laitimes

Why not use radar to insist on pure vision Tesla official: A truly smart car should be like a person

author:Chopping wood nets

On April 18th, it was reported that unlike the fusion perception scheme used by many domestic manufacturers, Tesla's automatic assisted driving system adheres to pure vision, and even millimeter-wave radar is cut, so why is it so? Tesla officially answered.

Human driving is completely based on vision, roads are also designed for biological neural networks and eyes, and humans do not have organs similar to radar, so a smart car can also imitate humans and rely on vision to become a qualified driver.

Why not use radar to insist on pure vision Tesla official: A truly smart car should be like a person

Tesla's automated driving assistance system simulates the process of human drivers driving, the camera collects road condition information, shoots real images, the visual neural network rebuilds the images captured by the camera into a 3D scene, and the fully autonomous driving computer analyzes and processes the scene information to make decisions.

And compared to the human eye, the vehicle camera field of view reaches 360 °, the detection range far exceeds that of humans, in the speed of driving reaction, the computing hardware is also better than the human driver, and the important thing is that the computer can always respond quickly and will not be as tired as the human brain.

Why not use radar to insist on pure vision Tesla official: A truly smart car should be like a person

So why not choose the combination of radar + camera? Musk once said: "If the information from the radar and the camera contradict each other, the intelligent driving system will be difficult to choose." It's better to choose only one and do it to the extreme. ”

Why not use radar to insist on pure vision Tesla official: A truly smart car should be like a person

Many car companies are doing intelligent assisted driving, and most of them are using a combination of "lidar + camera". It is indeed easier to stack a large number of sensors to improve the intelligent performance of the vehicle, but the cost of the sensor will also be paid by the consumer.

Numerous industry tests have also shown that simply increasing the number of sensors does not necessarily mean that the vehicle can better identify road conditions, but may lead to different information interfering with each other, wasting a lot of computing power, and resulting in 1+1

Therefore, Tesla has chosen a pure visual automatic assisted driving solution, I don't know if the official interpretation agrees?

[Source: Fast Technology]

Read on