If there is no Li Xiang in the Chinese automotive circle, there must be a lot less melon.
This time, Li Xiang, the founder of Ideal Auto, was on Weibo again, and Xia Yiping, CEO of Jidu Automobile, about the installation location of lidar, and "broke his head" online.
The cause of the incident came from a media communication meeting in Jidu, and AutoLab also participated in this online sharing. At the meeting, Jidu announced the latest progress of the first automobile robot concept car, the most important of which was the disclosure of two lifting lidars located on the front cover of the car.
Immediately after that, in the Weibo report of the "No. 42 Garage" about the Jidu Communication Meeting, Li wanted to comment that there was no difference in performance between putting one on the roof and putting 2 on the cover or bumper."
At the same time, Li Xiang also pointed out the various drawbacks of lidar on the cover, saying that the roof is the optimal solution. Lee would like to think that the only drawback of the lidar loading roof is that it collides with the "Mobile Suit - Gundam".
Many netizens said that Li Zong's remarks were a bit insulting, which was obviously the top of his head was swollen by someone, why did he pull on the Gundam? However, I searched the gundam's map, and it did look like a lidar on the top of my head, and it was suspected that I had purchased Velodyne Velabit.
For Mr. Li's "friendly discussion", Xia Yiping, CEO of Jidu Automobile, also quickly replied to the other party's defense friend on Weibo.
Xia Yiping proposed that the 120° FOV and the 180° FOV are still different, and there are still differences in product capabilities, functional experience, and safety.
Since then, Li Xiang has updated a comment, believing that if 2-3 lidar is used when the measurement is produced, it will most likely be placed on the bumper or fender, and believes that it cannot be placed in this position to pass the latest pedestrian collision regulations.
How different is the picture seen at 120° and 180°?
Math problems everyone will, 180 ° ≠ 120 °, but do not know the car, what is different, I still give you a simple drawing of a picture, convenient for everyone to understand.
In the picture, red is the angle of the ideal L9, and green is the angle of Jidu's first robot. Although the scale is not so precise, the field of vision that initially appears to be swept is indeed different, and each has its own blind spot.
Of course, this is only a blind spot indication from the horizontal FOV, and does not take into account the vertical FOV. Theoretically, the position of the front cover is placed by lidar, which will be partially obscured by the point cloud, because this piece is basically a plane.
However, we look at the placement position released by the set degree, and the set degree has made a micro-arch design in the position of the lidar arrangement, which not only solves the problem of the point cloud being obscured, but also makes the appearance look more super-running.
In addition, through the design drawings previously released by Jidu, the A-column angle of Jidu automotive robots is relatively small, and the design of SUVs such as the ideal L9 is very different, so the two cars naturally have certain differences in lidar arrangement.
Let's also look at a set of sensors, the same scene, the camera shot the picture, with 2 lidar and synthetic stitching after the picture.
Of course, it is also compared to the camera of the ideal L9 and the real picture of the lidar.
It can be seen that the lidar picture of the two cars is basically seen that the lidar picture is slightly larger than the front view camera screen. With the naked eye, it is really difficult to say how different the completeness and richness of the information on the two point cloud screens are.
In addition to the difference in perspective and blind spots, Li Xiang also mentioned the problem of pedestrian collision and vibration control. In order to integrate the lidar in the front cover position, the set degree is also specially designed to be lifted, and the overall cost is definitely higher than that of the direct COS Gundam.
So why does the set degree have to be close to the far?
The lidar arrangement should follow the functional scene
At this stage, in addition to the concentration of this kind of lifting and lowering in the front cover, the ideal L9 this kind of top lookout type; there is also a Xiaopeng bumper left and right corner of the layout; Huawei ADS bumper on the front left front left one.
There is also a Lotus ELETRE front, back, left, right, and a lidar arrangement dedicated to sweeping to the front and left of the noH.
It can be said that there is no fixed paradigm for lidar arrangements, and each family has chosen different layout schemes in order to adapt to its own functional needs. There may be differences in cost and difficulty of implementation between options, but there are no advantages or disadvantages.
In general, the field of view is higher at the top, and the placement cost is the most controllable, but the near-field low obstacle will have a certain blind spot, so the design orientation will be based on the perception of the front middle and long distance, such as improving the ability of the forward AEB, and improving the recognition of stationary obstacles under the ACC, and improving the safety of the pilot assistance driving function.
Lidar is arranged in the bumper position, the blind spot of vision in the vertical direction is smaller, and the identification of low obstacles in the near field is better, for example, at this stage, the L3 automatic driving model that sees the highway traffic jam scene (within 60km/h), such as the Mercedes-Benz EQS, will choose to integrate a lidar in front of the bumper position.
Therefore, where the lidar is integrated may not be so important, and there is no optimal solution, and it is a serious thing to pull out the function early for the big guys to slip.