Where is the front radar on HW2 cars located?

[AmpedRealtor]

I think it's pretty obvious that HW2 is insufficient. HW1 was insufficient. All of Tesla's hardware suites will be insufficient until they start incorporating the type of hardware being used by other manufacturers in autonomous testing, such as lidar and front/rear/corner radar. Audi has already developed a hidden, behind-the-bumper lidar solution. Tesla does not possess some kind of magic whereby they can get by with only one forward-mounted radar while other vehicle manufacturers need so much more. It's called redundancy, and Tesla has none in HW2. Redundancy is absolutely required for true autonomy.

I think Tesla knows this. I also think Tesla knows that it is selling something it cannot fully deliver, but since it has gotten away with it in the past why not try again? I love my Model S, but I trust Tesla's Autopilot claims about as far as I can throw my 5,000 lb car.

 

[lunitiks]

Audi has already developed a hidden, behind-the-bumper lidar solution.

Source please

 

[AmpedRealtor]

Source please

Not just Audi, but Hyundai too. Tesla is completely behind on sensor technology, there is no way EAP will work reliably or according to expectation with the current hardware suite.

This is regarding Hyundai: Hyundai Ioniq Autonomous Concept Debuts Hidden LiDAR System in L.A. - Motor Trend
This is regarding Audi from 2013: CES 2013: Audi Demonstrates Its Self-Driving Car

Tesla needs to get its sh*t together.

 

[mblakele]

Humans manage to drive cars without LIDAR. We use vision, mirrors, hearing, inner ear, etc., all tied together by specialized computing. Why can't Tesla succeed with a similar approach?

As I see it LIDAR and other expensive sensors are shortcuts that trade more expensive sensors for less processing power and less sophisticated software. In the past that was a useful trade-off for R&D, demos, and events like the DARPA urban challenge. However for a mass-market product it's less clear what the right tradeoff is, or will be. Systems that run on cheap hardware and focus on software often beat out systems that rely on more expensive hardware: history is littered with examples of this. LIDAR is getting cheaper, but its competition is very cheap and continues to improve through software. Will the price of LIDAR drop low enough and quickly enough to keep cheap sensors and sophisticated software from eating its market?

 

[AmpedRealtor]

Humans manage to drive cars without LIDAR. We use vision, mirrors, hearing, inner ear, etc., all tied together by specialized computing. Why can't Tesla succeed with a similar approach?

As I see it LIDAR and other expensive sensors are shortcuts that trade more expensive sensors for less processing power and less sophisticated software. In the past that was a useful trade-off for R&D, demos, and events like the DARPA urban challenge. However for a mass-market product it's less clear what the right tradeoff is, or will be. Systems that run on cheap hardware and focus on software often beat out systems that rely on more expensive hardware: history is littered with examples of this. LIDAR is getting cheaper, but its competition is very cheap and continues to improve through software. Will the price of LIDAR drop low enough and quickly enough to keep cheap sensors and sophisticated software from eating its market?

LIDAR offers no advantages over camera-based solutions? I wonder why everyone else - Audi, Hyundai, Nissan, Mercedes, Uber, Google, and more - all use LIDAR? I'm not sure I can believe that Tesla's software is so superior to every other vehicle manufacturer so as to require a less robust sensor suite. Google is pretty good at software, wouldn't you agree?

This is giving Tesla a little too much credit, in my opinion, especially given their track record. What is much more likely, in my opinion, is that Tesla is hamstrung on the costs and therefore has no choice but to get by with less. Cameras are cheap and so are ultrasonics. LIDAR not so much.

 

[mblakele]

LIDAR offers no advantages over camera-based solutions? I wonder why everyone else - Audi, Hyundai, Nissan, Mercedes, Uber, Google, and more - all use LIDAR? I'm not sure I can believe that Tesla's software is so superior to every other vehicle manufacturer so as to require a less robust sensor suite. Google is pretty good at software, wouldn't you agree?

This is giving Tesla a little too much credit, in my opinion, especially given their track record. What is much more likely, in my opinion, is that Tesla is hamstrung on the costs and therefore has no choice but to get by with less. Cameras are cheap and so are ultrasonics. LIDAR not so much.

I did not claim that LIDAR has no technical advantages. I specifically addressed the trade-off between its advantages and its cost. Apparently you didn't read my post very carefully. As a result you're replying to points I didn't make, and ignoring the questions I raised.

If you're open to reason, you might do some reading on the history of technology and how approaches based on high-cost hardware tend to lose out to commodity hardware plus software. You're probably reading this post on a prime example from that history.

Oh, and I know the question was rhetorical but I wouldn't agree that google is particularly good at software. They're about middlin' as companies in their league go. They've had some notable successes, and a large number of failures. They're good at capitalizing on the successes.

 

[Tcpsoft]

I found this thread when I was looking for the same thing (after a short drive in light snow disabled the TACC). I like pictures so this service doc helped me out. Now just need to figure a way to keep that area clear.

 

[Brian33]

Here is the Model 3 from the owners manual.

 

[Snerruc]

I think it's pretty obvious that HW2 is insufficient. HW1 was insufficient. All of Tesla's hardware suites will be insufficient until they start incorporating the type of hardware being used by other manufacturers in autonomous testing, such as lidar and front/rear/corner radar. Audi has already developed a hidden, behind-the-bumper lidar solution. Tesla does not possess some kind of magic whereby they can get by with only one forward-mounted radar while other vehicle manufacturers need so much more. It's called redundancy, and Tesla has none in HW2. Redundancy is absolutely required for true autonomy.

I think Tesla knows this. I also think Tesla knows that it is selling something it cannot fully deliver, but since it has gotten away with it in the past why not try again? I love my Model S, but I trust Tesla's Autopilot claims about as far as I can throw my 5,000 lb car.

I disagree. Hardware 1 cars have routinely surpassed cars with more complex systems in the past when tested against them. Presently we don’t know what the limits of Tesla’s approach are, but hardware 1 cars are still improving! Mine is much better than it was 6 months ago. That is one reason eap is having trouble getting ahead of ap. My HW1 car can drive itself 200 miles on a busy interstate with construction zones with no intervention from me, except to flip the turn signal when needed. In heavy crawl, it is a life saver. I can read a book. Follow mode appears to work to some extent at all speeds. That is the only way I can explain its behavior in construction zones. Hardware 2and 2.5 cars will keep improving for years, too.

 

[lunitiks]

I can read a book.

me 2

 

[WeberAuto]

Hi,
I'm new to the community but I wanted to add a pic from service information to further clarify the initial inquiry. Location of the front radar is in red.

 

[dsnows]

I think the hardware is adequate. Sensors are the easy part. It’s more about the computer processing powers to interpret all the data in real time and make sense of it. As the computers improve and the software gets smarter, Tesla will be fine.

 

[AWDtsla]

Your location is an improvement as compared to the gen 1 cars.
View attachment 218595

I'm surprised how few people know about this fix to this problem for AP1. I only discovered it myself last winter:

AP2 - Snow radar failure

IMHO, Tesla should have done a recall for this for AP1. Now about AP2+, I don't know...

 

[jlv1]

Now just need to figure a way to keep that area clear.

Has anyone tried what Tesla suggests at Winter Driving Tips

You can use Rust-Oleum NeverWet Top Coat on the front fascia of your car to help prevent ice buildup on the radar sensors used by Autopilot.

• Use Step 2 Top Coat only.
• A hazy finish may be present after the coating is applied, but after ice resistance is no longer needed it can be easily wiped off.