Waymo will beat Tesla FSD because of LIDAR

[Cybergirl]

there is no AI. Fight me.

I'm not so sure. What constitutes artificial intelligence? Alan Turing defined artificial intelligence (AI) as the ability of a computer to mimic human responses under certain conditions. Turing is best known for his work on the Turing Test, which is a benchmark for machine intelligence. It's almost impossible to distinguish FSD behavior from that of a capable human driver.

Sponsored

 

[M0unt41nm4n]

Nope. Data shapes the model and feature points, not the other way around. Feature points are literally the reaction to what the data demands in order to understand it.

So why are you saying nope? LIDAR is the additional data that helps make a richer model and provides both additional data AND features to make that better model. You just agreed with what I said.

Also I do not agree with you about the necessity for large data sets with diverse data. Smaller data sets with larger features can produce more accurate results. It's a trade off. The best is large data sets with large feature sets. Even Google ML says so:

https://developers.google.com/machine-learning/intro-to-ml/supervised

Some datasets are both large and diverse. However, some datasets are large but have low diversity, and some are small but highly diverse. In other words, a large dataset doesn’t guarantee sufficient diversity, and a dataset that is highly diverse doesn't guarantee sufficient examples.
...
A dataset can also be characterized by the number of its features. For example, some weather datasets might contain hundreds of features, ranging from satellite imagery to cloud coverage values. Other datasets might contain only three or four features, like humidity, atmospheric pressure, and temperature. Datasets with more features can help a model discover additional patterns and make better predictions. However, datasets with more features don't always produce models that make better predictions because some features might have no causal relationship to the label.

 

[Woodrick]

*Disclaimer: I work for Google, but not Waymo
*Overall opinion: I use the Cybertruck's FSD all the time, even to drive 2-minutes down the street; it is awesome because it's here NOW

After seeing many FSD threads this past week and enjoying the Cybertruck's FSD myself I wanted to share a brief crash course on autonomous driving

References:

• MIT's course on Artificial Intelligence: Implications for Business Strategy
• Lecture, "Understanding Robotics"
• Lecture transcripts attached for convenience

Robotics 101: AI in the Physical World, with Sensors and Actuators
At a high level robots are just automated devices that perform physical tasks in the real world using 1. sensors (inputs) and 2. actuators (outputs)



Tesla is Level 3 Autonomy, Waymo is Level 4



All Self Driving Cars "Map" Between Feature Space (virtual) and Physical Space (real)

• All of the sensors (cameras, LIDAR, radar, microphones, etc.) are used to create a map of our real world in a virtual world called "feature space"
• The car "thinks" and "acts" entirely in this feature space
• It needs the sensors to keep populating the feature space with objects, to try to mimic our physical world as much as possible
• This why the Cybertruck "drives over curbs" while Waymo "sees the curbs", Waymo's feature space is better populated, closer to our real world

Tesla FSD is Parallel Autonomy (human-in-the-loop), Waymo is Series Autonomy (no human, mostly)
In robotics there are two types of autonomy, series and parallel:

• Series autonomy (Waymo): either human in control or vehicle in control, not both
• Parallel autonomy (Tesla FSD): aka "Supervised" FSD

Tesla's Big Failure: No LIDAR
In one sentence, robots need "laser imaging, detection and ranging" (LIDAR) to accurately perceive depth, occlusion, etc.

Longer answer is that LIDAR facilitates building a 3D point map of the world. Initially cameras were meant only for object detection. Tesla admirably expanded camera-based computer vision into FSD, but the lack of LIDAR is still problematic







Tesla's Reason for Camera-Only Computer Vision FSD: Cost
LIDAR (and the now removed radar from early model Teslas) is expensive and it's not just the hardware. The Cybertruck FSD update boasts that its neural network / machine learning model replaces over "300K lines of explicit C++ code". That's a lot of software engineering hours and millions of dollars saved

My (unverified) opinion is that there are nowhere near as many LIDAR 3D map examples as there are computer vision image and video examples. This means training a LIDAR machine learning model is not (yet) feasible

Waymo cleverly overcame this limitation by transforming the LIDAR 3D point map into a more manageable format. I have not had the time to read all the research, but Waymo solved the LIDAR point map problem, where a mess of 3D dots inhibited using machine learning fully

All of Waymo's research came at an extreme cost and they had Google / Alphabet money. Tesla just couldn't afford to run at a loss and had no other massively profitable business units to draw from

Interesting hypothesis, just wrong though.
Also interesting that you didn't even mention the occupancy system that Tesla uses.

And even more, Tesla doesn't really have a problem with the occupancy network. They know where everything is at, it's what to do with that information that is the issue.

And it's interesting that lots of companies are now moving away from LIDAR, including the LIDAR manufacturers.

 

[Geebz]

Can lidar see lane lines, speed limit signs, and warning signs?

 

[jookyone]

I'm not so sure. What constitutes artificial intelligence? Alan Turing defined artificial intelligence (AI) as the ability of a computer to mimic human responses under certain conditions. Turing is best known for his work on the Turing Test, which is a benchmark for machine intelligence. It's almost impossible to distinguish FSD behavior from that of a capable human driver.

You are choosing a single featured test from 1950 (human communication) as a benchmark for AI which I would argue is both an antiquated definition and an extremely low bar.

 

[jookyone]

You just agreed with what I said.

Why not
My point was that the model itself is completely derived from the data set and its ability to describe diverse feature points. A model with no data to describe it, in detail or volume, is completely useless. Conceptually I would think that it was a foregone conclusion to have diverse data instead of billions of the same five documents (in response to Google's explanation). Data is gold.

 

[M0unt41nm4n]

Why not
My point was that the model itself is completely derived from the data set and its ability to describe diverse feature points. A model with no data to describe it, in detail or volume, is completely useless. Conceptually I would think that it was a foregone conclusion to have diverse data instead of billions of the same five documents. Data is gold.

Or both . Then magic happens!

 

[AbblePC]

Bring back Adobe Flash!! It still is better than all of today’s technology!

I still prefer my abacus to my iPhone calculator any day ?

 

[Cybergirl]

You are choosing a single featured test from 1950 (human communication) as a benchmark for AI which I would argue is both an antiquated definition and an extremely low bar.

You miss my point. To say "there is no AI" presumes that there's a way to prove it. How is cognition achieved, and how do you test for it? If it can't be distinguished from human behavior, does it matter what you label it? Intelligence is on a spectrum. One can only compare one form of intelligence with another.

 

[AbblePC]

You miss my point. To say "there is no AI" presumes that there's a way to prove it. How is cognition achieved, and how do you test for it? If it can't be distinguished from human behavior, does it matter what you label it? Intelligence is on a spectrum. One can only compare one form of intelligence with another.

Cool: I’m on a spectrum, must make me intelligent ?

 

[JackCypher]

If what you are saying was true, humans would need LIDAR to drive a car. It would also mean that humans missing sight in one eye would not be able to drive. Of course we know both of these are false. I respect anyone who has done work in this area, but it seems you have formed an unchangeable opinion about how FSD "must" work, and that's not a good scientific basis. On the contrary, Tesla jumped from fully-scripted FSD (C++) to fully video-trained AI using only vision within a short period, and it does not seem to have any problems with the distance of objects. It also drives incredibly smoothly.

You can close your eyes and walk, you can walk around in your house at night in the dark - with minimal or nearly any vision at all...you're way underselling what amazing animals we are.,

Human do way more than just see to navigate the world - you have a kinesics sense you 'know' where your body is, the position of your limbs the direction you're moving. That and an internal 'map' you have in your mind is how you navigate...it is not a simple as having to see to move around.

You can close you eyes, reach up and touch your nose...your world perception is way beyond vision.

 

[Crissa]

Can lidar see lane lines, speed limit signs, and warning signs?

Some of them can see the texture of the paint on the road and signs.

But generally, no.

-Crissa

 

[Deleted member 17810]

Also the premise of the thread, "will" is funny.

They're already doing it.

 

[cyberos]

This is the reason Tesla will win is large datasets. Models and feature points are completely 100% useless without vast quantities of data.

Nope. Data shapes the model and feature points, not the other way around. Feature points are literally the reaction to what the data demands in order to understand it.

Quite literally we can train that. With enough data showing video of tire tracks in snow and other visibly intuitive extractable data points, you can make a model for rain and snow driving. Human drivers don't have that type of intuition out of the box, they learn it.

There is no AI. It's a marketing term and what we have right now is machine learning. And machine learning performs best when there is more data. The goal is to emulate behavior and knowledge as if it knows every possible thing about a system while feeding it less than the entirety of the knowledge base. I've been in machine learning since before it was cool and there is no AI. Fight me.

I'll step into the ring ??

The MIT course came to a stunning conclusion about AI, that machines + humans collaborating in a "supermind" outperform both AI or humans separately. A supermind is actually classified as artificial intelligence, even though it is a human-machine collective

Not only that but AI existed long before machine learning (statistics based computing). In the London museum I saw a functioning computer built entirely out of mechanical pieces. It was designed in the 1800s (although actually built in the 2000s)

 

[TheMachinewon]

You can close your eyes and walk, you can walk around in your house at night in the dark - with minimal or nearly any vision at all...you're way underselling what amazing animals we are.,

Human do way more than just see to navigate the world - you have a kinesics sense you 'know' where your body is, the position of your limbs the direction you're moving. That and an internal 'map' you have in your mind is how you navigate...it is not a simple as having to see to move around.

You can close you eyes, reach up and touch your nose...your world perception is way beyond vision.

Funny how you humans think, once your take flight, all that touchy feely shit goes out the window. Especially in helo's. It's all about vision. Think of a vehicle as the same,, you'll catch on quicker.

Sponsored