Aircraft Taxi Guidance and Positioning Method Based on Onboard Forward-View Cameras

With the rapid development of the civil aviation industry and the increasing operational pressure on airports, the automation of aircraft taxi guidance has become one of the key technologies to enhance airport safety and efficiency. Traditional taxi guidance methods rely on voice instructions from ground controllers and flight crew. However, in low visibility, complex weather conditions, and high-density airport environments, this manual approach often fails to ensure the accuracy and reliability of the guidance. This paper constructs a video dataset containing complex airport scenes and proposes an autonomous taxi guidance line extraction system based on multi-task learning. Additionally, pose matching optimization is performed using Kullback-Leibler divergence (KLD) to align the forward view with the airport bird's-eye map for precise positioning. Experimental results demonstrate that the proposed method exhibits stronger robustness than existing approaches in complex environments, such as runway wear, curved and intersecting guidance lines, and multi-line interference.