AIRTIFICIAL INTELLIGENCE-BASED OF AIRCRAFT GROUND MOVEMENT TRACKER SYSTEM

• Main Author: Rangga Syuja Pratama, Teuku
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/67395
• Institution: Institut Teknologi Bandung

Airport digitalization is an effort to reduce airport expenditure, one of which is the use of a virtual ATC tower. Aircraft detection and tracking technology are needed to ensure the safety of virtual ATC tower implementation. This final research project presents a development of visual artificial intelligence capable of detecting and tracking aircraft ground movement in an airport. Additionally, the capability to provide a warning if two aircraft are close to each other should also be one feature that will be developed in this research. The methods are a combination of YOLOv4 object detection algorithm that has been trained using Image Dehazing Filter, Deep SORT object tracking algorithm, and coordinate system conversion from pixels to meters for aircraft separation calculation. The model is then validated by a recorded airport video in a fair condition. The trained YOLOv4 model has a mean average precision score of 95.92%, Deep SORT was able to track all aircraft in the video and the aircraft separation warning system was working as intended with an error of 5.09%. The designed model showed a potential implementation for an aircraft ground movement tracker in an airport.