Hacker News Reader

Comment by xaduha | Hacker News Reader

Lidar approach just has to be the winner in my opinion, simpler is better.

A future with high powered infrared lasers on top of every car will probably lead to a mysterious rise in people slowly losing their vision. Someone should do a study of the eyes and vision of people who work with LIDAR.

Why not instead use multiple radars (we already put 1-2 radars in cars with adaptive cruise control) to augment Tesla's vision based approach?

According to this article, lidar using eye-safe lasers is standard: https://photonicsreport.com/blog/is-lidar-dangerous-for-our-...

"If a lidar system ever uses a laser with a higher safety class, then it could pose a serious hazard for your eyes. However, as it stands, according to current information and practices, lidar manufacturers use class 1 eye-safe (near) infrared laser diodes."

And what if you are hit in the eyes with the outout of 20 of them at once?

Like on a busy highway.

What if you are hit all day long, for hours, every single day?

History is replete with studies which were used to presume different exposures would be fine.

You sound like the kind of internet commenter who whinges about how dangerous bicycles are on the road because someone in a car might swerve to avoid them and head-on collide with a tractor trailer.

The math makes this quite implausible for a random scenario. Perhaps a toll collector could introduce the right systematics to make the scenario not ridiculous on it's face.

Imagine pointing a laser at something more than five feet distant the size of an eyeball and then getting another laser to point at the same spot at the same time. And then another five. And those lasers are all attached to cars moving and the eyeball is moving too...

Put an IR filter in the car window glass?

You’re right that it needs further studying, to not end as asbestos.

On a side note, you eyes are hit with more powerful and full-spectrum radiation (from IR to UV) from a burning star every day for hours (or at least they are designed to).

Agreed, during the day.

Eyes fully dilated during night driving don't constrict from flashes of coherent IR radiation, though.

BTW: this is neat: https://en.wikipedia.org/wiki/Melanopsin

Phototoxic maculopathy induced by quartz infrared heat lamp

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5279082/

Lidar is lo-res, though it has other advantages. Sensor fusion is also not without complexity and problems. But camera-only systems seem like one of those bets that hangs out there for too long before it will have to be walked back, leaving a lot of customers with stranded technology and unmet expectations.

FMCW radar is more tenable for automotive applications, in my opinion. No lasers necessary.

Way too basic. They don't even produce an image.

Look at a recent video of FSD beta on YouTube and focus your attention on the visualisation. I can’t see how anyone can look at that and conclude that LiDAR is still necessary or even incrementally beneficial.

I watched it, you might have rose-tinted glasses on, it's not ready yet. Compare it to AutoX videos. And you can't expect others to follow this path, it's not like Tesla is going to make it available to others.

I didn't say FSD beta was "ready". I merely assert that it is demonstrating remarkable competence in spatial sensing. If spatial sensing isn't a problem—and clearly it is not—it makes no sense to add LiDAR, which only does point cloud spatial sensing. The real "hard problem" of self-driving is in the path planner, not spatial sensing.

I hadn't heard of AutoX. I watched a couple of their videos. Despite being videos cherry-picked by the developer they show a car being driven in cities at slow speed and on faster arterials seemingly unable to drive without ping-ponging. No evidence that these cars could ever drive on roads they haven't been directly trained on.

By comparison, thousands of videos of FSD beta are being posted by hundreds of real world vehicle owners driving intentionally challenging routes in dozens of distinct geographic jurisdictions. It is showing remarkable progress, but of course there's still limitations; understandable when exposed to such a massive long tail of edge cases.

E.g. https://www.youtube.com/watch?v=0g4TnvvuGqM

The current SOTA in monocular “spatial sensing” (3D object detection) is just much less reliable compared to LiDAR-only systems.

You can’t conclude that spatial sensing isn’t a problem because it didn’t crash with a monocular camera. The test is whether it accurately detects all relevant actors/objects in the scene and does so reliably.

From a couple FSD videos I’ve seen, the issues center around perception (failing to get the right lane lines, certain detections disappearing briefly, etc.)

> You can’t conclude that spatial sensing isn’t a problem because it didn’t crash with a monocular camera. The test is whether it accurately detects all relevant actors/objects in the scene and does so reliably.

Indeed, which is why Tesla have put a lot of effort into showing a visual representation of its object detection. Unlike many (most) other players in this space, Tesla isn't afraid to let people see what their system is capable of in literally any situation their customers wish to drive to.

Most of their competitors are hiding behind L4 which, because of how machine learning works, makes systems appear better than they are. Overfitting is a great way to make rapid gains, but much of it is illusory.

> From a couple FSD videos I’ve seen, the issues center around perception (failing to get the right lane lines, certain detections disappearing briefly, etc.)

Even if true, LiDAR wouldn't help as it cannot see painted lane lines and is extremely weak at object identification.

Looking at the most recent FSD beta (10.12+) as shown on YouTube by real customers, it's quite clear that detection is already damn close to exceptional. When FSD does fail, it's usually either out of an abundance of caution (it is extremely cautious around pedestrians) or because the path planner wasn't making a good decision. It's almost never because the car has failed to see something that LiDAR might have.

> Overfitting is a great way to make rapid gains, but much of it is illusory.

I'm not sure why you think this is the case for L4. Do you mean since it can be geo/weather restricted the learning task is easier? Surely weather plays a large role but I don't see why other L4 companies necessarily overfit more (also how does overfitting provide rapid gains?). If you overfit when training for your L4 system...it's not an L4 system.

> Even if true, LiDAR wouldn't help as it cannot see painted lane lines and is extremely weak at object identification.

True, but:

A) Again, the test is reliable detection which is simply out of reach atm from single-sensor input

B) HD maps help a lot with this issue since the lane lines become less important

When you limit your service area, it becomes feasible to train your models on pretty much (or literally) every road in the service area. Every single permutation of weird intersection it will ever come across. If your L4 car only ever has to drive in San Francisco, it doesn't matter that this ML model would have no hope when introduced to Salt Lake City. The model never has to spend a moment excluding any of the millions of potential things it has never been trained on and will never see.

Both the ease of data collection and the absence of contextual noise makes L4 machine learning an order of magnitude easier.

Can you point me towards any recent FSD Beta video where there was an important failure of object detection?

Here's a list of robotaxi companies https://en.wikipedia.org/wiki/Robotaxi#Testing_and_revenue_s..., some of them operational, Level 4.

I would be very surprised of any of them use AI vision without Lidars. That's what my point is about, Tesla is lagging behind. Wake me up when it gets Level 4 certification.

The SAE levels don't measure technical sophistication. In fact when it comes to machine learning, an L4 system is substantially easier to implement than L3 or L2 because the task is monumentally less difficult and you can overfit your way to perceived success.

Whether you (or I, or anyone else) thinks Tesla is lagging behind or not has no bearing on the actual reality, and it won't change whether or not Tesla's approach is ultimately successful. Hopefully we're both fortunate to live a few more years and see what happens first hand.

LIDAR isn’t simpler.

Why not? Sure it's all complicated enough, but conceptually measuring distance with a laser is simple.

Do that many, many times and you have a good enough picture of the surrounding area in a machine-readable 3d format, so you have a good starting point.

How is that not simpler than what Tesla is doing?

Trains and public transport was the winner 100 years ago.

Simpler is better, so you want to add another sensor that humans don’t have, to do a task that only humans can currently do safely? This makes no sense to me.

Reply on news.ycombinator.com