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

Waymo CEO dismisses Tesla self-driving plan: “This is not how it works”

arstechnica.com

Post reply on HN
Print view View original

251–260 of 581 posts

Re: Waymo CEO dismisses Tesla self-driving plan: “This is not how it works”

#251

Post by smaudet »

screye wrote: post #35

I worked with a self-driving product team a couple of years ago. Tesla's approach was looked upon with suspicion and often openly mocked. Tesla's entire self-driving strategy is led by Karpathy, who while brilliant, is an under-experienced researcher with a narrow expertise in large scale 2D CNNs. Tesla (and Comma.ai ?) seem to be the only groups that find Lidar unnecessary for L5 self-driving. I completely agree wit…

The only times I have ever had accidents or close calls in a car (where another driver was not at fault for e.g. reckless driving behavior) has been with cruise control or with auto keep lane - the former I spun out, the later it tried to drive me off of a highway at 70 mph repeatedly.

The sorts of assistive technologies that would actually help are not really being worked on, assistive has been misunderstood as meaning "allowing the driver to be lazier".

Sure - if Tesla pulls it off, it's great. But it's a gamble with more than just Teslas stock, that is what makes Elon an unethical sociopath here...

Waymo's approach is assuredly safer, there is still the possibility of technical glitches but it the approach doesn't create more problems with humans than it solves.

Yes, sure, humans can be dangerous, but if that's the problem then it's safest to just not get on the road period, kinda negates the whole point of an FSD though...

Re: Waymo CEO dismisses Tesla self-driving plan: “This is not how it works”

#252

Post by perryizgr8 »

screye wrote: post #35

I worked with a self-driving product team a couple of years ago. Tesla's approach was looked upon with suspicion and often openly mocked. Tesla's entire self-driving strategy is led by Karpathy, who while brilliant, is an under-experienced researcher with a narrow expertise in large scale 2D CNNs. Tesla (and Comma.ai ?) seem to be the only groups that find Lidar unnecessary for L5 self-driving. I completely agree wit…

> I avoid using cruise control and auto-lane-keep for the same reasons. Either I am in control or I am not.

Do you also turn off features like ABS, ESC, traction control etc?

Re: Waymo CEO dismisses Tesla self-driving plan: “This is not how it works”

#253

Post by fomine3 »

codebolt wrote: post #53
majewsky wrote: post #17

Earlier quoted context omitted.

> There are enough tail events while driving where you can only make "the right decision" if you actually understand what's going on. I have the feeling this bar disqualifies most human drivers, too.

Human drivers fail reliably, in ways that are understandable. Non-AGI artificial intelligence based on a mix of heuristics and dumb statistics (aka 'deep learning') will fail in highly unpredictable and to a human eye completely foolish ways. The day these cars are let on the road in my area is the day I start taking the train.

I saw many unexpected driver's action on car crash YouTube channel.

Re: Waymo CEO dismisses Tesla self-driving plan: “This is not how it works”

#254

Post by JohnJamesRambo »

anonytrary wrote: post #191
screye wrote: post #35

I worked with a self-driving product team a couple of years ago. Tesla's approach was looked upon with suspicion and often openly mocked. Tesla's entire self-driving strategy is led by Karpathy, who while brilliant, is an under-experienced researcher with a narrow expertise in large scale 2D CNNs. Tesla (and Comma.ai ?) seem to be the only groups that find Lidar unnecessary for L5 self-driving. I completely agree wit…

I think you hit the nail on the head. There is no such thing as "sort of self driving". It's either completely autonomous or I'm going to have to be paying some attention and I might as well be doing all of the driving at that point.

This is how I feel. My brain is pretty damn autonomous while driving. If I have to still pay attention to my “self-driving” car, that seems both more dangerous and taxing than just letting my brain do all the driving.

Re: Waymo CEO dismisses Tesla self-driving plan: “This is not how it works”

#255

Post by anonytrary »

darepublic wrote: post #190

Why not develop roads that act like tracks and can be guided by an algorithm. Seems like a simpler problem albeit less robust than autonomous vehicle

What happens when a deer crosses the road?

What happens when there's a fire along the highway (e.g California)?

What happens when there's debris and objects in the road?

What happens if a car needs to make an emergency stop?

Where do you place these new roads?

Do they require roads to be re-built or can they be added to existing roads?

What is the cost of replacing the entire highway and city infrastructure?

The AV companies decided that the simpler problem is to have AI that works on highways, then incrementally teach it how to handle more complex situations. This is the most additive solution, whereas your solution is not necessarily additive and does not necessarily preclude the need for AI.

Re: Waymo CEO dismisses Tesla self-driving plan: “This is not how it works”

#257

Post by raldi »

jiofih wrote: post #140
raldi wrote: post #14

He seems so obviously correct that I can’t fathom how any engineer could see things otherwise.

Humans, surprisingly, don’t have LiDar, yet they can drive.

I’m not sure what that has to do with my comment. Neither I nor the Waymo CEO said you need lidar to drive.

Re: Waymo CEO dismisses Tesla self-driving plan: “This is not how it works”

#258

Post by cycrutchfield »

joeblau wrote: post #245
kerng wrote: post #233

So much hate towards Tesla here. I'm more neutral - dont have a strong opinion on this. But from Google's messaging it seems that the Waymo CEO is maybe a bit defensive (scared?). Tesla has probably the long term better strategy which leads to more accurate up to date data to constantly improve their system. Tesla's crowdsourcing strategy might be the winning horse in this race.

The term you’re looking for is “Federated Machine Learning.” The more cars Tesla has, the faster they can train the neural networks edge cases.

That’s not really what federated ml means, unless they are doing the gradient updates from the cars directly.

Re: Waymo CEO dismisses Tesla self-driving plan: “This is not how it works”

#260

Post by anonytrary »

staticman2 wrote: post #202
paul7986 wrote: post #52

Im anti-robot car ... so many ppl will die for progress with this. I'm even more after my friend's Jeep Compass's AI system decided to lock up her brakes and steering wheel as she was turning into the bathroom area after the Golden Gate Bridge. The AI said nope im taking over your going to crash and thus you cant maneuver your way out of hitting the highway wall which she did and total that piece of junk; all bruised…

My Dad avoided a crash where he was nearly killed with children in the car and believes he did it because an angel took over the steering wheel and corrected his driving error faster than he could. How is this relevant to you friend's story, you may ask? It's relevant to show humans are not necessarily reliable narrators of why a crash occurred.

Your dad was in control of the wheel and made a subconscious/reflexive decision and avoided the crash. Whether he chalks it up to an angel or not is completely irrelevant. It is not about human narrative or beliefs, it is about whether or not a human was controlling the wheel.
Post reply on HN