汽車及零部件行業(yè)：顛覆汽車行業(yè)的競賽

1、10 December 2020Automobiles & Components Sector 汽車及零部件行業(yè)Sector ReportTable of Contents HYPERLINK l _TOC_250009 THE FUTURE OF THE AUTO INDUSTRY 3 HYPERLINK l _TOC_250008 SMART MOBILITY INTELLIGENCE 4 HYPERLINK l _TOC_250007 AUTONOMOUS DRIVING WILL SHAPE THE FUTURE 5 HYPERLINK l _TOC_250006 SENSING AN

2、D DATA INPUT THE HARDWARE 6 HYPERLINK l _TOC_250005 MOST BRANDS ARE MOVING TO LEVEL 3 7 HYPERLINK l _TOC_250004 A SOFTWARE-DEFINED ERA 10 HYPERLINK l _TOC_250003 GLOBAL ACCELERATION ON ELECTRIFICATION 12 HYPERLINK l _TOC_250002 KEY DISCUSSION ON DIFFERENT MARKETS 13 HYPERLINK l _TOC_250001 THE FAIRY

3、 TALE OF HIGH VALUATION 15 HYPERLINK l _TOC_250000 INVESTMENT SUGGESTIONS 18Dongfeng Motor Group (00489 HK) 19Nexteer (01316 HK) 22Great Wall Motor (02333 HK) 25THE FUTURE OF THE AUTO INDUSTRYPolicies to guide the China auto industry into a global powerhouse. Key policies continued to roll out in 2H

4、20 for the auto industry, with the New Energy Vehicle Development Plan 2021-2035 (the Development Plan) approved by the State Council to set the tone at the top for the industry. The overall target in 2025 is to achieve approximately 20% penetration rate for new energy vehicles (NEV), also requiring

5、 the average power consumption to lower to 12 kWh/100 km. While the Development Plan was more of a guidance base, the Energy-saving and NEV Technology Roadmap 2.0 (NEV Technology Roadmap) and Intelligent-Connected Vehicles Roadmap 2.0 (ICV Roadmap) provide more detailed requirements and timeline for

6、 the technological upgrade for the auto industry. The former relates more to the development timeline for NEVs, aiming for a 50% penetration rate by 2035. Also, average fuel consumption is to be lowered to 4.6/100 km by 2025. The ICV Roadmap provides more requirements related to requirements for sma

7、rt vehicles, setting out the basis and the timeline for full autonomous driving (See Table-1 for summary of key policies). We recognize that the future of the auto industry relies on the four primary forces, namely autonomous, connected, electric, and shared vehicles (the four forces). The scope of

8、the four forces is very extensive. This Sector Report attempts to discuss autonomous and electric vehicles at a high level, particularly focusing more in the context of China and OEMs.DatePolicyKey TargetsTable-1: Summary of Key PoliciesOct-20NEV Development Plan (2021-2035)新能源汽車產(chǎn)業(yè)發(fā)展規(guī)劃（20212035 年）A

9、15 year development plan for NEVs, targeting four key areas to help accelerate the industrys development, which includes 1) improving industry technology capacity by building new industry ecosystems; 2) accelerating industrial integration and development; 3) improving infrastructure; and 4) stepping

10、 up international market integration.The specific targets include bringing the average power consumption of new, purely electric passenger cars down to 12 kWh/100 km and raising the proportion of new NEVs in the sales of new vehicles to 20 percent by 2025.Oct-20Nov-20Energy-saving and NEV Technology

11、 Roadmap 2.0節(jié)能與新能源汽車技術(shù)路線圖 2.0Intelligent-Connected Vehicles Roadmap 2.0智能網(wǎng)聯(lián)汽車技術(shù)路線圖 2.0The new technology roadmap proposes overall goals for Chinas automobile industry by 2035, which include that the industry will realize electrification transformation; intelligent connected vehicle technologies will

12、 be mature and widely applied. Key targets are as followsNEVs are estimated to account for 50% of the countrys total sales by 2035;In 2025, 2030 and 2035, average fuel consumption of new vehicles, including NEVs, will reach 4.6L/100 km, 3.2L/100 km and 2.0L/100 km, respectively;High-level autonomous

13、 vehicles will enter the market by 2025, and will be widely used on highways by 2030. And by 2035, autonomous vehicles are expected to be able to run with other vehicles on the same road.The roadmap puts forward four development goals,The penetration rate of partial automation (PA) and conditional a

14、utomation (CA) level intelligent networked vehicles will continue to increase, reaching 50% in 2025 and over 70% in 2030;Over 50% of new vehicles will carry the C-V2X (cellular vehicle-to-everything) system. It will be basically popularized in 2030;In 2025, HA ICVs will first realize commercial appl

15、ications in specific scenarios and limited areas, and continue to expand its scope;Technologies like perception, decision making and drive-by-wire execution will meet the design demands of fully automated (FA) systems.10 December 2020Automobiles & Components Sector 汽車及零部件行業(yè)Sector ReportSource: NDRC,

16、 MIIT, Guotai Junan International.SMART MOBILITY INTELLIGENCEChina to push forward full automation. China issued its own set of definitions for autonomous driving. We believe it is important to understand the level of automation and to read this in tandem with the ICV Roadmap in order to get a full

17、picture of Chinas ambitions. MIIT issued levels of driving automation in Mar. 2020 which will be effective in 1st Jan. 2021. There are a number of points worth mentioning regarding the taxonomy of autonomous driving. Firstly, the set of standards issued by MIIT is highly similar to the ones issued b

18、y the Society of Automotive Engineers (SAE), with six levels of automation from level 0 to level 6, which represents a gradual process from no automation to full automation. Both SAE and MIIT defines human driver to be the one that controls the vehicle most of the time where the system only supports

19、 in level 0 to level 2, and the role of human driver and system starts to switch from level 3 onwards. Secondly, there are minor differences in regard to the monitoring of driving environment for level 0 to level 2, where MIIT defines that it is performed by both human and the system. Also, there is

20、 a difference in terms of human and system involvement in level 3 dynamic driving tasks. However, we believe that in substance, both standards recognize that the human driver will not be driving when automated driving features are engaged in level 3.Table-2: Level of Automation from SAE and MIITFall

21、backSystem performance ofcapabilitydynamic(drivingdriving taskmodels)driving environment / (Additional definition from ChinaStandard)Driver support features and Examplessteering andacceleration/decelerationName from SAE / (NameLevel from China Standard)Execution ofMonitoring of0No Automation (NA or

22、應(yīng)急輔助)2Partial Automation(PA or 組合駕駛輔助)These features are limited to providing warnings and momentary assistance.Human driverHuman driver /Some driving and system(System)Human drivermodesExample: Lane centering or adaptive cruisecontrol (“ACC”)acceleration support to the driver.Driver Assistance(DA o

23、r 部分駕駛輔助)1These features provide steering or brake/Example: Automatic emergency braking (AEB), blind spot warning and lane departure warningThese features provide steering and brake/ acceleration support to the driver.Example: Lane centering and ACCHuman driverHuman driver /(System)SystemHuman drive

24、r / (System)Human drivern/aHuman driverSome drivingmodesHuman driverSome driving/ (System)modesSystemlimited conditions and will not operateSystemConditional Automation3(CA or 有條件自動駕駛) unless all required conditions are metExample: Traffic jam chauffeurThese features can drive the vehicle under4High

25、 Automation(HA or 高度自動駕駛)These features can drive the vehicle under limited conditions and will not operate unless all required conditions are metExample: Local driverless taxi, pedals/ steering wheel may or may not be installedSystemSystemSystemSome drivingmodes10 December 2020Automobiles & Compone

26、nts Sector 汽車及零部件行業(yè)Sector ReportAll drivingmodesExample: Same as level 4, but feature candrive everywhere in all conditionsSystemSystemSystemall conditionsFull Automation(FA or 完全自動駕駛)5These features can drive the vehicle underSource: SAE, MIIT, Guotai Junan International.AUTONOMOUS DRIVING WILL SHA

27、PE THE FUTUREThe development of advanced driver assistance systems (ADAS) and autonomous driving (AD) are the largest domains that drive most research and development resources. The future of auto vehicles is changing as a result of the four forces, which will significantly alter the value chain of

28、the industry, with vehicles to be software defined. According to the Automotive Software and Electronics 2030 Report by McKinsey & Company (the McKinsey report), the software and electrical and electronic components (E/E) market is expected to exhibit a 7% CAGR during 2020 to 2030, from US$238 billi

29、on to US$469 billion. For components, that strong growth is mainly driven by software, sensors and power electronics, which should have faster growth, with CAGR ranging from 8% to 15%. In particular, growth of software and sensors has been attributable to the development and adoption of autonomous d

30、riving. As the McKinsey report shows this, the development of ADAS/ AD is the largest component for each sub-market. For example, ADAS/AD domain will represent 51.2%, 68.3% and 38.5% of software, sensors and electronic control unit (ECU)/ domain control unit (DCU) markets in 2030, respectively. Geog

31、raphically, the McKinsey report predicts that China will be the largest software and E/E market, being valued to up to US$161 billion in 2030.Figure-1: Automotive Software Market by ComponentFigure-2: Automotive Software and E/E Market byRegion, 2030USD bnPowertrain and chassisBod y and ene rgy84OS

32、and middl ewa re62433532151418854356569810080ADAS and ADInfotain ment, connectivity, security, connected servicesEUChina6040200202 0202 5203 0US, Canad a & Mexico781125068161Kor ea & Japa nRoWSource: McKinsey report, Guotai Junan International.Source: McKinsey report, Guotai Junan International.The

33、process of autonomous driving will go through three classic steps. According to Wevolver (a digital media platform for engineers), the process can be categorized as 1) sensing & data input, 2) computation & decision making, and 3) act & control. Basically, these processes can be likened to 1) the hu

34、man eyes, 2) brain and 3) limbs. The first step of the autonomous process comes from information received from sensors which include cameras, radar, LIDAR, ultrasound sensors and others. This will allow the vehicle to find its relative location on roads relative to other objects around it, a critica

35、l step for lower-level path planning. The second step involves passing on raw data captures in step one for something called simultaneous localization and mapping (SLAM) which is a constant update of the map of the vehicles environment, allowing for the system to start path planning. Further, in ord

36、er to perform SLAM, sensor fusion comes into play, which is the process of combining multiple sensory data inputs to achieve improved information. In this step, machine learning kicks in to execute algorithms that have already learned to perform a task from existing data. The final step is the actin

37、g of the vehicle based on the information processed in step one and two.Figure-3: The Process of Autonomous Vehicles St eeringAc c elerat ingSLA MPlanningBrak ingSignallingVehic le- to- I nf ras t ruc t ure C om m unic at ionVehic le- to- Vehic le C om m unic at ionMap D at aGN SS ( Global N av igat

38、 ion Sat ellit e Sy s t em )I MU ( I nert ial m eas urem ent unit )U l t ras ound Sens orsLI D ARR AD ARC am eras ( inc . Therm al C am eras )SEN SI N G & DATA I N PU TC OMPU TATI ON & D EC I SI ON MAKI N GSource: Wevolver, Guotai Junan International.AC T & C ON TR OL TH E VEH I C LE10 December 2020

39、Automobiles & Components Sector 汽車及零部件行業(yè)Sector ReportSee the last page for disclaimerPage 5 of 29SENSING AND DATA INPUT THE HARDWARESensors to gather information. ADAS/AD sensors are important as to gather accurate data from the surroundings in order to perform SLAM. There is a need to have a mix of

40、 passive and active sensors as each sensor has its own characteristics. The main types of sensors are discussed briefly below:Cameras Cameras are regarded as passive sensors as they detect existing energy, such as lights and radiation reflecting from objects. Passive sensors based on camera technolo

41、gy were one of the first sensors to be used on autonomous vehicles. Though mature, camera performance in low light or poor weather declines and it also generates a large amount of data, more than other active sensors combined, which could be an issue for computing power and cloud communication.Ultra

42、sonic sensors and radars These are active sensors as they will send detection signals to sense the environment. Ultrasonic sensors rely on ultrasound waves while radars use radio waves for ranging. They are commonly found in vehicles that achieve level 1 to 2 ADAS functions. Both ultrasonic sensors

43、and radars are relatively mature and are low cost. However, due to the fact that both sensors have long wavelengths (low frequency), the effectiveness decreases for longer distance detection and does not reveal information about the spatial shape of an object. Long-range radars are installed for hig

44、her automated functions such as ACC. A mix of short-, medium- and long-range radars are normally used together in AD.LIDAR LIDAR uses light in the form of a pulsed laser, which can send out 50,000-200,000 pulses per second to cover an area and compile the returning signals into a 3D point cloud. Whe

45、n moving to level 3 ADAS, vehicles typically have one more front-facing long-range LIDAR. The LIDAR sensor is also characterized by high resolution, wide angle, and high accuracy due to active distance measurement, which will be needed to detect and classify objects or track landmarks for localizati

46、on. Therefore, LIDAR is an important sensor for mapping purposes. The growth of the sensor market will be largely driven by the increasing demand for LIDAR, expecting an 80% CAGR during 2020 to 2030, according to a McKinsey report. The strong growth is due to this technology being not widely used in

47、 todays vehicles, but we should expect that LIDAR will be a key element equipped in vehicles that achieve ADAS level 3. One of the key concerns is the high price point for LIDAR, it is significantly higher than other types of sensors (see Table-3 for comparison of major sensors)Figure-4: Major Types

48、 of SensorsL o ng range RADAR Object detection, through rain, f og, dust. Signal can bounce around/underneathv ehiclesin f ront thatobstruct v iew.Source: Wevolver.CamerasA combination of cameras f or short-long range object detection. Broad spectrum of use cases: f rom distantf eature perception to

49、 cross traf c detection. Road s ign recognition.L I D AR3 D env ironment mapping, object detection.Short / Medium range RADAR Short-mid range object detection.Inc. side and rearcollision av oidance.UltrasoundClose range object detection. For object entering y our lane. For parking.10 December 2020Au

50、tomobiles & Components Sector 汽車及零部件行業(yè)Sector ReportTable-3: Attributes of Major SensorsSensorMeasurement distance (m)Cost (USD)Data rate (Mbps)Camera0-2504-200500-3,500Ultrasound0.02-1030-400 0.01Radar0.2-30030-4000.1-15LIDARUp to 2501,000-75,00020-100Source: Wevolver.See the last page for disclaime

51、rPage 6 of 2910 December 2020Automobiles & Components Sector 汽車及零部件行業(yè)Sector ReportIncreasing the number of sensors to achieve a higher level of ADAS/AD. As we move along to higher levels of ADAS/AD, the number of sensors will also grow. Figure-5 provides a visual presentation on the increase in numb

52、er of sensors on vehicles (an approximation). This is because companies take different approaches to the set of sensors used for autonomous driving and where they are placed around the vehicle. For example, Tesla relies more on cameras to detect road objects, such that the Model S has 3 forward faci

53、ng cameras. Whereas, the Volvo-Uber and Waymo uses 360 degree LIDAR to detect road objects. Moreover, the requirement for the number of sensors might change in the future upon new developments in sensor technology. This not only changes the sensors per car, it will also affect the price per car.Figu

54、re-5: Sensors Requirement and DistributionSource: McKinsey report.MOST BRANDS ARE MOVING TO LEVEL 3Automakers based in the three largest markets (China, the US and EU) have been pouring in investment to achieve higher level ADAS functions. While Tesla has been leading, the gap is closing, especially

55、 from its China peers. We will briefly discuss some examples by leading brands by geographical region in terms of ADAS/AD.China: Emerging brands ahead in terms of smart features. The Nio, Xpeng and Li Auto are all equipped with stronger ADAS functions, as they define their cars through software and

56、services. The Nio updated its NIO OS 2.7.0 in Oct. 2020, releasing Navigate on Pilot (NOP), which can automatically guide a car following the navigation route on ring roads and highways. It is the first commercial application of high precision mapping in China on the ADAS functionality of mass-produ

57、ced vehicles. Meanwhile, Xpeng will upgrade its autonomous driving software and hardware systems for its 2021 production models. The key highlight is that Xpeng will become the first car manufacturer to adopt LIDAR as a new hardware component for its next-generation autonomous driving architecture,

58、significantly improving its vehicles high-precision object recognition performance. We believe that both brands should be level 3 ADAS-ready. Along with Tesla, XPeng and Nio are the only automakers in China to offer an advanced ADAS system that can navigate on its own in certain environments.Figure-

59、6: Nio EC6 SUVFigure-7: XPeng P7 Smart Sports SedanSource: Nio.Source: XPeng.See the last page for disclaimerPage 7 of 29Europe: In Europe, autonomous driving is mainly driven by traditional OEMs, which are the German big three, Audi, BMW and Mercedes-Benz. Audi was the first that planned to launch

60、level 3 ADAS functions, the Traffic Jam Pilot, in the A8 back in 2017. However, due to immature regulation framework, Audi was unable to launch these advanced features in the A8. Moving to Mercedes-Benz, the new S-Class introduced in 3Q20 was equipped with a list of improved ADAS functions. The intr