• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Self-driving discussion thread (2026)

Kaido

Elite Member & Kitchen Overlord
Great video on the upcoming Rivian R2:

* $2,500 system (Tesla charges $8,000 & Lifetime is being retired this week in favor of a $99/mo subscription service)
* 11 cameras (65 megapixels total)
* 5 radars (including a front imaging radar)
* LIDAR (top of windshield)
* In-house compute platform
* Self-driving computer developed in-house by Rivian (800 TOPS platform with two of them for 1,600 TOPS)


Tesla has the best platform on the planet, but they lack LIDAR, which I think will hold them back.


1770645435577.png

1770645414145.png
 
Latest Tesla FSD with Grok:

Voice command demonstration, works better than Siri lol

 
I'm disappointed that MB has no direct successor to their L3 system available.
Would have been the time to raise the bar with even more L3 ODD, or even a limited L4, instead it's back to zero autonomy.

On the upside, Waymo is finally starting to break out of regional deployment and going (sort-of) global. I guess it will be restricted to high-value locations for quite a while. I wonder if they can get a leg into Asia next.

In the meantime, I remain unimpressed by these half-assed sorta-convenience functions, which are just going to stress you out, as you constantly anticipate them trying to kill you.
 
Now that Ford has announced their new EV strategy I suspect that BlueCruise 1.3 is all my 2023 Mach-E is getting. Why waste resources on a platform that you are no longer going to support?

Anyway, since this is the self driving thread it should mention the open source project called Open pilot brought to us by the folks over at https://comma.ai/. This open source project is not just for Ford vehicles. It works on a variety of differing manufacturers. I've never used it first hand but a few people on the Mach-E forum I hang on have and all of them claim it is better than Fords current implementation.

This is an option for people who want the tech but it was never offered in the vehicle they drive or for people like me who are getting left behind by the manufacturer and want to keep the tech but with improved performance and stability.

The only fly in the ointment is if Ford decides to send an OTA update to encrypt the CAN bus on older models of the Mach-E like they have on the 2025 model year. This will essentially render open source options unusable unless someone can break the encryption.
 
Last edited:
Now that Ford has announced their new EV strategy I suspect that BlueCruise 1.3 is all my 2023 Mach-E is getting. Why waste resources on a platform that you are no longer going to support?

Anyway, since this is the self driving thread it should mention the open source project called Open pilot brought to us by the folks over at https://comma.ai/. This open source project is not just for Ford vehicles. It works on a variety of differing manufacturers. I've never used it first hand but a few people on the Mach-E forum I hang on have and all of them claim it is better than Fords current implementation.

This is an option for people who want the tech but it was never offered in the vehicle they drive or for people like me who are getting left behind by the manufacturer and want to keep the tech but with improved performance and stability.

The only fly in the ointment is if Ford decides to send an OTA update to encrypt the CAN bus on older models of the Mach-E like they have on the 2025 model year. This will essentially render open source options unusable unless someone can break the encryption.

As much as I don't like it, I agree with Ford encrypting their stuff:

 
Now that Ford has announced their new EV strategy I suspect that BlueCruise 1.3 is all my 2023 Mach-E is getting. Why waste resources on a platform that you are no longer going to support?

Anyway, since this is the self driving thread it should mention the open source project called Open pilot brought to us by the folks over at https://comma.ai/. This open source project is not just for Ford vehicles. It works on a variety of differing manufacturers. I've never used it first hand but a few people on the Mach-E forum I hang on have and all of them claim it is better than Fords current implementation.

This is an option for people who want the tech but it was never offered in the vehicle they drive or for people like me who are getting left behind by the manufacturer and want to keep the tech but with improved performance and stability.

The only fly in the ointment is if Ford decides to send an OTA update to encrypt the CAN bus on older models of the Mach-E like they have on the 2025 model year. This will essentially render open source options unusable unless someone can break the encryption.

We have a subreddit for that!


Comma 4 hardware just dropped:


Over 300 vehicles supported:


Including Tesla:


And Rivian:

 
Tesla said it has rolled out its first cybercab.


I'm going to call bullshit for one very obvious reason:

If you solved self driving technology (which they havent) - a huge massive win on its own - and you are the only person who makes the product, why wouldnt you simply run a cybercab fleet yourself?

Companies love recurring revenue.

Musk companies are not strangers to rolling out their own networks of things (starlink, superchargers, tesla dealers). I dont see why a cybercab division responsible for running taxi fleets would be any different.
 
Musk companies are not strangers to rolling out their own networks of things (starlink, superchargers, tesla dealers). I dont see why a cybercab division responsible for running taxi fleets would be any different.

There are already over 8 million Teslas on the road. Like Uber, the end user would be on the hook for maintenance & whatnot. Great way to sell more cars!
 
If you solved self driving technology (which they havent) - a huge massive win on its own - and you are the only person who makes the product, why wouldnt you simply run a cybercab fleet yourself?

Companies love recurring revenue.

Here's how I see it: (market-talk, not personal judgement)

1. Musk is an excellent marketing hype man, as demonstrated by his wealth (~$850 billion) & his company (Tesla has a ~$1.3 trillion market cap). His actual cash is very little (which is why he had to get loans & equity from investors) and, like Bitcoin, the market value is built on hype. But, hype sells!

2. He's a visionary exaggerator who "silently" invests in long-term exponential systems, specifically: Supercharger billing (I believe every single US EV is now slated to come with his NACS system), Starlink (worldwide cell & Internet), and AI (primarily government contracts). He open-sourced Tesla patents. Nobody else wanted to make electric cars & nobody else wanted to build & maintain a charging network, which meant that literally all of the other auto manufacturers shrugged & handed him an endless conveyor belt of charging dollars lol. There are now over 36,000 Tesla charging stalls worldwide & less than a thousand Rivian Adventure Network chargers. They are also quietly rolling out utility-scale battery grids (MegaPacks & whatnot), which have the potential to be HUGE revenue sources in the future!

3. He sees the writing on the wall from China's EV explosion, which is why he pivoted Tesla to a robotics & AI company. I think it's odd they scrapped the S & X, but not the Cybertruck, which only sells ~1,500 units a month. But now they do spaceships, satellites, etc. The United States War Machine is an endless fountain of funding, which Musk is providing AI, Starlink, and shuttle rides for.

4. Self-driving IS coming, just not soon. My friend takes Waymo self-driving cars all the time out west! It will be in development for a long, loooong time. The thing is, Tesla has what no one else has: millions of car's worth of data!

5. Robotics is building at insane speeds due to AI. I've gotten more into professional robotics over the last year. They are really, really useful in specific use-case scenarios. Will we ever see a home unit that cooks & stuff? Maaaaybe lol. In theory, yes, but I think we'll see FSD before we see a useful humanoid home robot tbh.

6. Like everyone else, he's on a budget. He wanted to make the Model Y a new platform, but financial & investor constraints forced him to base it off the Model 3. Still a great car, but not quite his vision of what he really wanted it to be. So looping back to #2, rather than denying users access to Autopilot due to the huge capital cost ($8,000), he now charges $99 a month, which makes it more accessible to people, which creates yet another endless revenue stream. Despite what you may read online, Tesla FSD is HUGELY helpful. Many people where I live commute into NYC for hours in self-driving mode & wouldn't trade it for the world! "Imperfect" & "WIP" does not mean useless!

Right now:

1. Cybercabs & FSD are speculative

2. Robotics are speculative

3. Everything depends on Musk

He really crashed things financially for awhile there with his political antics, and yet he has managed to double his money since then & bring his stock price back up, so I have NO IDEA how the future will go lol. Still waiting on my 500-mile Cybertruck & $20k Model 2 hahaha
 
There are already over 8 million Teslas on the road. Like Uber, the end user would be on the hook for maintenance & whatnot. Great way to sell more cars!
Right but if you're running those cars then thats all taxable deductions yeah?

I dont disagree with much if any of what you said. But i would cite everything as the reason I'm skeptical lol. Tesla needs hype to keep people interested since they arent technically there yet (and never will be without lidar from what i understand the data requirements demands for self driving)
 
I’m curious how well lidar will work in heavy traffic with lots of lidar equipped vehicles all using the same frequencies?
 
Right but if you're running those cars then thats all taxable deductions yeah?

I dont disagree with much if any of what you said. But i would cite everything as the reason I'm skeptical lol. Tesla needs hype to keep people interested since they arent technically there yet (and never will be without lidar from what i understand the data requirements demands for self driving)

I'd imagine it'd be the same as how Uber taxes would work!

Personally, I don't think that Tesla can safely do universal highway & city full-self driving in any weather without LIDAR & Radar, at least not with steering-wheel-free robotaxis. LIDAR isn't perfect either, but given the current state of technology, it seems like the best option!

However, Tesla's Vision System DOES have quite a few tricky tricks, like high dynamic range, neural-net processing, prediction, image aggregation over videoframes to get a clearer picture, etc., which may be good enough at some point. Even right now, the latest release IS pretty awesome in actual practice!
 
I’m curious how well lidar will work in heavy traffic with lots of lidar equipped vehicles all using the same frequencies?

That is an issue, but modern LIDAR equipment uses coding & filtering (pulse coding, frequency hopping & modulation, sensor fusion to verify against other stuff like radar/ultrasonics/cameras, time-of-light timing windows, etc.).

Back in the day, my R/C toys used matched quartz crystals pairs, but now use spread-spectrum digital communication in a similar way as modern LIDAR to get a unique ID so that cross-chatter wasn't an issue.

Still an issue for LIDAR, but they have clever ways of filtering out unwanted signals these days!
 
Last edited:
Back
Top