On the evening of October 22, WM, a new car-making force, officially released its first all-scenario intelligent pure electric car, the WM M7, which attracted countless lights.
The most eye-catching thing about this new car of WM is its unique shape. There is a conspicuous bulge at the front of the roof, like wearing a small hat, which is more abrupt on the overall body, and is also evaluated by many spectators who are accustomed to smoothing the body as "ugly out of the sky".
WM M7 and conspicuous lidar on the roof
Design aside, this "hat" on the roof of the WM7 is actually hidden, it is the famous lidar. In fact, in addition to this "hat", the M7 is also equipped with 2 other lidars to achieve better environmental perception of the car.
The price of lidar has always been very expensive, and the WM M7 is equipped with 3 lidars at once, which makes people have to praise a "trench gas". With excellent performance in environmental perception, lidar has gradually become the standard of a new generation of smart car brands, and even has a hint of an arms race.
The world's first lidar mass production car Xiaopeng P5 is equipped with 2 lidars, and also hits the price of the vehicle to less than 200,000; the Polar Fox Alpha S Huawei HI version created by Huawei and Beiqi Blue Valley is equipped with 3 lidars and will be delivered in small quantities; Weilai's new generation production car ET7 will also be equipped with a lidar; the Great Wall WEY Mocha will be equipped with 3 laser radars...
It seems that overnight, lidar, which was once a necessary smart hardware for self-driving cars Robotaxi, began to be equipped with commercial mass production models by major car companies, so 2021 is also regarded by many investors as the first year of lidar mass production.
One, two and then three, the lidar is gushing, opening up the configuration competition between intelligent electric vehicle companies, and also opening up the blue ocean market of lidar 100 billion.
/ 01 /
The "Eye" of Smart Cars
From ancient times to the present, the word "intelligence" has seemed so ethereal that it is difficult for anyone to give an accurate definition.
For today's cars, "intelligence" is embodied in specific intelligent use scenarios such as valet parking and assisted driving at different levels, and the core representative ability is automatic driving.
If you want to make the car "smart", you must build an intelligent system with the trinity of "perception, decision-making, and execution". Among them, perception is mainly detected by various environmental monitoring sensors installed on the car, decision-making mainly relies on algorithms, software and chips, while execution is done by various controllers and components.
In the autonomous driving scenario, smart cars need to accurately identify the driving environment such as vehicles, pedestrians, traffic lights, and obstacles, and must use on-board sensors such as on-board cameras, ultrasonic radar, millimeter-wave radar, and lidar.
Different in-vehicle sensors have very different characteristics.
The camera can collect external data and image recognition according to the algorithm, the technology maturity is high, the price is cheap, and image recognition is also a path to achieve artificial intelligence, with greater potential. However, it is susceptible to the influence of bad weather, and the effect is poor in the complex environment of backlight and light and shadow, and it is difficult to accurately measure the distance.
Millimeter-wave radar measures the distance through millimeter waves, is not affected by the weather, and the detection distance can reach 200 meters. However, the cost is high, the ability to detect non-metallic objects is weak, and it is difficult to identify pedestrians.
Ultrasonic radar uses ultrasonic ranging, the cost is low, but the effective detection distance is usually less than 5 meters, the angle resolution is poor, and it is susceptible to weather.
Lidar through laser ranging, the highest cost, but the detection distance is long, the angle measurement accuracy is extremely high, can be built in real time 3D model, this function has no effective alternative.
These different on-board sensors, each with its own advantages, can complement each other to a certain extent. However, in the realization of the purpose of intelligent driving, the choices of various car companies are completely different, and basically two types of program routes have been formed:
The first type is represented by Tesla, which adopts the "pure vision + algorithm scheme" scheme, generally only using cameras and low-cost radar, relying on the strong algorithmic computing power of the decision-making layer to accurately judge the driving situation.
This approach relies on large amounts of data, as well as top-of-the-line image recognition algorithms, to create barriers to advantage at a fraction of the cost. However, the disadvantages are also obvious, the current image recognition algorithm is prone to misjudgment, which in turn leads to traffic accidents.
The second type is a perception scheme with lidar as the core, which is a common choice for automakers except Tesla. Lidar can measure distances at longer distances, supplemented by other sensors, which can basically achieve full-range perception of the whole vehicle. With longer distances of road awareness, cars have longer time to complete further detection, algorithm analysis, and decision-making, which can compensate for algorithmic deficiencies.
From the current point of view, the second type of route is the choice of most automakers, and lidar naturally becomes an indispensable perception device for L3 and above automatic driving systems.
/ 02 /
Lidar route battle
The performance of lidar is not the only factor considered by car companies, compared with performance, cost may be a parameter that car companies are more concerned about at this stage.
At present, lidar technology has actually been able to cope with environmental monitoring, but the reason why it is difficult to achieve vehicle-level mass production is that the cost is too high. As an accessory product to electric vehicles, it is only possible to commercialize if the price is reduced.
It is based on cost considerations that lidar technology now divides three technical routes: mechanical lidar with overall rotation, semi-solid-state lidar with partial viewing angles, and hybrid solid-state lidar with stationary transceiver modules.
The overall rotating mechanical laser mine has better performance, but the cost is higher, and it is not a suitable choice for electric vehicle companies at present. Because mechanical lidar can achieve scanning of the 360° field of view at the spatial level, it is most commonly used in Robotaxi projects such as Google Waymo and Baidu Apollo to achieve L4/L5 level autonomous driving.
The disadvantages of this form of lidar are also obvious. First of all, the high-frequency rotation and complex mechanical structure cause its average failure time to be only 1000-3000 hours, which is difficult to meet the minimum requirements of 13000 hours for vehicle-grade equipment. Secondly, mechanical lidar needs to be arranged at the highest point of the body to avoid obstruction, which has a great impact on the shape of the vehicle, and the raised radar is also more vulnerable to damage.
Considering commercial mass production, mechanical lidar is not a suitable path.
In contrast to mechanical lidar, solid-state lidar does not contain any mechanical moving parts and only faces a single point of view. Although the detection range is limited, this architecture is the most compact and extremely low cost, which is the best form for mass production.
Solid-state laser mines are mainly divided into two core routes, phased array scheme and Flash scheme, but both routes have problems in mass production.
The phased array scheme uses the interference effect of the wave to obtain a scanning effect of beams at different angles, and the current technological maturity is low, and there is no production of products;
The Flash scheme is not through scanning, but in a short period of time directly forward to the front of a large area of laser coverage, through a highly sensitive receiver to achieve the mapping of the surrounding image of the environment, the final data including depth and other 3D data.
This scheme is currently the most mainstream solid-state lidar scheme, but the detection accuracy and detection distance are insufficient, mainly used for lower-speed unmanned vehicles, such as unmanned takeaway vehicles, unmanned logistics vehicles and other automatic driving solutions with low requirements for detection distance, and temporarily do not have the ability to apply vehicle regulations.
Of course, there are several other solid-state laser mine routes, but they are all in the early stages of research and development, and there is a lack of short-term commercialization possibilities.
Before the technological breakthrough of solid-state lidar, semi-solid-state lidar was a common choice for smart cars. At present, the mass production of vehicle-grade products is mainly from the semi-solid-state lidar route.
Semi-solid-state lidar also only scans a certain range, but has some smaller moving parts, mainly MEMS galvanometers, rotors, prisms and other solutions.
Lanwo Technology customized for Xiaopeng car grade lidar HAP
Specifically, the representative brands of MEMS micro-galvanometer lidar include Innoviz, Sagitar Juchuang, Pioneer, etc., but no such products have passed the vehicle regulation level certification; the representative brands of the rotor scheme include Huawei, Valeo, Ibeo, Luminar, Innovusion (Tudtong), etc., and NIO ET7 uses Tutonic products; and the prism scheme is currently mainly used by DJI's Livox ( Lanwo) and the lidar used by Xiaopeng P5 is a Livox product.
Although there are many lidar routes at present, with the development of the industry and cost considerations, solid-state lidar with extremely high cost performance will become the final solution, but how long semi-solid-state lidar as a substitute will exist and what kind of market size will be achieved will still depend on performance and cost.
/ 03 /
When the 100-billion-level market opens, who is the biggest beneficiary?
At this stage, smart cars are still in the early stage of development, the future is bound to need more equipment, has been more accurate detection, the lidar market is facing an unprecedented opportunity.
There is no doubt that "smarter" will be the direction that most car companies have been pursuing now and in the future. Driven by this direction, automatic driving, as a landmark function of smart cars, will always become the focus of car companies.
In the face of the goal of higher-level automatic driving, lidar is more likely to be favored by car companies because of its powerful performance, as exemplified by the new generation of smart cars released by brands such as Weilai, Xiaopeng and WM.
Up to now, the mass production models of smart cars, including Tesla, are all L2 level autonomous driving on automatic driving. Lidar is considered to be an essential accessory to achieve L3 and above automatic driving, and it will be the first to be popularized in high-end autonomous vehicles with the maturity of autonomous driving technology.
What's more, the high cost factors that limit the popularity of lidar will also achieve a breakthrough.
First, the lidar product itself has a scale effect, that is, the higher the output, the lower the cost. When more and more car companies choose to be equipped with lidar, the output of individual lidar manufacturers continues to increase, and its price will also achieve a rapid decline, and the price decline will further promote the popularity of lidar.
According to the forecast of Lanwo Technology, as the lidar technology architecture changes from mechanical to hybrid solid to pure solid state, the production of lidar will rise from 100,000 units per year to 10 million units per year, and its unit price will also be reduced to less than 100 US dollars.
At present, Xiaopeng P5 is equipped with two LiDAR, and the sales price has dropped to 200,000. It is foreseeable that with the gradual decline in the cost of lidar, in terms of safety considerations, various automakers are expected to list lidar as a standard accessory to improve driving safety, and even popularize it in L2 and below cars, and this 100 billion-level blue ocean market is opening.
According to the forecast of Open Source Securities, by 2025, the global lidar market size will exceed tens of billions of DOLLARs. China has become the world's largest electric vehicle market, in the huge market of vehicle lidar, China is also expected to give birth to tens of billions or even 100 billion industry giants.
At present, the technical route of vehicle-mounted lidar products has not been fixed, and no lidar has been officially listed in China. Domestic participants, such as Huawei, have chosen the mirror solution; DJI's Lanwo Technology has chosen a unique prism solution; and Hesai Technology, which is preparing to go public, focuses on mechanical lidar.
Although the industry has reached a consensus that solid-state lidar after technological breakthroughs will be the best solution for commercial production, semi-solid state is undoubtedly the best alternative before technological breakthroughs.
Lidar will surely develop rapidly with the outbreak of smart cars, but who will become the ultimate winner is still uncertain, after all, lidar is not a simple technical performance, but also a test of corporate strategy and efficiency.
Before solid-state lidar products are officially rolled out on the market, there is a chance for every company to win. For lidar companies that already have production models, they are only closer to success.