Perception-Based Aircraft Landing via Runway Pose Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional aircraft landing systems face challenges in automating landings due to weather conditions, variability of airfields, and the need for expensive in-ground infrastructure, lacking robustness and flexibility in runway detection and guidance.
Innovation Solution
A machine learning model is used to identify runway features and determine aircraft pose, combining computer vision with sensors like LIDAR and RADAR, enabling autonomous detection and precise guidance without requiring extensive prior knowledge or infrastructure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional navigation systems (ILS, GPS) are used for automated landing, then landing guidance can be provided in some conditions, but the systems are subject to outages, inaccuracies, and interference, reducing reliability
Solution Approach 1:
The patent introduces an intermediary computer vision system that processes images from onboard cameras to detect runway features and calculate pose. This intermediary layer between the aircraft and runway provides redundant navigation capability that works independently of traditional ILS/GPS systems, thereby improving reliability through diversification of navigation sources
Solution Approach 2:
The system creates a virtual copy of the runway environment by detecting and mapping runway features (edges, markings, corners) from camera images. This virtual runway model serves as a backup navigation reference that can guide landing when primary navigation systems fail, enhancing system reliability
2Measurement precision
If expensive in-ground infrastructure is installed at airfields to support automated landing, then precise runway detection and guidance can be achieved, but the cost and adaptability to different airfields decreases
Solution Approach 1:
The system enables aircraft to perform self-navigation by using onboard cameras and computer vision algorithms to automatically detect and map runway features. The aircraft serves its own navigation needs without requiring external infrastructure, achieving both precise runway detection and universal adaptability across different airfields
Solution Approach 2:
The computer vision-based detection system is designed to work with passive visual features that exist on all standard runways worldwide. By utilizing universally present runway markings and geometric features rather than airfield-specific infrastructure, the system achieves multi-airfield compatibility while maintaining detection precision
3Device complexity
If visual imaging alone is used for runway detection, then the system is simple and cost-effective, but detection accuracy and robustness under various weather and visibility conditions deteriorates
Solution Approach 1:
The patent employs a composite sensing approach that integrates multiple sensor types (visible light cameras, infrared sensors, LIDAR) to detect runway features. Each sensor type captures different aspects of the runway environment, and their fused output provides robust detection across varying weather and visibility conditions, achieving reliability without excessive complexity
4Extent of automation
If traditional automated landing systems are implemented, then landing automation can be achieved in controlled conditions, but the system lacks robustness against weather variability and visibility changes
Solution Approach 1:
The system dynamically adapts to changing environmental conditions by continuously processing sequential camera images and updating the runway pose estimation in real-time. The computer vision algorithm adjusts to varying lighting, weather, and visibility conditions through adaptive feature detection and tracking, maintaining high-level automation across diverse operational environments
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of supporting an aircraft approaching a runway is provided. Example implementations involve receiving a sequence of images, captured by a camera onboard the aircraft. Example implementations involve applying a received image to a machine learning model trained to detect the runway or a runway marking in the image, and to produce a mask that includes a segment of pixels of the image assigned to an object class for the runway or marking. Example implementations may also involve applying the mask to a corner detector to detect interest points on the mask and match the interest points to corresponding points on the runway or the marking that have known runway-framed local coordinates. Example implementations may also involve performing a perspective-n-point estimation to determine a current pose estimate of the aircraft relative to the runway or the marking and outputting the current pose estimate for use during final approach.