One man put Tesla’s self-driving technology to the test, and the results were wild.
Earlier this month, YouTuber and former NASA engineer Mark Rober decided to see how far Tesla’s self-driving tech could be pushed by releasing a video titled ‘Can You Fool a Self Driving Car?’
He wanted to find out whether Tesla’s Autopilot, which relies on ‘simple cameras and image processing,’ could tell the difference between a real road and an optical illusion.
Using a Wile E. Coyote-style trap, he noticed a huge difference in two different self-driving technologies.
In the test, a Tesla on Autopilot drove straight through the lightweight, styrofoam wall without even hitting the brakes.
The high-speed footage showed just how easily the vehicle was tricked – a surprising result, considering the car’s camera system is supposed to recognise and respond to obstacles.
“I can definitely say for the first time in the history of the world, Tesla’s optical camera system would absolutely smash through a fake wall without even a slight tap on the brakes,” Rober said.
Rober compared Tesla’s camera-only setup with other self-driving systems that use ‘more advanced technology’ like LiDAR and radar. These systems are used by many of Tesla’s competitors and can scan the surroundings in 3D, identifying a real road from a fake one.
In this case, the LiDAR-tech vehicle had no problem identifying the fake ‘wall’ as a solid object and stopping in more than enough time.
Furthermore, the YouTuber ran several other tests at 40 mph, including one where both the Tesla and another car with LiDAR approached a dummy.
While the LiDAR-tech car performed well in all the tests, the Tesla had some issues.
In some instances, the Musk-backed company car noticed the dummy, but didn’t hit the brakes fast enough and ended up slamming right into it. Rober explained that this was an example of Tesla’s automatic emergency braking system which ‘only touches the brakes when it’s 100% there’s a problem, in order to avoid false positives.’
Mark Rober / YouTube
But when Autopilot is turned on, the car assumes the driver is ‘not paying attention’ and slows down sooner and more regularly.
In another round of experiments, Rober tested how the cars handled bad weather, using fog and rain to obscure the dummy.
Again, the LiDAR-equipped car handled it much better, with the Tesla making contact with the dummy in most situations.
However, the controversial video kicked off a heated debate among Tesla fans. Some argued that Rober should’ve used the company’s pricier and more advanced Full Self-Driving (FSD) software. Even with that in mind, the test raised real questions about Tesla’s camera-only approach when it comes to safety on the road.
