If there is one Tesla feature or breakthrough that its CEO, Elon Musk, is adamant about making you try, it would be its illusive Full Self-Driving and Autopilot.

During an earnings call last year, he encouraged shareholders to take him at his word about the automaker’s illusive Full Self-Driving (FSD) software.

Get expert insights and actionable trade alerts from veteran investing experts and hedge fund managers. Join TheStreet Pro today and get first month FREE🤑

“I would encourage anyone to understand the system better to simply try [Full-Self Driving] out, let the car drive you around,” Musk dared investors and analysts. “Once people use it, they tend to continue using it. So it’s vastly compelling.”

Truist analyst William Stein has documented his experiences as his own personal test subject on multiple occasions covering Tesla’s FSD, but one popular YouTuber became a guinea pig to expose one flaw with Tesla’s premier technology.

Mark Rober exposed Tesla’s worst flaw

In a new video on his YouTube channel, former NASA engineer and doohickey inventor Mark Rober exposed the downfall of Tesla’s ADAS technology using a test inspired by a classic Looney Tunes gag.

The test: see if a Tesla will be “fooled” into driving past a fake, painted wall.

Rober points out that Tesla’s ADAS features are primarily powered by an array of cameras, which he compared to an experimental Lexus fitted with a sophisticated LiDAR system by Luminar (a supplier of Polestar and Volvo).

Rober explained that compared to simple cameras, “there is no hiding from the LiDAR scan” since it uses lasers to measure distance and objects at speeds faster than the human eye can process.

More Business of EVs:

EV market share hits record highMercedes is leading a massive overhaul of its EV businessTesla VP shares critical update on long overdue EV

In the video, Rober puts a Tesla Model Y with Autopilot engaged against the LiDAR-equipped Lexus in a series of tests in different conditions involving a child-sized mannequin.

In the tests, the Tesla on Autopilot managed to stop for the mannequin in the middle of the road when the mannequin was standing in the lane, moving across the road, and while the driver was blinded by lights, but it couldn’t stop in simulated fog or heavy rain. In comparison, the Lexus was able to pass with flying colors through every simulated scenario.

However, the most jarring result was with the last scenario, which consisted of a wall with a fake picture of a road painted on it, akin to the Looney Toons gag.

The Luminar LiDAR Lexus’s sensors detected that the wall was, in fact, a wall, which triggered the car to stop well before it could meet a Wile E. Coyote-style ending. However, the wall looked much different from the viewpoint of the cameras on the Tesla Model Y. Essentially, the cameras detected the painted wall as if the road was continuing, causing the EV to smash through it and continue on like no big deal.

Thankfully, the wall was made of styrofoam, not bricks.

Given that it was an extreme demonstration, it illustrated the issue of cameras versus radar and lidar sensors, as they rely on the perception of potential obstacles rather than machine learning software that relies on data about potential obstacles.

Related: Polestar is not playing games with self-driving safety

In July 2024, the Wall Street Journal investigated many foundational flaws and shortcomings associated with Tesla’s Autopilot and FSD. By examining the details of more than 200 crashes involving Teslas equipped with Autopilot and FSD, they found flaws in Tesla’s reliance on car-mounted cameras and machine-learning technology.

When asked about an Autopilot crash that involved a Tesla slamming into an overturned tractor-trailer, Carnegie Mellon associate professor of electrical and computer engineering Phil Koopman told the Wall Street Journal in a video last year that the flaws in said machine learning are more than just simple “holes.”

“The kind of things that tend to go wrong with these systems are things like it was not trained on the pictures of an overturned double trailer,” he said.

“[…] A person would’ve clearly said something big is in the middle of the road. But the way machine learning works is it trains it on a bunch of examples and if it encounters something it doesn’t have a bunch of examples for, it may have no idea what’s going on.”

Tesla, Inc. is traded on the NASDAQ as TSLA

Related: Veteran fund manager issues dire S&P 500 warning for 2025