On the night of February 11, Chinese contestant Yan Wengang bagged the bronze medal in the Man’s Skeleton finals, marking the best record a Chinese team has ever attained in this field. The medal is a testament not only to the athlete’s hard work but also to the dedication of Wuhan University to quality inspection technology for the Beijing 2022 Winter Olympics.
Chinese Contestant Yan Wengang in the skeleton race
This technology, led by Wuhan University and Chongqing University, is an important part of the Winter Olympics security assurance. Its objective is to test the quality of the installation of temporary support frames for the Olympic venues in an intelligent way, of which Professor Zou Jingui’s team from Wuhan University shouldered a lot of responsibility.
Set in the Institute of Geomatics, the School of Geodesy and Geomatics of WHU, the team has long been committed to the application of high technology to solve engineering problems. Its members have accumulated a wealth of experience through numerous large national construction projects. As such, they are confident in doing a good job in the assignment within only a month, despite the challenging requirement of meeting the recognition accuracy rate of 95% for thousands of frame-supporting pins. Prof. Zou’s Ph.D. student He Yifeng and final-year Master degree student Gong Xiaoqiang are the main members of the team.
Pro. Zou Jingui’s team surveying in the stadium
The quality testing of support frames installation was made possible thanks to the method of frame-supporting pins inspection conceived by Zou’s team and its first key issue to solve is how to determine the success of pin installation. However, the pin structure is completely unified, and it is difficult to judge whether the installation is successful just by looking at it with the naked eye.
The conventional detection method is to manually knock one by one to ensure that the pins are in place, which is time-consuming and labor-intensive. Therefore, the team had to find a more advanced and efficient way for inspection automation. Inspired by Prof. Zou’s suggestion to create novel judging features, they decided to paint colors on the pins, resulting in an artificial addition of characteristic information. If the paint begins to peel, the pins were considered not to be successfully installed. By combining innovation, creativity and material, the intractable problem was solved in a smart way.
Creative pin installation inspection method
However, following the settlement of pin installation judgment, new problems emerged. There was a concern that due to the uneven brightness of the venue, the ordinary image processing method was not adequate, which made the automatic identification of paint information quite difficult.
With this in mind, the team ultimately employed FasterRCNN target detection network for the identification of the pins’ locations and categories in the image. This decision was made after careful consideration and repeated testing. With models and samplings collected, the team translated theory into practice and continuously improved detection technology through experiments, leading to a final detection accuracy rate of up to 97.2%, showing its validity and reliability for further adoption.
FasterRCNN target detection network for identification
In addition to the pin identification system, other equipment was furnished as well. As the need for pin identification was shifting from “successful identification” to “real-time efficiency”, He Yifeng proposed a hardware concept of a four-camera combination for inspection. With the four-in-one device in operation, four nodes in the frame of the support frame body were attentively photographed and all pins got inspected simultaneously at once. At the same time, the configuration arranged tripods of two sizes in consideration of inspection of pins in height, which facilitated real-time measurement and thorough evaluation.
As the research deepened, creative and progressive elements were added. One distinctive example lies in the utility of the real-time alert function. When unsuccessful pins are detected for installation, the intelligent voice alert functions as a reminder.
Four-camera combination device for inspection
Thankfully, all the hard work paid off when the inspection system for support frame installation finally passed the acceptance in the rehearsal on December 18, 2021. Sporting an integrated photo and detection function, its recognition rate reached 97.9% in the final inspection. Since then, the system has played the role of “Olympics medical examiner”, contributing to athletes’ security behind the glory.
Looking back to their preparation process, the team members find it with many twists and turns. They concluded, “New challenges and higher demands emerged continuously, but luckily new results were created under pressure. We are proud to make our own contributions to the Winter Olympics and grateful for the guidance from Prof. Zou. Luckily we made it!”
In addition to making a contribution from Wuhan University to this grand event, He Yifeng hopes to bring geomatics closer to the public and break down stereotypes surrounding the subject. In his view, the discipline of geomatics is not the same as a profession that suffers from hardship, fieldwork and lack of security. Instead, its innovative cross-fertilization with emerging disciplines is promising.
National Alpine Skiing Center in Yanqing
In fact, an instance such as WHU’s advanced surveying technology contributing to the Olympics is just a microcosm. In other positions, numerous WHUers are also dedicating themselves to this sports event, which is in line with the Olympic slogan “together for a shared future”.
Rewritten by Zhang Jing
Edited by Zheng Zicheng, Qin Zehao, Liang Tong, Zou Xiaohan, Sylvia and Hu Sijia
Source: https://mp.weixin.qq.com/s/IpKRHfjgwvAn1ergPDhS0w
