Economic Observer reporter Pu Zhenyu from Li Bin to He Xiaopeng issued a letter of delivery of 10,000 units, to He Xiaopeng in the air to Musk, and then to Li wanted to shell the group car CEO Wen Wei, in the past few years, the "war of words" of the leaders of new car companies has been common.
However, unlike the previous "war of words" around the macro technical route and sales targets, the recent debate between the big guys in the car circle points to a small component - lidar, and the topic of debate is also extremely trivial - where should lidar be installed? How many are the most suitable?
According to the latest exposure information of Jidu Automobile, its first car robot concept has two lidars placed on the front hood, and the lidar can be retracted to the inside of the hood when it is not working, which is somewhat similar to the design of the jump lights in the early years.
Xia Yiping Weibo
Ideal Auto founder Li Xiang first questioned the location and quantity of Jidu Auto LiDAR, and Jidu Auto CEO Xia Yiping publicly responded to Li Xiang, explaining why two lidars were used. Subsequently, He Xiaopeng, chairman of Xiaopeng Automobile, and Shen Hui, founder of WM Motor, were also mixed in.
Lee wanted to tweet
This round of war of words has a more special industry background. Guo Jishun, vice president of Joyson Electronics, told the Economic Observer that lidar has reached a stage that can be used, but it is certainly far from mature, "for example, a system, it already has this function, which can meet our minimum requirements for mass production, but then there are two paths to continuous enrichment of functions and continuous cost reduction."
Therefore, although 2021 is called the first year of lidar on the car, the number of lidar models that have really come to the market stage is still not much. No matter what kind of lidar layout scheme, it has not been fully verified by the market, which provides a huge space for the four car company leaders to argue.
In fact, the debate between the four head of the car company about lidar is also inevitably meaning to be their own new station platform. In addition to the automotive robots that Jidu Automobile has not yet listed, Ideal Automobile, Xiaopeng Automobile and WM Automobile have all launched new models equipped with lidar, and the location and quantity of each lidar are different.
Despite different views and solutions, one thing in common is that the four car companies all agree on the important role of lidar. At present, the automotive industry has come to the era of intelligent electric, and one of the most imaginative areas of intelligence is automatic driving, and lidar is currently the most cutting-edge automatic driving sensor hardware.
Compared with the predecessors of sensors such as millimeter-wave radar and ultrasonic radar, will lidar become the ultimate solution for autonomous driving sensors? In the case of the increasing maturity of lidar, does the pure visual autonomous driving technology route adhered to by a few car companies such as Tesla still have value?
Where and how many are the best?
One of the reasons Lee wanted to question Jidu's first car was the number of lidars. In Li Xiang's view, "putting one on the roof and putting two on the cover or bumper, there is no difference in performance, and even the single performance on the top of the head will be better."
However, Xia Yiping retorted, "There are still differences (one and two lidars), 120-degree FOV (field of view) and 180-degree FOV are still different, and the corner cases solved are not the same, so there is still a difference in product capabilities and experiences, and even safety."
Lee wanted to tweet
He Xiaopeng and Xia Yiping are more consistent in their views, "Two lidars are definitely far better than one in terms of performance, and the difference can be seen in the future in urban NGP and robustness."
Guo Jishun also believes that a single 120-degree FOV lidar does have limitations, such as in the open road of the city, when the vehicle makes an unprotected 90-degree turn, it needs the lidar to cover the range of 90 degrees on both sides to ensure sufficient safety.
In addition to the number, another reason Li wants to question the first car of Jidu Automobile is the location of its lidar, "the current position (on the front hood) of Jidu Automobile, the latest pedestrian collision regulations can not pass this pass", "in pedestrian collision, maintenance costs, and vibration control (linking the main body), the roof is optimal."
Xia Yiping did not object to Li Xiang's view that the roof was optimal, and He Xiaopeng, while indirectly agreeing with Li Xiang, explained why Xiaopeng Motors did not do so, "We first wanted to put two on the roof, but no matter how designed, they were similar to the rake ears of rabbits, and then gave up."
He Xiaopeng circle of friends
While Advocating the optimal roof, Li wants to deny that lidar placed on the roof will affect the appearance of the vehicle, "the shape of the roof lidar will need to adapt, because it is too similar to the mobile suit - Gundam."
The advantages of the lidar roof deployment scheme at the field of view level are recognized by the industry. According to the data, the lidar currently deployed on the roof of the car, mainly 1550nm long-distance products, can see farther and is more suitable for high-speed scenes. Lidar deployed in bumper/mid-grid locations is not without its advantages, and its perception function is relatively strong, which is more suitable for changeable and complex urban scenes.
Looking at the electric vehicle models that have been equipped with lidar, their lidar has two mainstream deployment areas, one is the bumper /middle network, of which Xiaopeng P5 and Jihu Alpha S use this scheme, and the other is the roof area, Weilai ET7, Ideal L9 uses this scheme, and WM M7 is deployed in the upper and lower areas at the same time.
Li wants to speculate that jidu automobile in the formal mass production link, can not escape the above two layout schemes, "lidar will not be arranged in the lid position shown in the picture, should only be the placement of the test car." If 2-3 lidar are used in the front, there is a high probability that it will be placed on the bumper and fenders, and the wind resistance and shape will be better, although the cost is higher."
A new path to differentiation for head players
"Differentiated value is the key to winning the market, for the new forces, their brand strength is relatively limited, the cost of service differentiation is too high, so most of them are desperately trying to build and emphasize differentiation in technology and function." In the view of Yan Jinghui, an analyst in the automotive industry.
From a functional point of view, intelligent driving is generally the key selling point of new car manufacturers. However, with the advancement of technology and the reduction of costs, the intelligent driving function of L2.5 or L2+ has become more and more a standard in new cars, and due to legal, ethical and other factors, the L3 intelligent driving function has not yet been truly landed. Therefore, from the perspective of realizing functions alone, the differentiated advantages of head new car manufacturers in automatic driving are gradually no longer prominent.
In the case of functional differentiation is no longer so prominent, the heap hardware configuration has become a new direction for new car manufacturers to seek to establish differentiated competitiveness of autonomous driving, and lidar is obviously a valuable breakthrough. Ben Ning, chief engineer of dongfeng technology center, told the Economic Observer that lidar is on the way to maturity.
It is precisely because it is not yet fully mature in terms of cost and technology, the use of lidar itself is a unique early move. Except for a few car companies, most car companies are not particularly anxious about mass production and listing of lidar models.
Zhu Jiangming, chairman of Zero Run Automobile, told the Economic Observer that lidar is better than millimeter-wave radar in some aspects, but the current products are not mature enough, and it is expected that the products will mature at the end of the year or next year, and the quality will be guaranteed. The 2024 products of zero-run cars are also equipped with lidar, but vision is still the main perceptual component of intelligent driving.
It is worth noting that the most explosive scene of this new armament race for autonomous driving that focuses on lidar is not the air war of words between the four car companies. It is the sentence shouted out by the head of the salon mech of Great Wall Motors' new high-end model: "How many lidar do you have?" If you have less than 4 pieces, please do not talk. ”
This sentence comes from November 2021, when the first model of the Salon brand "Mecha Dragon" was launched, and the two core selling points of the car in the field of automatic driving are focused on lidar, one is "the world's first 4 lidars" and the other is "the world's only laser full-view coverage".
In fact, Shen Hui also "Versailles" the number of lidars of A WM car in this round of war of words. He said, "The WM M7 has a horizontal viewing angle coverage of 330° through the layout of the overhead 1 + the two sides of the 2, which can take care of the cornerstone in most cases. In this way, perhaps 3 is the optimal solution for the current level of perception and the balance of safety."
Shen Hui Weibo
It is worth noting that lidar is not yet fully mature, but the current competition around it by new car manufacturers does not lack practical significance.
This is reflected in the fact that lidar is considered to be an important hardware basis for achieving high-level intelligent driving, although the current intelligent driving function above L3 level cannot be legally landed, but lidar can make the existing low-level intelligent driving function of the vehicle safer. For example, Guo Jishun said that close-range lidar can act as an angle radar and a blinding radar.
Is lidar the ultimate solution?
So, as lidar, which represents the most cutting-edge technology, further matures, will millimeter-wave radar and ultrasonic radar face the fate of being replaced and eliminated? In this regard, Guo Jishun gave a negative answer.
"The more scenarios (the higher the level of intelligent driving system) need to deal with, the more it needs heterogeneous sensors, that is, in different scenarios, sensors with different characteristics and different performances are redundant with each other." In the future, more advanced millimeter-wave radars may replace relatively less advanced millimeter-wave radars, such as 24G millimeter-wave radars being replaced by 4D millimeter-wave radars, but they should still look like a redundant relationship with each other. Guo Jishun said.
It is understood that lidar has advantages in detection accuracy and anti-electromagnetic interference, while millimeter-wave radar has advantages in anti-environmental interference capabilities. Lidar can not be used normally in some bad weather, sand and dust and other climatic environments, and is not suitable for use in some highly repetitive scenarios, while in low-speed, changeable urban scenarios, lidar can play a full role.
At present, most car companies are actively laying out multi-sensor technology routes such as lidar, but there are also car companies such as Tesla that adhere to pure visual intelligent driving technology solutions.
As early as May 2021, Tesla announced that all Model 3 and ModelY cars would no longer be equipped with radar, and would be replaced by Tesla's pure visual self-driving system that uses only 8 external cameras. In February, Tesla announced that it would introduce pure vision systems to the ModelS and Model X.
For the pure visual intelligent driving technology route, the industry has different views, of which the mainstream opinion is not to deny some of the advantages of the pure visual route itself, but it is difficult to rely on pure vision to achieve high-level intelligent driving.
Benning told the Economic Observer that pure vision is definitely a good technical route, because vision can provide 3D information, but it takes more time and algorithms.
Guo Jishun said that L2 and below with pure vision is feasible, but L3, L4 and other higher levels must have lidar, "pure vision to do high-level intelligent driving, that is to say, off the hand, off the eye, in my cognition is difficult to achieve, technically very difficult."
However, Guo Jishun also mentioned that in the past few years, the development of visual perception technology and the continuous improvement of the amount of deep learning data have indeed brought about the rapid progress of vision-based solutions for autonomous driving, and vision sensors have made great progress and performance has continued to improve, so the imagination space is also very large.
From an industrial point of view, the sensor is only an integral part of the intelligent driving system, and the commercialization of the entire system, overemphasizing one of the parts, is also considered a matter of little significance. "Hardware configuration simply represents a potential possibility, ultimately a balance of safety, cost, scenario and capability, still the barrel effect rather than sprint logic." He Xiaopeng said.