Tesla’s decision to rely solely on cameras for its Autopilot system has been a topic of debate in the autonomous driving industry. While most companies use a combination of sensors including cameras, radar, lidar, and ultrasonic sensors, Tesla has removed radars from its vehicles and deactivated them in existing models.
The company’s CEO, Elon Musk, has defended this strategy by claiming that Tesla’s vision plus neural net system is designed to mimic a human driver and can adapt to any road condition. However, critics argue that cameras have limitations compared to lidar sensors, especially when it comes to achieving level 5 autonomy, which allows for driving in any conditions.
A recent video by engineering YouTuber Mark Rober put Tesla’s Autopilot system to the test against a vehicle equipped with lidar technology. The tests showed that while the Tesla Model Y on Autopilot was able to detect and stop for obstacles in various scenarios, such as statics, moving objects, and bright lights, it struggled in heavy fog and rain.
One particularly interesting test involved a fake road wall painted to resemble a road, similar to something out of a Wile E. Coyote cartoon. The lidar system was able to detect the wall as an obstacle, while the camera-based system on the Tesla was tricked by the painted road.
This demonstration highlights the inherent limitations of relying solely on cameras for autonomous driving. While cameras can perceive obstacles based on visual cues, they lack the precision and reliability of radar or lidar sensors, which provide more accurate and data-driven information about the environment.
It is clear from these tests that Tesla’s current vehicles are not capable of achieving level 5 autonomy as Elon Musk has claimed. While cameras can be effective in certain situations, the addition of radars and lidars can significantly enhance the safety and reliability of autonomous driving systems.
In a conversation with Electrek during Tesla’s transition to vision-only systems, Musk acknowledged that “very high-resolution radars would be better than pure vision.” However, he maintained that such radars did not exist at the time. Despite the advancements in radar and lidar technology, Musk has been reluctant to incorporate them into Tesla’s vehicles due to his longstanding stance against their use.
Overall, while Tesla’s vision-only approach has its advantages, the tests conducted by Mark Rober highlight the importance of integrating radar and lidar sensors for achieving truly autonomous driving capabilities. As technology continues to evolve, it will be interesting to see how Tesla adapts its strategy in the future.
