Tesla’s decision to rely solely on cameras for its Autopilot system has been called into question after a recent video test revealed the technology’s limitations in various conditions. A camera versus lidar test, where a Tesla Model Y was pitted against a vehicle with a lidar system, showed that the Tesla’s Autopilot struggled to navigate through heavy fog and heavy rain, but performed well in other scenarios.
The issue lies in how cameras perceive potential obstacles. Unlike lidar sensors, which can detect walls regardless of their appearance, cameras can be tricked by fake road markings or other visual deceptions. This was demonstrated by a Wile E. Coyote-style wall with a fake road painted on it, which the Tesla’s Autopilot failed to recognize.
Critics argue that this approach will make achieving “level 5 autonomy,” which means functioning in any conditions, increasingly difficult. CEO Elon Musk has claimed that Tesla’s vision plus neural net system can adapt like a human driver, but experts say that radars and lidar sensors are necessary for safe self-driving.
A recent video by engineering YouTuber Mark Rober highlighted the problem, showing how the Tesla’s Autopilot struggled to stop in foggy conditions. Musk had previously stated that “very high-resolution radars would be better than pure vision,” but refused to acknowledge their use due to Tesla’s strong stance against them. The cost of retrofitting existing vehicles with radar and lidar systems has also raised concerns, as it may exceed the profit margins of Tesla’s cars.
As Tesla continues to develop its Autopilot system, many experts believe that a combination of cameras, radars, and lidar sensors is necessary for safe self-driving. With level 4 autonomy still in question, Tesla’s vision-only approach has raised significant concerns about the company’s ability to achieve true “level 5 autonomy.”
Source: https://electrek.co/2025/03/16/tesla-autopilot-drives-into-wall-camera-vs-lidar-test