YouTuber Tricks Tesla Autopilot With Painted Wall
A YouTuber has demonstrated a flaw in Tesla’s camera-only Autopilot system by tricking it into crashing into a painted wall.
Engineer and former NASA employee Mark Rober conducted the test on his YouTube channel, showing how Tesla’s reliance on cameras and artificial intelligence, rather than additional sensors like radar and lidar, can lead to dangerous misinterpretations of the environment.
Rober set up an experiment inspired by Looney Tunes, where a wall was painted to resemble an open road. When a Tesla on Autopilot approached, it failed to recognise the obstacle and crashed into it.
For comparison, Rober tested a lidar-equipped Lexus SUV from Luminar. The Lexus successfully detected the wall and stopped in time, highlighting the difference in how the two systems perceive their surroundings.
Tesla CEO Elon Musk has consistently defended the company’s camera-based approach, arguing that it mimics human vision. However, critics argue that without additional sensors, Tesla’s system may struggle with certain real-world scenarios.
Tesla’s Autopilot and Full Self-Driving systems have been linked to multiple crashes, some fatal, and have been the subject of state and federal investigations. Despite improvements, experts continue to question whether a camera-only system can ever be as safe as human drivers.
YouTuber Mark Rober tested Tesla’s Autopilot with a painted wall.
The Tesla failed to detect the obstacle and crashed.
A lidar-equipped Lexus SUV successfully stopped before the wall.
Source: INSIDE EVs