A group of scientists from the Vietnam Aviation Academy used images from cameras and machine learning models to detect and warn of foreign objects that could cause unsafety at the airport.
The image processing technology application system was developed by the research team over 2 years with the desire to support aviation safety.
To do this, the team sketched a 3D model on the computer to simulate the actual airport, including the entire terminal, aircraft, runway, tunnel, lighting system (simulating day and night)... In reality, the team arranged cameras to detect objects along the runway.
Different scenarios were built for the computer to detect foreign objects on the simulated runway. The data source was built by the team from collecting available images at runway locations, taxiways, and aprons at domestic and international airports, combined with images taken by students and lecturers during their internship.
When data is fed into the computer, it will learn all the objects in the photo set. For example, metal roofs, water tank covers, antenna dishes, pet birds... even passenger items such as ballpoint pens, suitcase handles, document clips... all pose potential safety risks. When foreign objects are introduced into the model runway, the camera will capture images, send them to the server for analysis, processing and issuing warnings.
When testing on a machine learning model with images in well-lit conditions, it can detect foreign objects with over 99% accuracy. As for noisy images, i.e. in low-light, dusty, rainy, windy conditions... the model operates with lower accuracy, averaging about 70 - 80%. As a result, the machine learning model recognizes the shape, size and location of the object.
Currently, the group's product only detects objects on the ground. Dr. Dung said that he will continue to research and develop similar functions for objects in the air.
The machine learning model to detect foreign objects was tested by the team on an airport model. Photo: NVCC
According to Dr. Nguyen Thanh Dung, Deputy Director of the Academy and head of the research, testing the system on an airport model is very different from a real airport. The reason is that the distance from the camera position (meeting safety conditions) to the object (side length over 3 cm) on the runway is very large, sometimes up to hundreds of meters. Therefore, the camera system needs a higher resolution to identify the object and a computer system with faster data processing speed.
Mr. Dung said that the technology to detect foreign objects in airports is applied by many countries, but the price is very expensive. In 2017, the total investment in the foreign object detection and warning system (FOD - Foreign Object Debris - FOD) was quoted at 486.2 billion VND for Noi Bai airport and 509.7 billion VND for Tan Son Nhat airport.
In Vietnam, "automatic systems to detect foreign objects have not been used, mostly manual methods are used. That is, airports mobilize people to control and collect foreign objects in runways, taxiways, and parking areas," said Dr. Dung.
Dr. Nguyen Thanh Dung, head of the research. Photo: Ha An
According to Associate Professor Dr. Bui Van Hong, Director of the Institute of Technical Education (Ho Chi Minh City University of Technical Education), foreign object detection systems in the aviation sector using camera systems have been researched and applied in practice by developed countries around the world. This technology is combined with short-wave radar systems at some airports around the world to detect foreign objects. However, the effectiveness of these systems has not been evaluated beyond the manufacturer's announcement. However, to apply in Vietnam, the cost is high and the technology is not proactive.
He believes that the group's research is the basis for designing, installing, exploiting, maintaining, mastering domestic technology, and minimizing costs if applied in practice. Therefore, he expects the system to be completed by the research group, tested and applied at domestic airports.
Ha An
Source link
Comment (0)