LiDAR - NASA confirms cameras better than LiDAR - Musk is correct

[mikes_fsd]

Camp 1: Elon firmly believes that camera vision and machine learning are all you need to do FSD. He believes this because humans only use 2 eyes and a brain to drive and because he has tremendous confidence that machine learning and AI have improved enough to do the job. Tesla "fanboys" automatically trust and believe Elon is right just because he is Elon.

Camp 2: Everybody else

Here you go with your false statements again.
I've listed other experts and leaders in the field that have come out stating that LiDAR is not the answer, but that does not fit your narrative.
You put the entire "Camera only" approach on Elon's shoulders.
Either way, your group think approach of LiDAR is going to crash and burn along with Waymo and anyone that demands that LiDAR must be used.

You can stop with the lies, we have enough fake news as it is.

 

[diplomat33]

Here you go with your false statements again.
I've listed other experts and leaders in the field that have come out stating that LiDAR is not the answer, but that does not fit your narrative.
You put the entire "Camera only" approach on Elon's shoulders.

I think Levandowski has also said that he does not think lidar is needed. Who else?

My point is that Elon is the only major auto maker that I know of that is making a serious attempt at doing FSD with just cameras. So yes, I put the entire "camera only" approach on his shoulders.

Either way, your group think approach of LiDAR is going to crash and burn along with Waymo and anyone that demands that LiDAR must be used.

 

[mikes_fsd]

Elon is the only major auto maker

Waymo is not an automaker neither is MobileEye!

Automakers can't even switch over to BEV platforms, and you expect them to accomplish FSD on their own?!

 

[diplomat33]

Waymo is not an automaker neither is MobileEye!

Right, Waymo and Mobileye are autonomous driving companies.

And answer my question: you said there are plenty of experts who also reject lidar. So who, other than the cheat Levandowski is saying that lidar is not needed? Which auto maker or SDC, other than Tesla, is trying to do autonomous driving for cars with just cameras?

 

[mikes_fsd]

So who, other

Sebastian Thrun original person running predecessor to Waymo.
His stance is that with the advancements in computer vision LiDAR is not necessary. This is the same person who led the team that won 2005 DARPA challenge in self driving. (2nd place in 2007)

I've linked to the articles in other threads (in reply to you), I guess you weren't that interested.

 

[diplomat33]

Some good information that I think sheds some light on lidar and why Tesla does not use it.

TLDR: Lidar is awesome but when Tesla started work on FSD, they could not afford to put lidar on 500k cars so they were forced to go with camera vision and try to make it work.

Why did LIDAR take off with self-driving cars? In a word: mapping.
LIDAR allows you to generate huge 3D maps (its original application!), which you can then navigate the car or robot predictably within. By using a LIDAR to map and navigate an environment, you can know ahead of time the bounds of a lane, or that there is a stop sign or traffic light 500m ahead. This kind of predictability is exactly what a technology like self-driving cars requires, and has been a big reason for the progress over the last 5 years.

Object Detection
As LIDARs have become higher-resolution and operate at longer ranges, a new use-case has emerged in object detection and tracking. Not only can a LIDAR map enable you to know precisely where you are in the world and help you navigate it, but it can also detect and track obstacles like cars, pedestrians and according to Waymo, football helmets.

Modern LIDAR enables you to differentiate between a person on a bike or a person walking, and even at what speed and which direction they are going in.

The combination of amazing navigation, predictability and high-resolution object tracking has meant that LIDAR is the key sensor in self-driving cars today, and it’s hard to see that domination changing. Unless…

Camera-Powered Cars
There’s a number of startups out there approaching the problem of self-driving cars using purely cameras (and perhaps radar), with no LIDAR in sight. Tesla is the biggest company of the bunch, and Elon Musk has repeatedly pushed the idea that if humans can perceive and navigate the world using just eyes, ears and a brain, then why can’t a car? I’m certain that this approach will achieve amazing results, especially as other talented teams work toward this goal, including Comma and AutoX.

It’s important to note that Tesla has an interesting constraint that may have factored in to their decision: scale. Tesla hopes to ship 500k cars a year very soon, and can’t wait for LIDAR to come down in cost (or be manufactured in volume) tomorrow, it needed to happen yesterday!

Source:
An Introduction to LIDAR: The Key Self-Driving Car Sensor

 

[Uncle Paul]

I guess you can make any case you want, but the reality is that this is Elon's company, and he gets to make the decisions.

If you feel strongly that Lidar would be better accepted by the buying public, then start your own car company.

First show us a currently available to the consumer car, that uses Lidar to provide better self driving than Tesla.

 

[Jeff Hudson]

Well stated Uncle Paul.

 

[diplomat33]

I guess you can make any case you want, but the reality is that this is Elon's company, and he gets to make the decisions.

If you feel strongly that Lidar would be better accepted by the buying public, then start your own car company.

First show us a currently available to the consumer car, that uses Lidar to provide better self driving than Tesla.

Your post comes across as a pretty snarky.

The fact is that Autopilot is the #1 reason why I bought my Model 3 (and Tesla makes the best EVs). I felt like Autopilot was the best general driver assist on the market and with OTA updates promising future "FSD" features, I knew it would get even better. No other automaker was offering that. But I can love my Model 3 and love AP and hope for better features and still admit the benefits of lidar.

 

[Uncle Paul]

Agree with you on this. Currently Lidar does not offer enough benefits to offset it's liabilities, including costs and packaging issues.

Elon knows lots about Lidar. Uses it for Space X.

Problem is that I find it hard to stomach your constant arguments that he has made a poor choice, and that you know better.

 

[diplomat33]

Problem is that I find it hard to stomach your constant arguments that he has made a poor choice, and that you know better.

Chill. I never said that I know better than Elon. But were not allowed to discuss what we think of Elon's decisions?

Obviously, the cost of lidar back in 2016 was a key factor. Elon's decision not to use lidar makes total sense form a business point of view. But there a lot of companies who use lidar for autonomous driving and have better autonomous driving than Tesla has. So I think it is a fair question to ask whether Tesla made a mistake not using lidar. Now if Tesla does achieve autonomous driving soon then Elon will be totally vindicated for sure. I certainly hope Tesla is successful.

 

[Uncle Paul]

Like a dog on a bone

 

[guitarman]

The issue for me is that for FSD, is the main problem the decision making when objects are known, or the fact that objects are not detected and thus the decisions are incorrect? The former is solved by the neural net while the latter is solved by more accurate sensors. I'm in the Musk camp because humans have limited sensors (only two eyes) and yet can do reasonable FSD (when attentive!). So clearly from that perspective, it's a NN issue. This is where I believe Musk is putting most of his resources, while LIDAR is really adding more sensor capability, but not decision making capability. If and when Musk is successful, LIDAR can only help his NN approach. The opposite is not true.

 

[ai4px]

A serious comment... you know the biggest thing that is missing from the AI which will do FSD? Anticipation. At present my car will go slow for somone who is actively turning right. A human would /anticipate/ them being out of the lane in a few seconds and proceed. The present level of AI only /responds/ to lane incursions there is no anticipation. The ability to predict and anticipate will do more for self driving than trying to split hairs between lidar and camera systems. I say the way to go is more intelligence and the present level of cameras... heck it -is- how we humans drive and it's worked fine for more than a century.

 

[caligula666]

Stating the obvious, Tesla is not camera only, it is using radars and ultrasonics. Are some of you expecting Tesla to get rid of them? If not, why?

 

[diplomat33]

Stating the obvious, Tesla is not camera only, it is using radars and ultrasonics. Are some of you expecting Tesla to get rid of them? If not, why?

No, I do not expect Tesla to get rid of the radar and ultrasonics.

"Camera only" is a short hand for Tesla's approach in contrast with the lidar approach. Tesla has a front radar for TACC but the main sensor that Tesla is using for autonomous driving are the 8 cameras. Tesla's entire autonomous driving approach is really based on training the computer vision to understand the world around the car. In contrast with the "lidar approach" which is short hand for the other autonomous driving approach which also uses cameras and radar but involves lidar as a primary sensor for vision.

 

[caligula666]

« Scientists Just Found a Way For Machines to See Through Clouds And Fog »

Scientists Just Found a Way For Machines to See Through Clouds And Fog

Obviously not ready in any mean for production but interesting research and potential

 

[cucubits]

« Scientists Just Found a Way For Machines to See Through Clouds And Fog »

https://www.sciencealert.com/new-visualisation-system-can- peer-through-clouds-and-fog

Obviously not ready in any mean for production but interesting research and potential

Your link doesn't work and I can't be bothered to debug but I am slightly curious.

 

[caligula666]

Your link doesn't work and I can't be bothered to debug but I am slightly curious.

Should be fixed now!

Scientists Just Found a Way For Machines to See Through Clouds And Fog

 

[cucubits]

Thanks. It is interesting. So it's essentially a laser which passes through low density materials.
Wonder what LIDAR stands for JK, I did see that at the end of the article they mention this system could complement LIDAR and cover its limitations.

I can see this together with a good LIDAR system pushing the boundaries.