OT-ish: Self driving levels

Investment Analysis Clubs Macro Economic Trends and Risks
1

The discussions of self driving vehicles often involve different levels of self driving. Level 2, level 3, level 5. So what are these levels?

I didn’t know, so I looked them up. Here’s a quick summary I found that helped me.

https://www.synopsys.com/automotive/autonomous-driving-level…

Level 0 - no automation
Level 1 - things like adaptive cruise control which can sense the car in front of you
Level 2 - partial automation (Tesla autopilot or GM Super Cruise are examples)

…

Level 5 - full automation. No steering wheel, no brake pedal, no accelerator pedal. The car does it all. And no limits on where the car can operate.

1 Like
2

The URL is a good reference on the different levels of autonomous driving. However, the timelines of the predictions should be disregarded. The web page states:
The 2019 Audi A8L arrives in commercial dealerships this Fall.

This means that this page was written in 2018, 4 years ago. Audi’s Traffic Jam Pilot feature was cancelled for the US market due to legal issues.

The web page states at the beginning:
<1>Researchers forecast that by 2025 we’ll see approximately 8 million autonomous or semi-autonomous vehicles on the road.

That’s only 3 years away. I hope I’m wrong, but it doesn’t feel like the proliferation of autonomous driving is that close by.

That said, it’s also possible that the automakers may publicly downplay their progress if and when they’re close to launching level 4 autonomous vehicles. People will be less motivated to buy new cars that they believe are about to become obsolete. They will keep their old cars longer, or they will buy old used cars to tide them over for a few years. While leasing is another option, I’m sure that this would become MUCH more expensive when the leasing companies take into account the expected accelerated depreciation.

3

Once the car companies can produce Level 5 vehicles, car sales to individuals will disappear.

This has already been stated by Musk and Barra (head of GM). The new business model for manufacturers will be to create and manage vehicle networks that function essentially as public taxis.

1 Like
4

Once the car companies can produce Level 5 vehicles, car sales to individuals will disappear.

Doubtful.

First off, there are people with special needs, such as wheel chairs, sports equipment, big families, towing etc etc. Just because “a” car can be made that does level 5 doesn’t mean all possible configurations will be instantly made that fill all these needs right away at a price point that is reasonable.

It will take a decade or two to replace cars on the road in any case.
I’ve also used Uber and Lyft a couple of hundred times and it isn’t always more convenient than just driving yourself.

Mike

2 Likes
5

First off, there are people with special needs, such as wheel chairs

Easily handled because the interior of a vehicle is easily changed to a variety of special configurations to suit almost any needs.

There will be a variety of vehicles with layouts appropriate for various groups.

1 Like
6

Easily handled because the interior of a vehicle is easily changed to a variety of special configurations to suit almost any needs.

To transport a passenger in a wheelchair, you need a door with a height in the neighborhood of 56 inches. A bit more is better. Width isn’t as much of a problem, but plan on 26 inches or more.

Almost no current vehicles that aren’t a bus meet those requirements.

No sedans. No SUVs. No minivans.

Kind of hard to easily change the interior to meet a need if you can’t GET to the interior.

There will be a variety of vehicles

When?

–Peter <== who knows a thing or two about transporting people in wheelchairs.

7

There are templates:

https://www.londonblacktaxis.net/wheelchair-accessible-taxis…

Jeff

8

There are templates:

Yes, there are.

How many black London taxis have you noticed in NYC?

There are several aftermarket up-fitters who modify minivans from Chrysler, Toyota and Honda to be accessible to people in wheelchairs. They invariably start by cutting out the floor from the firewall to the rear axle and inserting their own custom made lowered floor. A 10 - 12 inch drop is typical. With the lowered floor comes extensions to the side doors. Then you have to put the original driver and passenger front seats onto removable pedestals to get those seating positions back to the right place. They’re removable so that a wheelchair user can also drive (typically with hand controls) or use the front seat position and make room for a second wheelchair in the back. The original 3rd row seat typically remains, although it may be moved rearward to give a little more room in the middle for the wheelchair user. And let’s not forget the vehicle’s suspension. You’ve got to give the suspension a significant lift to make room for that lowered floor.

An alternative approach is to cut out what amounts to a trench down the middle of the minivan from behind the front seats all the way to the back. The wheelchair user can then enter from the rear of the van. This is occasionally done to a Ford Transit Connect or the little Nissan minivan.

A few cut the roof out and install a raised roof for the needed height. This has the down side of leaving the wheelchair user sitting well above any other passengers. This is typically used for taxis, as the additional roof height can be useful for other passengers as well. A fold down bench seat goes in the middle row to be used when a wheelchair is not present.

None of these are a terribly good solution. The structure of the vehicle is compromised by all of the cutting. Less so with the roof mods, but still an issue. The quality of the work is universally horrible. Fit and finish are barely there. The modified side become noisy and poor fitting.

All of this cutting generally rules out any EV, as the floor is usually the location of choice for the batteries.

One company, Mobility Ventures produced a complete vehicle designed around wheelchair users. It used Ford running gear. The vehicle itself was probably best used as a taxi. Heck, it bore more than a passing resemblance to the iconic London Taxi. It looks like the company stopped producing cars several years ago. I occasionally still see one on the road. Fit and finish was somewhat better than the hacked up minivans, although typically pretty spartan. I rented one for a week once. It worked fine. Sucked up the gas, though.

So what cars currently available in the US do have adequate access for wheelchair users? Ford Transit. Mercedes Sprinter. Gotta make sure you get the mid-height or high roof on the Transit. They have a low roof version that is too low. And don’t get a Chevy Express (or it’s GMC badged twin). They’re not tall enough.

And guess what? The passenger versions of these full size vans are legally buses. They qualify for an exemption from a commercial driver license if used for your family, but they are still buses. And they’re also big. In every dimension. Won’t fit in most parking garages. Won’t fit in your garage at home, either. (Unless you’ve got an RV sized garage.)

–Peter

9

Once the car companies can produce Level 5 vehicles, car sales to individuals will disappear.

I think that’s nonsense.

Oh, sure - far more people will choose not to own their own cars if there are Level 5 AV’s out there, and robotaxis become a thing. But a lot of people use their cars not just to move themselves around, but a lot of their things as well - and it’s massively inconvenient to not be able to store those belongings in your car when not using them.

Take kids. When my kids were younger, our family car always had two car-seats (among tons of other stuff, but focus on the car seats). If I head off to the zoo (for example) with the two kids, if I’m using a robotaxi instead of a personal car I’ll have to take those car-seats with me all day while we’re walking around. Which would be a major inconvenience - repeated hundreds of times over the course of a year.

Vehicles that individuals use for work - that are loaded up with tools and equipment and materials - aren’t going to switch over to TaaS. Heck, even as a lawyer I frequently had multiple meetings in the course of a day that were “chained trips,” where I needed to be able to keep materials from one meeting in my car while I went to another meeting.

And, of course, it kind of goes without saying that the luxury market will still be out there, for people who would prefer to simply have their own private vehicle - with interior appointments to their taste and comfort rather than someone else’s.

Absolutely there will be a lot of people that robotaxis work great for. But there will still be a non-trivial market for personally owned vehicles.

Albaby

8 Likes
10

Once the car companies can produce Level 5 vehicles, car sales to individuals will disappear.

I think that’s nonsense.

Think what you choose, but it is a rational decision for business. What people want vs what works for business (and/or govt) are different things.

When my kids were younger, our family car always had two car-seats

Car seats can be built-in (designed in by choice to ALL cars), so not an issue.

Vehicles that individuals use for work - that are loaded up with tools and equipment and materials - aren’t going to switch over to TaaS.

Again, a minor issue. Specifically configured tradesman vehicles can be leased via the service, so the person gets the same trade vehicle every time. Vehicle stored and maintained by the car company and sent to where it is needed when requested by the user. Costs more reserve it, but it is a tax writeoff.

the luxury market will still be out there, for people who would prefer to simply have their own private vehicle - with interior appointments to their taste and comfort rather than someone else’s.

Same arrangement as with any proprietary or trade vehicle. The company owns and operates the vehicle, but the user can have it set up as they choose.

11

Think what you choose, but it is a rational decision for business. What people want vs what works for business (and/or govt) are different things.

It’s not a rational decision for business. If there’s a market for people to own their own cars, in addition to the fleet/robotaxi model, then the automakers are going to sell cars to that market.

Car seats can be built-in (designed in by choice to ALL cars), so not an issue.

Yes, it is. You need different car seats for different ages. An infant needs a different car seat than a toddler, than a four year-old. Etc.

The issue isn’t just limited to car seats, of course. It’s any situation where people are using their car to do things that may involve picking up or using things at one location that they won’t want to schlep around with them at another. If I stop off and pickup my dry cleaning on the way to work, I don’t want to have to bring the dry cleaning up into my office with me and then back down with me - I want to leave it in my car. If I’m picking up my kids on the way home from work, I don’t want to have to bring a car seat and stroller with me into my office and back down again. Etc.

Specifically configured tradesman vehicles can be leased via the service, so the person gets the same trade vehicle every time.

It’s not a question of getting the same vehicle. It’s that the vehicle stores all of their tools and equipment, so that they don’t have to completely load it up every morning, unload everything when they get to a job site, then reload everything again when they leave the job site, and unload everything again when they return home. If I’m a plumber doing 4-5 jobs in the course of the day, I need one vehicle that will stay with me all day - that has all my tools and gear and materials.

Same arrangement as with any proprietary or trade vehicle. The company owns and operates the vehicle, but the user can have it set up as they choose.

Yeah, I don’t think that’s going to work for the luxury segment. You’re not saving enough money by using a robotaxi compared to owning your own car for someone who has money to forego having their own car - always there in their driveway, and always in the condition that they have maintained. No matter what the company who owns and operates a shared vehicle does, after many tens of thousands of miles of use as a robotaxi the interior is going to be a little worn down and damaged. If I’m a wealthy person, it’s totally worth it to me to spend a few K more per year to have my own car, rather than a shared one.

Albaby

6 Likes
12

Yeah, I don’t think that’s going to work for the luxury segment.

The top end of the luxury market already doesn’t need a robotaxi – they have drivers.
Who also act as security

Mike

13

Easily handled because the interior of a vehicle is easily changed to a variety of special configurations to suit almost any needs.

Even if this were the case (which Peter has shown is false, unless you think an angle grinder and welder count as easily) how many modified Uber/robotaxis are going to be available/nearby when you need one? Of course you could reserve one in advance – just not as convenient.
Same thing goes for a car with a bike carrier, or kayak roof rack or any of the other things people need/want for their lifestyles. Will the robotaxi come with a ski rack and chains for driving 100+ miles to a ski resort?

Sure, making the service handle all these things is possible but certainly is not economical at the infancy of the robotaxi market, so they won’t exist for a long time. The TaaS companies will focus on what Uber/taxis do today – airport trips and city center trips where parking is difficult or expensive.

Mike

2 Likes
14

The TaaS companies will focus on what Uber/taxis do today – airport trips and city center trips where parking is difficult or expensive.

I think this depends a bit on whether Musk is right about autonomy or not.

Okay, I probably shouldn’t personalize it to Musk. But Tesla’s approach is that they’re going to be making an AI driver that they will be able to download into all the many Teslas that have been built and sold. Straight to Level 5 autonomy - every Tesla will be able to drive itself to any destination (within reason) that a human driver could drive to, without driver intervention. AV’s are a software update (or maybe also a chip replacement) away for countless existing drivers.

Other companies are working through Level 4 autonomy - vehicles that will be geofenced, rely on lots of external data sources (like super-accurate mapping) and non-visual sensors (like LIDAR), and at first be owned and operated by a company providing TaaS exclusively, rather than being sold to individuals. For that kind of vehicle, running as a taxi makes a lot of sense. They’re going to crawl before they walk - run limited Level 4 service in defined areas, which they gradually expand out as the systems get more capable. And they can avoid frustrating owners who natural expect that a car which they purchase (rather than a TaaS) should be able to go anywhere.

Albaby

1 Like
15

Again, a minor issue. Specifically configured tradesman vehicles can be leased via the service, so the person gets the same trade vehicle every time. Vehicle stored and maintained by the car company and sent to where it is needed when requested by the user. Costs more reserve it, but it is a tax writeoff.

  1. A tradesman wants his own tools. Unless you are going to unload and reload the vehicle every morning and night, this is a non-starter. And I don’t want to be the plumber who rents the vehicle for the day and get to the customer’s house and find there’s no 1/4” driver for the ratchet set.

And B. How does the person get the same trade vehicle every time unless the car company reserves it, typically 5 days a week, M-F, in which case it’s hard to see how this is any less expensive for the car company than for the individual.

Same arrangement as with any proprietary or trade vehicle. The company owns and operates the vehicle, but the user can have it set up as they choose.

Who makes money here? The car company has to reserve the vehicle for the user which means they don’t have multiple users except rarely and occasionally at the margin. The customer has a vehicle which can’t potentially be used by others, so no special tools, datasets, diagnostic devices can be left in the vehicle unless he wants them to be gone. And the customer has to pay a higher price to give the “car company” some sort of profit for their time and trouble.

It’s clear you are caught up in the excitement of watching the sideline cheerleaders rather than understanding the basics of the game. Nobody wants to pay more for a vehicle which is less convenient than they can have owning it themself.

2 Likes
16

It’s clear you are caught up in the excitement of watching the sideline cheerleaders rather than understanding the basics of the game. Nobody wants to pay more for a vehicle which is less convenient than they can have owning it themself.

It will be interesting to see what the rate of uptake is. For an ‘ordinary’ user, who’s fine taking any vehicle, there’s a chance that the costs of doing TaaS will be as much as 20-30% lower than the cost of self-ownership of an equivalent vehicle. For folks who need particular types of vehicles, much less the exact specific vehicle every day, the business model just might not work out at all.

The original poster was wrong to think that’s enough of an incentive to get everyone to shift, and thus eliminate all direct sales to consumers. I think that there will be a fair number of takers for TaaS as opposed to self-ownership of the same car, but certainly not all of them.

Albaby

17

Sure, making the service handle all these things is possible but certainly is not economical at the infancy of the robotaxi market

All the configurations don’t have to be optimal as long as you can make them work. Let’s say you had three standard configurations, a “Smart-sized” small personal vehicle for up to two people, a “sedan” model based roughly on the type of design that London taxis use - which can easily accommodate a wheelchair along with other passengers as a “people mover”, and a second design roughly configured based on a pickup truck which could handle “carrying” and towing applications.

Sure, you could create scenarios which these couldn’t accommodate, but I suspect these three configurations would accommodate 99.9% of all non-commercial applications. As these will not “belong” to anyone, all the frills which are applied to vehicle to make them “macho” or attractive could be left off - further simplifying the design to make it purely pragmatic.

As these services would be available nation-wide (in all towns and cities), it is tempting to say that buses/trains and planes would be used for the majority of long-distance travel.

The major challenge would be “load-balancing” to minimize times when there were no vehicles available, yet keep inventory levels moderate.

Jeff

18

I think this depends a bit on whether Musk is right about autonomy or not.

He could easily be wrong.
Or he could be right but just off by a decade or more.

But Tesla’s approach is that they’re going to be making an AI driver that they will be able to download into all the many Teslas that have been built and sold.

I’ve been a FSD beta tester for 10 months now. While some of what it can do is impressive, IMO it is not “close” to being robotaxi quality. I just got the latest update a couple of days ago and haven’t really tested it against all my known fails yet.

But, assuming that the technology gets to L4 and/or L5 I still think that, at least in the US, people like their cars. And as they get electrified and more reliable, people will like them even more. Some people will ditch them for Uber/robotaxis. But the first thing I see young people do that go everywhere with Uber do after they get a good job it to go buy a car…so they can go to the beach, mountains, wherever, when they want, with their stuff.

Mike

1 Like
19

But the first thing I see young people do that go everywhere with Uber do after they get a good job it to go buy a car…

That is because there is no effective alternatives to a car in most of the world, particularly the US. Far less need for a car in many European cities–because they have a good non-car transit system. They also have decent train transit, so less need for a car to take a longer trip.

20

There will be a variety of vehicles

When?

Depends when the L4-L5 vehicles get certified.

Given the base of the vehicle, a wide variety of vehicles can be effectively designed and built on thata single base at a reasonable price (for the car company).

Remember: The self-driving cars can also be automated delivery vehicles for many businesses, so need need to pick stuff up and bring it home. It is delivered wherever/whenever you chose for delivery.