Search results

I was.

I don't think this is generally correct. You could do it for some amount of sample data (ie the test cases that you use as proof), but you can't guarantee functional equality for all cases of input, since we can't explain why NN:s produce the output it does. Perhaps you could potentially...

There is a video showing the service area from a year ago here: https://rollout.autoura.com/cities/asia/wuhan

That must be only in Wuhan, right? https://rollout.autoura.com/platforms/apollo

No one knows what ”AGI will require” tbh. There isn’t even a clear definition of what people mean when they say “AGI”.

There is zero chance of that happening. Western "media" got that wrong. Baidu said "Baidu will seek cooperation opportunity with Tesla on Robotaxi based on Tesla’s specific need and timetable." It's a misconception that they would use Tesla tech, it's the other way around if anything. They...

I doubt that's going to happen. It's basically reverse car-rental. If the car isn't returned to you in the same condition, who deals with that? Damaged exterior/interior? How do you prove it? I wouldn't rent out my car if there are no guarantees about that. In fact I probably wouldn't rent out...

The Zeekr MIX doesn't come with a wheelchar ramp built in to the vehicle if that's your question. However, we can't rule out that Waymo might retrofit some percentage of the fleet if they go with that vehicle for a wide roll-out.

Low MPI means that ”I” is high. Miles/Intervention.

If it looks like a tree, and walks like a tree...

MobilEye is not a believer in pure e2e. I think they argue their position quite well. https://www.mobileye.com/blog/autonomous-decisions-the-bias-variance-tradeoff-in-self-driving-technology/ Back in November 2022, the release of ChatGPT garnered widespread attention, not only for its...

I think the perception might classify the object as a tree and not a trailer which would explain why it wants to go around it all the time. But the stupid tree is always one step ahead....

Yeah it's not a slam dunk by any means, but from what I've read 90% or the revenue is in large metros, so clearly geofencing isn't a problem. It also goes hand in hand with the fact that depots and maintenance crews that are fairly fixed geo-wise. I like the reasoning around worker costs in...

Phil Koopman wrote the following about this incident. "A Waymo vehicle was involved in a serious crash in Los Angeles. Significant damage and an injured autonomous specialist (possibly acting as a safety driver). Reported details are incomplete, but include: - Struck from behind by a vehicle...

Yeah right... So why don't they do highway L3 then? That would clearly increase the take rate massively. To me personally its worth 10x the "city fsds".

Again, I don't think you understand the difference between what Mercedes offers in terms of normal L2 and what they offer as L3. You can use MB L2 on the autobahn in the rain at night just fine.

If I learned anything from buying FSD in 2019, it’s that you never buy anything based on what someone says it may be able to do later. The hardware will be obsolete if we get to usefulness for these robots in 10-15 years.

Yeah, unless the Waymo was going in reverse the other driver is probably at fault...

Not while watching a movie instead of the road, no. At what speeds can you do that in a Tesla again? No? Oh, I forgot Elon is autonomy-challenged and that they added SUPERVISED in their latest goal-post relocation…. Do you understand that MB have a very capable L2 too that works under all...

Yeah that's never going to happen. Elon said as much. The boards don't even have the same form factor, have different wiring and different cameras.

Omar "Tesla master pumper" Quazi and his dad should get a life. The ODD for the the MB system is the same as basic Autopilot - limited access highways. The feds should really force the vendors to disallow activation outside the ODD described in the owners manual. I'd pick the MB driver assist...

Google ”Zeekr Mix” for the new platform.

Sooo… How many Lidars will Tesla buy in Q2? More? less? That will be an interesting data point.

Deploying a (poor) L4 system shouldn't qualify in that category imho. RA isn't a failure, it's a system limitation. I'm pretty sure regulators will be concerned quickly with a bunch of Robotaxis stranded on limited access highways. (moderator edit)

Not according to every robotaxi company on the planet. Different challenges and higher stakes game tho. You can't just stop and wait for remote assistance on the highway. That's illegal (and unsafe).

I find that outcome incredibly unlikely.

Regulators shouldn't legislate based on certain technology or solutions, they should legislate on performance (ie safety) and they likely will. Current science is the nail in the coffin for hw3/hw4 autonomy. Unless there is a few major breakthroughs that system will never be autonomous at...

Exactly. It's a classic Elon non-denial denial. He's going to need something more than cameras if he's serious about autonomy this decade. It's really that simple.

MB said to journalists at the Bejing Auto Show recently that they plan to do 80 km/h (50 mph) by the end of 2024 in Europe at least.

Progress in computer vision basically and sensor tech in general. I personally think it will be a while, if ever. Too many weird failure modes (an image of a road on a truck etc). I think Lidar will drop in price faster than there is progress in computer vision, now that more OEM:s (esp in...

Tesla CEO Elon Musk has said that lidar sensors are a “crutch” for autonomous vehicles. But his company has bought so many from Luminar that Tesla is now the lidar-maker’s top customer. Tesla accounted for “more than 10%” of Luminar’s revenue in the first quarter of 2024, or a little more than...

Tesla only counts air bag deployments as "crash". There are likely many many more.

Good journalism. Factual reporting and with two experts arguing pro:s and con:s.

Theoretically. There is no way of archiving software that warrant a humanoid form factor with current ML practices in my opinion. Robotics companies are betting that someone will provide a bunch of breakthroughs. In general, it helps to pretend the demos aren't performed by a humanoid robot to...

I don't count Tesla as.a player as of today. No one will do business with Elon, tbh, unless they can prove it works beforehand. Mike Ramsey, VP of automotive, transportation and cross-manufacturing at Gartner said said thei pretty much here recently...

Yes, I am aware of how Tesla currently works with maps for driving and navigation respectively.

Softbank lead the round. MS has invested before (probably mainly Azure credits) and Nvdia is new - I'm guessing they are keen on providing the onboard compute and perhaps other hardware/software products. I noticed that Wayve updated their product page: https://wayve.ai/product/wayve-ai-driver/...

You main thesis seem to be that HD-maps can improve performance. I think that's likely too. If maps meaningfully improves the performance of the ADAS/TaaS-solution that Tesla lets you subscribe to, then they should add the necessary hardware to the cars to collect and maintain such map data and...

I think one needs to consider how auto regulation works (eg NHTSA), how insurance works, how manufacturer liability works etc before suggesting hobby-level suggestions. The solution is clearly not to have the customers "mod" their cars' behaviour like described in Adequate Self Driving. If you...

A few examples of failures doesn't invalidate the idea that maps add safety and other value such as ride quality. Waymo's chief scientist on mapping: (31:50)

You said exactly that though. “Maps are bad” “Maps are good, will improve performance” I’ve explained in my first post why this isn’t feasible to do on-car and so on.

So you’re saying: 1) Maps are bad so TV is better without maps. 2) Maps where you drive would be better Which one is it? Maps good or maps bad? Crowd sourcing of maps is not a novel idea. Mobileye has done it for 5-6 years or more. These are no HD-maps though. You need better sensing and...

They are in a bounded game world where they play each other in a simple game with a clear reward. They train vs each other in simulation in millions of games and then the result is put into physical robots. With “real world” robotics you cant do reward based sim training really and you can’t...

Trying to explain ML and robotics in that sub was clearly a mistake. So much hopium, so little critical thinking.

The next decade I think is going to be really interesting in robots, the emergence of the robotics industry has been in the waiting for 10, 20 years without really emerging other than for pre-program behavior and stuff like that. And the main issue is, again, the Moravec paradox, how do we get...

Hard to say. I don’t know if he’s there anymore. He's clearly been instrumental in setting the risk profile and perseverance though. Shotwell seems great. The success depends on starship. Elon has bet the company on it. That’s known matter and doesn’t require major scientific breakthroughs...

So your reasoning is that Elon is not lying. He's just incredibly stupid, wrong and naive over ten years? And that happens to align with his goal of inflating the share price? I mean, it's not impossible just very improbable... 🤣🤣 However, I'm thinking a man with his resources would have access...

Narrow task robotics are better solved with an industry robot. Cheaper and faster by two orders of magnitudes. This is all smoke and mirrors. Try to disregard from the human form factor and judge the actual value produced.

I think the technical definition of the above would be "moving the goal posts". Tesla PR: This is a humanoid robot. It will do actual work in Tesla's factories next year. Hidden context: "as a counterweight in an elevator". The statement is true, so therefore Elon is not pumping the stock...