Tesla’s FSD is not ready for prime time and won’t be without LIDAR.
Cartoon Prank Crashes Tesla
Motor Trend reports Cartoon Prank Crashes Tesla, Awkwardly Rehashes NHTSA Investigation Findings
Before we get into the very-serious issue of whether or not Tesla’s Autopilot driver assistance system preemptively shuts itself off, ceding control back to the driver, in the moment before a crash, we bring you YouTube proof, of sorts. Mark Rober put out a CrunchLabs video recently highlighting the key drawbacks to Tesla’s driver assistance technology, namely that the hardware package (sensors) is woefully lacking given the operational scope of Autopilot (adaptive cruise control paired with a self-steering lane-keep function that can accelerate, brake, and steer your car at a set speed on freeways, with supervision) and, on models so equipped, Full Self Driving (which has received a name change appending “Supervised” to the end to remind users that they still need to pay attention when this fully autonomous feature is running). What are we on about, here?
Most other vehicles with similar capabilities (or less) rely on a mix of cameras, short-distance ultrasonic sensors, radar sensors, and even lidar to map their surroundings in real time and respond to other traffic, pedestrians, and roadways. Tesla dropped most of that in favor of a cheaper, simpler array of visual cameras. You needn’t be an engineer to surmise this isn’t a very good idea, given cameras face the same basic limitations as human eyes: They can’t “see” well through fog, rain, snow, or dirt—especially if any of that blocks the camera lens itself—nor do they have any redundant backups to check their work (see: radar, lidar, and other sensors that can offer a different view of an object or something “seen” by a camera). This is where Teslas getting confused by sunsets or sunrises, tunnels, and the like come into play. Lidar or radar would know that something is or isn’t actually in the path of the vehicle, serving as a check on the cameras’ vision.
YouTube Mark Rober slickly demonstrates Tesla’s approach by comparing his own Model Y’s Autopilot performance against a lidar-equipped test vehicle in a series of evaluations involving an unfortunate child-sized dummy. We won’t spoil the video—watch it for yourself—save for the big ending, since it is of great importance to NHTSA’s Autopilot investigation. The two vehicles are driven at a wall painted to resemble the road continuing ahead—an optical illusion, of sorts, familiar to anyone who grew up watching cartoons on Saturday mornings. Even the human eye can detect the ruse, at least once closer to the faux wall, and obviously radar and lidar can see it, too, since it’s a solid object in the roadway. The Tesla? Well, it’s fooled—smashing through the (foam!) wall without slowing, demonstrating that its cameras saw only a road continuing on, and are thus easily tricked.
While that’s both amusing and damning at the same time, the video shows another phenomenon: Autopilot shuts off just before the Model Y impacts the fake wall. As electrek points out, Tesla fans rushing to react to the CrunchLabs video made several misinformed or misleading assertions about Autopilot’s use in the clip—with the comments-section back-and-forth inadvertently bringing attention back to the phenomenon at the center of a NHTSA safety investigation first opened in 2022.
Tesla has repeatedly marketed Autopilot in such a way that suggests its capabilities are greater than they are, often by not correcting popular thinking that the system truly is like a self-driving, well, autopilot feature
Can You Fool Self-Driving?
LIDAR No. Tesla’s FSD, yes, three ways.
Question Number Two: Do Teslas shut off their driver assists seconds before crashing?
Answer: Yes
Please play the above video. It included full details on how Mark Rober snuck a LIDAR device into Disneyland to map the pitch black ride called “Space Mountain”.
Using LIDAR he mapped the ride, showing it is outside the park perimeter.
Wow, congrats.
Six Tests
Rober then switches from how he mapped Space Mountain to the self-driving tests. He concludes “My Tesla is more Wile E Coyote and less Road Runner.”
Rober states “This story was originally published in June 2022, and has since been updated to include new developments.”
Regardless, without LIDAR Tesla can never fix what it cannot see. At a minimum, that is smoke, rain, and snow.
Musk cut corners to save money. Tesla will never make it as a robo-taxi except in good weather.
The video is hilarious. Play it.
Related Posts
May 24, 2025: BYD Overtakes Tesla in Europe for the First Time, Sales Jump 169 Percent
BYD sold 7,231 battery electric vehicles in April versus 7,165 by Tesla, according to data from Jato Dynamics.
June 1, 2025: Waymo Surges to 10 Million Paid Rides, Tesla Has Zero
Hello Elon Musk, you are now 10 million rides behind Waymo.
July 2, 2025: Tesla’s Global Vehicle Deliveries Plunge 13.5 Percent in the Second Quarter
Tesla is struggling everywhere. Expect things to get worse.
My understanding of LiDAR is it requires extremely detailed mapping. Waymo has very specific boundaries where it operates as its LiDAR system is ineffective outside the mapped area. Waymo may be ahead in scale at this time, but given Musk probably will overcome his deficiencies with currently incorporated technology, then scaling service will be to his advantage making it possible he overtakes Waymo. He entered space industry way late and is now the dominant player in cost effective space travel and freight delivery. Seems a gentleman who can autonomously fly to the international space station to include docking, make serious advances for people with nervous disorders, design and build humanoid robots, economically bore tunnels under congested cities, and probably other things that do not come to mind at the moment, should be given a little more space to perfect his fully self-driving automobile rather than assuming he has failed at this point in the development.
I very sure that image across the road will fool 99 percent of himan drivers
I doubt it, except perhaps at night
Equipping a robotaxi with Lidar technology can involve costs ranging from $7,000 to $10,000 at low scale, according to a recent analysis. The specific cost depends on the number and type of Lidar sensors used, as well as the accompanying sensor fusion and compute modules.
Here’s a more detailed breakdown
Musk F’d Up
I don’t like that Mark Rober guy. There’s something off about him.
If a recent article I read is true. Then The problem with Lidar is that it is currently prohibitively expensive. Costing many times more than the car itself. Companies using it are losing obscene amounts of money on every RoboTaxi Ride. If I understand correctly it adds up to $1 Million in costs to every RoboTaxi. How do companies lose $Billions on taxi rides? They use Lidar. But without it you cannot trust a RoboTaxi. When will the cost of adding road worthy Lidar to a car become $25k or less?
Equipping a robotaxi with Lidar technology can involve costs ranging from $7,000 to $10,000 at low scale, according to a recent analysis. The specific cost depends on the number and type of Lidar sensors used, as well as the accompanying sensor fusion and compute modules.
Here’s a more detailed breakdown
There is temptation to combine two trends – electric cars and self-driving cars.
Electric cars were invented more than 100 years ago and still require subsidies and encouraging government regulation to compete, plus they compete poorly in terms of market share and product life.
Self-driving cars have been possible for a much shorter time, and its arguable they still have not been proven to exist. The features are not subsidized and government regulation does not favor their adoption, yet high end vehicles are sold with those features.
An environmentalist trying to limit CO2 emitted by gas autos or limit CO2 emitted by power plants for charging stations would probably not be excited about empty cars driving themselves around because its cheaper than downtown parking.
EV and SD – two different ideas.
If I recall, the first electric car was also one of the first cars
“The first electric cars used lead-acid batteries, invented by Gaston Planté in 1859″
Is the insurance being underwritten in the afternoon at the London pubs?
Matson Suspends Shipping EVs Citing Hazards of Lithium-ion Batteries
— I would not trust car systems in any role other than “assistant only, just in case“. Not to mention that such systems may be hacked in the future, and your neighborhood geek might be able to control your car while it’s taking the kids to school.
— Last winter thousands of electric cars were immobilized in Chicago’s winter, with the owners waiting hours at charging stations.
— The huge weight of EVs will soon require more frequently repaving some roads. This weight also makes them dangerous to the other cars.
— Batteries don’t last for long, and their disposal will further burden the environment, perhaps a lot more than current low-emission gasoline cars.
I think the current technology is a lot less mature than the enthusiasm of the tech bros who promote it so fanatically. With both cars and computers nowadays it feels like buying a house with 10 doors and 30 windows nobody asked for.
Fortunately, this is changing quickly. Solid state batteries are being produced by PowerCo under license from Quantum Scape (QS) as we chat. PowerCo is VW groups EV battery company. They have three giga factories under construction that will produce solid state batteries for 2 million vehicles per year when they are fully built out.
Solid State batteries solve the big problems that current batteries have:
Charging time reduced to 12 minutes.Reduce weight by >30%.Are not as temperature sensitive.Increased range.Fire safety dramatically improved.Pure lithium Cathode and Anodeless design with a ceramic separator is easily recyclable. (No Cobalt!)Batteries lose less that 5% of their charging capacity through 1,000 full discharges. = 300,000 miles.I would not buy one of todays EV’s because of the issues inherent in thefirst iterations of EV batteries. Sure they are fun to drive with the silent and swift acceleration. They are also great for commuter cars. But that is not my driving pattern.
Europe and China are going to leap ahead of the US given the insane 50% copper tariffs and the oil industries control of Trump and his neanderthal sycophants. They will keep us from developing our grid and charging infrastructure and drive costs to astronomical levels.
>
In the US good technology gets swept up by ideological warfare. Oil fuel good. Electric bad. Why?
WTF is technology a culture wars topic.
Yes US will be left behind and have to invade more and more countries to seize their oil when it has plenty of other options to power it’s people
Lefteris’s post is a good example of lumping EV and self-driving into the same category. It is possible to be one or the other without being both. Mish’s headline story was about Telsa, one case where “both” is true, but one would ask “why?”. I’d say “self driving” featues today correlate with “high end”, and Tesla today correlates with “high end”. The electric part is coincidental, convenient because battery management requires better computer programming so EV products need programmers anyway.
Old news. Musk admitted during the Bidem administration that any feature related to autopilot should be considered experimental. Musk admitted this in congressional testimony. Musk doesn’t care if people die over using new technology. To him that’s part of risk inherent in new technology.
Our leg’s blood vessel have valves which send blood one way to the heart. Trump’s valves failed the test. They send blood two ways.
Messing with the Fed has consequences . The last president that tried to revoke the authority of the Fed ended up assassinated.
Just some varicose veins in the legs. No big deal.
A minor problem, eh?
Defective valves increase blood pressure. Check your left hand and left ankle blood
pressure (while lying). Check your right hand and right ankle blood pressure. Check
if your “Ankle Brachial Index” is normal.
That was the ” Wile e. Coyote Test” that used the old Tesla software and was highly criticized for not following normal test protocols and looked like it was setup to make Tesla fail. The company that makes the Lidar system Mark Rober used in the video initially used the video to promote its product but after the controversy about the test dropped the video from its site quickly. Kyle Paul recreated the test using standard test protocols and complete transparency and using up-to-date Tesla FSD software and Tesla passed the test. Nevertheless Mark Rober’s video pops up occasionally even if it has been thoroughly debunked.
We have a real world test going on in Austin and it is far too early to see who has the better taxis system both from the ease and safety side but also from the cost side because after all, the goal of both Waymo and Tesla is to make money. The two have widely different systems and philosophies and Austin is the perfect place for them to duke it out. I think this is a “winner take all” test. Personally I think Tesla will win but let’s see.
Mark Rober’s video is an old news. After the slump, JP will cut rated to ease the debt burden, before the 2026 election. Tulsi charges are essential for Nov 2026. Epstein counterpunches are weak. Arresting Obama on AI video please MAGA. X is essential to preserve the reps. TSLA is essential to fight inflation. FSD drive the drunk back home.
is it really debatable that a system that relies on cameras PLUS Lidar will be safer than one that relies only on cameras? it’s a redundancy that can save lives in heavy rain or snow or fog or weird road reflections n the desert.
lidar can “see” things visual-only cameras cannot. so why is this even being discussed?
“Why is this even being discussed?” Because this is an economics blog that discusses costs and benefits
Yes, a system with cameras and Lidar will be safer than one with just cameras.
But it would also be safer if the speed limit across the country was maxed out at 45mph.
Or if all new cars were required to encase their passengers with Mad Max-styled cages.
But those all involve trade-offs between costs and what people want when they drive. Everyone will have a different opinion, thus the marketplace
but the article doesn’t discuss the price difference between lidar+ cameras and just cameras, though it says that Teslas uses a lower cost system. I assume, but don’t know, that the Lidar + cameras has fewer cameras, but and don’t know if it’s a $250 difference or a $5,000 difference.
the things you list have significant performance and comfort tradeoffs, not just price. i get your point, it’s a fair one, but i would say a system that gets out of whack in storms, fog, or in desert glare, sunsets, is unsuited to the task at any price and Teslas aren’t cheap.
no camera or visual sensor is as good as the human eye. not even close, and the improved reaction time of computers don’t really make up for that difference, and the AI can’t come close to the human mind in terms of understanding the different situations that come up while driving. so having sensors that aren’t just worse versions of the human eye makes sense.
That video is a scary scenario for a Tesla FSD, even if it was set up specifically in a fake situation to fool the system.
But my mom is an insurance agent and so handles hundreds of human-driven car crashes each year. Her accounts over the dinner table of the stupidity that actually transpires in many of these cases is almost inconceivable. Makes me not want to drive home afterwards with these other drivers, but what are you going to do?
I recently visited a friend in Austin and we used the Uber app to call for multiple non-supervised Waymo rides. It was super easy to use and an awesome experience! Until you got near the bar you were visiting and the drunk locals kept jumping at the car so it would slam on its brakes. A human driver would know they were screwing around and keep driving steadily through.
All these systems have their different costs and benefits
They need to fit them with those side-flamethrowers they were making in Brazil.
Please! Do not fix Full Self Crash. The nation needs spontaneous comedy now more than ever. Just pad Teslas with 24″ foam and limit speeds to 5 mph. Think of the joyous mayhem in Walmart parking lots.
I’ve always wondered why we didn’t have nerf-covered cars.
Fatality is less expensive than long-term injury?
-or-
It would be inefficient to maintain nerf-covered roads for pedestrians to bounce on?
Put pedestrians in those big sumo suits.
Systems should be fail-safe so folks are not hurt or killed. Sensors can be fooled, fail or jammed. Software will never be bug free due to the infinate number of situations that might occur but are not programmed to deal with.
I remember watching I believe “Frontline” on PBS probably around 25 years ago. The episode was about top Universities from around the US competing in an autonomous robot contest. The Robot moved slower than walking speed and had to navigate around a couple of obstacles on a course. Most could barely do it. Now compare that against FSD technology today, it’s stunning how good the technology is. It’s not perfect but it’s pretty close and will eventually be.
Tesla Navigating NYC.
https://youtu.be/OT4oyJhskBs?si=7IBLcxi7bADNffz8
I remember my school’s Sunrayce solar vehicle entry from 25 years ago… newest entry still looks about the same.
Some technology, like self driving, advances a little every year with faster computing, better sensors, standard software, … and some technology, like solar energy, finds a ceiling quickly then stops advancing.
While toddler is trying to walk he needs support, lidar, radar, etc till he mature to walk alone with vision.
My lidar stumps are still just barely visible.
– key drawbacks to Tesla’s driver assistance. > Keyword being “Assistance” Let’s say for discussion purposes, Tesla is not interested in the current Technology being used for this, and Cameras alone, with “Assistance” will get them by for now, without heavy investment and time. Neither of which wishes to spend or have the time to do so. His cars sell, and their Name “Tesla” is Solid.
>> Meanwhile He’s watching what others are finding through investment and newer technologies, while quietly conducting His Own. Musk is taking the time to do it right, the right “Personally Created” and “Legally Protected” Musk Way (Think Space X). When He is ready to introduce His Ideas, they will be way ahead of where any competition may stand overall, would be my guess. Sounds like a Musk Move to Me…
– Most other vehicles with similar capabilities. > Musk doesn’t want to be anywhere near that name or thought for His Release of His done the Musk Way, or Awesome!!
– Tesla dropped most of that in favor of a cheaper, simpler array of visual cameras. > Musk bought Time, and Saved Money while doing so.
You needn’t be an engineer to surmise this isn’t a very good idea. > Everyone should know Musk knows this as simple and basic FACT.
– This is where Teslas getting confused by sunsets or sunrises, tunnels, and the like come into play. > Not Musk at all, but perhaps His Competitors are very much so, and extremely likely imo.
– Tesla has repeatedly marketed Autopilot in such a way that suggests its capabilities are greater than they are, often by not correcting popular thinking that the system truly is like a self-driving, well, autopilot feature. > The love for Tesla is they are always seeking more innovation, and ideas, and stretching things to see how far, fast, high Etc. Things will go!! People can’t wait to see what’s next, after He finishes with the Cars. Small driving/flying type vehicles (think 2-Passenger) for example, or Perhaps “Pod Travel” oops… (think Bank Machines for money withdrawal), on a specific, and designed/defined lower level airspace. For “Common Spaces” so they continually go back and fourth. Grocery Stores, Hair, Lunch, Workouts Etc. Eventually into small designed locations for business, say Medical Lane, Hospital Lane Etc. Many were already built upon Hill’s if I’m not mistaken?
So much potentially going on right now…
Interesting conundrum…
Musk was just used like a cheap whore to get Trump elected. Then given an impossible task to control guvment spending and humiliated. After that he was shown the door.
Ya think he is gunning for revenge?
Everything in me wants to short TSLA because it is no longer selling many cars, but goodness he has always had a bunches of aces up his sleeve. It seems like all he has to do is put lidar into his system and its a go.
Problem is he alienated his buyers on both sides of the aisle…
>
He doesn’t have software for the lidar. Sure, you could strap one on for looks…
I always trust the evaluation of a guy who peppers his Youtube with gimmicks and childish pranks.
PS: I have a friend who couldn’t be more happy with Tesla self-driving tech, the predecessor of FSD.
PS2: Tesla model Y, not a van, Cybertruck, too long, heavy. Tesla needs some model in between.
I found self driving to be alternately terrifying and useless.
The car gives control back to the driver before a crash so the driver can be held responsible and not tesla.
Just out of curiosity i wonder if all these radars and such are dangerous radiation sources. Any studies done etc.
ps im not a conspiracy type person. In my 59 years i have come to realize we make things only to discover they are harmful later. Ddt would be a classic example. Throw in so Teflon etc.
LIDAR does not put out any considerable radiation. If radar did, we would all have cancer by now since they’re essentially just sound waves.
Lidar is light. Radar is radio. Both emit electromagnetic radiation from different parts of the spectrum. Not harmful at low power, but crank them up and focus and you have a laser and a microwave.
You’re thinking of SONAR. There isn’t much of a market for aquatic robotaxis… maybe you’re thinking of BatTaxies I rode in a BatLyft once… had masks for the driver and me, Batman soundtrack music, and even shined the bat signal on the sidewalk when you opened the doors. I am unsure whether he was navigating by sonar.
Do more research on DDT.
Musk’s SpaceX while developing the reuseable booster by getting it to return to a hard land zone softly, suffered a few failures, and learned from them to refine the designs and technology. What with sort sighted focus on the failures seemed like a useless venture turned out to be a very cost-effective advance for space travel utilizing first stage boosters repeatedly. Thinking of Tesla as a failure at this time is very likely thinking of the Falcon 9 booster return landings as a failure before their design perfections.
Not arguing the point about LiDAR ‘seeing’ in fog rain and snow, I personally use a system similar to LiDAR for my vision, and I have limited vision in poor weather. It could be with more use; flaws may be determined in the LiDAR system that will need refinement. Radar with proper end devices ‘sees’ quite well in very poor weather in maritime and flight usage. We may be surprised at what comes of the design improvements for Tesla fully self-driving. No doubt more work needs doing by Tesla but, in my opinion, it is too early to count out the design team.
Every once in a while I see a Buick Rendezvous on the road.
Pontiac Aztec!
Which brand is ready for full self driving or even close to what Tesla offers that can be purchased by consumers? None. It’s easy to pick on Tesla but reality is that no one in the consumer market is even close to what Tesla has to offer.
Waymo 10 million rides
BYD better than Tesla
But OK you cannot buy a Waymo
Lidar is TOO close to LIE-DAR.
And now that we’ve heard from the grade-school cohort….
It will eventually be ready. Tesla is an AI and robotics company, Cathie Wood just loaded up on TSLA with a $2600 prediction by 2030. I’m all in on that as well.
Cathie Wood, vaporware, again, again. Cheers, time to short, different views are what markets are for!