On April 28, it was reported that the battle between pure vision and laser radar has been going on for a long time. Jin Yuzhi, CEO of Huawei Smart Car Solution BU, said in a recent interview with the media that there are two main reasons why cars must have laser radar in the future to move towards L3 and L4 automatic driving.
Why do we need a LiDAR to achieve L3 and L4? ”The host asked.
Jin Yuzhi stated that technically speaking, for any obstacle, the principle of LiDAR determines that there is no need to recognize the obstacle, but it must be able to detect the obstacle, know it is an obstacle, and stop or avoid it accordingly.
Pure vision requires recognizing obstacles, which requires a lot of training. If there is not enough data, or if the amount of data is very large but there are always some things that have not been seen before, it may be unfamiliar, and this extreme and small corner case will bring risks.
Secondly, he also pointed out that the camera of the pure vision scheme is limited in some scenes, such as in the environment of very dark light and rainstorm, its definition will be affected.
Vision is essentially a simulation of human eyes, and not being able to see clearly can bring risks, but through the fusion of sensing capabilities such as lidar and millimeter wave radar, these can be effective in extreme scenarios.
It is understood that pure visual solutions mainly rely on image data captured by cameras and processed and analyzed through deep learning algorithms to achieve environmental perception and decision-making.
At present, the main representatives include Tesla's FSD, Israel's Mobileye's SuperVision, China's DJI Auto, Horizon's Mono series, and Xianglong Technology's Erlang Shen Sanmu visual system, among other mainstream pure visual perception solutions.
