Chapter 2134: Exaggerated new generation lidar technology
“For example, we have improved the accuracy of the detection sensors installed in the autonomous driving system. First of all, we have improved the original lidar, greatly extending its lidar ranging range.
As we all know, the detection distance directly affects the response time of the system for identifying and warning obstacles. The longer the detection distance, the more data information can be detected, which is conducive to improving overall safety.
In addition, the longer the detection distance, it means that the speed of the vehicle can be greatly improved. In the past, self-driving cars must be controlled within a speed range, which may be limited by detection technology. Beyond this speed, the automatic
The driving system will automatically exit and let the driver take over.
Our new generation of lidar can detect farther distances, which is basically three to four times the detection range of these lidars currently on the market. This also means that our autonomous driving system can detect distances hundreds of meters away.
Detect obstacles that may appear ahead, so you can plan ahead without having to react temporarily when you arrive.
At the same time, it also means that autonomous driving technology supported by the new generation of lidar can be applied to faster vehicles or other means of transportation. For example, high-speed rail can reach a speed of 350 kilometers.
, some can even reach 400 kilometers. Such a fast speed leaves the driver with very little time to react. If this system can be applied, then the high-speed train will be able to detect and react to the road conditions ahead more accurately while traveling at high speed.
It’s faster. This helps reduce some accidents and problems that occur because people are limited by various conditions and react too slowly.”
At this point, Zhou Yonghui took a breath, and then continued: "In addition to the detection range, the scanning frequency and detection resolution of lidar are also the basis for measuring the quality of lidar.
This time, we have put a lot of effort into these two aspects and greatly improved the performance in these two aspects.
The so-called scanning frequency refers to the detection speed of the same object or surrounding environment within a period of time. The higher the scanning rate, the more detailed and timely the surrounding environment information obtained by lidar detection, and the surrounding environment can be detected in real time.
small changes in.
The difference in scanning frequency may not make much difference in low-speed processes, but in high-speed and ultra-high-speed environments, scanning frequency is crucial. The higher the scanning frequency, the more detailed the detection of high-speed moving objects, so that we can learn
More detailed shapes of high-speed moving objects around you can tell the autonomous driving system to handle and avoid them.
As for detection resolution, it refers to the detection ability of laser radar for small objects.
One thing everyone should know is that the objects detected by lidar are not what we see, but point clouds. The contour shape of the object can be known by the distance between the object's contour surface and the lidar.
The higher the detection resolution, the more detailed the outline of the detected object we can get.
The current lidar technology on the market can achieve high ranging resolution for close-range targets, but its resolution will also drop sharply as the distance to the detected object increases.
Therefore, in order to achieve longer-distance detection, it is not just as simple as increasing the laser power, but also requires essential improvements to the ranging core.
These lidar products on the market have very low detection resolution for long-distance targets and environments, which will cause some problems when the vehicle is driving.
As for our new generation lidar, its detection resolution has been greatly improved. Its detection accuracy of 100 meters can reach centimeter level. What does it mean? That is to say, we can identify the ground on the ground at a distance of 100 meters.
A small pit or a falling iron nail, so you can avoid it in advance."
This is too exaggerated. After hearing Zhou Yonghui’s introduction, some people at the scene couldn’t help but exclaimed.
Haha, Wu Hao and the others smiled when they heard this. They were indeed a little surprised by such a high detection accuracy.
"It's a bit exaggerated, but this is the result of our actual test, without any adulteration. Moreover, the detection accuracy is very high. The size and contour of the target object we detected are basically consistent with the size and contour of the actual object. There is no
There is a deviation." Zhou Yonghui responded with a smile.
After saying this, Zhou Yonghui glanced at Wu Hao, and then continued: "To improve the performance of the new generation of lidar, we have also greatly improved the anti-interference performance of the new generation of lidar.
In particular, the detection rate of objects of different colors and the anti-interference of ambient light and other light have been greatly improved.
The original detection of lidar is to use the time between the emission of light and the object being measured, the reflection and reception, and the time between the return and return, so as to obtain the distance between the lidar and the detected object, and then the object in front can be known through countless rays of light.
The size is reduced to information.
Then this brings a new problem. As we all know, objects of different colors and materials have different light reflectivities, so the detection values obtained are naturally very different.
For example, dark objects have a high reflectivity unlike white objects. Dark objects can absorb most of the light. It is obviously unrealistic to expect lidar to have the same detection power for dark objects as white objects.
.
Therefore, how to balance the direct detection differences of objects of different colors and materials, and how to reduce the impact of dark-colored objects on lidar detection, has been the focus of our research.
Fortunately, Huangtian paid off. After our continuous research and optimization, we finally improved our performance in this area by improving sensor technology, optical adjustment, and optimization processing algorithms. As of now, we have greatly improved our performance in this area.
In this regard, we can completely defeat a lidar product on the market.
In addition, there is also the anti-interference ability of ambient light, which is also the focus of our research.
Lidar uses light to detect objects, so it will naturally be affected by other light. For example, direct sunlight, artificial lights, and other ambient lights will have an impact on the detection performance of lidar.
Therefore, we have also put a lot of effort into this aspect. On the one hand, we have improved the laser light source in lidar so that it can be as different from natural light as possible. On the other hand, we have used intelligent processing algorithms to perform calculations and filtering.
Removing noise from other light sources besides lidar makes the laser detection data obtained clearer and more accurate."
