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

Sub-$200 Lidar could reshuffle auto sensor economics

spectrum.ieee.org

Post reply on HN
Print view View original

551–560 of 582 posts

Re: Sub-$200 Lidar could reshuffle auto sensor economics

#551

Post by leptons »

moogly wrote: post #539
leptons wrote: post #535

Earlier quoted context omitted.

Why would you assume I "forgot" about any of that? It's implied. That's what "infer" means in that sentence. Of course it requires a brain and nervous system. Maybe you don't know what the word "infer" means?

That fake indignation doesn't change the fact that you equated two cameras with two eyes, and if we're going that hard on semantics, you used the word produce for the cameras. > One camera can't really produce depth/distance information, but two cameras sure can.

This pointless conversation you created is over. If you get your jollies by intentionally misunderstanding comments and pointing out your miscomprehension, then you're talking to the wrong person. I really can't imagine what you think you're getting out of this interaction.

Re: Sub-$200 Lidar could reshuffle auto sensor economics

#552

Post by JumpCrisscross »

galangalalgol wrote: post #143
zelphirkalt wrote: post #119

Since lidar has distance information and cameras do not, it was always a ridiculous idea by a certain company to use cameras only. Lidar using cars are going to replace at least the ones that don't make use of this obvious answer to obstacle detection challenges.

The reasoning is cynical but sound. If the system uses only the sensing modes people have, it will make the mistakes people do. If a jury thinks "well I could have done that either!" You win. It doesn't matter if your system has fewer accidents if some of the failure modes are different than human ones, because the jury will think "how could it not figure that out?"

> If a jury thinks "well I could have done that either!" You win

“A federal judge” recently “rejected Tesla's request to overturn a $243 million jury verdict over the 2019 crash of an Autopilot-equipped Model S” [1]. If a human supervising still incurs liability, human-like errors, particularly if Waymo and BYD aren’t making them, is a poor defense.

[1] https://www.reuters.com/world/us-judge-upholds-243-million-v...

Re: Sub-$200 Lidar could reshuffle auto sensor economics

#553

Post by kube-system »

throw10920 wrote: post #540
kube-system wrote: post #483

Earlier quoted context omitted.

Humans have both issues. There are many human failures which are distinctly a vision issue and not attention related, e.g. misestimation of depth/speed, obscured or obstructed vision, optical focus issues, insufficient contrast or exposure, etc.

> There are many human failures which are distinctly a vision issue and not attention related Which are a tiny minority. The largest causes of crashes in the US are attention/cognition problems, not vision problems. Most traffic systems in western countries (probably in others, too, but I don't have personal experience), and in particular the US, are designed to limit visibility problems and do so very effectively.

That sounds more like a personal opinion, because I don’t think that data is particularly easy to objectively collect.

Regardless it is irrelevant to the point. Whatever the number may be, lapses in human visual perception are responsible for some crashes

Re: Sub-$200 Lidar could reshuffle auto sensor economics

#555

Post by rhubarbtree »

culi wrote: post #424
rhubarbtree wrote: post #421

Just to be clear, this article is talking about the possibility that this might happen one day. LiDAR remains prohibitively expensive AFAICT.

This is not quite true. It depends what you're talking about. Automotive LiDAR sensor prices typically range from $150–300 today for standard units. Mid-range ADAS systems (Ls+/L3) sensors are about $600-750 and the long-range units used by robo-taxis like Waymos are about $1,500–6,000 or more per sensor https://www.fleetowner.com/technology/article/55316670 The ~$75k per sensor in 2015 refers to the long-range senso…

I think the sensible cost to take is the latest cost in a Waymo model, as that's what "works" for automated driving [1].

I see numbers anywhere between $10K-$30K for that.

[1] to a limited extent in a limited area - it's inaccurate to say that Waymo vehicles are fully autonomous.

Re: Sub-$200 Lidar could reshuffle auto sensor economics

#556

Post by Someone »

tsimionescu wrote: post #195
Someone wrote: post #167

Earlier quoted context omitted.

> Since lidar has distance information and cameras do not, it was always a ridiculous idea by a certain company to use cameras only Human eyes do not have distance information, either, but derive it well enough from spatial (by ‘comparing’ inputs from 2 eyes) or temporal parallax (by ‘comparing’ inputs from one eye at different points in time) to drive cars. One can also argue that detecting absolute distance isn’t n…

Human/animal vision uses way more than parallax to judge distances and bearings - it uses a world model that evolved over millions of years to model the environment. That's why we can get excellent 3D images from a 2D screen, and also why our depth perception can be easily tricked with objects of unexpected size. Put a human or animal in an abstract environment with no shadows and no familiar objects, and you'll see…

> it uses a world model that evolved over millions of years to model the environment. That's why we can get excellent 3D images from a 2D screen

That doesn’t require millions of years of evolution. We can ‘evolve’ it way faster on computers.

For an example, see https://depth-anything.github.io/.

I also think we don’t need good depth estimation to avoid collisions while walking around. The problem is scale-invariant except for the fact that deceleration is superlinear (doubling your speed more than doubles stopping distance), but at walking speed, that effect isn’t very large.

Decent depth estimation is needed for judging foot placement, but that’s at relatively close range.

At driving speed, that changes, but I think you can still get away with rough estimates.

(I’m not saying one shouldn’t use LiDAR, just arguing that we don’t know whether “LiDAR is necessary” is true. Yes, cameras cannot reproduce all aspects of human vision yet, but they also can surpass many aspects of human vision. Examples are resolution and field of view)

Re: Sub-$200 Lidar could reshuffle auto sensor economics

#557

Post by utopcell »

runjake wrote: post #328
zelphirkalt wrote: post #119

Since lidar has distance information and cameras do not, it was always a ridiculous idea by a certain company to use cameras only. Lidar using cars are going to replace at least the ones that don't make use of this obvious answer to obstacle detection challenges.

Karpathy provided additional context on the removal of LiDAR during his Lex Fridman Podcast appearance. This article condenses what he said: https://archive.is/PPiVG And here's one of Elon's mentions (he also has talked about it quite a bit in various spots). https://xcancel.com/elonmusk/status/1959831831668228450?s=20 Edit: My personal view is that LiDAR and other sensors are extremely useful, but I worked on aircra…

From the article:

Karpathy’s main points: Extra sensors add cost to the system, and more importantly complexity. They make the software task harder, and increase the cost of all the data pipelines. They add risk and complexity to the supply chain and manufacturing. Elon Musk pushes a philosophy of “the best part is no part” which can be seen throughout the car in things like doing everything through the touchscreen. This is an expression of this philosophy. Vision is necessary to the task (which almost all agree on) and it should also be sufficient as well. If it is sufficient, the cost of extra sensors and tools outweighs their benefit. Sensors change as parts change or become available and unavailable. They must be maintained and software adapted to these changes. They must also be calibrated to make fusion work properly. Having a fleet gathering more data is more important than having more sensors. Having to process LIDAR and radar produces a lot of bloat in the code and data pipelines. He predicts other companies will also drop these sensors in time. Mapping the world and keeping it up to date is much too expensive. You won’t change the world with this limitation, you need to focus on vision which is the most important. The roads are designed to be interpreted with vision.

Re: Sub-$200 Lidar could reshuffle auto sensor economics

#558

Post by utopcell »

utopcell wrote: post #557
runjake wrote: post #328

Earlier quoted context omitted.

Karpathy provided additional context on the removal of LiDAR during his Lex Fridman Podcast appearance. This article condenses what he said: https://archive.is/PPiVG And here's one of Elon's mentions (he also has talked about it quite a bit in various spots). https://xcancel.com/elonmusk/status/1959831831668228450?s=20 Edit: My personal view is that LiDAR and other sensors are extremely useful, but I worked on aircra…

From the article: Karpathy’s main points: Extra sensors add cost to the system, and more importantly complexity. They make the software task harder, and increase the cost of all the data pipelines. They add risk and complexity to the supply chain and manufacturing. Elon Musk pushes a philosophy of “the best part is no part” which can be seen throughout the car in things like doing everything through the touchscreen.…

So the argument is pretty much: it should be sufficient to use vision only, and that it is too difficult / expensive to do otherwise.

Re: Sub-$200 Lidar could reshuffle auto sensor economics

#559

Post by fragmede »

imtringued wrote: post #122
fragmede wrote: post #60

Earlier quoted context omitted.

It's fair to point out that comma.ai is SAE level two system, however it's not geofenced at all, which is an SAE level 5 requirement. But really that brings up the fact that SAE's levels aren't the right ones, merely the ones they chose to define since they're the standards body. A better set of levels are the seven I go into more detail about on my blog. As far as distinguishing shadows on the road, that's what rada…

Your autonomy scale is pretty arbitrary and encodes assumptions about the underlying technology and environments the vehicle is supposed to implement and operate in. The SAE autonomy scale is about dividing responsibility between the driver and the assistance system. The lowest revel represents full responsibility on the driver and the highest level represents full responsibility on the system. If there is a geofence…

Thanks for reading! They are arbitrary, but they're just as arbitrary as SAE's levels, it's just that I'm not a standards body or anything. I am more interested in the underlying technology than the politics of it all because that's what's more interesting to me. Maybe if I update to call them milestones or some other analogy rather than levels I can get it to stick.

Re: Sub-$200 Lidar could reshuffle auto sensor economics

#560

Post by aylons »

venusenvy47 wrote: post #455
aylons wrote: post #299

Earlier quoted context omitted.

Binocular vision is not only relevant for driving (well, maybe for the steering wheel, but that's not the point).

It gives us depth perception. And moving the eyes and/or head gives the depth perception over a wide field of view.

What I mean is that binocular vision just give us depth perception for a meter or so - about around where our hands can touch.

Moving the head/body goes a little further, but that was not my point.

Post reply on HN