As a representative product of mechatronics technology development, industrial robot is the core equipment of the entire intelligent manufacturing. In China's "Made in China 2025" manufacturing power strategy, robots are one of the ten key development directions planned. In 2018, China's industrial robot sales reached 15.64 units, a year-on-year increase of 14.97%. Industry experts agree that in the next 10 years, China's industrial robot installed capacity will show an exponential growth, with an annual installed capacity of more than 2 million units.
Research on target recognition and intelligent detection of industrial robots was carried out earlier in the world. Fanuc, Staubli, YASKAWA and other companies launched their own intellectual property rights decades ago. Smart industrial robot. The basic structure of intelligent industrial robots is similar, mainly including industrial robots, vision-based target recognition and intelligent detection systems. The research and application of intelligent robots abroad have developed early and developed very quickly, and they have advantages in terms of adaptability, cost, and scope of application.
In China, research in this area is still just beginning. Although some achievements have been made in some colleges and scientific research institutions, there are still many problems that need to be faced. The core problem is: core components (such as CCD, CMOS and other imaging) Sensors) basically rely on the United States, Japan, South Korea and other countries. Domestic manufacturers are mainly engaged in OEM, which is very unfavorable to the development of smart robots and visual technology in China. Therefore, intensifying efforts to develop China's intelligent robot technology, research and development of industrial robot target recognition and intelligent detection, whether it is for China's development of industrial robot independent innovation technology, or the cultivation of industrial robot talents, has very important significance.
As an advanced robotic sensory system, machine vision can not only perceive the surrounding information, but also make optimized decisions based on it. With the rapid development of industrial robots and the complexity of application tasks, the requirements for machine vision are becoming higher and higher. Developing a vision system with better performance and meeting application tasks will be one of the development directions of machine vision.
The technology of combining vision technology with a single industrial robot is relatively mature, but it cannot meet the needs of smart factories. The multi-robot coordination system has stronger functions, can complete complex tasks, has higher efficiency, and has good development prospects.
Deep learning methods such as CNN are widely used in the field of image processing. The CNN network structure is highly intelligent and does not require humans to extract features. Therefore, the use of deep learning to optimize the vision of industrial robots and provide the recognition ability of the vision system has great room for development.
Industrial robot technology and visual technology have developed rapidly at home and abroad. Industrial robots equipped with visual technology can replace people to complete visual recognition, guidance and positioning, appearance inspection, high-precision inspection, etc., and can greatly improve production efficiency and production automation. , Significantly reduce production costs. Industrial robot vision system has broad application prospects, and it is also an inevitable direction for the development of industrial robot target recognition and intelligent detection technology.
