by - The Tesla owner repeated Mark Rober’s fake wall test on the road.
- In some tests, they achieved the same result, while in others they did not.
- The difference seems to lie in the newer versions of FSD.
Earlier this week, Mark Rober sparked a major online battle by testing driving technology. In a video titled Can You Trick a Self-Driving Car? he compared lidar with optical systems, such as those used by Tesla. The result? Instant criticism, praise, confusion — basically, the entire internet lost its mind. Now someone else has performed the same test, and, unsurprisingly, the results are both familiar and slightly different.
In a short summary, lidar can see more clearly and accurately compared to optical systems. This should not be a big surprise, as it is ultimately a highly detailed radar system that can detect objects in complete darkness.
However, when Rober highlighted Tesla’s failure to detect a wall that looked like a real road, brand supporters came out “with shovels,” so to speak. It’s fair to acknowledge that Rober’s test did not use full self-driving (Supervision), but only Autopilot.
More:
This is where Kyle Paul, the owner, enters the scene. He decided to repeat the same test under the same general parameters, but this time using FSD instead of Autopilot. He printed his own wall that looked like the real road it was standing on and drove his Model Y towards it several times.
In each test, Tesla did not detect the wall until Kyle was literally inches away from it. As Rober suggested in an interview, the capacity of parking ultrasonic sensors probably detected the wall, not the autonomous driving technology.
Next, Paul changed the situation by bringing in the Cybertruck for the same test. Interestingly, it passed the test with flying colors, stopping on its own every time it began to approach the wall. What’s the difference? Besides the obvious fact that FSD is more advanced than Autopilot, it lay in the Tesla HW4 hardware. The 2022 Model Y did not have it since it operated on HW3 hardware.
Missing key aspects
A significant comment noted the tests that Paul did not conduct. For example, he did not test FSD with a dummy or in the rain — two factors that could provide a more realistic notion of how the system performs in everyday conditions.
Nevertheless, this should at least help quell the noise around Rober’s video. It’s clear that there is some truth in the criticism, and those who continue to challenge Tesla’s approach to autonomous driving are not entirely wrong.
