At present, machine vision technology mainly focuses on 2D image analysis and recognition. With the continuous upgrading of industrial automation, as well as the continuous development of hardware devices such as sensors and AI algorithms, machine vision has achieved a leapfrog upgrade from 2D to 3D, and 3D machine vision technology has begun to enter the industrial application arena.
As a "world factory", China must change from primary production and manufacturing to cutting-edge manufacturing. Precision manufacturing is essential to develop cutting-edge manufacturing. The precision requirements of precision manufacturing industry also promote the promotion of 3D machine vision in industrial application scenarios. In addition, the disappearance of the labor force demographic dividend and the recurrence of the current epidemic are also promoting the manufacturing industry to replace labor with machines. Various trends indicate that machine replacement is imperative, and technology is constantly developing in the direction of precision and intelligence.
3D machine vision came into being under the "dual power" of technological development and market demand. The 3D machine vision system uses optical technology to simulate the human vision system. The 3D point cloud with precise coordinates can locate the high-precision 3D digital model of the target object. It breaks through the limitations of 2D vision, such as light source changes, dependence on the contrast on the surface of the target object, and inability to process height information, and provides professional technical support for industrial destacking, automatic sorting and other "pain point" application scenarios.
Its main working principle is: the infrared laser transmitter emits near-infrared light, after the reflection of the object, the pattern is received by the infrared image sensor (IR CIS), and the position of the object (Z-axis) is calculated through the algorithm; Then the visible light image sensor collects the object information in the two-dimensional plane (X and Y axes); The information of the two image sensors is summarized to a special image processing chip to obtain three-dimensional data, and finally realize the expansion from the plane "horizon" to the three-dimensional "horizon".
3D vision technology can obtain the depth information of objects, and then through the accurate positioning of algorithms, the machine can use and control materials more accurately in the production process. It can be used in almost any scene of industrial production, such as destacking, sorting, detection, measurement, obstacle avoidance, etc. Driven by the industrial application demand, 3D machine vision technology has developed from single scene application to full production line production empowerment.
Contrary to the hot development trend of 3D vision technology, its market application scale does not catch up with 2D at this stage. In 2021, the 2D vision market will be about 10.78 billion yuan, and the 3D vision market will be about 1.151 billion yuan, but the annual growth rate of the 3D vision market in 2021 will exceed 100%. It can be said that 2D machine vision technology is the current mainstream, while 3D machine vision technology is the future trend.
The industrial logistics scene has the characteristics of large scene differences, rich categories and frequent additions. The pain points such as the wide variety of SKUs for a long time make it difficult to implement machine vision solutions in batches. With the help of 3D vision technology, Hongbao Automation has developed a "AI+3D+industrial robot" solution targeted at the application pain points of the industrial logistics industry.
First, use the AI algorithm model to quickly complete the application building, then receive the near-infrared light reflected by the turnover box through the infrared image sensor, calculate the position of the turnover box through the algorithm, and finally use the industrial mechanical arm to grab and unpack the turnover box. In the whole process, the industrial robot synchronously completes the visual ranging, positioning, detection and route planning within 180 seconds to achieve accurate destacking. In some complex stacks, industrial robots equipped with 3D vision technology can also accurately identify the placement posture of objects (such as close fitting, random placement, stacking, etc.), and accurately segment turnover box targets.
In the era of Industry 4.0, application scenarios are important resources for machine vision manufacturers to develop 3D machine vision business. In the future, in addition to ensuring the normal operation of the existing 2D vision customer base, Hongbao Automation will continue to adapt to the development of the times, continue to learn and improve the application and development of 3D vision technology, further break through industrial applications, strive for rapid development, and provide high-quality technical services for more manufacturing enterprises.