So, when are we changing this forums name from Technology to it’s actual purpose of late “every click and rage bait post about Tesla and Musk so people can circlejerk worse than reddit”?
It’s literally nothing but bullshit about Tesla and Twitter. All day long. No one cares!
I want to know about some actual tech, not the drama.
Unfortunately, all the current tech news is either people running naked scams or people debunking them.
The tragedy of our modern era is how much money we’ve invested in selling people a box labeled “Newest Life Changing Gadget” that’s just full of rocks.
Check out the podcast TrashFuture. They do a bit about a shitty tech enterprise every episode, sometimes twice a week. From Juicero to Neom, the list of awful tech bullshit is limitless.
Here is an alternative Piped link(s): https://piped.video/xvqQ4F7Yf2o?si=3tMs2_aHpnfqKCBu
Piped is a privacy-respecting open-source alternative frontend to YouTube.
I’m open-source, check me out at GitHub.
Without LIDAR, this is a fool’s endeavor.
Do you have lidar on your head? No, yet you’re able to drive with just two cameras on your face. So no lidar isn’t required. Not that driving in a very dynamic world isn’t very difficult for computers to do, it’s not a matter of if, it’s just a matter of time.
Would lidar allow “super human” driving abilities? Like seeing through fog and in every direction in the dark, sure. But it’s not required for the job at hand.
Do you have lidar on your head?
Nope,
And that’s exactly why humans crash. Constantly.
Even when paying attention.
They don’t have resolution in depth perception, nor the FOV.
You have eyes that are way more amazing than any cameras that are used in self driving, with stereoscopic vision, on a movable platform, and most importantly, controlled via a biological brain with millions of years of evolution behind it.
I’m sorry, you can’t attach a couple cameras to a processor, add some neural nets, and think it’s anything close to your brain and eyes.
A lot of LIDAR fans here for some reason, but you’re absolutely right.
There’s just not a good amount of evidence pointing that accurate depth perception only obtained through LIDAR is required for self driving, and it also won’t solve the complex navigation of a real world scenario. A set of visible spectrum cameras over time can reconstruct a 3D environment well enough for navigation and it’s quite literally what Tesla’s FSD does.
I don’t know why someone would still say it’s not possible when we already have an example running in production.
“But Tesla FSD has a high disengagement rate” - for now, yes. But these scenarios are more often possible to be solved by high definition maps than by LIDAR. For anyone that disagrees, go to youtube, choose a recent video of Tesla’s FSD and try to find a scenario where a disengagement would have been avoided by LIDAR only.
There are many parts missing for a complete autonomous driving experience. LIDAR is not one of them.
I remember watching a video talking about is there a camera that can see as well as a human eye. The resolution was there are cameras that see close but not as well and they are very big and expensive and the human brain filters much of it without you realizing. I think it could be done with a camera or two but I think we are not close to the technology for the near future.
I don’t know why people are so quick to defend the need of LIDAR when it’s clear the challenges in self driving are not with data acquisition.
Sure, there are a few corner cases that it would perform better than visual cameras, but a new array of sensors won’t solve self driving. Similarly, the lack of LIDAR does not forbid self driving, otherwise we wouldn’t be able to drive either.
challenges in self driving are not with data acquisition.
What?!?! Of course it is.
We can already run all this shit through a simulator and it works great, but that’s because the computer knows the exact position, orientation, velocity of every object in a scene.
In the real world, the underlying problem is the computer doesn’t know what’s around it, and what those things around doing or going to do.
It’s 100% a data acquisition problem.
Source? I do autonomous vehicle control for a living. In environments much more complicated than a paved road with accepted set rules.
You’re confusing data acquisition with interpretation. A LIDAR won’t label the data for your AD system and won’t add much to an existing array of visible spectrum cameras.
You say the underlying problem is that the computer doesn’t know what’s around it. But its surroundings are reliably captured by functional sensors. Therefore it’s not a matter of acquisition, but processing of the data.
I wish this was talked about every single time the subject came up.
Responsible, technologically progressive companies have been developing excellent, safe, self-driving car technology for decades now.
Elon Musk is eviscerating the reputation of automated vehicles with his idiocy and arrogance. They don’t all suck, but Tesla sure sucks.
Dont let them know about that I don’t want my radar detector flipping out for laser lol
K and KA band are used for blind spot monitoring and would make radar detectors go nuts until filtering got worked out, cars that use Lidar will set them off as well though they’re more rare still
Even with LIDAR there are just too many edge cases for me to ever trust a self driving car that uses current-day computing technology. Just a few situations I’ve been in that I think a FSD system would have trouble with:
-
I pulled up at a red light where a construction crew was working on the side of the road. They had a police detail with them. As I was was watching the red light the cop walked up to my passenger side and yelled “Go!” at me. Since I was looking at the light I didn’t see him trying to wave me through the intersection. How would a car know to drive through a red light if a cop was there telling you to?
-
I’ve seen cars drive the wrong way down a one way street because the far end was blocked due to construction and backtracking was the only way out. (Residents were told to drive out the wrong way) Would a self driving car just drive down to the construction site and wait for hours for them to finish?
-
I’ve seen more than one GPS want to route cars improperly. In some cases it thinks a practically impassible dirt track is a paved road. In other cases I’ve seen chains and concrete barriers block intersections that cities/towns have determined traffic shouldn’t be going through.
-
Temporary detour or road closure signs?
-
We are having record amounts of rain where I live and we’ve seen roads covered by significant flooding that makes them unsafe to drive on. Often there aren’t any warning signs or barricades for a day or so after the rain stops. Would an FSD car recognize a flooded out road and turn around, or drive into the water at full speed?
In my opinion, FSD isn’t attempting to solve any of those problems. Those will require human intervention for the foreseeable future.
Or there are better other ways to tell a FSD car that the road is closed. We could use QR code or something like that which includes info about blockade, where you can drive around it, and how long it will stay blocked. A FSD should be connected enough to call home and give info to the servers, those then update the other FSD cars, et voila tadaa.
Musk’s vision is (was?) to eventually turn Tesla’s into driverless robo-taxis. At one point he even said he could see regular Tesla owners letting their cars drive around like automated Ubers, making money for them, instead of sitting idle in garages.
Well FSD is supposed to be Level 5 according to the marketing and description when it went on sale. Of course, we know Tesla’s lawyers told California that they have nothing more than Level 2, have not timeline to begin building anything beyond Level 2, and that the entire house of cards hinges on courts and regulators continuing to turn a blind eye.
Lol, ok. Your anecdotal experience can totally be believed over all the data gathered over years. Great. Thanks.
The worst it will do is pick the wrong lane and detour a bit to get back on track.
https://www.cbsnews.com/amp/news/tesla-car-crash-nhtsa-school-bus/
Counter-counterpoint: I’ve been using it since 2019. I think you’re exaggerating.
-
It aggressively tries to center itself, always. If you’re in a lane and it merges with a second lane, the car will swerve sharply to the right as it attempts to go back to the middle of the lane.
-
It doesn’t allow space for cars to merge until the cars are already merging. It doesn’t work with traffic; it does its own thing and is discourteous to other drivers. It doesn’t read turn signals; it only reacts to drivers getting over.
-
If a motorcycle is lane-splitting, it doesn’t move out of the way for the motorcycle. In fact, it assumes anything between lanes isn’t an issue. If something is partially blocking a lane but the system doesn’t recognize it as fully “your lane”, the default is to ignore it. The number of times I’ve had to disengage to dodge a wide load or a camper straddling two lanes is crazy.
-
With the removal of radar, phantom braking has become far, far worse. Any kind of weather condition causes issues. Even if you drive at sunset, the sun can dazzle the cameras and they don’t detect things that they should be able to - or worse, they detect problems which aren’t there.
-
It doesn’t understand road hazards. It will happily hit a pothole at 70 MPH. It will ignore road flares and traffic cones. When the lanes aren’t clearly marked (because the paint has worn away or because of construction), it can have dramatic behavior.
-
It waits so long to brake, and when it brakes it brakes hard. It accelerates just as suddenly, leading to a very jerky ride that makes my passengers carsick.
The only time I trust FSD is when it’s stop-and-go traffic. Beyond that I have to pay so much attention to the thing that I might as well just drive myself. The “worst thing it can do” isn’t just detour; it’s “smash into the thing that it thought wasn’t an issue”.
I have only been driving a Tesla for a few days in 2022, but i fully agree with you, i wanted to specifically test the FSD and i had so many incidents where it tried to get into an appearing turning lane, even tough it should go straight, just straight up slowed to 10kph in a Tunnel where speed limit was 50, and there were blind corners because of “bad vision conditions” even the cruise control was annoying, it felt like my steering input was basically just a “suggestion” that i sometimes really had to force through against the will of the car because otherwise bad shit would’ve happened… Sports mode steering made that only slightly better in the dual motor Model Y
Overall I enjoyed driving the ID3 more actually… at least that had solid and responsive steering that felt - compared to the Tesla - like driving a sports car… and i’ve driven the ID3 directly after the Tesla.
Only good thing about the Tesla was acceleration.
It doesn’t read turn signals
It does in the FSD beta (somewhat). It even brakes and allows them in if it detects that they have a signal on. It doesn’t understand merges as well, but it’s still better than regular autopilot. All your other points are pretty valid. I am constantly taking it out of AP Ana putting it back in during a city drive, even though I have “FSD”.
Tesla’s software is not safe:
https://www.washingtonpost.com/technology/2023/06/10/tesla-autopilot-crashes-elon-musk/
I wonder how this statistically compares to non-Tesla crashes?
Edit: quick Google/math shows average rate of lethal automobile crashes at 12 per 100,000 drivers. Tesla has supposedly sold 4.5million cars. 4.5million divided by 17 deaths from the article = 1 death per 200,000 Tesla drivers.
This isn’t exactly apples-to-apples and would love for some to “do the math” more accurately, but it seems like Tesla is much safer than a standard driver.
The other confounding factor is we don’t know how many of these drivers were abusing autopilot by cheating the rules (it requires hands on the wheel and full attention on the road)
Your statistical analysis is so bad that it’s not even wrong. It’s just a pile of disparate data strung together with false assumptions.
So all of those Teslas were sold in America? And all 4.5 million of those Teslas have Autopilot? And they’re in Autopilot mode 100% of the time?
You forgot the most important issue: Tesla drivers are not representative of the average driver. They have more money and more education. They live in places with nicer weather. These all contribute to lower crash rates without self driving. I bet high end Mercedes have lower crash rates too, because people don’t defer maintenance and then drive them crazily in the snow.
Compare apples to apples and I bet Teslas have average crash rates for luxury cars.
The true comparison is in miles per accident. Fatal accidents will be higher for older model cars. Not all Tesla cars have FSD. In many situations FSD is not available even on equipped cars. There is nothing to indicate from the current data that Telsa FSD is safer or more dangerous than the median driver.
This isn’t necessarily true either. The NHTSA Standing General Order data shows that Tesla reports a large number of crashes (which they get to cherry pick in a LOT of cases) under ADAS use compared to other brands. Taking conservative rollout numbers from companies like Honda shows that the crash per ADAS equipped vehicle rate is significantly higher.
The real red flag in all all of this is that Tesla’s own reported marketing numbers for ADAS crashes wasn’t declining with newer releases over time. A rate that doesn’t improve as the CEO claims the software is already performing better than humans should instantly discredit the software, its performance, and any claims about new or improving features.
