By Chen Nan Yang
China’s output of automobiles reached 9.345 million in 2008, surpassing the United States to become the number two automotive market in the world. According to Xu Liuping, president of market analysis company Chana Auto, output will increase approximately 10% in 2009, even during the global financial crisis. The Chinese government has taken several important measures to boost the industry, including a tax cut for small cars, subsidies for new-energy and low-emission cars, a 10-billion-yuan package to encourage technological innovation, and other policies. Forecasts from Chinese associations and experts regarding increased growth ranged from 5–12%.
Until recently, machine vision has played a small role in the Chinese automotive industry. Chinese machine-vision companies began to enter the field a few years ago and more part manufacturers and automobile makers have recently turned to machine vision to improve efficiency.
Realvision Tech, for example, developed a machine-vision system in 2003 for the country’s largest spark plug producer, Torch Spark Plug. Shanghai General Motors turned to machine vision after wheels repeatedly got stuck in the assembly line, causing substantial rework and undermining manufacturing efficiency.
Quality control has been another driving factor. As concerns over the safety and quality of Chinese automobiles increased, the government released its first car recall regulation in October 2004, and some 1.6 million automobiles have since been recalled. The regulation pushed the Chinese machine-vision market forward as automobile makers found it increasingly important and valuable to use machine-vision technologies to control quality, improve safety, and trace problems.
The Chinese government is currently developing a national recall law expected to be put into effect before the end of 2009. The enforcement of the new law will no doubt give more momentum to the Chinese machine-vision market than the current half-voluntary regulation in which the government imposes a very small fine on automobile makers who refuse to recall their products.
Vision hot spots
According to the Chinese Automobile Quality Report released by the China Association for Quality (CAQ) in February 2009, CAQ received 1764 complaints from Chinese automobile buyers in the fourth quarter of 2008, increasing 15% compared to the same period of 2007. Among the problems, gear boxes accounted for 29.9%, engines 27%, and tires 14%.
The gear box and engine problems had topped the complaints in the Chinese automobile market for several years. The gear boxes and engines formerly designed for drivers from developed countries do not work well in China, where driving habits are different and many roads are still rough. In the 48,400 vehicles Shanghai GM recalled between 2004 and September 2008, gear box and engine problems took the biggest share. As a result, Chinese automobile manufacturers such as Shanghai GM and engine producers such as Wuxi Cummins have used machine vision to improve the quality of gear boxes and engines.
Wuxi Cummins uses a Windows XP Professional-based machine-vision system from Realvision Tech to reduce the mistakes during assembly of the middle shells of engine superchargers. To ensure the correct shell has been used during assembly, Realvision uses 5 cameras to read 12 features of a shell. These features enable the vision system to differentiate between more than 30 types of middle shells and warn when the wrong one is being used. The best measurement precision of the system is 0.01 mm, and the check time for each middle cell is 1–2 s (see Fig. 1).
Once a quality problem is found, the manufacturer must trace the source and history of the part and identify the supplier.
To trace automotive parts through the entire manufacturing process, suppliers must mark each part with machine-readable information such as part number, supplier number, and production number. Two-dimensional coding, especially direct parts marking (DPM), has become important for manufacturers because the code has large data capacity and high fault tolerance.
For example, Shanghai GM’s powertrain parts manufacturing department uses the DPM system from machine-vision system integrator Tongyu Auto Tech to control quality and trace problems. Tongyu’s system, which has been used by other Chinese automotive parts producers, performs three functions: marking, reading, and managing.
To improve marking quality, Tongyu chooses different marking machines for various automotive parts. For instance, it uses Telesis Technologies’ laser or needle marking to mark the valve body of a gear box.
Tongyu uses code readers such as DataMan 700 and DataMan 7500 from Cognex in key points of manufacture to control quality and record information for later tracing. The DPM/code reader system is connected to a manufacturing execution system (MES) so users can trace defects and other problems.
To ensure that complex automobile assembly lines are kept running, Shanghai GM has instituted numerous machine-vision inspection stations for its Cadillac and Buick models. One problem was that wheels would frequently stick inside a sealed track, which could stop the entire production line. Traditionally, one would open a window in the sealed track and send in a technician to find the wheel causing the delay.
Instead, Realvision Tech developed a Windows XP-based system that uses a high-frequency fluorescent lamp as a light source and two 768 × 576-pixel CCD cameras. Users can set the software parameters for each wheel type. The system identifies any variation from the template, captures an image, and sends the information to the data center (see Fig. 2).
Flexibility of system design
Chinese machine-vision system designers are showing adaptability in dealing with different situations in the automotive industry. For example, many Chinese automobile parts suppliers do not have an MES—or do not have a good one—and it is difficult to incorporate them into a problem-tracing system. To address this issue, Zheng Wenjiang, a consulting engineer, designed a web-based platform allowing different parts suppliers to share the same tracing system via the Internet.
Tongyu Auto Tech also customizes its system. For example, the company can use Windows CE to customize different programs or user interfaces to combine its machine-vision system with the MES from various users.
Micro Vision Imaging Technology uses VB and C++ to develop its vision system. Other platforms are also available. It used LabVIEW from National Instruments to develop a system that checks the temperature-control instrument panels of Passat and Sagitar models for Shanghai Volkswagen and FAW Volkswagen.
The system integrator chose an NI frame grabber and Sony FireWire 1024 × 768-pixel color camera with an 118 × 88.5-mm field of view. The vision system included a PLC system to control the movement of instrument panels and a computer for the machine-vision and user interface.
Chen Nan Yang is a contributing editor based in Shanghai, China.
Chana Auto, Chongqing, China
Cognex, Natick, MA, USA
Micro Vision Imaging
Technology, Wuxi, China
Austin, TX, USA
Realvision Tech, Shanghai, China
Robert Bosch, Stuttgart, Germany
Shanghai GM, Shanghai, China
Sony Electronics, Tokyo, Japan
Circleville, OH, USA
Tongyu Auto Tech, Shanghai, China
Torch Spark Plug, Zhuzhou, China
Wuxi Cummins, Wuxi, China