As DevOps technology is more widely used in all walks of life, from the perspective of the company, it is not only the technology of the DevOps platform that is investigated, but also whether the compatibility of the DevOps platform and enterprise products and whether the corporate culture can provide DevOps The soil of landing, behind this is a series of changes in business models, ways of thinking, and management concepts.
In the latest high-end dialogue podcast column "Hackers and Painters" launched by Authing Identity Cloud, Wu Xinghui, CEO of Chasing Geometry, Huang Anjie, CEO of Zengguang Intelligent, Erin Wu, partner of Momo Technology, Ye Yangsheng, co-founder of Xiangong Intelligent, and Dorabot DevOps are in charge. Yuan Jingyang conducted an in-depth discussion on "DevOps Hardware Combat", which was hosted by Li Yuhang, the co-founder of Authing . Comprehensively interpret the application scenarios of DevOps in the hardware field, including robotics, aerospace and other fields, and provide practical guidance for the implementation of DevOps projects.
This article is based on the content of the second issue of "Hackers and Painters" "DevOps Hardware Actual Combat", which only accounts for 1/10 of the dialogue content. For all the content, please click on the audio to listen to the full story, or search for the podcast "Hackers and Painters" in Xiaoyu and Himalayas to listen.
01 How to define DevOps in the hardware field?
Ye Yangsheng, co-founder of Xiangong Intelligence, believes:
"Compared with software, hardware has a longer development cycle. After the software is written and submitted, edited, it can be deployed to the enterprise. The hardware still needs to be boarded, placed, and tested. The whole process has been running for more than a month. .
A good manufacturing enterprise cannot do without quality management. The essence of quality management is to prevent problems that have occurred from happening again , and all problems must be verified retrospectively.
Good hardware products are designed, that is, quality management starts at the research and development stage, instead of paying attention to product quality after product testing. For DevOps in the hardware field , to a certain extent, it also improves the efficiency of quality management and realizes automated operation and maintenance . For example, when we purchase a SaaS product, we will consider whether Authing has integrated this product to improve integration efficiency. "
Huang Anjie, CEO of Augmented Intelligence, believes:
" The process of quality improvement is a process of accumulation and learning . It is very difficult for the industry to do things right at one time. In many cases, if you do not step on some pitfalls, you will never be able to make a good product. For product testing, do verification in advance , It is of course the best to find risk points and test out problems in advance. But sometimes it is necessary to have an improvement mentality and increase the speed of iteration and response to solve product problems.
So for DevOps, the best way is to prevent what happened before from happening again, and solve the problem in the next product iteration.”
Wu Xinghui, CEO of Light Tracking Geometry, believes:
"The development of industrial software has been realizing the pre-judgment of industrial links .
For example, in the earlier era when there was no simulation software, many physical tests had to be done for product verification. However, with the popularization of simulation processing software, the product development cycle and cost have been greatly reduced. In the design stage, the product can be verified in the first time, and problems can be found and solved. After the simulation processing software, there will be some other software, such as some virtual experiment software such as Siemens, which strengthens data analysis, solves the problems that occurred in the production process in the past, and accelerates the process of product digitization.
Nowadays, with the increasing popularity of the Internet of Things, indicators such as equipment operating parameters, status parameters, and environmental parameters will be fed back to monitor the operating status of production equipment, and help engineers make predictions in advance based on these feedbacks to avoid problems.
Software defines the world . As far as the entire industrial digital evolution process is concerned, DevOps is an important path for enterprises to migrate to the cloud, which will make industrial quality and testing more and more digital and software-based.
However, the development of hardware may not have reached this stage. In the future, whether industrial software will directly run through the entire manufacturing industry chain and realize virtualization of the entire life cycle is actually a question worth thinking about. "
Yuan Jingyang, head of Dorabot DevOps, believes:
"A software-defined analogy can be drawn from DevOps.
Comparing the DevOps definition mentioned in the software to the hardware is fast iteration and fast verification . For example, building a drawing management platform similar to Git, as long as someone modifies and submits the mechanical drawings and circuit diagrams, a series of changes can be triggered. The simulation process uses automated simulation testing to verify some basic or complex problems, and then combines some iterations of regression test cases for known problems to gradually improve the hardware system, so that the hardware system can be improved with the lowest cost and the shortest time. Time to efficiently find problems. "
02 What problem does DevOps want to solve?
Erin Wu, partner of Benmo Technology, mentioned:
" DevOps transforms the traditional manufacturing process by using agile development capabilities to make hardware . There are two main problems in hardware development:
On the one hand, the review cycle is long : In the past, it often took a one-month development cycle to find some problems in the production process. After the review was completed and the problems were fixed, another cycle was needed.
With DevOps, there are three aspects:
1. It can improve the efficiency of hardware product development information flow and shorten the online time;
2. The project time node can be predicted more accurately;
3. Quality process management, full process control The entire R&D process includes every small node from hardware to drawings to product BOM. If there is a problem, it can be solved in time to ensure the quality of the final product.
On the other hand, the reproducibility of the problem is low : the physical environment in which the hardware operates is difficult to reproduce, which makes it difficult or expensive to reproduce the hardware R&D problems. If software can help hardware record more physical environment information, it can help hardware reproduce problems, locate problems, and solve problems. "
Li Yuhang, co-founder of Authing, believes:
"Constrained by national conditions and experience, some foreign SaaS software is not suitable for the Chinese manufacturing industry. Some companies have bought foreign SaaS products, but the R&D department believes in their own experience and does not trust these software, resulting in a low usage rate. "
Wu Xinghui, CEO of Light Tracking Geometry, said:
"Constrained by the R&D cycle and established thinking, there is still a long way to go before the implementation of DeveOps hardware. Enterprises must have great perseverance and determination to step on many pitfalls, and reflect on and iterate products from these pitfalls .
For example, when the software is actually implemented on the hardware, many processes cannot be connected. However, due to the development cycle of several months and the cost of tens of millions a year, it is impossible for the software company to stay on site for R&D support. It often happens that the business department can't use it after making great efforts to purchase the product. "
03DevOps platform landing practice plan
Ye Yangsheng, co-founder of Xiangong Intelligence, said:
"Xiangong's self-developed low-code platform build logic is to build a platform first to solve the smallest problems in the system , then solve other problems, and finally build a complex system. After a period of time, some problems in the system will be discovered. Then it is necessary to continuously adjust, modify the process or business documents, etc., and iterate and improve the system to better meet the business needs.
We are a complete machine product. After the release, there will be problems with the automatic compilation and test operation of the machine, which will be fed back to the log analyzer to locate the problem, and then the experience of common problems will be summarized and solidified into the log analyzer. After the test is completed, it is sold to customers. "
Huang Anjie, CEO of Augment Smart, said:
"Augmented smart products are a combination of software and hardware. There will be controllers and mechanical parts inside. Through more sensors and the method of exporting data from the product itself, through some white box tests, and instant feedback on system operation problems, it can be faster . Through automated testing, it is ensured that even if R&D is constantly fine-tuning and modifying the system, it will not cause the problems encountered in the past.
Every time we launch a new product, we will redefine the data operation of this product based on previous test cases, and avoid some problems encountered in the past in time. "
Wu Xinghui, CEO of Light Tracking Geometry, said:
"Light Chasing Geometry itself is a software company, so agile development tools such as teambition and Tapd are often used for project management, and GitLab is used for codes. Recently, DevOps is gradually being practiced internally.
In the project management process, we found that the difference between software and hardware is actually not that big . The project management of the software development link is from PRD, UI design, programmers write code, test to release. Hardware project management, mechanical design, simulation, factory processing, assembly testing and release. To some extent, links such as mechanical design and factory processing can be compared to the link of programmers writing code in the software field, but it is more complicated and cross-cutting. "
Yuan Jingyang, head of Dorabot DevOps, said:
"Dorabot is also a company that combines software and hardware. Since the robot involves many modules, there will be special monitoring tools to collect and organize the logs of each module, and will record the scene of the problem for use after the problem is fixed. regression testing.
Our DevOps platform is also based on GitLab. At the same time, in order to reproduce the problem to the greatest extent, we will run the test program on the same model of robot equipment that has the problem, and perform regression testing by simulating and replaying the previously recorded scene. . "
About Xiangong Intelligent
Founded by the champion team of RoboCup, a small soccer robot competition in the world, Xiangong Intelligent is a product portfolio provider with intelligent controller and digital software as its core. The company is headquartered in Shanghai and its business covers more than 20 countries and regions around the world.
About Tracking Geometry
Light Chasing Geometry is an efficient mechanical design drawing management and collaboration platform. It applies the Git concept to the management of mechanical drawings according to local conditions, solves the most daily and inefficient multi-person collaboration problems around drawings and related technical documents, and explores modern machinery. Efficient way for design teams to collaborate
About Augmented Intelligence
Zengguang Intelligent is a national-level high-tech enterprise integrating R&D, production and sales. With the industry-leading SoftForce® high-precision force control technology, it provides global customers with more precise and reliable precision motion control systems and one-stop intelligent electric motors. actuator products.
About Benmo Technology
Benmo Technology is committed to using direct drive technology to remove the reducer in the robot joints, making the original delicate and noisy traditional robots reborn. Momo owns a complete set of technologies in robot joints from sensors, drivers to motor bodies, and the unique direct drive solution also provides product choices with differentiated performance for many industries.
About Dorabot
Dorabot (blue fat robot) is an intelligent unmanned warehouse solution provider. It uses technologies such as robot perception (vision), motion planning, planning and reasoning, autonomous navigation, multi-machine collaboration, and machine learning to provide logistics, express delivery, and e-commerce. Scenarios such as warehousing, seaports, airports, and advanced manufacturing provide solutions that combine software and hardware including loading and unloading, transportation, single-piece separation, loading, and sorting.
About Authing
Authing is a developer-centric full-scenario identity cloud product that integrates all mainstream identity authentication protocols to provide enterprises and developers with comprehensive and secure user authentication and access management services.