LETI: LETI Researcher Came Up with a Way to Avoid Airplane Accidents on the Runway

Today, aircraft are equipped with synthetic vision systems that allow pilots to compensate for poor visibility in various weather conditions. Such systems make the “picture” clearer by increasing the contrast of object contours. However, accidents on the runway still often occur due to foreign objects. These can be various kinds of obstacles – both moving and static: elements of trim, luggage, vehicle, animal. Airlines lose about $13 billion a year due to punctured airplane tires on the runway alone. That is why segmenting objects on the runway to alert the crew should complement the existing visibility enhancement functionality.

Denis Andreev, a Ph.D. student of the Department of Television and Video Engineering at LETI, proposed to expand the functionality of synthetic vision systems. He has developed a cascade of methods that allows assessing visibility conditions, compensating for them, increasing the range of vision and distinguishability of objects, and segmenting objects on the runway. The approach proposed by the scientist can also provide the trajectory of a moving object.

“There are ground systems that are installed on the runway. They are quite effective and find obstacles of a rather small size – about 10×10 cm. The disadvantage of such systems is the need to modernize the entire airport infrastructure and the high cost of such conversions, which only large airports can afford. Now, such systems are installed only in the largest airports of the world – Bangkok, Tel Aviv, Vancouver,” says Denis Andreev.

The new approach does not require major changes to the onboard computers: it is supposed to change the software module responsible for the cameras and add connections for the correct visualization on the multi-function display.

“In my paper, I applied classical methods of digital image processing: methods of enhancing local contrast, segmentation of objects, an upgraded method of background subtraction. Also, I used neural networks to follow the points and build a projection of an object. With their help, we fix reference points on the runway to track changes in the camera angle relative to the runway and thus make object compensation.”

Denis Andreev, a Ph.D. student of the Department of Television and Video Engineering at LETI
The project is being implemented as part of a Ph.D. thesis under the guidance of Nikolay Lysenko, Professor of the Department of Television and Video Engineering at LETI, and is at its final stage: the cascade of methods is already fully implemented in a software package.

Comments are closed.