Aircraft Pose Estimation Using Onboard Vision and Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies face challenges in automating aircraft landings due to variability in weather and visibility conditions, limitations of external navigation systems, and the need for specialized equipment at airfields.
Innovation Solution
The implementation of a method using computer vision heuristics or machine learning to determine the pose of an aircraft relative to a runway, utilizing a variety of sensors such as visual imaging devices, LIDAR, and RADAR, without requiring expensive in-ground infrastructure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If external navigation systems (ILS, GPS) are used, then navigation capability is provided, but they are subject to outages, inaccuracies, and interference
Solution Approach 1:
The patent introduces an intermediary computer vision system that processes images from onboard cameras to determine aircraft pose relative to the runway. This intermediary system bridges the gap when external navigation systems fail, providing an alternative method for position determination using image processing and machine learning models trained on runway visual cues.
Solution Approach 2:
The system creates a virtual copy of the runway environment by training machine learning models on labeled images with ground truth poses. The model learns to map visual features of the runway to aircraft position and orientation, effectively copying the navigation function from physical external systems to an onboard computational system that processes visual information.
2Measurement precision
If specialized equipment and in-ground infrastructure are installed at airfields, then pose estimation accuracy is improved, but implementation cost and complexity increase
Solution Approach 1:
The system enables aircraft to determine their own pose using only onboard cameras and computational algorithms. The machine learning model processes images captured by the aircraft's own sensors to estimate position and orientation relative to the runway, eliminating the need for external infrastructure and making the system self-sufficient.
Solution Approach 2:
The patent designs a universal system that can operate at any airfield without requiring specialized infrastructure. The machine learning model is trained to recognize runway features and determine pose using only standard onboard imaging devices, making the solution applicable to diverse aircraft types and airfield environments without customization.
3Measurement precision
If machine learning models are trained on labeled images with ground truth poses, then pose prediction accuracy is improved, but training data requirements and processing time increase
Solution Approach 1:
The system performs preliminary action by pre-training the machine learning model offline using large datasets of labeled images with ground truth poses. This preliminary training phase prepares the model in advance so that during actual aircraft operation, pose estimation can be performed rapidly using the pre-learned mappings from visual features to position and orientation.
Solution Approach 2:
The patent segments the pose estimation task into distinct components handled by specialized machine learning models. Different models are trained for specific aspects of pose determination (e.g., position estimation, orientation estimation), allowing each model to be optimized independently and processed efficiently during operation.
Data Source
AI summary
A method is provided for supporting an aircraft approaching a runway on an airfield. The method includes receiving a sequence of images of the airfield, captured by a camera onboard the aircraft approaching the runway. For at least one image of the sequence of images, the method includes applying the image(s) to a machine learning model trained to predict a pose of the aircraft relative to the runway. The machine learning model is configured to map the image(s) to the pose based on a training set of labeled images with respective ground truth poses of the aircraft relative to the runway. The pose is output as a current pose estimate of the aircraft relative to the runway for use in at least one of monitoring the current pose estimate, generating an alert based on the current pose estimate, or guidance or control of the aircraft on a final approach.