Lidar VS pure vision, the measured results are unexpected: who is better?
Updated on: 36-0-0 0:0:0

On the track of autonomous driving technology, a long-standing focus of controversy is: which is superior, lidar or pure vision solution? This topic not only aroused heated discussions among manufacturers in the industry, but also affected the hearts of the majority of users.

In China, Huawei and other car giants have stood in line with LiDAR solutions, and they firmly believe that LiDAR is the first choice to ensure driving safety in various complex environments. Lidar has won the favor of these technology pioneers due to its high accuracy and strong environmental perception capabilities.

However, Elon Musk, who is at the helm of Tesla, has a diametrically opposite attitude to this. He has repeatedly criticized lidar in public and on social media, arguing that this technology is not the way of the future, and that only pure vision solutions are the ultimate answer to autonomous driving. Musk's insistence has undoubtedly added more gunpowder to this technology battle.

In order to visually show the advantages and disadvantages of the two schemes, Mark Rober, a well-known foreign blogger, designed a unique comparative test. He selected a Tesla Model Y in Autopilot mode and conducted a series of rigorous tests against a vehicle equipped with a lidar system.

The test results showed that Tesla's pure vision solution was able to accurately identify and avoid obstacles in most situations, and performed well. However, in adverse weather conditions such as fog or heavy rain, its performance is greatly reduced. In contrast, the LiDAR system can still maintain stable performance in these extreme environments, showing stronger environmental adaptability.

In the final scene of the test, Rober set up a wall with a fake road painted on it to simulate the difference between the camera and the lidar. The results show that the camera mainly relies on the visual perception of obstacles, so it is easy to be confused by the pattern on the surface. LiDAR, on the other hand, detects obstacles based on actual data and is not deceived by superficial illusions. In short, lidar focuses more on the presence of the obstacle itself than on the camouflage on its surface.

This test result undoubtedly adds a convincing force to the lidar solution. While cameras are capable of autonomous driving in most cases, the addition of radar can undoubtedly further improve the safety and reliability of the system. Musk has admitted in private messages that high-resolution radar may be superior to pure vision systems, but he believes that such radars do not exist in reality. However, as technology continues to advance, this view may be upended in the future.