Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Waymo pauses Atlanta service as its robotaxis keep driving into floods

techcrunch.com

Post reply on HN
Print view View original

271–280 of 498 posts

Re: Waymo pauses Atlanta service as its robotaxis keep driving into floods

#271

Post by fragmede »

quickthrowman wrote: post #259
Muromec wrote: post #215

Earlier quoted context omitted.

As much as one could expect waymo to train on it, one could also expect a functioning city to not have flooded streets

It would be a massive waste of resources to build out every city with a drainage system capable of handling any amount of rain. Houston had ~30 inches of water dumped on it during a somewhat recent hurricane, designing and building infrastructure for that level of storm is not realistic. I’m not familiar with storm sewer capacity design, but I’m confident they aren’t designed to flawlessly handle a 1 in 500 or 1 in 1…

Tell that to Fukushima.

Re: Waymo pauses Atlanta service as its robotaxis keep driving into floods

#272

Post by ai-x »

mschulkind wrote: post #124
ai-x wrote: post #98

Earlier quoted context omitted.

I thought about this and I think it boils to how the model is trained. Tesla trains it models from actual drivers purely based on (input) Vision and (output) actuators - Brake, Steering, Accelerators. Human output is based on what they and the camera sees. So, it's a 1:1 match. If Waymo were to do that, it'll muddle the training set. The Lidar input may override camera input. I always struggled when Musk mentioned Li…

This is an interesting viewpoint, but isn't it also solveable? Just because the human in the scenario only took vision as input, why does that matter to the training data and the model? The actions are the same. To put it another way, what about all the cultural context the human had, or the sounds, smells, past experiences at the same intersection, etc? Even Tesla can't record this, but I'm not sure that matters.

E.g If the driver brakes because they saw a pothole, and Lidar captures someone biking 200m away on their own path, it may mistakenly put more weight on brake causation to the 200m away object (because large moving object) vs the pothole.

I'm exaggerating, but I hope you get the point. It isn't even conflicting sensor signals about the pothole, but conflicting information about the causation. With vision only there is no conflict for the training data. This was my Aha moment. Multiple Sensors are absolutely important for fallback and extra safety, but screws up training that are based on Human Drivers

I think Elon himself doesn't understand this and hence can't articulate it, while just repeating whatever his ML engineer has said.

Re: Waymo pauses Atlanta service as its robotaxis keep driving into floods

#273

Post by hnthrow0287345 »

fwip wrote: post #246
jfim wrote: post #245

Earlier quoted context omitted.

There's a long tail of unpredictable events in the AV industry that you end up seeing, especially since the cars in aggregate end up driving more than one could over a lifetime. At a previous employer, we've seen anything from cars getting mooned, a SUV slowly driving past the AV, the rear window roll down, and someone poke their head out and start throwing dollar bills at the AV, a convention of people dressed up in…

A flooded road is a very predictable event, though.

My guess is this was brought up but getting the product out there was more important to the business so it got ignored.

Now that it's a problem for them, they get to hide behind an "oops sorry, let's fix the really obvious thing now", almost like taking "if it ain't broke, don't fix it" to malicious levels.

This jives with CRUD software in general, where people are not usually rewarded for preventing future issues and instead rewarded for waiting until it's a visible problem and then fixing it.

Re: Waymo pauses Atlanta service as its robotaxis keep driving into floods

#274

Post by fragmede »

overfeed wrote: post #270
happyPersonR wrote: post #257

Earlier quoted context omitted.

This guy knows what he’s talking about. Born and raised in GA, it wasn’t until I moved to CA, the bay specifically, after college that I realized things like flood warnings multiple times a month and, flooded out roads during the summer weren’t just part of life lolll My ex moved to ATL from Seattle, and it was just WILD watching her go… “you guys have RAIN, here… like it comes down HARD” When Waymo came here and als…

> My ex moved to ATL from Seattle, and it was just WILD watching her go… “you guys have RAIN, here… like it comes down HARD” So it makes sense to first rollout to a place with frequent, lighter rain - no? As an outsider, Waymo's approach seems to be solving challenges step-by-step, and the criticism in this thread is asking why it hasn't already solved the hardest cases. > The data was there but it wouldn’t surprise…

You don't understand! Google is trying to do something difficult, and because they haven't solved all possible theoretical problems with it, they should just give up and go home and never try anything difficult ever.

Re: Waymo pauses Atlanta service as its robotaxis keep driving into floods

#275

Post by estearum »

autoexec wrote: post #248
RealityVoid wrote: post #225

Earlier quoted context omitted.

This feels disingenuous to the extreme. Yes, chances are that some people will die run over by a Waymo. Put enough miles in one and someone will die eventually. Compare the numbers to human drivers. Would you, if they had LESS fatality rates than human drivers, say that the difference is "lives saved"? - I don't think you would. In 5 years, after someone is eventually fatally injured you'll just jump up and say "AHA!…

> In 5 years, after someone is eventually fatally injured you'll just jump up and say "AHA! Told you Waymos are unsafe!" That'll depend on the circumstances. If someone is killed because of a mistake a human wouldn't have made (like driving into oncoming traffic or down a light rail track) it'll be entirely their fault. Even if they do something humans sometimes do but never should like running a red light I'd argue…

It is honestly kind of funny how clearly your comments read like motivated reasoning, and then one just looks at your username.

Re: Waymo pauses Atlanta service as its robotaxis keep driving into floods

#276

Post by jvanderbot »

VoidWhisperer wrote: post #8
jvanderbot wrote: post #5

Snark aside, there will probably always be conditions in which waymo is not the right answer. Are they going to do hurricane evacuation? I think removing the driver just necessitates this.

While this is going to be an overly optimistic scenario: Imagine how smooth a hurricane evacuation would go if _everyone_ used a self-driving car to do the evacuation - atleast there might be less gridlock than there is during any usual hurricane evacuations. And assuming the self driving cars don't do something stupid that causes every car behind it to essentially lock up and stop moving That said, I know a scenario…

Until the first traffic cop has to manually direct because lights are out or there was an accident

Re: Waymo pauses Atlanta service as its robotaxis keep driving into floods

#277

Post by mxkopy »

ACCount37 wrote: post #120
etempleton wrote: post #90

This is really my bear case against AI. I am not against it. I actually think it is really neat! But we have been working on driverless cars for how long and spent how much? And still things like a flooded roadway completely throw them. Tesla failed to deliver driverless cars but now is pivoting to the much more complex fully autonomous robots. And we can’t get AI to stop hallucinating facts, but any day we are going…

The "responsible adults" know that chasing perfection gets you nowhere fast. A part of growing up is learning to put up with "good enough". A car that only fails in a road conditions edge case is good enough for the vast majority of cases. You accept that, and issue a manual override for when that edge case pops up. Then you add that edge case to your training sets. Then the issue never comes up again. If you think t…

The false equivalence of emotional maturity with being able to chase production is really telling.

When people think of autonomous driving as a solved problem it evokes something very specific. It means vehicles can drive on their own, without guidance. Until you solve AVs you don’t have a claim to present whatever you actually have as such. There’s no “good enough” for AVs, you’ve either solved them or you haven’t.

Re: Waymo pauses Atlanta service as its robotaxis keep driving into floods

#279

Post by fastball »

venussnatch wrote: post #87
fastball wrote: post #72

Earlier quoted context omitted.

This is why I personally feel like Tesla's approach is more likely to "win". The fundamental blocker to self-driving cars is not sensing / sensor fusion, it is intelligence. And the Tesla approach seems much more likely to achieve functional intelligence than Waymo's.

You can have intelligence with lidar. You can have even more intelligence with both.

You can have your cake and eat it too?

Sensor fusion isn't free. Lidar requires more power consumption and more onboard compute. Cycles that could be spent on "intelligence" are instead being spent on sensing.

Re: Waymo pauses Atlanta service as its robotaxis keep driving into floods

#280

Post by sarchertech »

dhbradshaw wrote: post #123

To me this doesn't seem like a disaster but just the kind of thing that happens as you role out a service and expose it to new challenges. Presumably they haven't had the chance to do a lot of flood training but now they have that chance. The huge advantage they have over people in general is that ideally if they figure this out then it will stay figured out. Then they can slowly role out and watch for the next hitch…

> The huge advantage they have over people in general is that ideally if they figure this out then it will stay figured out. Then they can slowly role out and watch for the next hitches from new situations.

This is also a huge disadvantage because any flaws in the software that don’t show up in a slow rollout will be present in every single car.

It’s a contrived example, but say a new billboard campaign rolls out that causes every car to immediately veer away from it.

Post reply on HN