FSD v11 Released to Customers.

[Ogre]

Tesla will just waltz in to any new area with ZERO establishment cost.

I wouldn’t say “Zero”, Tesla is just investing in solving for the entire map all at once.

 

[charliemagpie]

Not the RD. My take is on rolling out to new zones. As opposed to mapping and testing and all the other stuff competition does, the Tesla car just turns up and ready to take a fare. Anywhere.

 

[firsttruck]

I wouldn’t say “Zero”, Tesla is just investing in solving for the entire map all at once.

Tesla pure vision approach (like 10s of millions years of land animal navigation of the physical world) using 8-12 cameras VS competitors dependence on cameras PLUS minimum 4 RADARs/LIDARs PLUS Hi-Def maps (all cities, towns, rural roads, parking lots, highways, etc) PLUS all the Hi-Def maps updates for just 10 years of all places means the competitors spread to more locations will be much slower and much more costly.

 

[ED_SFO]

I believe the downfall of pure vision with teslas approach is that our own eyes fails us. We've been through a blizzard, foggy weather, storms, our own eye sight can only go so far. With a camera behind a windshield with wipers going full tilt cannot see any better than us, and most likely worse , other sensors like lidar and radar can see past these trivial things. To me teslas pure vision approach is the easiest and fastest way to their goal. Even though it's been slow going, there's a reason why, pure vision is only good in good weather

It definitely takes more time and resources to analyze video feeds, and lidar data, and radar data and then incorporate all those data streams into an output that the driver computer can recognize. Teslas been promising all this for over 7 years is getting close...but not closer than their peers.

 

[Stuck4ger]

Luckily for us Tesla drivers there's a difference between vision system capability and human eye vision capability. And the former will only get better.

 

[CyberGus]

I believe the downfall of pure vision with teslas approach is that our own eyes fails us. We've been through a blizzard, foggy weather, storms, our own eye sight can only go so far. With a camera behind a windshield with wipers going full tilt cannot see any better than us, and most likely worse , other sensors like lidar and radar can see past these trivial things. To me teslas pure vision approach is the easiest and fastest way to their goal. Even though it's been slow going, there's a reason why, pure vision is only good in good weather

It definitely takes more time and resources to analyze video feeds, and lidar data, and radar data and then incorporate all those data streams into an output that the driver computer can recognize. Teslas been promising all this for over 7 years is getting close...but not closer than their peers.

If the cameras are blinded, then the car should pull over and stop. While Lidar and Radar can penetrate bad weather in total darkness, neither can read a street sign, or tell if a traffic light is green or red.

Since roads are designed for people with eyeballs, the only sensor your car cannot live without is the camera.

 

[firsttruck]

I believe the downfall of pure vision with teslas approach is that our own eyes fails us. We've been through a blizzard, foggy weather, storms, our own eye sight can only go so far. With a camera behind a windshield with wipers going full tilt cannot see any better than us, and most likely worse , other sensors like lidar and radar can see past these trivial things. To me teslas pure vision approach is the easiest and fastest way to their goal. Even though it's been slow going, there's a reason why, pure vision is only good in good weather

It definitely takes more time and resources to analyze video feeds, and lidar data, and radar data and then incorporate all those data streams into an output that the driver computer can recognize. Teslas been promising all this for over 7 years is getting close...but not closer than their peers.

RADARs/LIDARs can not read painted traffic signs, detect color traffic lights, read electric digital speed signs, read electric digital text signs, and may more required.

There is also a cost consideration. There are 60-70 million vehicles sold every year, these vehicles are operated in varying climates but most total operation miles are in fairly good conditions with the normal worse being a normal rain storm.

Should we pay huge amount extra cost so every vehicle is able to operate in a blizzard or extreme dense fog that probably much less then 1% of driven miles occur. In case of blizzard, weather forecast can normally give some pre-warning and blizzards probably never occur on 99.99% of driven miles in most places in the world.

 

[Ogre]

PLUS all the Hi-Def maps updates for just 10 years of all places means the competitors spread to more locations will be much slower and much more costly.

Yes.

I just disagreed with the idea that it is “zero”. Tesla has spent billions on their vision system and will continue to spend billions on their vision system for at least another 10 years.

But Waymo has spent billions making making 3 cities work.

Very clear Teslas billions will scale better.

 

[Ogre]

other sensors like lidar and radar can see past these trivial things.

LIDAR and Radar can only see half the road. You will always need a vision system. Roads are designed around drivers who can see.

 

[firsttruck]

Yes.

I just disagreed with the idea that it is “zero”. Tesla has spent billions on their vision system and will continue to spend billions on their vision system for at least another 10 years.

But Waymo has spent billions making making 3 cities work.

Very clear Teslas billions will scale better.

Google Self-Driving Car Project/Waymo has spent many more years on autonomy than Tesla and more money on autonomy than Tesla and Google/Waymo has less to show.

Google Self-Driving Car Project/Waymo
started January 2009)
https://en.wikipedia.org/wiki/Waymo

 

[Crissa]

I wouldn’t say “Zero”, Tesla is just investing in solving for the entire map all at once.

And robotaxis will need places to charge and get cleaned and repaired.

That costs a non-zero amount.

-Crissa

 

[charliemagpie]

As time rolls on, Tesla will continue to rollout Robotaxi's, rendering the average cost to establish a new area as close to zero to make the point moot.

In the meantime, dozens or even hundreds of autonomous gated taxi companies around the world will be boutique with coffee machines etc, or various luxury standards, or niche specialized needs.

Unequalled, Tesla will be geared to rollout at tsunami speed, and establish itself worldwide as the lowest cost no frills mass transit company.

Just have to google 'how many Ubers' around the world. On paper, it will be a matter of production weeks to put that out of business, from there, the potential change in how we get around is amazing.

just spilling some thoughts lol

 

[Ogre]

First Tesla needs to get cars that let me go places while I sleep.

Then they need to make cars I can hail from my phone. Then they need to figure out how to get those cars down the driveway which Google, Apple, and every mapping company thinks is 500 feet past where it actually is.

This IMO is the fundamental flaw with most Robotaxi projections. When I talk to a human driver I can say “It’s the first driveway past Mathew’s Road, the dome way in the back”. Or if I’m a passenger “It’s the next driveway”.

Not that many Robotaxis will be coming this way, but this is a more general problem. There are a lot of places that just don’t follow the rules quite right.

 

[WHIZZARD OF OZ]

Not the RD. My take is on rolling out to new zones. As opposed to mapping and testing and all the other stuff competition does, the Tesla car just turns up and ready to take a fare. Anywhere.

FARE Enough!