having a fleet of 50 cars. If the peak demand for 100 cars only lasts for an hour a day and the rest of the time 50 cars is more than adequate, nobody will invest in a fleet of 100 cars. For an hour a day there will be long waits. Some people will accept that. Others will keep their private cars.

My kids are 17 and 19, so that is no longer a concern for me, but I don't see how you couldn't have minivans that already have 1, 2, or 3 car seats in them ready to go. In the app you could choose the vehicle of your needs. During peak times of course your wait might be a little longer, but not unreasonable.

The supply:demand equation is similar for a fleet of robotaxis as it is for a fleet of taxis or ride-shares: There are times and places when the demand is so great that you cannot get a ride for love or money. Robotaxis will not change that because nobody is going to invest in a fleet of robotaxis large enough to meet peak demand. People can dream of a utopian future when there's always a car available when you need it, but that won't happen because the car costs money and somebody has to put up that investment and they won't invest without hope of profit, and if there are that many cars competing for riders none of them will make enough money to operate.

Some people use ride-shares now because for them the inconvenience is worth the savings. Other people have private cars because for them, the cost of a car is worth the convenience. This will not change with robotaxis. Yes, there will be customers. But a lot of people will keep their private cars.

As for robotaxis making the roads safer, autonomous cars will make the roads safer, whether they are robotaxis or private self-driving cars.

 

[powertoold]

Having a fleet of 100 cars costs more than having a fleet of 50 cars. If the peak demand for 100 cars only lasts for an hour a day and the rest of the time 50 cars is more than adequate, nobody will invest in a fleet of 100 cars. For an hour a day there will be long waits. Some people will accept that. Others will keep their private cars.

Ubiquitous robotaxis will be priced such that profit is made. So I wouldn't worry about no one investing in a fleet to address peak hours.

The main point of discussion was whether paying for a robotaxi service that can handle peak hours would be as expensive as owning your own individual car. Based on current ride-sharing model, we know people are willing to pay a lot for rides, and robotaxis (especially electric) will reduce the cost many fold (even after considering the fleet to address peak hours).

Also, much of the traffic caused during rush "hour" (which is actually 2-3 hours) is caused by variability in human driving: people cutting in and out of lanes, some driving too fast / too slow, poor merging, etc.

 

[EVNow]

Ubiquitous robotaxis will be priced such that profit is made. So I wouldn't worry about no one investing in a fleet to address peak hours.

You should - because all the many examples we have now shows that people don't invest for peak consumption - but avg + some extra.

Take parking lots, for eg. You get the point ...

ps : Urban living is where taxis make sense compared to private cars. Parking spaces are too expensive, traveling distances are small, people are generally healthier etc. Suburban living and taxis don't mix. What could make sense is families keeping one car instead of two or more. Esp., if working from home because a permanent thing rather than a covid blip.

 

[powertoold]

You should - because all the many examples we have now shows that people don't invest for peak consumption - but avg + some extra.

Take parking lots, for eg. You get the point ...

ps : Urban living is where taxis make sense compared to private cars. Parking spaces are too expensive, traveling distances are small, people are generally healthier etc. Suburban living and taxis don't mix. What could make sense is families keeping one car instead of two or more. Esp., if working from home because a permanent thing rather than a covid blip.

This discussion was spawned when Microterf said that we underestimate the disruption of robotaxis. Sure, there will be people who will find individual ownership more convenient and economical (because of their needs), but a ubiquitous and great robotaxi service would suit a large majority of needs.

In cities where there is a great public transit system, people own much fewer cars. A great and reliable robotaxi service would be even better than a great public transit system (my guess).

Tokyo population = 14 million
Cars registered = 3 million (probably many of these are for business-related hauling)

Japan: number of motor vehicles in use in Tokyo 2021 | Statista

As of March 31, 2021, the number of registered motor vehicles in Tokyo, Japan, reached a decade low of approximately 3.09 million vehicles, down from around 3.22 million vehicles in 2012.

www.statista.com

 

[EVNow]

This discussion was spawned when Microterf said that we underestimate the disruption of robotaxis. Sure, there will be people who will find individual ownership more convenient and economical (because of their needs), but a ubiquitous and great robotaxi service would suit a large majority of needs.

In cities where there is a great public transit system, people own much fewer cars. A great and reliable robotaxi service would be even better than a great public transit system (my guess).

Tokyo population = 14 million
Cars registered = 3 million

Japan: number of motor vehicles in use in Tokyo 2021 | Statista

As of March 31, 2021, the number of registered motor vehicles in Tokyo, Japan, reached a decade low of approximately 3.09 million vehicles, down from around 3.22 million vehicles in 2012.

www.statista.com

As part of solution to climate change I've been discussing this issue for over 2 decades.

The basic issue in US is the sub-urban sprawl. That is where public transport (which includes train, buses, taxis, robotaxis) have issues.

ps : Growng up in India hardly anyone had cars - so I know how life is without cars.

 

[powertoold]

The basic issue in US is the sub-urban sprawl. That is where public transport (which includes train, buses, taxis, robotaxis) have issues.

Have you considered that with ubiquitous robotaxis, software can aid with many of the issues we're discussing? For example, in your robotaxi app, you can say, "I need to get to work by 8am every day and willing to ride with up to 2 others." The software can figure out your location and group you with others on your same block or area?

Plus, when robotaxis are solved, we might not even see the same vehicle form factors anymore. The robotaxis could be a van with individual separate pods, so you don't have to deal with smelly humans

 

[EVNow]

Have you considered that with ubiquitous robotaxis, software can aid with many of the issues we're discussing? For example, in your robotaxi app, you can say, "I need to get to work by 8am every day and willing to ride with up to 2 others." The software can figure out your location and group you with others on your same block or area?

Plus, when robotaxis are solved, we might not even see the same vehicle form factors anymore. The robotaxis could be a van with individual separate pods, so you don't have to deal with smelly humans

You know apps don't solve everything - have you considered that ?

For eg., can I say - "I need to drop off my kid in pre-school between 8:45 AM and 8:55 AM and my other kid needs to catch the bus that comes sometime between 8:40 and 8:50 AM. If we miss the school bus I need to drop off the other kid in school by 9:00 AM" ?.

ps : I plan never to go back to work in offices. I'm going to WFH permanently

 

[powertoold]

For eg., can I say - "I need to drop off my kid in pre-school between 8:45 AM and 8:55 AM and my other kid needs to catch the bus that comes sometime between 8:40 and 8:50 AM. If we miss the school bus I need to drop off the other kid in school by 9:00 AM" ?.

Yup, that's why I said that robotaxis won't do much disruption (in terms of displacing cars) until they are ubiquitous and have less than 5 minute wait times, all of which are possible once fsd is "solved."

As for the business case, it's clear that cars are mostly idle over their lifetimes, so even factoring peak times (2-3 hours a day), using a robotaxi service would still be more economical than owning individual cars.

Let's say rush hour is actually 2-3 hours and most commutes are 30 minutes, then over the span of rush hours, the robotaxi can make 4-6 trips. You add centralized trip planning with software, you might get more than 6 trips.

Alsooo, your kids won't need to take the school bus anymore!

 

[nativewolf]

Yup, that's why I said that robotaxis won't do much disruption (in terms of displacing cars) until they are ubiquitous and have less than 5 minute wait times, all of which are possible once fsd is "solved."

As for the business case, it's clear that cars are mostly idle over their lifetimes, so even factoring peak times (2-3 hours a day), using a robotaxi service would still be more economical than owning individual cars.

Let's say rush hour is actually 2-3 hours and most commutes are 30 minutes, then over the span of rush hours, the robotaxi can make 4-6 trips. You add centralized trip planning with software, you might get more than 6 trips.

Alsooo, your kids won't need to take the school bus anymore!

I've written before on the use case and I think many underestimate the untapped demand for transportation, both urban and rural.

Take the elderly. Many older folks are just not comfortable driving, vision is failing, reaction times are slow, etc. They know this and work to minimize drives. People with substance abuse are another large group of untapped riders. Handicapped...oh...most people have no idea how horrible it is to be handicapped and older in the USA and not really better in Japan or Europe. Robo wheelchair accessible vehicles, a dream for many. Anyhow, on and on. I think transporation gets completely disrupted and as the marginal cost of solar gets to 0 we'll see the marginal costs of robotaxis become very very low in daylight hours. Going to encourage lots of older people to have lunch together, which they would like to do anyway. So many ways this could be helpful. With increased telecommuting we could see whole new transportation patterns emerge.

 

[JHCCAZ]

I've written before on the use case and I think many underestimate the untapped demand for transportation, both urban and rural.

Take the elderly. Many older folks are just not comfortable driving, vision is failing, reaction times are slow, etc. They know this and work to minimize drives. People with substance abuse are another large group of untapped riders. Handicapped...oh...most people have no idea how horrible it is to be handicapped and older in the USA and not really better in Japan or Europe. Robo wheelchair accessible vehicles, a dream for many. Anyhow, on and on. I think transporation gets completely disrupted and as the marginal cost of solar gets to 0 we'll see the marginal costs of robotaxis become very very low in daylight hours. Going to encourage lots of older people to have lunch together, which they would like to do anyway. So many ways this could be helpful. With increased telecommuting we could see whole new transportation patterns emerge.

I am in the category that will need to reduce or completely stop driving in the foreseeable future.
(You could call me "neo-elderly" )
Therefore I have a particular interest in eventual L4 capability, not just really good L2 that many here say would be perfect for them. I've always loved driving but I have to be a realist.

I'm afraid I'm less enthusiastic than you are about the Robotaxi Dream.

Like others have said, I would much rather have my own car available on a whim, and (after I buy it) for the cheapest per-mile operating cost, and no need to clean everything out for the next person or to worry about what the last person left for me. Call me anti-social, but for whatever strange reasons I'm not particularly charmed by the thought of time-sharing the car (mine, or any of Elon's or Sergey's fleet) with incontinent elderly people, meth-heads / other assorted drug addicts, bar-hoppers etc.

Perhaps when I get old enough to not know the difference, I would indeed look forward to gumming my lunch with my fellow carpoolers from the House-of-Ramps. But (to steal a line from Gladiator) Not yet... not yet.

 

[diplomat33]

This is a great discussion between Anguelov (Head of Research at Waymo) and Alex Wang (CEO of Scale AI) on perception, and autonomous driving.

 

[daniel]

Ubiquitous robotaxis will be priced such that profit is made. So I wouldn't worry about no one investing in a fleet to address peak hours.

So the cost of manufacturing an L5 car will just automatically fall to whatever level allows an unlimited number of cars to be sold that will sit idle for 23 hours a day so they're available for everybody who wants one at peak demand times???

The main point of discussion was whether paying for a robotaxi service that can handle peak hours would be as expensive as owning your own individual car. Based on current ride-sharing model, we know people are willing to pay a lot for rides, and robotaxis (especially electric) will reduce the cost many fold (even after considering the fleet to address peak hours).

If people made their car-buying decisions on transportation cost alone, everybody would drive a Honda Civic or a Toyota Corolla. There would be no BMWs or Mercedes or Porsches, etc. etc. People buy cars for far more than to have the cheapest possible transportation. In this case, convenience will keep people owning cars.

Some people don't own cars, and some people do. This will not change just because robotaxis replace Uber.

Also, much of the traffic caused during rush "hour" (which is actually 2-3 hours) is caused by variability in human driving: people cutting in and out of lanes, some driving too fast / too slow, poor merging, etc.

That's part of it but certainly far from all. And there are also weather events that create demand for rides. People who might otherwise walk or ride a bicycle or take a bus want a taxi when it rains.

 

[Microterf]

So the cost of manufacturing an L5 car will just automatically fall to whatever level allows an unlimited number of cars to be sold that will sit idle for 23 hours a day so they're available for everybody who wants one at peak demand times???



If people made their car-buying decisions on transportation cost alone, everybody would drive a Honda Civic or a Toyota Corolla. There would be no BMWs or Mercedes or Porsches, etc. etc. People buy cars for far more than to have the cheapest possible transportation. In this case, convenience will keep people owning cars.

Some people don't own cars, and some people do. This will not change just because robotaxis replace Uber.



That's part of it but certainly far from all. And there are also weather events that create demand for rides. People who might otherwise walk or ride a bicycle or take a bus want a taxi when it rains.

People keep thinking about the vehicle economy the way it currently is.

Let's says tesla is the first to L4/5. In fairly short order, the vast majority of their vehicles produced won't be sold, but will be put into their robotaxi network. Same thing For mobileye and their OEM partners.

At least that is how I envision it playing out. It will be awhile before the demand is met for the robotaxis they can manufacture.

People think they are special and they have legitimate needs that can't be met with robotaxis etc. I'd be willing to bet there are a million carless people in NYC who moved there thinking they would always own a car but figured out otherwise when they factored all the costs in.

 

[ThomasD]

The county I live in has a population of around 25,000 people. With a population density of 35 people per square mile and a land mass area of 586 square miles. I don't see a robotaxi service wanting to serve such a low population density

 

[daniel]

People keep thinking about the vehicle economy the way it currently is.

Let's says tesla is the first to L4/5. In fairly short order, the vast majority of their vehicles produced won't be sold, but will be put into their robotaxi network. Same thing For mobileye and their OEM partners.

At least that is how I envision it playing out. It will be awhile before the demand is met for the robotaxis they can manufacture.

People think they are special and they have legitimate needs that can't be met with robotaxis etc. I'd be willing to bet there are a million carless people in NYC who moved there thinking they would always own a car but figured out otherwise when they factored all the costs in.

The demand for taxis/ride-shares is not going to change significantly with the advent of robotaxis. Robotaxis will compete with existing taxis and ride-shares, with some people preferring robotaxis, and others fearing them. It is preposterous to imagine that anywhere near Tesla's entire car production could go into its own, brand-new ride-hailing service.

Further, there is nothing disruptive to the ride-hailing business about robotaxis. They will compete with Uber and Lyft and taxis, but they will be indistinguishable from them other than the novelty. It is ludicrous to imagine that people who keep their private cars in the face of the availability of taxis and ride-shares will suddenly sell them because now robotaxis have been added to the existing ride-hailing mix.

Not to mention that the supposition above (that Tesla is the first with L4/L5) is wildly speculative on the face of it.

Some people live where ride-hailing services are not economically feasible due to low population density. Far more people simply like the convenience of a private car. You think people are going to abandon their BMWs and Mercedes and Mustangs to take a robotaxi that they have to wait for? You think soccer moms are going to be willing to wait for a robotaxi van to pick them and their four crying kids up from the grocery store while struggling with a shopping cart loaded with food that needs to go into the fridge or the freezer? While the ice cream melts and the butter gets soft and the milk goes bad in the sun?

When I broke my arm and was on pain killers I had to take an Uber to the grocery store. Then when I was ready to leave I had to call for another and wait for it. Let me tell you, I was so glad when I was able to drive again!

 

[Microterf]

The county I live in has a population of around 25,000 people. With a population density of 35 people per square mile and a land mass area of 586 square miles. I don't see a robotaxi service wanting to serve such a low population density

I live in a low population area. We will be some of the last to get them when they are avaliable. This is true.

 

[Microterf]

The demand for taxis/ride-shares is not going to change significantly with the advent of robotaxis. Robotaxis will compete with existing taxis and ride-shares, with some people preferring robotaxis, and others fearing them. It is preposterous to imagine that anywhere near Tesla's entire car production could go into its own, brand-new ride-hailing service.

Further, there is nothing disruptive to the ride-hailing business about robotaxis. They will compete with Uber and Lyft and taxis, but they will be indistinguishable from them other than the novelty. It is ludicrous to imagine that people who keep their private cars in the face of the availability of taxis and ride-shares will suddenly sell them because now robotaxis have been added to the existing ride-hailing mix.

Not to mention that the supposition above (that Tesla is the first with L4/L5) is wildly speculative on the face of it.

Some people live where ride-hailing services are not economically feasible due to low population density. Far more people simply like the convenience of a private car. You think people are going to abandon their BMWs and Mercedes and Mustangs to take a robotaxi that they have to wait for? You think soccer moms are going to be willing to wait for a robotaxi van to pick them and their four crying kids up from the grocery store while struggling with a shopping cart loaded with food that needs to go into the fridge or the freezer? While the ice cream melts and the butter gets soft and the milk goes bad in the sun?

When I broke my arm and was on pain killers I had to take an Uber to the grocery store. Then when I was ready to leave I had to call for another and wait for it. Let me tell you, I was so glad when I was able to drive again!

Once again. You are looking at it from a perspective of what's currently available and not what is coming.

Uber/Lift is great for certain situations, but bad for others. You don't have privacy like you would in a robotaxi, you're limited by who is willing to pick you up, etc. The cost of robotaxis will be much less than uber/Lift, and that will be a very compelling argument for most people.

Nobody is going to sell their car to switch over to using solely robotaxis, but they will not buy a new car and instead use the service.

I'm not assuming that Tesla will get there first. I don't know if they will EVER get there. I'm saying that SOMEONE will get there, and I have a hard time seeing a future where what I've described doesn't come to fruition once they are available to people.

I definitely don't think all Mercedes/BMWs will stop being used, all I'm saying is that for people who have wide access to them, it will be the most affordable option, and for the vast majority of people (>70%) the value proposition will be too high to resist.

Assuming that all of us own Teslas in here, were fortunate enough to be in positions to afford what is considered a luxury car, and we chose to spend our money on that instead of a cheaper car, and the rest on other stuff. I mean hell, I (and a lot of us) spent more money on software that we don't know if we'll ever see a benefit from than some people spend on a car.

I don't think robotaxis will be like current ride hailing. I think it will be a subscription. I believe you will pay something like 200-300 a month for access to the cars whenever you need them. Will it be more inconvenient than having your own car, of course, but for the vast majority of people, paying $200 a month instead of $800 a month for a car payment/insurance/gas is more than worth the difference in convenience. With the algorithms, you'll have a car ready every time you need to go to work, or home from work, etc.

 

[mark95476]

Pretty amazing how BMW is conspicuously absent in ALL of Mobileye's marketing material. IAA Munich just wrapped up and BMW was nowhere to be seen when Mobileye was mentioned. Now Mobileye has Moovit or some other brand that nobody has heard of.

Its been a while, but I thought Intel bought Mobileye expressly for their BMW design win. It's 2021 and this BMW+Intel+Mobileye robotaxi didn't happen.

BMW Teams Up With Intel, Mobileye for Autonomous Car by 2021

BMW AG is teaming up with chipmaker Intel Corp. and camera-software company Mobileye NV to bring self-driving cars to the road by 2021, becoming the first major automaker to set a specific date to produce a fully autonomous vehicle.

www.bloomberg.com

July 1, 2016

(edit: Bloomberg article title since inlining is blocked: 2016-07-01/bmw-teams-up-with-intel-mobileye-for-self-driving-car-by-2021)

@powertoold In the video, starting at 25:35, the Intel CEO introduces the Mobileye robotaxi and explicitly says it is "capable of SAE L4 full autonomous driving".

So Intel did use L4 to describe the robotaxi. I was right. I think I am owed an apology.

So, more of the same rehashed Intel/Mobileye marketing spiel that we've seen years before. They replaced all references of "BMW" with some noname company that people have already forgotten.