Automated Sensor Data Labeling via ADS-B Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The production of training data for machine learning systems, particularly for detecting moving objects in image data, is a time-consuming and error-prone process that requires significant human effort, with estimates suggesting it takes around 800 human hours to label one hour of image data.
Innovation Solution
The system utilizes Automatic Dependent Surveillance-Broadcast (ADS-B) position data to automatically determine and label data points in sensor data, such as images, representing moving objects, thereby reducing the need for manual labeling and enhancing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual labeling by humans is used to produce training data, then the accuracy and reliability of labeled data is improved, but the time consumption and labor cost increase significantly
Solution Approach 1:
The system enables automatic self-labeling of training data by using machine learning models to process sensor data and generate labels without human intervention. The model autonomously identifies moving objects, determines their positions, and creates training data entries, thereby eliminating the time-consuming manual labeling process while maintaining acceptable accuracy through continuous model improvement
Solution Approach 2:
The patent replaces the mechanical human labeling process with an automated computational system. Instead of humans manually reviewing and labeling image data, the system uses algorithms to automatically process sensor inputs, identify objects, and generate labels, thus substituting human labor with automated mechanical processing
2Manufacturing precision
If manual review and labeling of image data is performed, then the quality of training data is improved, but the productivity and speed of data production decrease
Solution Approach 1:
The system enables continuous automated processing of sensor data to generate training data in real-time or near real-time. The machine learning model continuously processes incoming sensor inputs, identifies moving objects, and produces labeled training data without interruption, thereby maintaining high productivity while ensuring consistent data quality through ongoing model refinement
Solution Approach 2:
The patent replaces the slow manual labeling process with high-speed automated computational processing. The system uses machine learning algorithms to rapidly analyze sensor data and generate training entries, dramatically increasing the speed of data production from hours to seconds while maintaining quality through algorithmic consistency
3Reliability
If human hours are used to label training data, then the accuracy of object detection is improved, but the cost and complexity of the process increase
Solution Approach 1:
The system eliminates the need for complex human coordination and manual labeling processes by enabling self-service automated labeling. The machine learning model autonomously performs object detection, position determination, and data labeling tasks that would otherwise require complex human workflows, thereby reducing process complexity while maintaining detection accuracy through algorithmic processing
Data Source
AI summary
Described are systems and methods for generating training data that is used to train a machine learning system to detect moving objects represented in sensor data. The system and methods utilize position data received from a target vehicle to determine data points within sensor data that represents that target vehicle. For example, a station at a known location may receive Automatic Dependent Surveillance-Broadcast (“ADS-B”) data (position data) corresponding to a target vehicle that is within the field of view of a station sensor, such as a camera. The position data may then be correlated with the sensor data and projected into the sensor data to determine data points within the sensor data that represent the target vehicle. Those data points are then labeled to indicate the location, size, and/or shape of the target vehicle as represented in the sensor data, thereby producing training that may be provided to train a machine learning algorithm or system to detect moving objects, such as aircraft.