Onboard Aircraft Ice Detection Using Cameras and Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing aircraft ice detection methods are time-consuming and inefficient, particularly due to manual ground-based inspections, which can lead to delays if icing is detected after departure.
Innovation Solution
An onboard ice detection system using pre-existing cameras and machine-learning models to analyze image data for ice presence, integrating data reduction and feature extraction to identify ice patterns, reducing the need for ground-based equipment and enabling real-time inspections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual ground-based ice detection is used, then ice detection can be performed, but it is time-consuming and leads to delays
Solution Approach 1:
The patent replaces manual mechanical inspection methods with an automated optical detection system using cameras and machine learning algorithms. The system captures images of aircraft surfaces and uses computer vision to automatically detect ice, eliminating the need for manual visual and tactile checks while significantly reducing detection time.
Solution Approach 2:
The system enables the aircraft to self-detect ice conditions through onboard cameras and processing systems. The machine learning model automatically analyzes images and determines ice presence without requiring external inspection equipment or personnel, allowing the aircraft to perform its own ice detection.
2Reliability
If ground-based deicing equipment is used, then ice mitigation can be achieved, but it requires additional ground equipment and infrastructure
Solution Approach 1:
The patent extracts the ice detection function from ground-based systems and relocates it to the aircraft itself. By placing cameras and processing systems onboard, the system eliminates the need for complex ground-based detection equipment and infrastructure while maintaining reliable ice detection and mitigation capabilities.
Solution Approach 2:
The system uses pre-existing onboard cameras that serve multiple functions, including ice detection, to reduce the need for dedicated ground equipment. The machine learning model can analyze various image types and conditions, providing universal ice detection capability across different scenarios without requiring specialized ground infrastructure.
3Measurement precision
If manual walk-around inspection is performed, then ice detection is possible, but it is challenging due to aircraft size, height, and poor weather conditions
Solution Approach 1:
The patent replaces difficult manual inspection with automated optical systems. Cameras mounted on the aircraft capture images of surfaces that would be hard to reach during manual walk-around inspections, including high and hard-to-access areas. The machine learning model automatically analyzes these images regardless of weather or lighting conditions, eliminating the physical challenges of manual inspection.
Solution Approach 2:
The system uses cameras as intermediaries to observe ice conditions on aircraft surfaces. The machine learning model acts as an intermediary that processes visual information and determines ice presence, allowing detection without direct human contact with difficult-to-reach surfaces and immune to poor weather conditions that hinder manual inspection.
4Measurement precision
If information about icing is available only after walk-around check, then ice detection is achieved, but delays occur if icing is detected after departure
Solution Approach 1:
The system performs ice detection continuously or at frequent intervals using onboard cameras before departure and during flight. By having ice detection capability already in place and operational, the system can detect ice conditions early and provide timely warnings, eliminating delays associated with post-departure detection and allowing preventive action to be taken.
Data Source
AI summary
A method of detecting ice on a surface of an aircraft includes obtaining a first series of images captured by one or more cameras onboard the aircraft. The method also includes performing data reduction operations to generate a second series of images. The method further includes generating feature data based on the second series of images, where the feature data is indicative of changes over time of pixel data of the second series of images. The method also includes generating, based on the feature data, input data for a trained classifier and determining, using the trained classifier, whether ice is present on one or more surfaces of the aircraft.