Hot track is stuck in the quagmire of "low penetration rate"? See how machine vision companies "refute rumors"-EEWORLD

Collect

Machine vision is a key core technology for realizing industrial automation and intelligence. Previously, the "2022 Machine Vision Industry Development Blue Book" jointly compiled by Gaogong Consulting and seven participating units including Entropy Technology, Tuyang Technology, Visionbit Robot, Mekamind, Synad, Lingxi Robot, and Vision Science and Technology Robot was officially released.

According to statistics from Gaogong Robot Industry Research Institute (GGII), in 2021, the investment and financing scale of China's machine vision industry exceeded 4.635 billion yuan, with 43 capital operation cases. The financing rounds were mainly concentrated in round B+ and below, and the number of financing cases below round B+ accounted for about 82%. As one of the key technologies to promote the intelligence of high-end equipment, machine vision has attracted much attention from the capital community. The continuous injection of capital will provide the source of power for the rapid development of China's machine vision in the next 2-3 years.

The penetration rate of industrial scenarios is only 5%

As one of the hottest sectors in the past year, the scale and frequency of financing in machine vision have also attracted a lot of envy. However, in contrast to the high frequency of financing is the low penetration rate of machine vision in industrial scenarios.

Machine vision is the "eye" of intelligent equipment and can provide visualization solutions for various industries. Therefore, machine vision has a very wide range of downstream applications. However, from the perspective of actual application, the penetration rate of my country's machine vision industry in industrial scenarios is still at an extremely low level, with an overall penetration rate of about 5%.

Compared with the huge scale of industrial scenes, the machine vision industry still has a lot of room for development.

Zhou Hongpu, general manager of Visionbit Robotics, believes that there are two main reasons for the low penetration rate. First, 3D vision is a relatively new track, but no matter "new" or "old", it must ultimately be demand-oriented. Only when customers look for this "demand" can direct orders be generated; second, 3D vision is not "easy to use" enough. Now many 3D vision scenes basically belong to the "stack of people mode", but in fact, what companies need to solve is how to quickly replicate and promote in simplified scenes.

"It is actually difficult to define the penetration rate. Take lithium batteries for example. Each factory has a different process level. For example, there are many manufacturers in the front, middle and back ends of the entire production line. So overall, there are unmet demands and opportunities for import substitution." Lei Yu, vice president of Lingxi Robotics, said.

Lei Yu also revealed that both 3D vision manufacturers and 2D vision manufacturers are based on lithium battery production technology, which will inevitably face some new challenges, such as space layout, efficiency, etc. At present, Lingxi is already a direct supplier of a lithium battery manufacturer.

As a machine vision company, what is the secret behind being able to bypass integrators and directly supply terminal companies? Lei Yu explained: "Whether to supply directly or integrate, it mainly depends on the terminal. When customers plan their production lines, they will find the best way to divide the work. Different lithium battery manufacturers will have different ways of seeking cooperation and subcontracting levels. Even for the same manufacturer, different development stages will lead to different changes in demand, and all kinds of forms exist."

Standardization breakthrough

The low penetration rate of machine vision is actually closely related to the non-standard nature of the industry. Due to different application scenarios, customers prefer customized and other non-standard products, which directly leads to the lack of unified standards for industry products. However, major vision companies have completely different views on standardization.

Mecha-Mind CEO Shao Tianlan believes that to make standardized products, every aspect must meet the requirements of most customers without obvious shortcomings, which is much higher than meeting the needs of individual customers. The visual technology chain is long, and there are high requirements for core components, whole machine design and production, algorithms and software. The investment required to form standardized products is extremely huge.

"Our product philosophy is to make a replicable standard product from 0 to 1," said Hu Tian, CTO of Synaide . Synaide, which focuses on the logistics industry, also sees that products in this field can be standardized, from data collection equipment, sorting equipment, monomer separation equipment to robotic arms, all can be standardized.

Hu Tian also said that there are indeed many non-standard customizations in the field of machine vision, but this has a lot to do with the fields that each company has been deeply involved in. Different scenarios and attributes will lead to different product demands. If a good niche can be found, product standardization can be promoted, allowing companies to focus on one breakthrough, and thus achieve a double breakthrough in "product" and "growth".

Wang Zhiqiang, industry director of Enzhi Technology , emphasized that once a product is standardized, from demand understanding to product definition, it will involve many parameters such as speed, accuracy and environmental adaptability. Once these are converted into product definitions, all development of the company must be done around this product definition.

Differences in hardware and software

The application fields of machine vision products are very wide, the more common ones include robotics and factory automation, 3C electronic products, production and processing, and automobile manufacturing.

So what is the proportion of machine vision in industrial and consumer categories? What are the differences in the choice of software and hardware in these two categories? In this regard, Yu Shufan, general manager of Vision Robotics, believes that industrial products are standardized products that meet fragmented needs, which is also the core of industrial machine vision.

Yu Shufan said that the difference in software and hardware is related to the scale, and the difference in scale also represents the different iteration methods and directions in software and hardware. For example, in industry, it will pay more attention to product segmentation, and the same is true for software. In the industrial sector, it may pursue stability, efficiency, flexibility and fragmentation issues; in consumer products, relatively speaking, the attributes of generality and platformization will be more important.

As for the application proportion of software and hardware, no matter in terms of "quantity" or "amount", Yu Shufan believes that the proportion of consumer applications is relatively large, which may be 9:1 compared with industrial applications.

"From another perspective, because industrial product customers care more about reliability and whether it can actually solve customer problems, this means that its project cycle will be longer. However, the sales cycle of consumer products is very short, and the customer's purchase decision-making process is relatively different. This also determines that when companies are engaged in industrial and consumer products, the entire business logic, including R&D logic and product definition, will be different." Nong Keke, BD Director of South China at Tuyang Technology, said.

In fact, industrial and consumer categories involve more than just software and hardware issues. From a larger perspective, they also concern "domestic breakthroughs."

Wang Zhiqiang said that when enterprises go deeper into more segmented fields, they will find that there is no so-called "domestic substitution". In the field of industrial robots, Entropy Technology focuses more on metal cutting scenarios, which requires very high precision. This is also the current Entropy Technology solves some unmet needs in these segmented scenarios from the product dimensions of upstream optics and electronic circuits.

The price must be brought down

At present, price is regarded as a key point for domestic machine vision manufacturers to break through. As a local brand, they have obvious advantages in resource integration, especially in terms of price. However, as the price competition enters a new stage, some peers begin to worry whether domestic manufacturers will fall into the quagmire of price wars.

Hu Tian is optimistic about this. He said that in terms of hardware products, the gap between domestic and foreign products is still quite large, perhaps ten times, so it is necessary to lower the price. This has both advantages and disadvantages. It is good for integrated applications because the cost of the product is lower, the cost of R&D will also be reduced, and the cost to customers will also be reduced, so the market volume can be very wide.

Zhou Hongpu has a different view. He thinks that places with fierce price competition must belong to relatively simple scenarios, such as logistics. This is a typical industry that is very sensitive to prices. If the price is high, this industry cannot accept it. Moreover, in the logistics industry itself, there are more wolves than meat. Everyone is competing, so price competition will naturally be more intense.

For this reason, Zhou Hongpu also called on everyone to explore new areas and not always focus on the same "fish in the pond". In addition, if you want to form your own unique competitive advantage, you need to think more from the perspective of the customer and create more "added value".