Hacker News Reader

Comment by peppertree | Hacker News Reader

After using FSD 13 for 2 weeks I'm convinced we are close to solving self driving. Too bad the everyone lost interest and now robotics is the hot new thing.

View on HN

As someone who worked in V&V for AV systems for a decade, it’s exactly the kind of thinking displayed here that has held back real assessment of AV safety for years.

There is absolutely no meaningful signal about a system’s safety that can be derived from one person using a system for two weeks.

At best it can only demonstrate that a system is wildly unsafe.

There is a very large chasm of 9s between one person being able to detect an unsafe system in two weeks of use and actually having a truly safe system.

And it only takes a (near) accident in 5 more minutes' driving to completely negate that.

Your observation from this short time window isn't enough to prove the usefulness of something as serious as life and death.

I’m not sure if you’re generalizing to a specific region in your assessment but regardless, I doubt this is anywhere close to a solved problem given the crashes/incidents (so far) still associated with the tech and the dependencies IIRC on street signs and other markers.

re: region, I’d like to see it take on more challenging conditions, like in India for example where things are chaotic even for human drivers. I doubt that it’ll survive over here.

> Too bad the everyone lost interest and now robotics is the hot new thing.

Self driving is robotics. Simple as that.

Quite simple robotics actually. Especially if you use Lidar. Basically IF (object present) THEN (do not go there) style of simplicity. Of course in reality there are lots of cases to consider, but each one of these cases is not rocket science.

Building a robot that can cook or fold a t-shirt, for example, is much harder.

Note that Nvidia is also working on self driving. The Jetson robotics platform is based on the same SoC as the DRIVE platform, but is a separate product.

Although the idea of self driving is obviously cool I think it's good that robotics take priority (if such a thing is possible) e.g. think of it like the invention of the washing machine as a liberating force on the world.

Have you been a passenger in a Waymo? My only ride felt safer than every uber / Lyft driver I have ever had pretty much, so wondering how it compares to a beta thing you have to be able to take over in an instant.

Last time I was in SF I took 3 waymo rides and attempted a fourth. The attempted one was cancelled after 15 minutes of waiting for it being 2 minutes away. As best as I can tell, the waymo was stuck at an intersection where power had been lost and didn't understand it needed to treat it like a 4 way stop.

2 rides went fine though neither was particularly challenging. The third though the car decided to head down a narrow side street where a pickup in front was partially blocking the road making a dropoff. There was enough space to just squeeze by and it was clear the truck expected the car to. A few cars turned in behind the waymo, effectively trapping it in as it didn't know how to proceed. The dropoff eventually completed and it was able to pull forward

Cars are robots without arms

Waymo already solved self driving years ago. Tesla still has a long way to go.

Is there anyone even close to Waymo in this game? Is Waymo going to own the entire market?

Cruise is close to Wayno, but nobody is willing to invest in Cruise anymore

Cruise doesn't exist anymore (or rather shortly won't). The teams are being folded into GM to work on other things.

Why would waymo own the entire market? Sure, they might be the first ones there, but every year recreating what is "good enough" should be cheaper and cheaper.

Baidu Apollo. They also have a commercial fleet without in-car safety drivers (they use remote operators for real time monitoring, though, so hard to say how hands-off it really is)

Waymo works in US grid cities on highly modified cars. I know people love hating Musk, but it is still very much up in the air if Waymo will be a better solution than what Tesla or Wayve is doing.

What does "grid" have to do with anything at this point? Mapping was done a million years ago and try and see if "grid" helps you understand the lane and traffic light system in San Francisco (which tourists need to figure out in real time - they are hard enough on the locals.)

Nice easy intersections. Wide two way streets. Put a waymo in Rome and I will be impressed.

I ride Waymos in San Francisco that traverse longer twisting "two-way" roads in the San Francisco hills (look at the neighborhoods around Mount Davidson). In these cases, the road, while two way, more often than not only has space which allows a single car to pass at a time; the rest of the space is taken by cars parked on either side of the roadway. The Waymo cars. at least during my rides, handled these situations well.

While it's not Rome, the operating areas for Waymo, at least in San Francisco, are not all grids of modern wide streets either.

That’s not gonna impress me either. There were zoox cars on Lombard st in sf I think. Windy streets are not the challenge. Putting your money where your mouth is - that’s the challenge.

That seems to be thing now: get some deployed mass out there. And at this point it seems that each new city is still a significant investment (well within Waymo but still).

I'm still puzzled on why Waymo insists on not having any remote driving or any remote advising cars on where/how to get themselves out of a situation. Yes that would cost a little more - but this is early stages so it's just a little more money at this point (lol) - in exchange for avoiding embarassing PR bullshit about cars self-honking at each other or rides stuck in infinite hesitation loop or not knowing what to do when there is a traffic cone on the hood. I haven't seen any convincing arguments for not having that. Anyone heard a legitimately good tech or liability reason? I doubt I would have missed it but...

Isn't that the difference though? I've never even seen a Waymo and I've been successfully driven by Teslas many times.

>n February 2024, a driverless Waymo robotaxi struck a cyclist in San Francisco.[132] Later that same month, Waymo issued recalls for 444 of its vehicles after two hit the same truck being towed on a highway

I am not entirely sure that is solved. And certainly not years ago. And it is only close in US where the data are trained. Doesn't mean it could be used in Japan ( where they are doing testing now ) driving on the different of the road with very different culture and traffics.

Citing a specific (tragic) incident isn’t really great evidence in re: safety. You have to normalize by something like accidents/mile driven and compare to comparable services (taxi/uber etc) - having said that I couldn’t quickly find any sources either positive or negative on those stats (besides Waymo PR docs) so I’m not saying you’re necessarily wrong. just wanted to point out the obvious flaw with citing anecdotal evidence for something like this.

You could easily use the same logic to say humans haven’t solved driving yet either!

A big part of me believes the only extra safety they give is they drive much slower. This in itself might be the solution for human deaths on the road.

Crashes per mile is multiple times lower than the human rate for both Waymo and Tesla. If your definition of solved is that there will be 0 collisions ever then the problem will never be solved. But if we have a system that is much better at driving than most humans, I think that qualifies it as good enough to start using.

In a very small subset of cities, road conditions, weather condition, &c. Basically US grid cities with 300 days of sun per year

That’s not the limiting factor, fwiw. It’s an operations problem for them at this point. Freeways are a big contentious point as well.

It 100% is a limiting factor, weather conditions and crazy roads/poor infra will definitely impact self driving cars, idk how it could not be a factor... Go drive in eastern Europe after a snow storm, you miss one snow covered sign and end up on the wrong side of a high speed road, &c.

It's like learning to code in JS on a 2024 MacBook pro and thinking you can "just" transfer your skills to cobol on 1970s hardware because both are "programming"

Extreme weather is a problem, as is snow. No doubt.

I’m simply talking about “300 days of sun” as being the limiting factor. You extrapolated the rest.

Waymo is pretty good (but not perfect) as far as safety, but there's too many ways it can get stuck. Including vandalism from humans like "coning". And if a significant number of them are on the road, it could gum up traffic when that happens.

I still think it'll do well because even if you need to hire 1 person to remotely monitor every 10 cars (I doubt Waymo has anywhere near that many support staff) it's still better than having to pay 10 drivers who may or may not actually be good at driving. But to really take over they'll need to be much more independent.

Reply on news.ycombinator.com