Home » News » In the Anti-Covid19 Pandemic Process, AI-based Radio Frequency Device Debugging Robots Fully Enter the Mass Production Stage

In the Anti-Covid19 Pandemic Process, AI-based Radio Frequency Device Debugging Robots Fully Enter the Mass Production Stage

After Xinghuo Technology released the world’s first artificial intelligence-based 5G filter intelligent debugging robot in 2020, the company has successively used artificial intelligence technology to solve the automatic debugging problems of antennas, filters, multiplexers, and circulators. After nearly a year of verification, good results have been achieved.

In addition to debugging, the company also provides software and hardware integration solutions for products such as intelligent test, sorting, loading and unloading materials, laser engraving, and tape and reel packaging.

 Technical Challenge

In the production process of radio frequency devices, the most human-dependent link is debugging. For a long time, the debugging of radio frequency devices has basically used manual hand-held grinding heads, screwdrivers, scalpels and other tools, to adjust the physical structure of the devices, and to carry out real-time monitoring of changes in device performance through radio frequency measurement equipment, and to make corrections little by little.

In addition, in the mass production of radio frequency devices, affected by material inconsistencies, processing errors, assembly errors and other factors, each product must be repeatedly debugged by skilled workers to achieve the consistency and stability of the radio frequency device performance, so the debugging work has high requirements for the proficiency and practical experience of workers. However, it takes at least half a year to train an experienced debugging worker. The production cycle, cost and capacity bottlenecks caused by manual debugging have greatly hindered the large-scale production of products.



Xinghuo Technology’s Unique Effective Weapon

Xinghuo Technology proposes the General Couple Matrix technology for the first time in the world, and extends the technology from the filter field to multiple radio frequency devices such as multiplexers, antennas and circulators, extracting the parameters of radio frequency devices, and it uses Deep reinforcement learning technology to train AI to control the movement of the robot from scratch. Xinghuo Technology has established a complete set of virtual simulation environment for robots, using the Xinghuo intelligent cloud platform, using 56 GPUs for collaborative training, and the algorithm has been iterated tens of millions of times to achieve good convergence. The production efficiency of a robot is about 5 times that of labor.


About Us

Nanjing Xinghuo Technology Co., Ltd. was established in April 2019. It is a high-tech enterprise established under the strong support of the Venture Capital Fund of Aviation Industry Corporation of China and the Nanjing Municipal Government. Relying on the strong technical power of China Electric Technology and Chinese Academy of Sciences, the company utilizes deep learning, reinforcement learning, imitation learning and other algorithms, builds virtual engineers to replace manual work based on the Xinghuo intelligent cloud platform, and realize the design and debugging of integrated circuits. The company always insists on independent innovation, owns core intellectual property rights, and has the same development progress as Google, and its research findings are at the world’s leading level.