In-Sight vision system: a good helper for car model recognition on automobile production lines

By Li Wenbing, FAW Car Co., Ltd.
 

    FAW Car Co., Ltd. (hereinafter referred to as FAW Car) is a holding subsidiary of China First Automobile Group and the first joint-stock listed company in China's car manufacturing industry. The company's main business is the development, manufacturing and sales of passenger cars and their accessories. The factory has four major process plants, including stamping, welding, painting and assembly, with an annual output of hundreds of thousands of various passenger cars, including Hongqi, Pentium, Mazda6 and other car series.
 

    FAW Car is committed to making the best cars, making driving more enjoyable, making employees happier, making society more harmonious, and building a first-class brand. Based on this concept, the production line adopts advanced standard automation processes and focuses on the application of new technologies to improve efficiency and product quality, reduce waste, and save labor intensity.
 

    The automobile production line assembles various parts such as body panels, engines, transmissions, electrical appliances, etc. in a certain order (sheet metal stamping, welding, engine assembly, painting process, vehicle assembly, etc.). During the production process, various transmission and clamping devices are required to assemble and transfer to the next process position according to different workpiece structures and installation requirements.
 

    Usually when different models of vehicles are transferred and exchanged in the workshop, there is a read/write head between the sled and the vehicle to transmit the corresponding data to the control room. However, complex environments may cause signal loss, resulting in inaccurate positioning and false landing. The model matching of the sled is wrong and there is no verification step, and the landing status of different models is unclear. This will cause assembly errors and damage in the next process, and it will take a lot of time to correct, affecting the production progress of the entire workshop.
 

    At this time, FAW Car engineers learned that Volkswagen used Cognex products to solve similar problems. FAW Car engineers also visited Volkswagen in person and saw many application examples of Cognex in Volkswagen.
 

    “Cognex smart cameras are used everywhere on the production line to detect whether products are missing or not. QR code recognition and tracking are widely used on engine production lines, and they are also used to guide robot measurement in auto parts factories.” FAW Car engineers pointed out, “These are all reliable and accurate classic applications that we have witnessed, and the results are very good. Some are even strongly recommended by foreign production line manufacturers.”
 

    Therefore, the engineers of FAW Car found Cognex and discussed with them the solution to the problem. "The technicians of Cognex demonstrated to us the powerful visual recognition capability and technical application parameters of the In-Sight vision system," said the engineer of FAW Car. "When I saw that the product can accurately recognize fuzzy features, I felt it was very advanced and it was exactly what we needed."
 

    Next, Cognex technicians and FAW Car engineers worked together to study how to use Cognex smart cameras to identify vehicle features, make judgments and distinguish them. "After on-site inspections, we shaded some glass windows in response to the full-day production situation (the glass corridor has strong direct sunlight in the morning and evening, which affects machine recognition), and added light sources to the characteristic parts to reduce environmental interference." FAW Car engineers said, "The most important thing after that is to add the camera's photo transmission signal to the system and equip it with on-site result display equipment. This process was very smooth and simple, and the adjustment work did not take much time."
 

    "As far as the customer applications we have come into contact with are concerned, in most cases users are concerned about communication issues, and the In-Sight vision system does a very good job in this regard," said a Cognex technician. "Each camera has at least two I/O signal lines, as well as 232 communication and network port communication lines. The corresponding module combination can be added according to needs."
 

    The smart camera of Cognex In-Sight vision system supports almost all communication methods, and can realize information command interaction between PLC and camera through input and output commands, while 232 communication is more suitable for transmitting data such as device coordinates and characters. At present, the more widely used communication method is network transmission, because it supports communication protocols that directly transmit data in network mode (Profinet, TCP/IP, Ethernet/IP, Ethernet native, Ethernet string), and can also adapt to Profibus applications through conversion components, directly interact with PLC through configuration files, and debug programs.
 

    After the In-Sight vision system is installed on the production line of FAW Car, the on-site process is as follows: 1) The vehicle passes through the lifting platform, transmission line, rotating table, and transplanter from the welding workshop to the skid conversion position; 2) The transplanter takes over the car body, the original skid returns, and the skid of the next station is in place. The transplanter goes to the next station, the lifter takes over the car, the transplanter withdraws, and the lifter falls; 3) The car body falls onto the skid, the lifter continues to fall into place, the sensor generates a signal, and the PLC triggers the camera to take a picture; 4) The camera processes the image and transmits the result to the PLC, and the image is transmitted to the touch screen through the switch for display; 5) The PLC gets the result, and if it is qualified, it continues to be transmitted. If it is wrong, it will fall again or change the skid for processing.
 

    Since FAW Car started using the In-Sight vision system, the camera can automatically perform image recognition when the car body falls onto the sled, and alarm for vehicles that have not fallen in. The system will then program the lift to lower the vehicle again until it falls in, easily solving the problem of the previous sled falling on the wrong side.
 

    The In-Sight vision system perfectly solves the problem of vehicle model identification on FAW Car's automobile production line. "In the past, if we saw that the transmission line stopped but there was no alarm, we had to send people to the site to check the situation. The monitoring room was very far away from the workstation, so it was troublesome to send people to run there, and sometimes there was no time to spare." Mr. Zhou, a staff member in the monitoring room, said, "Now it's much easier. I can see the production situation on the production line in the control room."
 

    "After installing the In-Sight7200 smart camera, we have almost never missed any vehicle model identification, achieving a false alarm rate of 0.05% (residual non-error), while the previous error rate was around 1%, which has greatly improved the efficiency and quality of the production line," FAW Car engineers emphasized. "The performance of the In-Sight vision system is amazing, and the results are even more satisfactory. We plan to carry out the same transformation on other production lines in the future, and I think we will continue to choose Cognex products."