Tesla safety has once again caused controversy. Instead of a purely visual solution, why not look at infrared technology?

Recently, Tesla's high-speed driving caused serious casualties and accidents, and there was a lot of discussion on the Internet. After the accident, there were those who supported Tesla and those who supported the car owners. Some analysts said that sensor or controller failure may cause loss of control during acceleration.

And we all know that Tesla has always been a staunch practitioner of pure visual perception. Tesla uses 8 vision cameras to calculate the distance between the vehicle and obstacles through different camera focus angles.

However, its purely visual perception is currently quite controversial. On the one hand, visual recognition cannot be 100% accurate; on the other hand, because this solution introduces too many variable parameters, the calculation results are prone to errors.

(Tesla’s 8 cameras provide pure visual perception)

In this regard, different radar providers or car companies provide different solutions. For example, "Sagitar Juchuang", a provider of autonomous driving lidar (LiDAR) environmental perception solutions, recently released the solid-state blind-filling lidar E1, which It has a large field of view and a wider sensing range.

For another example, Eian released the Protoss architecture last week, which involves infrared remote sensing technology, which can make up for the shortcomings of pure visual perception and radar.

Because both cameras and radars obtain information based on visible light. It's just the difference between the light reflected by the received object and the light emitted by itself. The disadvantage of both is that when the visibility of the environment is low or the target is occluded, the difficulty of obtaining the target object increases greatly.

To give a very simple example, in bad weather with heavy rain and fog, if there is a small animal walking on the front side of the car, it will be difficult to identify it by pure vision and radar.

Infrared technology has the characteristics of super transmittance, and has stronger anti-interference ability under severe weather conditions such as haze.

In addition, infrared technology uses thermal radiation emitted by objects to image, and does not rely on light imaging, so it has stronger imaging capabilities at night. For drivers with special characteristics who need to run frequently at night, infrared technology has great auxiliary advantages.

In addition, infrared technology is more accurate in identifying living things. From an infrared perspective, the difference between living things and non-living things is very obvious, which can better assist behavioral judgments in emergency situations. For example, at complex intersections, it can avoid sudden reactions. Pedestrians or small animals that break in will cause damage.

Of course, infrared technology is not without its disadvantages. For example, it only has brightness information and no color information, so the image details are not clear. At the same time, due to the small number of infrared detector arrays, the resolution is relatively low.

The best solution is to supplement it with radar and cameras. For example, the following figure is very clear. Aion's Protoss architecture uses mainstream sensing technologies such as cameras, millimeter-wave radar, lidar, and infrared remote sensing. Make accurate judgments with enough data.

In fact, infrared detection is mostly used in military vehicles. Special vehicles usually travel day and night, and the road conditions they travel on are also relatively complex. Infrared assisted driving technology can provide help, and can detect distant obstacles early and detect them in complex road conditions. Make driving decisions as early as possible to ensure safety.

Part of the reason why it is rare for civilian use is that it is expensive. But to achieve high-end autonomous driving, the introduction of infrared detection technology can achieve higher-dimensional perception, which can be said to be a must for high-end smart cars.