The rise of domestic automobile R&D management tools
Latest update time：2024-08-30
Reads：
In the first half of 2024, the domestic market share of local automobile brands further climbed. Among the top 10 sales list, 6 are Chinese local brands,
among which BYD, Chery, Geely and Changan are all in the top five.
Image source: China Passenger Car Association
Although the market share of local brands continues to increase, the profitability gap between local car companies and their foreign counterparts has not narrowed proportionally.
In the first quarter of 2024, BYD, China's most profitable automobile brand, had a net profit of 4.569 billion yuan, ranking just among the top 10 global automobile companies.
The combined net profit of the top 11 most profitable Chinese automakers was 15.462 billion yuan, ranking seventh among global automakers and only one-third of Toyota's net profit.
What does this mean? The combined earnings of all profitable Chinese automakers (excluding those that lose money) in the first quarter of 2024 were about one-third of the earnings of Toyota, a Japanese automaker, during the same period.
No wonder Teacher Jia Ke asked, are we busy for nothing?
Source: Automotive Business Review
Source: Automotive Business Review
The net profit per vehicle of local car companies is at a relatively low level globally. Except for Great Wall Motors, whose net profit per vehicle can enter the top 10 and exceed 10,000 yuan (I suddenly understand the previous high-pressure fuel tank incident a little better, respect for Boss Wei), other local car brands maintain a net profit of several thousand yuan. Among the new car-making forces, only Ideal and Seres are not in a state of "losing money on every car sold". No wonder the bosses of car factories have come out to reverse the price recently, because the price really can't be reduced.
If you want to increase the profit per vehicle,
exporting the entire vehicle
and
reducing costs and
increasing efficiency
are the ways to solve the problem.
China does not have an advantage in traditional fuel vehicles, but China's new energy vehicles have great advantages in technology and price, which can increase the profit per vehicle.
Research by Spanish bank BBVA shows Chinese cars are often sold in Europe for double their sticker price at home, with BYD's entry-level Dolphin costing just over £30,000, almost three times its £10,900 price in China.
However, this advantage is currently in a period of dynamic change. As the United States and the European Union increase tariffs on cars produced in China, this price cost advantage will be compressed.
One solution is to build factories locally, similar to Volkswagen's In China, for China strategy. In the future, Chinese cars will also need to build more factories overseas. Toyota Group, the world's highest net profit automaker, has produced about 10 million vehicles per year in recent years, 70% of which are overseas.
So far, Chinese automakers have built complete vehicle, parts or components assembly plants in nearly 30 countries around the world. Recently, BYD may build a plant in Turkey or Hungary, while Dongfeng is discussing the establishment of a plant with Italy, a long-established auto powerhouse.
Although the car factory has been localized, if it adopts completely localized employment, parts supply, etc. like local car factories, then the cost advantage of the car factory is not obvious.
At this time, having experience in making cars in China has become a major advantage. It can be foreseen that overseas Chinese factories will still use a large number of parts produced in China. At the same time, it will also drive some important industrial chains to go overseas simultaneously and provide parts nearby.
For car companies and parts manufacturers,
the higher the degree of automation in machining, the higher the efficiency of management, R&D, and operation
, which means that the company needs fewer employees and has lower requirements for the quality of personnel, making
it more likely to build factories locally and have greater cost advantages compared to its peers
.
The improvement of soft efficiency in management, R&D, and operation often means that enterprises adopt more advanced productivity tools. The automotive R&D software tool chain plays an important role in this. Compared with the localization rate of automotive parts, which is about 50%, the localization rate of software may be lower. This article also briefly sorts out the foreign and local companies in the automotive software tool chain.
Definition of the Automotive R&D Tool Chain
In the process of vehicle and parts development, software tools to improve R&D and management efficiency include but are not limited to demand management, project management, architecture design and MBD, code repository and CICD, static code scanning, testing, etc. However, due to the lack of alternative tools for architecture design and MBD in China, this article does not cover this part (
if domestic manufacturers are catching up with MBD, and the author missed it, readers are welcome to add it
).
Domestic Alternatives for the Automotive R&D Tool Chain
Demand Management
Requirements management includes requirements acquisition, requirements decomposition, baseline management, change management, traceability management, coverage display, requirements compliance, etc.
In the field of demand management, before 2021, the domestic market was almost entirely occupied by overseas brands, such as Germany's Siemens Polarion and the United States' IBM Doors.
After 2021, some domestic tools have already possessed a certain degree of competitiveness in this field.
Take Yunti Technology's MappingSpace as an example. Unlike other tools in this field in the past, which all start from a management perspective, MappingSpace has made bold innovations to the tool based on thinking about the human brain's way of thinking.
The most intuitive feeling is that this tool uses a lot of
mind maps
. In the past, requirements were either written as a document or a list, but MappingSpace combines the two and has made a lot of changes on this basis. For example, the advantage of list-style requirements is that it conforms to the habits of users who are accustomed to Excel or Jira, but the disadvantage is that the connection between the requirements is not clear enough. For example, 5 requirements related to car windows were written a month ago, and 3 more were written today. The positions of these requirements in the table are very different, and it is difficult to connect them. Even the contradictions between the requirements written before and after cannot be seen. But in MappingSpace, these 8 requirements are all in a mind map, and the last 3 requirements and the first 5 requirements are organized according to the logical order of thinking, which easily solves the problem of contradictions or inconsistencies between the previous and next requirements.
In addition to the mind map of requirements, MappingSpace has another great advantage: high efficiency. Since you can always see multiple requirements on a mind map page at the same time, operations on requirements are all achieved through very simple box selection + batch operation, which is very different from the previous experience of modifying requirements one by one. The time used to modify one requirement can now be used to modify N requirements, and the efficiency increases exponentially.
However, my favorite is the traceability display in MappingSpace, which is based on the display required by the V model or W model. It is very intuitive and can easily view the coverage of requirements, output coverage reports, etc.
project management
Since the entry of new forces into car manufacturing in 2014, there has been a wave of Jira entering the automotive industry. The agile development concept represented by Jira has gradually penetrated into the automotive industry, leading to extensive discussions on the integration of agile development and ASPICE. Before the release of Ones in 2017 and PingCode in 2020, there were almost no domestic project management tools based on agile development that were truly widely recognized from a product level. The emergence of Ones and PingCode has brought opportunities for the replacement of domestic agile project management software.
Take Ones as an example. Since its release in 2017, Ones has completed multiple rounds of financing worth hundreds of millions of yuan. It has strong enough strength to compete with Jira in the Chinese market. In fact, they are indeed widely used to replace Jira in the fields of finance, Internet, and automobiles. It natively provides a variety of R&D management models such as agile and waterfall to meet the management needs of enterprises of different sizes. Their development path is also quite similar to Jira. Ones itself focuses on the management of R&D teams of more than 100 people. By acquiring Tower, it has made up for the management needs of R&D teams of dozens of people. Through the acquisition of Weizhi Notes, Ones has quickly obtained technical supplements and a large number of users in the fields of personal notes and corporate knowledge bases. The experience of these users in Weizhi Notes has been quickly transformed into real capabilities on Ones wiki. As a user, I can clearly feel the rapid improvement in the capabilities of Ones wiki modules. The acquisition of Sifou Community provides Chinese developers with a technical community platform for more extensive discussions.
When I first used Ones, the most obvious feeling was that compared to the complex and layered project configuration of Jira, the project configuration of Ones was much simpler, and it was easier for the team to use it out of the box. The native built-in test module was smoother to use than the third-party test plug-in in Jira. For example, unlike in Jira's zephyr plug-in, there was no need to understand the difference between complex test cycles and test plans. The test part of Ones was more in line with the usage characteristics of domestic users.
When our team initially switched to Jira, another important reason was price. At that time, Jira was facing the discontinuation of the server version, which forced us to switch to the DC version with an annual subscription, and the cost of use increased a lot.
Code repository and CICD
Code repositories are obviously the most basic tools in software development. Before the amount of code was not so large, frequent changes were not required, and multi-person collaborative development was not an option, many companies used tools such as SVN. However, with the exponential growth of the amount of code, the requirements for multi-person collaborative development are getting higher and higher, code submission needs to be linked with CICD, and more and more Internet practitioners are entering the automotive industry, which has promoted the increasing use of Git in the automotive industry. Although both are version management systems, there are the following differences between Git and SVN:
-
1 Git is distributed, SVN is not : This is the core difference between Git and other non-distributed version control systems, such as SVN, CVS, etc.
-
2 Git stores content as metadata, while SVN stores it as files: All resource control systems hide the file's meta-information in a folder like .svn, .cvs, etc.
-
3 Git branches are different from SVN branches: branches are not special in SVN at all, in fact, they are just another directory in the version library.
-
4 Git does not have a global version number, while SVN does: This is by far the biggest feature that Git lacks compared to SVN.
-
5 Git's content integrity is better than SVN: Git's content storage uses the SHA-1 hash algorithm, which ensures the integrity of the code content and reduces damage to the version library in the event of disk failure and network problems.
Domestic code repositories can be divided into two categories. One is the "pure code repository" tool represented by Gitee and Jihu Gitlab. The reason for this is that the most core and well-known part is the code repository (of course, the current functional coverage is also getting wider and wider). Gitee is a code hosting platform under the Open Source China Community Team, and its functions are closest to Microsoft's Github. With Github banning Russian and Iranian developer accounts, many domestic open source projects are currently uploaded to Gitee simultaneously. Jihu Gitlab can be considered a domestically controllable company, with all Gitlab's irrevocable code authorizations, as well as various secondary developments based on the domestic ecosystem, such as adaptations for WeChat, DingTalk, and Feishu, etc.
The other type is a one-stop DevOps platform represented by Jiandanyun ezOne.work. The birth of these teams is inseparable from China's Internet process. In the era of the rise of BAT, China still lacks sufficiently excellent commercial software that not only carries modern software engineering methods, supports the business needs of rapid iteration of Internet business, but also supports the large-scale collaboration of tens of thousands of engineers in giant companies. Therefore, each of BAT has established its own R&D efficiency team and built its own R&D platform. Moreover, at that time, only these giant companies had the technical strength and resources to build these basic software engineering platforms with high technical barriers. Up to now, domestic software engineering tools have a certain origin with these technology giants. Some of them are the internal products of these giant companies that are commercialized externally (such as BAT's own DevOps commercial products), some are the original team's entrepreneurship to establish professional companies (such as Jiandanyun ezOne.work), and some are products acquired by BAT and digested and transformed (such as Coding.net), all of which have a strong Internet gene.
Static code scanning
As the amount of code in cars increases significantly, static code scanning functions are becoming more and more important. In the automotive field, Coverity, QAC, and Polyspace, as traditional static code scanning tools, are now increasingly facing challenges from domestic tools, such as Corax from Feiyu Technology or Qi'anxin Code Guard.
In actual testing, we mainly focused on the two languages C and Java. The two indicators of scanning accuracy and false alarm rate are particularly important for static scanning tools. After all, no one wants to check for a long time and find that there is nothing wrong with the code, but the tool reports an error. Corax performed very well in scanning accuracy and false alarm rates for these two languages, comparable to Coverity. In particular, its vulnerability display function is accompanied by corresponding explanatory text and jump interactive operations. It is the easiest-to-use xAST tool we have experienced so far. In addition, Corax is the first domestic open source SAST manufacturer, and its completely self-developed analysis engine allows us to customize rules and implement the application of internal specific rule sets.
It is worth mentioning that Corax has also launched some functional modules in AI-assisted code. Its AI function can even be privately deployed in our intranet environment, which ensures that our core code assets will not be leaked. At the same time, AI can help us determine false positives and generate available defect repair solutions, greatly improving the work efficiency of internal personnel.
Overall, although foreign tools started earlier, their products are mature and internationally recognized, they are expensive, and they are mainly supported by agents, so it is difficult to provide in-depth services, and some products do not support Chinese. In contrast, domestic tools provide more timely feedback and support, and their product designs are more in line with Chinese people's usage habits, and have surpassed foreign products in some aspects.
test
According to different test types, they can be roughly divided into unit testing, integration testing and functional testing. According to the role of software tools in the testing process, they can be divided into test case generation tools, test execution tools, coverage analysis tools, etc. Foreign tools, such as Tessy, can be used to assist in unit testing, integration test case generation, and can also be used to execute test cases and generate test case results.
Domestic testing tool manufacturers such as Nebula Test complete automatic plugging and test case generation by scanning the source code locally on the customer's site. During the test case execution process, the test case and code can be traced one by one, which can greatly reduce the time for development engineers to find problematic code and improve the efficiency of bug fixing. In addition, by statistically analyzing the test case coverage of the code, test engineers are guided to further improve and improve the test case coverage. Nebula Test's products have been used in many major projects, such as Beidou Navigation, Tiangong-2, six major military projects, Huawei Automobile, 5G base stations and other core projects, ensuring the reliability and security of the system, replacing the products of foreign competitors, and showing superior performance.
(The above information is not complete, and the summary of each product may not be accurate. However, one thing is certain: domestic tool manufacturers are summarizing the experience of their predecessors, using newer technical means and more advanced productivity to quickly launch solutions suitable for the Chinese market. We also welcome more domestic manufacturers to contribute and recommend innovative solutions to the industry through this platform.)
/END/
京公网安备 11010802033920号