Text|Intelligent Relativity
Author|kinki
Just after the Spring Festival, the car circle became active again. On the recent Xiaomi Investor Day, Lei Jun shared the latest progress of Xiaomi's car manufacturing, saying that the first mass production will be achieved next year, and he also set a goal, saying that "(Xiaomi Auto) will strive to enter the world's top five in 15-20 years. .”
Here is Lei Jun's bold words and lofty ambitions, but Xiaomi Car has been "seeing flowers in the fog" there. The real car has never appeared, but the gossip continues. Before and after the Spring Festival, a group of design renderings of Xiaomi cars were circulated on the Internet. Several sets of lidars on the roof and rear of the car can be vaguely seen. Some media said that there is a high probability that Xiaomi cars will be equipped with five sets of lidars. Roof, left and right fenders and both sides of the rear.
So far, the new energy vehicles on the market are equipped with at most 4 lidars. From this point of view, Xiaomi is really relentless in piling up materials. It may be because Hesai Technology, the manufacturer of Xiaomi’s lidar, is the capital of Lei Jun’s family. One of the investment projects. However, as the "eyes" of smart cars, the more lidars, the better. When the industry is moving toward "mass production explosion", who is paying for lidars?
01 LiDAR mass production "in progress"
Starting from 2021, car companies have gathered together to announce the "onboard" of lidar. By 2022, many models equipped with lidar will begin mass production and delivery, such as Xiaopeng P5, Weilai ET5, Weilai ET7, etc. 2022 is the "first year of mass production" of lidar.
From cost reduction to mass production delivery, lidar seems to be only one step away from "explosion". Many people in the industry also regard 2023 as the "explosion year" of lidar. In this context, Hesai Technology, the industry leader mentioned above, also submitted an IPO after the new year, and is expected to become the first domestic LiDAR stock.
"Tianshi" means that the lidar market is indeed growing in volume. Guotai Junan pointed out that the shipment volume of lidar in 2022 is about 200,000-300,000 units, and it is expected that the shipment volume in 2023 will reach 1 million units. The French consulting company Yole predicts that by 2027, the global lidar delivery volume is expected to reach 5.3 million units. Obviously, lidar is moving from the stage of technological breakthrough to the stage of mass production and on-board vehicles.
And behind this is the new energy vehicle market that has continued to explode in recent years. High-end automatic driving has become the "future" of automobile development, and lidar is an indispensable sensor in the automatic driving system.
Therefore, although the L4/L5 level of autonomous driving has not been implemented quickly, under the clear trend of intelligentization, consumers have already expected autonomous driving. In order to meet the market demand, car companies have also proposed new technologies at the level of intelligent driving. Concepts, such as L2+ level, L3.5 level, etc., at the same time as smart cars move from L2 to L3 and L4, the speed of LiDAR boarding is naturally accelerated simultaneously.
The "geographical advantage" benefits from the outbreak of new energy vehicles, and related industrial chains such as lidar have also received more policy support, encouraging the laser industry to achieve technological breakthroughs, industrial development and application expansion, providing enterprises with a good production and operation environment .
But whether all lidar companies can achieve "human harmony", I am afraid there is still a question mark. In addition to policy support and market demand, the main reason for the rush to "mass production" of lidar is the consideration of "cost reduction".
Even the head of Hesai Technology, according to its latest prospectus submitted, from 2019 to the first nine months of 2022, its net losses reached 120 million yuan, 107 million yuan, 245 million yuan and 165 million yuan respectively, still stuck in the The embarrassment of "increasing income but not increasing profits".
On the one hand, enterprises must maintain a large R&D investment to achieve technological innovation. In the past few years, Hesai Technology’s R&D investment was almost equal to the total gross profit. The competition in the industry has intensified, the overall cost of the industry has dropped, the price of lidar has also entered a downward channel, and the gross profit margin of enterprises has declined.
Although the gross profit margin has decreased, it is still a common choice for the lidar industry to "increase volume" of low-end ADAS assisted driving, because mass production can drive revenue scale. Taking Hesai Technology as an example, its revenue in 2021 will be 2021. An increase of 73.3%.
The return of funds can continue to maintain the company's R&D investment. Otherwise, once the company cannot get enough orders in one technology cycle, it will not have enough financial strength to cope with the innovative R&D of the next technology cycle, so that the company will fall into a vicious circle. In fact, This is also the main reason why overseas lidar companies have encountered a cold winter, bankruptcy and mergers.
Therefore, even if domestic lidar companies themselves are still trapped in the dilemma of "low gross profit and large losses", enterprises must first win orders to ensure that their own funds can support hard investment. From this point of view, the lidar industry is also It's not completely "human harmony", but the current income is sacrificed first in exchange for an uncertain future.
But it is precisely because car companies and lidar manufacturers have the same "mass production" goal, car companies need more "technical support" to capture the consumer market, and lidar manufacturers need more "production scale" to support research and development. The current new energy vehicle market It also seems to be caught in a vicious circle - whoever has more lidars has stronger technical strength.
The number of on-board lidars has also changed from 1, to conventional 2-3, and then to the evolution of Xiaomi cars with 5. At the end of 2021, the person in charge of the salon mech brand under Great Wall even shouted, "4 (lidar) or less , please don't talk." However, who should pay for the ever-increasing number of on-board lidars?
02 Are more "eyes" better?
For a long time, lidar has been called the "eye" of smart cars. Its working principle is to irradiate the target by emitting electromagnetic waves and receive its echo, thereby obtaining information such as the distance, azimuth, and height from the target to the electromagnetic wave emission point. , so as to achieve the purpose of obstacle avoidance.
Theoretically, the more lidars a vehicle has, the fewer blind spots it will have and the stronger its ability to assist driving. However, whether more lidars are better is not necessarily the answer.
First of all, cost should be considered, which is the most direct factor. Although the current price of lidar has been lowered a lot, from tens of thousands of yuan a piece to about 1,000 US dollars, but if consumers are to pay for it, it is natural that it is enough.
Secondly, it is cost-effective, that is, how many lidars are most suitable for loading? We all know that autonomous driving is mainly divided into three steps: prophet, decision-making, and execution. Lidar is responsible for perception, while decision-making depends on algorithms. The algorithm of waiting for information is the key.
The degree of "intelligence" of a car cannot be judged only from the number of lidars. Computing power is the core, otherwise there may be system overload, freeze, downtime and other situations.
Judging from the current number of lidars mounted on most vehicles, it is basically about 2-3, and 4 may be infinitely close to the golden balance point between computing power level and hardware configuration. No matter how many lidars are below L3 As far as the driving system is concerned, it does not bring equivalent additional benefits.
Finally, we have to consider another situation, that is, the technology of lidar is constantly iteratively updated, and the product is more likely to have a "late-mover advantage".
比如当下部分只在车顶装载一颗激光雷达的车型，可能会存在一定的“感知盲区”，但这样的视觉盲区其实也可以通过产品升级来解决的。目前一颗前向激光雷达的水平视场角一般在90°-120°之间，但结合比如专为近场感知设计的侧向补盲激光雷达，就可以让车辆前向感知水平视场角扩大到180°以上。
虽然当前侧向补盲激光雷达并不能替代传统前向激光雷达，但正如激光雷达从机械式发展至纯固态式，通过固态激光雷达360°的视场角，有效地减少了雷达的搭载量，谁又能说未来不会出现“一个顶几个”的激光雷达呢？
另外，既然智能车的感知都依赖于算法来决策，随着智能系统算力的升级，未来智能车所需“眼睛”数量是多了，还是少了，也不一定。
结合这几点来看，当前各大新能源车企加快了激光雷达“量产上车”的节奏，是因为市场仍存在“竞赛需求”，在硬核技术暂时难以实现突破的背景下，不同车型之间只能通过“堆料”来比拼。
但随着智能驾驶行业的持续发展，以及消费者意识的不断提升，消费者是否还愿意为智能车的“过度配置”买单，就很难说了。
03 量产后的故事怎么讲？
结合以上几点来看，通过汽车搭载量增多而刺激行业增长可能是个伪命题，至少在L4以上智能驾驶尚未落地的背景下，激光雷达的应用场景仍有一定限制。
但在当前，考虑激光雷达厂商降本、持续研发等需求，“量产”又已势在必行，那么量产以后，汽车市场和消费需求会不会发生什么新的变化？
首先，“量产战”何时结束或由头部车厂说了算。自激光雷达进入“量产元年”后，当前行业的产能已远高于利用率，比如大疆旗下激光雷达公司Livox，其HAP激光雷达单条产线的年产能可达20万颗，但去年1-10月，其前装交付量（不含选装）约为2万台左右。
高工智能汽车研究院监测数据显示，2022年1-10月中国市场乘用车前装标配搭载激光雷达7.42万台，这样的需求仅一家企业，一条产线就能满足起来，总体来看，池塘鱼太多，饲料却不够分。
但纵观“车载雷达”以外的市场，比如智慧物流、智慧交通、机器人等，跟乘用车市场的需求相比，还不是一个量级。在这样的背景下，车载激光雷达的量产规模将如何膨胀，智能汽车企业的需求仍是关键。
因此，激光雷达厂家的发展规模，很大程度取决于乘用车市场销量的递增，以及L2+和L3前装渗透率的提高。一台车不能无限搭载激光雷达，但搭载激光雷达的车型可以越来越多。
其次，回到消费者层面，什么时候为激光雷达买单才不算“过度配置”？就目前智能驾驶的发展程度来看，在大城市的市中心路段中，基本用不上L2.5/L3辅助驾驶功能，而在高速公路上，数次驾驶事故后，消费者对辅助驾驶功能也多少抱着点“怀疑”的态度。
但激光雷达“前装量产”的节奏不仅没有停下来，还在加快，车企称之为“预埋”，即当前还不能实现“点对点”的智能驾驶，但前装激光雷达可以始终确保汽车在最远探测距离下的实时测量精度达标，实现稳定、高质量的数据采集质量。这意味着，这些数据的采集，其实也是为优化智能驾驶背后的算法提供助力。
纵观各大车企的“智能驾驶”时间表，如小鹏、理想等，大多均称会在2025年前后实现全面自动驾驶、无人驾驶，那么也可以理解为，当前购买L3级别智能驾驶车型的消费者，可能是提前三年为智能驾驶系统买单。
那么三年后车主是不是就能马上开启“自动驾驶”？假如技术发展顺利，车主也还需要额外购买、订阅其自动驾驶包，特斯拉国内FSD目前的订阅价格为6.4万，蔚来 NIO Pilot全配包订阅价格为3.9万，订阅价格均超过3万元。
提前三年为激光雷达买单，再加上未来可能支付的自动驾驶包，才有可能达成畅享L3自动驾驶的可能，值不值得就得由各消费者自行判断了。
但不得不提的是，正如上文所言，智能汽车“眼睛”的技术方案有可能会不断迭代，对激光雷达厂商来说，大幅度的研发投入正是为了让产品能始终在市场占有一定位置，比如从前向激光雷达到补盲激光雷达的迭代。
还有4D毫米波雷达，以“纯视觉”解决方案为代表的特斯拉，在去年被曝光或将在新车中搭载一款高分辨率的4D成像雷达。
虽然并非国内车企仍以“组合传感器”为主流，但4D毫米波雷达毕竟有成本优势，车企在未来会否采用价格更低的技术路线，智能汽车又会不会因此而下调价格，也是未知之数，只是对消费者来说，汽车像手机一样越来越便宜，也未必不可能。
总体来看，在没有创新性技术出现之前，以及L4以上自动驾驶未能落地之前，激光雷达赛道仍属于1.0时代，只有当其成本能降低到100美元以下，能让中低端车型也开始搭载，激光雷达的2.0时代才会真正到来，要让消费者大方地掏出钱包，“价格”始终才是关键。
*本文图片均来源于网络
此内容为【智能相对论】原创，
仅代表个人观点，未经授权，任何人不得以任何方式使用，包括转载、摘编、复制或建立镜像。
部分图片来自网络，且未核实版权归属，不作为商业用途，如有侵犯，请作者与我们联系。
•AI产业新媒体;
•澎湃新闻科技榜单月度top5;
•文章长期“霸占”钛媒体热门文章排行榜TOP10;
•著有《人工智能 十万个为什么》
•【重点关注领域】智能家电（含白电、黑电、智能手机、无人机等AIoT设备）、智能驾驶、AI+医疗、机器人、物联网、AI+金融、AI+教育、AR/VR、云计算、开发者以及背后的芯片、算法等。