If Tesla increased their car price by $1000 but included LiDAR would you pay?

Not unless it makes the car drive better? People's boner for LIDAR is bizarre.

Yes

No. [Tesla FSD Outperforms Rivals In China’s Toughest Test Yet](https://insideevs.com/news/767000/tesla-fsd-outperforms-chinese-brands/) [Tesla beats Chinese FSD competitors despite training restrictions or LiDAR - NotebookCheck.net News](https://www.notebookcheck.net/Tesla-beats-Chinese-FSD-competitors-despite-training-restrictions-or-LiDAR.1067719.0.html) "The surprising result was Tesla’s vision-based system outperforming its lidar-equipped Chinese competitors such as BYD, Xiaomi, Xpeng and Huawei ..."

No, why? What is LiDAR going to do that the current system currently does not? The obsession around a sensor is so strange. I’ve been using FSD for years now and V14 for the last 3 months. Not once has there been a situation where a LiDAR sensor would have been the solution to the problem. My main issues with FSD are route planning and stop sign delays. Distance measuring and object detection is not one of them.

That would be the point. LiDAR seems more than Vision can. Having one at least at the front would give Tesla a huge advantage IMO.

No. What do I want that for?

No, using auto driving tech anywhere but long highway stretches is lazy. Should be geo locked.

Only if it makes the car drive better. Every car you can buy with lidar is worse than Tesla FSD, so that's far from a guarantee.

Then why are all the cars you can buy with lidar worse than Tesla FSD?

Hmm I’d have to disagree. LiDAR is the next step to a fully autonomous car. Which is teslas goal. It will improve insane amounts from debut after dat is collected and updates are pushed. It’s gonna make all previous cars obsolete.

the love for LIDAR is weird

No.

If FSD is already this good, why would you ever want to add in LIDAR and increase the cost of the car for no benefit. I’m confused.

Redundancy.

No, it's not the next step.

What real problems with FSD do you think LiDAR would solve? It doesn't solve route planning. It doesn't solve stop sign delays. My biggest complaint about FSD is missing speed limit signs and going too fast in construction zones - does LiDAR solve that? What FSD problem do I have that LiDAR would solve?

Fsd was proven that it would drive into a wall if it was painted to look like the road ahead. There are flaws in cameras just like there are more flaws in lidar. For 1000$ more tho to have both? Cmon why not

Why? What does LiDAR do in you opinion that makes it the next step? Do you have any experience implementing LiDAR systems? (I do.)

If you want redundancy add more cameras. If you train on both cameras and LiDAR, your model isn't going to work well when one of them goes away, so I don't see where the redundancy comes from.

All Lidar could provide is exact distance to objects, but you do not need that to drive a car. No human drivers have that information anyway, it is simply noise. You just need to know approximately where everyone are, and what they are doing, and then keep within the lines and behave as people expect you to, that is driving. The real power of the FSD software is it's ability to interpret the pictures it gets, and translate that to actual knowledge of the situation it is in, the behavior of the other parties, and understand the context they are in. Lidar can not even see the difference between a brickwall, a dog and some trash in the street. Moreover, Lidar data is not trainable with ther neural net, it is simply not usable for anything that makes the system the greatest in the world. Lidar could at best provide a more precise auto park experience, but i do not feel that is needed. I do not understand why people still claim that a vision based approach is not the way to acchieve results, when it has been clear for a while that it is the best way to acchieve it. Even Xpeng who was very vocal in the opposition agains a vision based approached has since come around, and are actually now showing real promise.

Lidar won't detect if a coyote paints a hole in the ground.

I've been in the camp of LIDAR being unnecessary, but this morning FSD gave up on me in the middle of driving because I was pointed at the sunrise. If it would help it in those scenarios, it would be worthwhile.

Nope, a newer version of FSD passed that wall test. You don't need lidar for that. You just need intelligent software.

No

1k is negligible on the scale of 40k that these cars cost. Adding lidar would raise the price at least 5k.

Because they have worse software. If Tesla added more sensors they'd leapfrog themselves where they're currently at even.

You're guessing. You don't actually know that. It could be a negligible improvement, or even net-worse due to the added compute costs.

Yes, because having access to more sensor data would allow them to create a safer experience. Right now the full reliance on cameras to do all the work is kind of frightening. Look at Waymo for example their automous system uses a combination of all sorts of sensors from LiDAR/Cameras/IR Cameras/Sonic sensors/etc. Having access to all this sensor data would allow you to program a better fit for all cases. Right now the self reliance on just cameras scares me when there are environmental obstacles. Let's say a dust storm, snow, rain impairs the cameras or even poorly lit areas are encountered. The system currently will probably immediatelly pause saying it cannot proceed but now you are hoping the driver was still paying attention enough to where they can react when the FSD gives up. I still never understood why they don't make the cameras have self cleaning/heating system where they will clear debris off. Tesla started off in the right direction but greed made them pivot just like what happened with Apple. They started to focus more on profit than actual quality. It is upsetting that this is the direction they are going and I am wavering because of it.

YES! My mother died in a car accident and anything that reduces serious car accidents is well worth $1000. LiDAR may not be a total game changes but it would move the needle in the direction of safety.

Two senses is better than one.

Maybe, but ultimately NO, it wouldn't sway my purchase. Up to this point Tesla has been steadfast on being camera only with their autonomy systems, to suddenly offer up a LiDAR setup to me would be very odd. IF they choose to change course then they likely are a few revisions away from it actually being prime time ready and better than the camera only FSD setups. Which by then, will the car then need better / newer LiDAR technology? We all know the history of HW2.5 > HW3 > HW4.

Watch rivian's AI day presentation.

Why would I do that? Their system is way worse. I'd rather listen to the makers of the better system, so Tesla.

There's a reason why aircrafts use LiDAR.... There are no lights in the sky other than the ones on the aircraft.. If an airplane used only cameras would you be comfortable with sitting in that aircraft? Do note that LiDAR can detect various things cameras cannot detect such as temperature and it can see in the dark.... In all cases LiDAR would make the car drive better. Being able to detect temperature changes from afar would allow for the car to adjust to best fit the climate. Being able to detect wind direction/wind particle and predict storm conditions would only make the car smarter with adjusting. The thought that they could use the LiDAR to detect road conditions by testing if the road's traction level to adjust suspension and speed is something that really would save a lot of lives.

Software

Exactly. It's the software that matters. Adding lidar doesn't solve it. Making intelligent software does.

"Lidar sees more than vision can"? Let's push on that, because it's totally wrong. Resolution is very low. Can't see stoplight colors. Can't read roadsigns. Can't see lane striping. On the other hand, lidar can see in the dark. Cameras require headlights unless they are expanded-spectrum types that can see near-infrared. And last time I checked, cars came with headlights. Lidar can also report the distance of every object it sees. But, so can a camera feeding a neural network. So that's moot. Both cameras and lidar have issues with heavy fog and rain. Well here's a news alert: get off the damn road if conditions are that bad. Even if you had magical sensors that could see through the heaviest precipitation, *everyone else on the road does not*. Get off the road. At this point, lidar brings precisely nothing, is expensive, and is unsupported by software. For automobiles, it is useful for validation testing only.

> I'd rather listen to the makers of the better system, so Tesla https://mastodon.social/@rodhilton/109572674700288958 > He talked about electric cars. I don't know anything about cars, so when people said he was a genius I figured he must be a genius. > > Then he talked about rockets. I don't know anything about rockets, so when people said he was a genius I figured he must be a genius. > > Now he talks about software. I happen to know a lot about software & Elon Musk is saying the stupidest shit I've ever heard anyone say, so when people say he's a genius I figure I should stay the hell away from his cars and rockets. Can't even get auto-wipers half right. And my Tesla just *loves* to randomly slam on the breaks in the middle of the freeway if a wisp of snow dares to blow across the road.

I think I should have been more specific. I did not mean replace Vision with LiDAR but compliment it. One sees certain objects better the other one sees what the first missed. Then the Vision Software on top. Best of both worlds?

Humans have been driving using vision only for decades and decade. And if you did implement LiDAR for redundancy, which system does it prioritize in the event of the two systems disagreeing?

Id guess the cameras would be prioritized. It wouldn’t make sense the other way.

and your blind trust is somethin!

this is written in whole by AI.

Why would I want Lidar? No.

What’s written by AI? The software ?

When going though a tunnel and exiting on the other end, the glare blinds the camera for a split second. When there is thick fog, rain or snow storm , cameras cannot see through. Cameras are great at some things, but lack others like human vision would. If we are making something superior to a driver why only rely on sensors a regular driver would have?

My point exactly- above. Vision is limited and requires support - LiDAR

> Fsd was proven that it would drive into a wall if it was painted to look like the road ahead. I said "real problems." > For 1000$ more tho to have both? Cmon why not It's not that simple. It's not like you add LiDAR and suddenly the hardware knows what to do with it. You'd have to retrain the entire model to interpret the new data. I'd expect massive regressions in driving quality as LiDAR was introduced that would take many iterations to smooth out. Just to avoid the risk of smashing into a wall that's painted to look like the road? And assuming you've got both, what happens when the camera and the LiDAR disagree?

LiDAR can't see through much of this as well. Waymo shuts down all operations in these situations for this reason. HD Radar is a far superior sensor for this situation as it actually can see through them all, not LiDAR. but most people on the internet who think they are experts have no actual experience using LiDAR. Furthermore, you can't drive on LiDAR or Radar alone as you can not see things like road markings, signs, or anything other than point cloud data. This is not a safe driving environment. It's just pure ignorance from inexperienced people on the internet to think that LiDAR would allow for driving in low visibility situations. You ALWAYS need Camera's to drive autonomously. LiDAR just provides precision distance measurements in the form of a pointcloud. It can not be used without Camera's for autonomous driving. If Tesla's Camera's are down, so are Waymos. LiDAR is stupid if precision distance is not needed (sub cm distance accuracy), which it is not needed for driving as Humans distance estimation is FAR worse than Camera only distance estimation using stereoscopy.

Perhaps the best solution includes: vision for general 360 view, signs etc LiDAR for 3D Mapping Radar for what you can not see ?