[jandrewrogers]

The long-term view of LIDAR was not so much that it was expensive, though it was at the time. The issue is that it is susceptible to interference if everyone is using LIDAR for everything all the time and it is vulnerable to spoofing/jamming by bad actors.

For better or worse, passive optical is much more robust against these types of risks. This doesn't matter much when LIDAR is relatively rare but that can't be assumed to remain the case forever.

[tim333]

I hadn't heard that criticism. You can get multiple Waymos near each other without them crashing into things.

[vachina]

When I see Waymos fail they usually fail together

[catgirlinspace]

Doesn’t mean they’re failing because of interfering lidar though. If it’s something like them failing due to the road being blocked or something, it makes sense they’d fail together. Assuming they’re on the same OS, why would one know how to handle that situation and another not?

[consumer451]

I am just some schmoe, but optics alone can be easily spoofed as any fan of the Wile E. Coyote has known for decades. [0]

What's crazy to me is that anyone would think that anything short of ASI could take image based world understanding to true FSD. Tesla tried to replicate human response, ~"because humans only have eyes" but largely without even stereoscopic vision, ffs.

[0] https://www.youtube.com/watch?v=IQJL3htsDyQ

[wongarsu]

But optical illusions are much less of an issue because humans understand them and also suffer from them. That makes them easier to detect, easier to debug, and much less scary to the average driver.

Sure, someone can put up a wall painted to look like a road, but we have about a century of experience that people will generally not do that. And if they do it's easy to understand why that was an issue, and both fixing the issue (removing the mural) and punishing any malicious attempt at doing this would be swift

[consumer451]

> and punishing any malicious attempt at doing this would be swift

Is this a joke? Graffiti is now punishable and enforced by whom exactly? Who decides what constitutes an illegal image? How do you catch them? What if vision-only FSD sees a city-sanctioned brick building's mural as an actual sunset?

So you agree that all we need is AGI and human-equal sensors for Tesla-style FSD, but wait... plus some "swift" enforcement force for illegal murals? I love this, I have had heath issues recently, and I have not laughed this hard for a while. Thank you.

Hell, at the last "Tesla AI Day," Musk himself said ~"FSD basically requires AGI" - so he is well aware.

[wongarsu]

Intentionally trying to create traffic accidents is illegal. This isn't an FSD-thing. If you try to intentionally get humans to crash their cars you are going to get into trouble. I don't see how this suddenly becomes OK when done to competent FSD (not that I'd count Tesla among them)

[consumer451]

If I understand your argument correctly, then posting a sign that it is incorrect.. like a wrong way highway on-ramp sign, would be illegal? That sounds correct.

But what if your city hired you to paint a sunset mural on a wall, and then a vision-only system killed a family of four by driving into it, during some "edge case" lighting situation?

I would like to think that we would apply "security is an onion" to our physical safety as well. Stereo vision + lidar + radar + ultrasonic? Would that not be the least that we could do as technologists?

[wongarsu]

I did only talk about malicious attempts being punished, and the sunset mural would not be malicious

[dham]

That was autopilot not FSD. Autopilot is a simple ADAS system similar to Toyota Safety sense or all the other garbage ADAS systems from Honda, Kia, Toyota, GM etc. FSD passed this test with flying colors

https://www.youtube.com/watch?v=TzZhIsGFL6g

[solumunus]

Why isn’t the solution a combination of both?

[ycui1986]

everyone uses cellphone that transmit on the same frequency. they don't seem to cause interference. once enough lidar enters real word use. there will be regulation to make them work with each other.

[jandrewrogers]

Completely different problem domains. A mobile phone is interacting with a fixed point (i.e. cell tower) that coordinates and manages traffic across cell phones to minimize interference. LIDAR is like wifi, a commons that can be polluted at will by arbitrary actors.

LIDAR has much more in common with ordinary radar (it is in the name, after all) and is similarly susceptible to interference.

[CamperBob2]

No, LIDAR is relatively trivial to render immune to interference from other LIDARs. Look at how dozens of GPS satellites share the same frequency without stepping on each others' toes, for instance: https://en.wikipedia.org/wiki/Gold_code

Like GPS, LIDAR can be jammed or spoofed by intentional actors, of course. That part's not so easy to hand-wave away, but someone who wants to screw with road traffic will certainly have easier ways to do it.

[addaon]

> No, LIDAR is relatively trivial to render immune to interference from other LIDARs.

For rotating pulsed lidar, this really isn't the case. It's possible, but certainly not trivial. The challenge is that eye safety is determined by the energy in a pulse, but detection range is determined by the power of a pulse, driving towards minimum pulse width for a given lens size. This width is under 10 ns, and leaning closer to 2-4 ns for more modern systems. With laser diode currents in the tens of amps range, producing a gaussian pulse this width is already a challenging inductance-minimization problem -- think GaN, thin PCBs, wire-bonded LDs etc to get loop area down. And an inductance-limited pulse is inherently gaussian. To play any anti-interference games means being able to modulate the pulse more finely than that, without increasing the effective pulse width enough to make you uncompetitive on range. This is hard.

[CamperBob2]

I think we may have had this discussion before, but from an engineering perspective, I don't buy it. For coding, the number of pulses per second is what matters, not power.

Large numbers of bits per unit of time are what it takes to make two sequences correlate (or not), and large numbers of bits per unit of time are not a problem in this business. Signal power limits imposed by eye safety requirements will kick in long after noise limits imposed by Shannon-Hartley.

[addaon]

> For coding, the number of pulses per second is what matters, not power.

I haven't seen a system that does anti-interference across multiple pulses, as opposed to by shaping individual pulses. (I've seen systems that introduce random jitter across multiple pulses to de-correlate interference, but that's a bit different.) The issue is you really do get a hell of a lot of data out of a single pulse, and for interesting objects (thin poles, power lines) there's not a lot of correlation between adjacent pulses -- you can't always assume properties across multiple pulses without having to throw away data from single data-carrying pulses.

Edit: Another way of saying this -- your revisit rate to a specific point of interference is around 20 Hz. That's just not a lot of bits per unit time.

> Signal power limits imposed by eye safety requirements will kick in long after noise limits imposed by Shannon-Hartley.

I can believe this is true for FMCW lidar, but I know it to be untrue for pulsed lidar. Perhaps we're discussing different systems?