Do Size and Shape matter in Self-Driving?

[voyager]

Thus far, there are literally hundreds of threads about autonomous cars on the Tesla forum,
mostly about technological developments, company's claims and pitches, the implications
economically, regulatory and legally, about testing and pilots, user experiences and a lot of
what I call 'semantics' (how to define and value the many different aspects in VA).

So far, it looks like the 'spatial dimension' is missing, which is odd since going from A to B still
requires some sort of physical transport mode... that we now decided to make driverless.

It as if car makers and AV developers have jointly decided that vehicle size does not matter,
shouldn't matter. Now I know that an AV setup is that costly, that only more expensive,
therefore bigger cars are used for (preliminary) testing.

The bigger the AV:
- the less space there is to maneuver through traffic
- the less margin there is to take evasive action
- the harder it is to scan - sensor - image its vicinity
- the more likely they will cause accidents (say IIHS, NHTSA and ETSC)

This is an older article (which shows you how rapid developments are),
but it basically says that smaller vehicles are easier to deploy autonomously.
Might do without tens of dollars in AV setups.

Autonomous Vehicles Should Start Small, Go Slow

Self-driving vehicles can already work well on campuses where traffic moves slowly

spectrum.ieee.org

THE beautiful catch of course is that greening the car (the smaller, the less kWh needed)
and making it driverless can be two sides of that same highly coveted coin.

What say you?

 

[nvx1977]

probably makes sense in small-geofenced applications. As the article explains, size is not a safety concern IF such small AVs run in low-speed areas. Given Tesla is not solving for narrow application of AVs (with a possible exception for Boring Company tunnels), size of car doesn't seem that important. I also don't think the size range of typical consumer vehicles have that much variation on perception and planning. A bus, yes.

 

[diplomat33]

Thus far, there are literally hundreds of threads about autonomous cars on the Tesla forum,
mostly about technological developments, company's claims and pitches, the implications
economically, regulatory and legally, about testing and pilots, user experiences and a lot of
what I call 'semantics' (how to define and value the many different aspects in VA).

So far, it looks like the 'spatial dimension' is missing, which is odd since going from A to B still
requires some sort of physical transport mode... that we now decided to make driverless.

It as if car makers and AV developers have jointly decided that vehicle size does not matter,
shouldn't matter. Now I know that an AV setup is that costly, that only more expensive,
therefore bigger cars are used for (preliminary) testing.

The bigger the AV:
- the less space there is to maneuver through traffic
- the less margin there is to take evasive action
- the harder it is to scan - sensor - image its vicinity
- the more likely they will cause accidents (say IIHS, NHTSA and ETSC)

This is an older article (which shows you how rapid developments are),
but it basically says that smaller vehicles are easier to deploy autonomously.
Might do without tens of dollars in AV setups.

Autonomous Vehicles Should Start Small, Go Slow

Self-driving vehicles can already work well on campuses where traffic moves slowly

spectrum.ieee.org

THE beautiful catch of course is that greening the car (the smaller, the less kWh needed)
and making it driverless can be two sides of that same highly coveted coin.

What say you?

I get the idea that a smaller vehicle might be safer (easier to take evasive action etc). But I am not sure it would make that big of a difference. Cruise uses the Chevy Bolt. Argo uses the Ford Focus. They are not big cars. Waymo's Pacifica is bigger but it would be no different than any other Pacifica driven manually. And a really small car, like those microcars, would be more agile to avoid collisions but they also would not hold a lot of passengers which would not be profitable for a robotaxi service. And microcars may do poorly in a collision so they could be less safe if they do get into a collision. Ultimately, you need to make your FSD safe no matter what the size of the car.

I think the purpose of the AV is more important because it will inform the design of the vehicle. Is the AV going to be a consumer car? A robotaxi? Delivery vehicle? If you are building a consumer AV, then you need to make the car attractive to consumers. So your AV will likely need to be an affordable, stylish midsize sedan or crossover. If you are building a robotaxi, then you need to think about passenger capacity. If you are building a delivery vehicle, then your AV will need lots of cargo space so it will likely need to be big like a big van or semi truck. The PerceptIn pod that was used on the campus of Purdue is a great example of this. The design works great for moving 1-2 students around a big university campus. But it would be a poor design for a consumer car to drive on the highway.

Keep in mind that right now, companies are still very much in a testing/development phase. They don't need a fancy, custom designed vehicle just to do testing. Any commercial vehicle will probably do fine for the purpose of testing. That's why we see companies simply retrofit existing vehicles like a Pacifica, Ford Focus or a Chevy Bolt. But we are seeing companies like Cruise, Zoox and Waymo in the process of developing custom built vehicles because they are thinking ahead to what type of AV they want to deploy when the FSD is "finished". We see that these vehicles are very different than consumer vehicles. They are more like boxes on wheels. The shape makes sensor placement better. Also, since the vehicles are driverless, they don't need space for a driver. So the entire interior can be designed with the passenger in mind.

 

[ThomasD]

I think you will need a variety of vehicles. Different shapes and sizes. I don't see the Zoox vehicles being for long travel or operating outside of cities. I don't think you will see a Zoox vehicle take people to the ski resorts after landing at the Denver Airport These vehicles will most likely need remote operators. Say you are running 24/7 then you will need monitors 24/7. You will need office space and maintenance facilities. Depending on how many vehicles you have operating you may need more than one maintenance facility/customer service facility.

 

[voyager]

I get the idea that a smaller vehicle might be safer (easier to take evasive action etc). But I am not sure it would make that big of a difference....
And microcars may do poorly in a collision so they could be less safe if they do get into a collision. Ultimately, you need to make your FSD safe no matter what the size of the car.

I think the purpose of the AV is more important because it will inform the design of the vehicle. Is the AV going to be a consumer car? A robotaxi? Delivery vehicle? If you are building a consumer AV, then you need to make the car attractive to consumers.

Keep in mind that right now, companies are still very much in a testing/development phase. They don't need a fancy, custom designed vehicle just to do testing.... They are more like boxes on wheels. The shape makes sensor placement better. Also, since the vehicles are driverless, they don't need space for a driver. So the entire interior can be designed with the passenger in mind.

All valid and obvious points. Question is where/when will big AVs form an unnecessary big risk for pedestrians
and cyclists, when can they be considered too small that they constitute a huge risk for who's onboard the AV?
Tesla Cybertruck too big, and the Swedish podbike (below) too small. Easy to see that with its easily scanable,
rounded and sloping contours and its tiny footprint, it may make for a great AV, albeit not safe.

Waymo abandoned its own what I labelled 'helmet-on-wheels' principle when they started experimenting with Chryslers Voyager and Jaguars i-Pace. Most of you will probably know that 360 degree vision motorcycle helmets are already available. Rear vision is projected on the inner visor. New Bell Helmets Incorporate 360-Degree Video Camera « MotorcycleDaily.com – Motorcycle News, Editorials, Product Reviews and Bike Reviews

I think you will need a variety of vehicles. Different shapes and sizes. I don't see the Zoox vehicles being for long travel or operating outside of cities.

My thought too: differentiate. The thing is this: instead of trying to make AV tech work on a given car,
look at the car first and have it physically adapt to the available AV tech. For instance, the boxier
the car shape, the bigger the blindspots and for the AV tech to see around its corners.

Again the catch is that a more FSD-suited vehicle may also make for a greener electric vehicle, since
it will feature a sleek footprint and low vehicle mass. Without compromising safety and comfort (too much).

 

[diplomat33]

Question is where/when will big AVs form an unnecessary big risk for pedestrians
and cyclists, when can they be considered too small that they constitute a huge risk for who's onboard the AV?

The AV needs to be reliable at detecting pedestrians and cyclists no matter what their size.

Waymo abandoned its own what I labelled 'helmet-on-wheels' principle when they started experimenting with Chryslers Voyager and Jaguars i-Pace.

The reason Waymo abandoned the Firefly pod is because regulations don't allow vehicles without steering wheels at speeds above 25 mph. That speed is too slow for commercial robotaxi service. So Waymo needed to use cars with steering wheels if they were going to deploy robotaxis on main roads with higher speeds than 25 mph. Also, the Firefly is too small to be viable as a robotaxi for families. So it made sense to use the Pacifica and the Jaguar.

 

[voyager]

Self-driving pilots are taking place in several cities, Waymo notably in NYC.
Those need to be evaluated, obviously. Before any AV system will receive
approval/certification, we will be years further.

In the mean time what I have written below can't hurt. Also because of its easier mingling
with micro-mobility and its 'greener EV' implications. Electrification has turned cars into
large electric appliances (btw, something Musk also referred to). Gone is the complicated,
ICE-related engineering. The car is the easy part, FSD is much more complicated.

 

[ThomasD]

In order for these robotaxis to work you need thousands of them in cities and the suburbs. You need short wait times. Less that 15 to 20 minutes. They also need to take where you want to go. Not some geofenced area. There needs to be a variety of vehicles. I want to go to the restaurant 30 minutes away from my home. Then on Saturday I need to go to home depot and pickup some plywood and 2x4s. Some of these AV companies have been at it for years and yet they can't even take people to and from Airports

 

[voyager]

Yup, there is too much a one mold debating for AVs. Different needs, different sizes and shapes, I'd say.
WSJ estimated in a June 2021 article that so far $80 billion has been spent on AV development, which
is underway for the better part in the past decade. Time to come up with alternative strategies.

Again, the car as the oversized electric appliance it has become forms the easy part when it comes to FSD.
The FSD itself the hard part. So, why not give 'reformat the car before any AV tech goes in' a shot?
Two pictures as food for thought.

So, do we need "Big = Beautiful" to be and feel protected in today's traffic? And doesn't this at the
same time present an unnecessarily higher risk for vulnerable road users as long as FSD hasn't been
figured out? And wasn't FSD supposed to protect us from getting into accidents in the first place?

 

[voyager]

As long as FSD is not perfect, it helps to have a more committed driver.
Below: more driver involvement since he/she is seated curbside for better view
of what matters most: vulnerable road users, like pedestrians and cyclists.

 

[voyager]

https://twitter.com/x/status/1511421876895109131

 

[voyager]

When technology is supposed to take over control, the car basically becomes the 'app', the application.
Why not reformat the car first to optimize the use of autonomous vehicle technology?

The bulkier and boxier the vehicle, the bigger the blind spots, the more sensors needed.
As Aurora already acknowledged.

Below: rounded and sloping vehicle contours reduce the need for sensors.
Besides, the sleeker the vehicle, the nimbler it is and the more space to take evasive action.

 

[Daniel in SD]

When technology is supposed to take over control, the car basically becomes the 'app', the application.
Why not reformat the car first to optimize the use of autonomous vehicle technology?

The bulkier and boxier the vehicle, the bigger the blind spots, the more sensors needed.
As Aurora already acknowledged.

Below: rounded and sloping vehicle contours reduce the need for sensors.
Besides, the sleeker the vehicle, the nimbler it is and the more space to take evasive action.

Can't see around corners with that design and it still has blind spots near the vehicle. It also looks difficult to get in and out of and not very roomy. Dedicated AVs with no manual controls will look like the the Zoox and Cruise vehicles, a room on wheels.

 

[voyager]

Not all driverless taxis have to come down to looking the same, rectangular-dimensioned vehicles.
There is enough room to cater to a number of target groups. As far as ride-hailers concerned,
my guess is that there is room for more rickshaw-type NTVs.

 

[ThomasD]

That Zoox vehicle doesn't look like it has a lot of space with the seats jutting past the door openings.

 

[Daniel in SD]

That Zoox vehicle doesn't look like it has a lot of space with the seats jutting past the door openings.

https://twitter.com/x/status/1470918299550580736

 

[ThomasD]

Are they operating door to door

 

[voyager]

Information technology went from bulky and cumbersome to lightweight, sleek, versatile….

Whilst Personal Mobility is transitioning from already bulky and cumbersome to full-fledged
overweight and wide, because of all the batteries… Can’t we do something? Yes, we can.
Come up with a car equivalent of the smartphone...

 

[Daniel in SD]

Information technology went from bulky and cumbersome to lightweight, sleek, versatile….

Whilst Personal Mobility is transitioning from already bulky and cumbersome to full-fledged
overweight and wide, because of all the batteries… Can’t we do something? Yes, we can.
Come up with a car equivalent of the smartphone...

A car has to fit a human inside and meet safety standards. There are plenty of sleek AVs that do package delivery.