Source | New Intelligent Driving (ID:AI-Drive)
Author | Rob
When everyone is talking about intelligence, what are the advantages of Xiaopeng?
The wave of automotive intelligence is surging and "software-defined cars" have gradually become a high-frequency hot word in the automotive industry, and consumers are also looking forward to cars to continue the behavior habits and experience of smart phones.
From OEMs to traditional parts companies, from HMI software companies to AV algorithm software companies, operating system companies, from basic software companies to AI chip companies... "software-defined car" has penetrated into all aspects of the automotive industry, changing or even reconstructing the vehicle's software and hardware architecture design, development organizational framework, talent construction, product definition perspective, as well as the supply relationship and business model of the entire industry.
This trend is particularly obvious in 2021, for example, for OEMs, in terms of improving software capabilities, expanding the internal R & D team has been a routine action, those who have the heart will also establish strategic alliances with software companies, and those who have the ability will directly bypass Tier-1 and choose to establish their own independent research and development capabilities in software, algorithms, chips and other all-technology boardwalks.
In fact, at present, there are not many car companies that adhere to the full-stack self-developed technical route of software and hardware, and Xiaopeng Automobile is a firm fan of them, and with actual achievements, it has initially proved the feasibility of this model.
ADAS and smart cockpits open the way first
From the beginning of its establishment, Xiaopeng Automobile has held high the banner of "intelligence" to attack the market, so compared with traditional OEMs, it has eliminated the time cost and historical burden of strategic transformation.
In the view of He Xiaopeng, CEO of Xiaopeng Automobile, from the perspective of products, the intelligence of the car is divided into "in-car intelligence", "body intelligence" and "outside the car intelligence".
In-vehicle intelligence mainly includes on-board intelligent large screen, instruments and other in-vehicle hardware, the operating system is the soul of in-car intelligence, and voice is the core way to achieve human-computer interaction in the car;
The focus of body intelligence is automatic driving technology and understand the user's hardware, He Xiaopeng believes that the system with automatic driving usage rate as the dimension and the utilization rate is higher than 85% can be called three or more levels of automatic driving;
Out-of-vehicle intelligence refers more to the interconnection between cars and cars, cars and roads, cars and other transportation facilities, from the perspective of product operation, the driver will become the first batch of people to produce data, and the passengers will also become important generators of data.
After several years of continuous precipitation and exploration, the intelligent cockpit and ADAS have now become the outposts of Xiaopeng Motors under the wave of "car intelligence".
In 2021, Xiaopeng Automobile carried out large-scale OTA upgrades in January and June, focusing on the opening of high-speed NGP navigation driver assistance functions and VPA parking lot memory parking functions.
In March last year, Xiaopeng Motors once used the P7 to carry out the NGP expedition challenge activities of 3675 kilometers from Guangzhou to Beijing, and through the 110 effective record data it collected, Xiaopeng NGP achieved an average of 0.71 takeovers per 100 kilometers, a success rate of 94.41% in lane change, a success rate of 92.76% in ramp passing, and a tunnel passing success rate of 94.95%, which was quite eye-catching.
VPA parking lot memory parking is another highlight of Xiaopeng Automobile's intelligence.
According to the introduction, P7 through the owner of the common parking lot and parking space memory and driving route learning, without relying on the parking lot transformation, can achieve the parking lot and parking space of the autonomous parking, has the ability to avoid obstacles, planning and control of the parking lot.
In December 2021, the memory parking function of the VPA parking lot was upgraded again, and the parking roadmap sharing function was added, becoming the industry's first parking assistance function that can share the route.
However, the auxiliary driving system is only part of the intelligent car, in order to reflect the product differentiation and branding, the current competition between car companies is more fierce, but also in the field of cockpit intelligence.
According to a survey by Yiou Think Tank, as of October 2021, the penetration rate of smart cockpits in Passenger Cars in China is 50.6%, of which the penetration rates of Central Control Color Screen, Intelligent Voice System and OTA are 97.9%, 86% and 50.9%, respectively.
It can be seen that the central control color screen, intelligent speech recognition system, and OTA remote online upgrade have become the standard of the current smart cockpit.
Xiaopeng Motors regards voice as a key way to achieve human-computer interaction in the car, and tries to create a perceptual fusion interaction scheme with voice interaction as the core.
In July 2021, Xiaopeng Automobile launched the new timbre of the intelligent voice assistant Xiao P in the Xmart OS on-board intelligent system, which can achieve advanced interactive capabilities such as accurate recognition, rapid response, continuous dialogue, and semantic interruption of the voice assistant, and also makes the voice have a temperature, and can express emotions such as dissatisfaction, anger, fear, and sadness.
According to the international common voice quality evaluation method MOS evaluation, in the case of a full score of 5 points, the new AI sound score of Xiao P is as high as 4.49 points, which is the highest score of moS voice quality evaluation in the car intelligent voice assistant.
Dig deep into the moat
Under the wave of automotive intelligence, Xiaopeng Automobile is currently taking the lead in intelligent cockpit and ADAS because of its early entry time and firm "intelligent" strategy.
The more critical reason behind this to support Xiaopeng Automobile's leadership lies in the self-research path of software and hardware it has chosen, and the continuous investment in funds and talents for this purpose.
In fact, whether smart cars need software and hardware self-research, the industry has different views before.
At the very beginning, many people think that smart cars do not need to be self-developed, relying on integration, coupled with the high technical difficulty of full-stack self-development, huge input costs, not short research and development cycle, more good at doing chassis, driving car companies, but also after a long period of wait-and-see and inspection, they slowly chose to play the slogan of "software and hardware self-research" in the past two years.
"The core of smart cars is operations, OTAs, services, software, iterations, data-driven, I think some car companies will move slowly, but this action needs a suitable gene, needs a big enough team, they need more time to catch up." He Xiaopeng once said so in an interview with the media.
Xiaopeng Automobile has indeed invested a lot in research and development. According to the financial report, from 2018 to 2020, the research and development expenses of Xiaopeng Automobile increased from 1.051 billion yuan to 1.726 billion yuan, and in the first three quarters of 2021, the research and development expenses of Xiaopeng Automobile have increased to 2.763 billion yuan.
He Xiaopeng also revealed, "In the darkest hour, we cut a lot of people, but we didn't start with the intelligent driving team." I want to work hard to raise it, and I believe it will be successful in the future. "At present, the intelligent driving of Xiaopeng Automobile is still growing rapidly, and as of the end of the third quarter of 2021, the number of Xiaopeng R&D team exceeds 4,000 people, an increase of more than 100% over the end of 2020."
Deloitte China believes that on the surface, the "software-defined car" is more numerous and valuable than the mechanical hardware of the software (including electronic hardware) in the car, and behind it is more reflected that the car is gradually transforming from a highly mechatronic mechanical terminal to an intelligent, scalable, and sustainable iterative mobile electronic terminal.
This means that from monitoring and controlling the temperature of the battery pack of electric vehicles, to running the application on the central control screen, to the human-computer interaction experience, to the realization of intelligent driving car perception, path planning and other functions, a large number of hardware devices and corresponding massive software development and data operation processing work will be introduced.
Therefore, for car companies, the choice of software and hardware self-research, can quickly perceive user needs, develop a unique intelligent solution at the same time, through the mastery of data, algorithm closed-loop capabilities, to achieve rapid iteration and push of products and functions, to achieve continuous evolution of performance.
Intelligent cars can be divided into perception layer (sensor), architecture design, decision-making layer, map, key hardware (steering, braking), data and other parts, while Xiaopeng Automobile has now realized self-research in architecture design, operating system, middleware and algorithms.
It is worth mentioning that in December 2021, Xiaopeng Motors' latest electrical and electronic architecture, X-EEA 3.0.
As the "skeleton" of the car, the automotive electrical and electronic architecture (hereinafter referred to as the "EEA") is to integrate various sensors, ECUs, wiring harness topologies and electronic and electrical distribution systems to distribute computing, power and energy.
An EEA suitable for intelligent cars is like opening up the second pulse of Ren Dou when practicing martial arts, so that the various parts of the body have the basic ability to work together, so as to realize the intelligent functions of the whole vehicle.
Bosch has divided the EEA into three major phases: distributed electrical and electronic architecture – [cross] domain centralized E&E architecture – vehicle centralized E&E architecture.
Guosen Securities believes that at present, most of the traditional car companies' electronic and electrical architecture is in the transition stage from distributed to [cross]domain concentration, and the electronic and electrical architecture X-EEA 3.0 adopted by Xiaopeng G9, due to the realization of central supercomputing and regional control, has come to the 3.0 era, which can achieve a full vehicle without OTA, multi-controller synchronous refresh, and the entire upgrade time is also controlled within 30 minutes.
From the distributed ECU and local OTA in EEA1.0 to the functional domain controller and vehicle OTA in EEA 2.0, and then to the sensorless OTA of EEA3.0 and less than 30 minutes of vehicle upgrade time, Xiaopeng Automobile is a step by step, with countless time, patience, money and talents, digging a wide and strong moat.
What is the next breakout point?
In fact, the concept of "software-defined car" can most trigger car companies, investment people are excited about the fact that the previous profit model of the automobile industry is likely to be changed in one fell swoop, that is, from a hammer hardware sales model to a "continuous hardware upgrade, subscription service" approach, the game is subverted at the same time, the industry's overall interest chain will also be re-dispersed and distributed.
This model was first explored by Tesla, but now in China, Xiaopeng has become one of the first explorers.
The Boston Consulting Group, a world-renowned enterprise management consulting organization, has predicted that in the near future, the automotive industry will be divided into three types of car companies:
Service car companies, active transformation, service car companies with user contacts will dominate the terminal market;
OEM car companies, enterprises that have lost user contacts, are facing elimination or becoming foundries;
The new foundry, the third type of player outside the industry, will also actively attack and actively seek new roles in the foundry.
Judging from the results of "automotive intelligence", Xiaopeng Automobile has gained a lot in 2021 and has initially explored the feasibility of its software and hardware self-research roads.
As early as the 2021 Q1 earnings call, Xiaopeng Motors announced its XPILOT software sales revenue, when He Xiaopeng said, "I believe we are the only car company in China that has realized a full-stack self-developed automatic driving assisted driving software for separate charges."
In the earnings call for Q2 2021, Xiaopeng Motors revealed that about 97% of the P7 delivered already supported XPILOT 2.5/XPILOT 3.0.
In the third quarter, Xiaopeng Automobile said that during the reporting period, the software penetration rate of XPILOT3.0 of Xiaopeng P7 was close to 20%, and software revenue increased month-on-month.
In addition, in the third quarter, Xiaopeng Automobile's monthly utilization rate of high-speed NGP and high-speed mileage penetration rate exceeded 60%, assisting users to drive 5.51 million kilometers of highway mileage.
As of December 31, 2021, Xiaopeng Automobile's total delivery of 98,155 smart cars in the whole year has reached 263% year-on-year, ranking first in sales among new car-making forces.
"At present, under the massive Chinese driving scenario, we have accumulated a number of data such as cornell cases and the closed-loop capabilities formed, which will become an important competitive advantage of Xiaopeng in intelligent assisted driving, and open up the gap with other friends." He Xiaopeng thinks.
Next, according to the plan, Xiaopeng Automobile will also launch XPILOT 3.5 in 2022, expanding the scope of use of NGP to urban road scenes with more complex road conditions, and will continue to update its driving assistance system XPILOT to version 4.0.
The above data show that software charging, as a new profit model for smart cars, has officially landed in China, and the domestic market can also provide fertile soil for its development and growth.
According to the overall development goal of China's automotive technology, by 2025, L2/L3 intelligent networked vehicles will account for more than 50% of annual sales of automobiles, and L4 vehicles will begin to enter the market.
Northeast Securities believes that in 2030, China's intelligent and connected car market space will exceed 1 trillion yuan, and the compound annual growth rate is expected to reach 26.22%.
Therefore, He Xiaopeng said on the earnings call in Q3 2021 that in the second half of 2022, Xiaopeng Automobile will begin to explore the Robotaxi business.
In order to achieve this goal, in the short term, the strategic goal of Xiaopeng Motors is to improve the robustness and safety of autonomous driving algorithms by generalizing scene operations.
Ultimately, Xiaopeng Motors will change the user's travel experience and create business value through its front-loading and mass production of Robotaxi's hardware and software capabilities, as well as future cooperation with various travel operators.
However, at this stage, for Xiaopeng Automobile, the next thing to do may be how to reduce the cost of the sensor to further enhance the market competitiveness of the product.
Previously, some media asked He Xiaopeng: When everyone is talking about intelligence, what are the advantages of Xiaopeng?
He Xiaopeng summarized it as Xiaopeng Automobile can achieve the best ability compound at a limited cost, "Our team is very large, has done a long time, experienced almost seven years of automatic driving research and development, three years of mass production practice, and there are countless challenges in practice." Xiaopeng P5 has been able to do automatic driving in a relatively reduced cost, which is a huge difficulty. ”
summary
Before 2020, there are still many people in the automotive industry who are worried about the development potential of new energy vehicles.
However, in 2021, the market penetration rate of new energy vehicles is unexpected by most people, and in just one year, it has achieved a rapid climb from single digits to double digits, prompting the industry chain to be excited, cautious and optimistic, and wait-and-see.
Automotive intelligence is also facing a similar situation at present, and Xiaopeng Automobile, as a firm user and the first card entry, may also become a taster of the sweet fruit of "automotive intelligence".