Domestic self-driving technology criticizes Tesla, understands Tesla, and finally integrates into Tesla
Domestic unmanned vehicles: Black Tesla, understand Tesla, and finally be Tesla!
As we all know, in the early days of the field of unmanned driving, there were two main factions, one was pure visualism, such as Tesla, who used cameras to collect information, and then relied on pure visual computing to complete the driving of unmanned vehicles.
The other faction is represented by Huawei and other groups, mainly domestic enterprises, "lidar" faction, which collects data, combines cameras, integrates different types of sensors, and completes unmanned driving through computing.
In the early years, the LiDAR faction complained about Tesla's visual algorithm, saying that its algorithm was flawed, such as in the dark, snow, ice and snow, rain, fog and other weak light, the camera was interfered by the light, and the amount of information obtained was very small.
The most important thing is that because the cameras are all 2 D, without any distance and other data, even the most advanced algorithms cannot make up for the shortcomings, Tesla's unmanned driving is not suitable for them, and even LiDAR is not as good as ......
However, when domestic vehicles are equipped with this kind of lidar, Tesla's approach has been recognized, after all, it is much more expensive than cameras.
In addition, the excess data and the maintenance and maintenance costs required to keep the vehicle on the vehicle for a long time are taken into account, and the cost is even higher than that of a conventional camera.
In addition, compared with pure vision, the unmanned system using lidar has not changed in essence, so everyone agrees that if there is a good algorithm, then this result will not be bad.
So, in order to save money, they bought Tesla's camera, which makes sense - because Tesla's computing power is too strong, and it can also solve some hardware problems through computing.
Since then, local companies in the United States have followed Tesla's lead and become Tesla, that is, they have removed the lidar and adopted pure image processing technology.
With visualization, a lot of resources can be saved; In today's competitive society, price is the most important competitiveness, and if the price is too high, the car cannot be sold, and everything else is empty talk.
Secondly, due to the development of cameras, the chips of ISPs are now able to adapt well to low-brightness light, so this problem called "lack of information" has gradually been overcome.
At the same time, with advanced scientific and technological means such as AI and OCC placeholder network, 2D images can be converted into 3D images through computer assistance, which is rich in content and far exceeds the existing LiDAR images, and has great application prospects in mineral resource exploration.
This means that some people used to think that Tesla only adopted the camera because of the ** problem, but now this view seems very ridiculous.
Musk believes that humans can drive cars with both eyes, and so can cameras, but if they can't drive, it's a problem with technology, and it has to be improved.
What Musk said seems to be an understanding of technology.