Tesla's trillion-dollar market value has made the smart electric vehicle route more determined, but in the field of automatic driving, China's new car manufacturers have no intention of following Tesla.
Nowadays, New car-making forces such as Weilai, Xiaopeng and Weima have begun to "rush to install" lidar, and 2021 has become the first year of mass production of lidar routes.
It is predicted that the global shipments of front-loading models equipped with lidar will be in the order of thousands in 2021, and there will be an exponential increase in 2022-2023 - reaching tens of thousands of units / year, and 2025 may be the next exponential jump - reaching hundreds of thousands of units / year.
Lidar has become a must for China's new car manufacturers to advance autonomous driving.
In 2021, more and more smart electric vehicle brands choose lidar solutions, from Weilai ET7, Xiaopeng P5 to WM M7, 2021 can be called the first year of lidar route mass production.
At the same time, the new forces in the traditional brand - GAC E-An, Zhiji and Jihu have also released their own lidar mounting plans.
Among them, the Xiaopeng P5 and the Polar Fox αS will be delivered one after another this year, and Ideal also plans to release a second model and be equipped with lidar.
Next year, mass production models equipped with lidar, including NIO ET7 and WM M7, will also be delivered, while new car brands, including Zero Run, also plan to put into production of lidar solutions in 2023.
In terms of the number of lidars, Weilai ET7, Xiaopeng P5, Zhiji L7 two, Weima M7 reached three, intelligent hardware is also caught in the inner volume.
<h2>Why lidar</h2>
Not long ago, Tesla's notification to new users testing its fully autonomous driving (FSD) system explicitly mentioned that its electric vehicles will use pure vision systems and completely eliminate on-board radar.
No matter how determined Tesla is on a purely visual route, or even ready to ditch the on-board radar equipment altogether, for now, it has no followers.
That seems to be the biggest disagreement between Tesla and other smart-electric car makers.
Not only most OEMs, but also for Tier1, lidar is necessary to achieve assisted autonomous driving L3 or above.
Regarding the safety of the two perceptual routes, the most common research data comes from ElecFans, which can reach 99.999999% of the lidar autonomous driving system, while sensors such as cameras and millimeter-wave radar can only guarantee 99%.
<b>Although the difference in safety is less than</b> <b>0.1%,</b> <b>the installation of lidar means that its perception is not constrained by light conditions, its ranging ability is strong, and its ability to distinguish objects such as pedestrians and columns is stronger. </b>
A very typical phenomenon in tesla assisted driving accidents was the inability to detect stationary objects, such as a white stationary vehicle parked in front of it.
The potential for accidents caused by this gap comes at the cost of life-threatening.
Some people say that lidar must be necessary to achieve automatic driving?
At least at this stage, who would be willing to take this risk at the cost of their lives. This is also where Chinese car manufacturers are more cautious than Tesla's "technology madman route".
Previously, the reason why lidar was not widely used was because of its high cost.
The Audi A8, which was first equipped with lidar, was equipped with Valeo's 4-line SCALA lidar in 2017, claiming to match L3 level assisted driving, which is the world's first pass-through lidar. However, as Audi's flagship model, not many users have really experienced the Audi A8 equipped with lidar, and its functions are also very limited.
Its real role is to bring about the trend of lidar mass production pre-installation.
On the other hand, from the current point of view, the cost of lidar is not unacceptable. <b>Lidar has a Moore's Law similar to that of computer chips: every 18 months, the price of doubling the resolution of lidar sensors is halved. </b>
At least 5 years ago, lidar was considered expensive equipment.
However, judging from the market size of lidar in the field of advanced assisted driving released by Hesai Technology, 2021 has increased significantly compared with previous years.
The increase in market size also means a decrease in costs.
Huawei's vehicle-grade medium- and long-range 96-line lidar product has reduced the cost of lidar to $200.
At the same time, with the gradual maturity of lidar MEMS technology, in the case of many mass production models of vehicle regulations, the cost of the front-loading fixed point is completely different from the previous blue roof cylindrical mechanical lidar. <b>Some lidar suppliers revealed to Pin Play that the cost of solid-state lidar using MEMS technology can be reduced to less than $500. </b>
The company said that <b>the new generation of technology breakthroughs + large-scale mass production orders are the two major conditions for cost reduction. </b>
That is to say, with the increase in the number of lidar mass production front-loading models, it is also driving the cost of lidar down.
Another major reason for the use of lidar by China's new car manufacturers is that their understanding of China's complex road conditions, including the difference in The perception of full-scenario road conditions between China and the United States, has also led to different choices between China's new car manufacturers and Tesla, which was born in the United States, in terms of perception schemes.
Shen Hui, the founder of WM, gave an example in an interview <b>with the media: when communicating with American suppliers, he found that Americans defined highways as closed scenes and made full-scene displays on high speeds, while China would not consider highways to be closed scenes, and some highways were dangerous. </b>
Therefore, the new forces of Chinese car manufacturing generally choose the lidar scheme, which can be said to be a conservative choice, or it can be said that they are more willing to pay a certain cost to get safer results.
<h2>Solution for WM M7</h2>
Of course, the real determination of automatic driving capabilities is not only lidar, equipped with lidar multi-fusion perception system, bringing about the complementarity of perception capabilities, which means that in the identification of vehicles, identification of human bodies, identification of any object and measurement of distance, including in rain, snow and other bad weather recognition ability, through lidar and other sensors complementary, effectively improve the accuracy of the vehicle's perception of the surrounding environment.
<b>So, hardware redundancy is very necessary before L5 really arrives. </b>
At present, from the layout of automatic driving perception hardware, WM has achieved this forward-looking. Taking the recently announced WM7 with a lidar scheme as an example, in its perception scheme, in addition to the industry's most three autonomous zoom high-precision super-visual solid-state lidar, it also includes 32 sensor configurations including 8 million pixel high-definition cameras, with a perception range of 330 °.
Three intelligent solid-state lidars are distributed on the forward roof of the WM7 and the front fenders on the left and right sides, and each lidar unfolds a 120° field of view, which is responsible for forward, left rear and right rear detection, forming a 330° perception area. WM M7 is also equipped with seven 8-megapixel high-definition cameras, its farthest detection distance of more than 600 meters, close to 4K pixel imaging capabilities, can accurately identify pedestrians, vehicles, traffic signs, and pile barrels and other irregular objects. It can fully perceive the different road conditions of parking lots, streets, main and secondary roads, highways and other scenarios, and fully meet the needs of intelligent perception of the whole scene.
<b>WM hopes to achieve L5 presets on hardware, which almost makes the hardware of the vehicle enterprises fall into the inner volume. </b>But for users, it is like buying a 5G mobile phone when the 5G network has not yet rolled out, and when the network meets the requirements, you can experience it one step ahead.
In addition to hardware redundancy, in order to meet L5, it is also necessary to have a chip with high computing power, powerful algorithms, and a flexible electrical and electronic architecture to ensure that sensors and actuators are more intelligent.
<b>In the final analysis, the improvement of perception, computing, decision-making capabilities, as well as the blessing of wire control chassis technology, are very important for achieving higher levels of automatic driving. </b>
In addition to intelligent hardware redundancy, WM M7 is also equipped with 4 of the latest Orin-X vehicle-level autonomous driving chips, with a computing power of up to 1016TOPS, through the supercomputing platform and continuous optimization of algorithms, to make rational decisions for multi-scenario autonomous driving.
WM M7 adopts a self-developed new central domain control electronic and electrical architecture, which is based on the three-dimensional layout of the three-in-one trinity of execution layer, window layer and ecological layer, and the so-called evolution is based on this, which can not only ensure more efficient and secure in-vehicle communication, but also ensure higher frequency OTA upgrades.
Whether it is hardware redundancy or software redundancy, in fact, in the end, these are only to determine the basic research and development capabilities and technical capabilities of an L4 and L5 level autonomous driving, and in addition, how to define a smart electric vehicle truly reflects the real strength of an automobile company in terms of automatic driving and intelligence.
Although it has achieved the top hardware reserve capabilities, WM is also very clear that intelligence is not stacked by hardware, it needs to continue to help users solve new problems faced by various possible use scenarios, that is, it can evolve and grow, which is an important condition for a smart electric vehicle to be intelligent.
WM put forward a concept - cognitive redundancy, in fact, after the intelligent hardware fell into the inner volume, a car manufacturer should have the ability to open ideas on the basis of hardware capabilities.
After the internal volume, in the end, the new car-making forces must understand what is the meaning behind this arms race.
<h2>What is the real meaning of lidar</h2>?
After the release of the WM M7, there was an interesting debate about the styling.
Three laser radars are distributed on the roof of the car, allowing the WM M7 to break through the mainstream aesthetic of traditional cars, and WM named it a unicorn, but many people still can't understand it.
In fact, if you look at the unmanned vehicle being tested on the road, the large wide angle, mechanical lidar with surround view function, not only is the cost difficult to reduce, it is not beautiful enough, the cost of a car is often millions, for the public, neither economical nor practical.
This is also the reason why many car companies choose solid-state lidar. However, in order to maximize the viewing angle advantage of solid-state lidar, some compromises also need to be made in terms of styling.
The earliest Audi A8 chose to install a circular arc 4-line lidar in the middle of the bumper, which can reach 145° (horizontal viewing angle) and 80m detection distance. The two lidars of the Xiaopeng P5 are mounted on both sides of the bumper to cover the field of view in the forward direction. The NIO ET7 is to install the lidar on the top of the A-pillar on both sides.
In fact, lidar on the roof, can not only prevent being hit, but also expand the detection range, of course, will also cause wind resistance, endurance will be affected.
<b>Between aesthetics, endurance and safety, lidar brings irreconcilable new problems to the styling design of intelligent electric vehicles. </b>At least at this stage, a perfect solution is missing. And it is precisely this that can make us more fully understand the inherent car-making concept of a car-making company.
Obviously, the WM M7 is more concerned about the perception ability of the "unicorn".
And lidar and higher levels of automatic driving, for users, landing or experience.
Although there is still a period of time before the real arrival of L4 and L5, from the current point of view, unmanned parking, intercity high-speed and other scenarios can be more and more safe.
On the other hand, the difference behind the choice of autonomous driving perception route also reflects the difference in the concept of China's new car manufacturing forces and Tesla.
Tesla, while advocating technological progress, hopes to collect a large amount of data and its own powerful algorithms to complete the evolution of autonomous driving. But in the market, it also promotes the strategy of continuously reducing costs to achieve price reduction. Although lower-priced technology products are more popular with the market, but when technological progress has not developed to a certain stage, the cost and price cannot reach a sufficient balance, behind a series of auxiliary driving accidents, it just shows that the misleading behind excessive publicity, the price paid is heavy.
WM emphasizes the universality of science and technology, the high-end intelligent hardware to the mass market, <b>in the process of the intelligent electric vehicle market really from the spindle shape into the dumbbell market, both the product puts forward high hardware level, high technical level requirements, but also to ensure a sufficiently safe cautious attitude, for car manufacturers, consideration is not only technical strength, but also a car manufacturer for greater social responsibility determination. </b>
This more cautious concept is the true meaning of lidar for users, and it is also the real value of scientific and technological inclusion.