China Post Securities pointed out that at this stage, humanoid robots have entered large-scale mass production, and demand for vision solutions focuses more on both capacity supply and cost control. It is speculated that consumer-grade 3D vision products are expected to become mainstream in demand.
The Zhitong Finance App learned that China Post Securities released a research report stating that machine vision products can be roughly divided into two categories: consumer grade and industrial grade according to the performance requirements of downstream application scenarios. Industrial-grade machine vision requires relatively higher technology, accuracy, stability, etc., and the overall cost is higher. Consumer-grade vision solutions require relatively low accuracy and are more in demand for cost control. At this stage, humanoid robots have entered large-scale mass production. Demand for vision solutions is more focused on both capacity supply and cost control. It is speculated that consumer-grade 3D vision products are expected to become mainstream in demand. As future humanoid robot application scenarios become more complex and diversified, the possibility of requiring industrial-grade machine vision products is not ruled out.
Related targets in the industry chain: Obi Zhongguang (688322.SH), Opt (688686.SH), Ling Yunguang (688400.SH).
The main views of China Post Securities are as follows:
Vision sensors help humanoid robots perceive the world. Vision sensors, or machine vision, are designed to use machines to perform visual recognition and judgment tasks. About 80% of human information is obtained through human eye perception, and future robots will be the same as humans, and a large amount of information will be obtained through visual perception. Currently, many humanoid robots use 3D vision solutions to ensure the humanoid robot's ability to sense the environment and interact. In the future, as humanoid robots continue to develop, and their interactive functions and application scenarios become richer, manufacturers of pure camera solutions at this stage may also introduce 3D vision sensors into their new products through iterative updates.
Machine vision is a computer or image processor and related equipment to simulate human visual behavior and complete the information obtained by the human vision system. It is widely used downstream, and can mainly perform functions such as positioning, identification, measurement, and inspection. Currently, it is mainly used in industrial scenarios of intelligent manufacturing, and it is also used in medical and intelligent transportation scenarios. The upstream machine vision industry chain mainly includes suppliers of machine vision component equipment such as light sources, lenses, industrial cameras, industrial controllers, image processing software, etc.; the midstream industry includes inspection, measurement, positioning, identification systems/positioning guidance systems, etc., and various types of vision equipment; the downstream of the industrial chain is mainly the terminal industry, which is a supplier of comprehensive production line solutions for various industries. China's machine vision market is currently in a stage of rapid growth and is expected to surpass 56 billion by 2027.
Machine vision technology can be divided into two major categories: 2D and 3D according to different imaging dimensions. Traditional RGB cameras using 2D imaging technology can only capture texture information on the surface of an object, and there is no distance information from the object to the camera. 3D visual perception technology effectively complements the shortcomings of 2D vision technology, making more complex and intelligent functions possible. 3D visual perception technology mainly includes time-of-flight (ToF) method, binocular stereo vision, structured light method, etc.
Risk warning:
The scale of humanoid robots falls short of expectations; the risk of changes in the visual perception technology path of humanoid robots; competition for consumer-grade vision products increases the risk; and the risk that 3D vision technology does not meet expectations.