AI Model Update via Micro-Doppler Measurements for UAV Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless communication systems face challenges in efficiently updating artificial intelligence (AI) or machine learning (ML) models for object recognition, particularly in resource-intensive micro-Doppler spectrum processing, which can lead to low accuracy and excessive data storage usage.
Innovation Solution
A method and apparatus for wireless communication that allow a wireless communication device to transmit micro-Doppler measurements associated with a first AI/ML model and receive an update to configure a second AI/ML model, which can improve prediction accuracy and reduce signaling overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If micro-Doppler spectrum processing is used for object recognition, then measurement precision is improved, but use of energy increases
Solution Approach 1:
The patent segments the object recognition process into two phases: a training phase where the AI/ML model is trained offline using micro-Doppler spectrum data, and an inference phase where the trained model performs rapid classification. This segmentation allows computationally intensive processing to occur during training while keeping runtime energy consumption low.
Solution Approach 2:
The patent performs preliminary action by training the AI/ML model in advance using extensive micro-Doppler spectrum processing. The trained model parameters and features are stored for later use, eliminating the need to repeat intensive processing during actual object recognition tasks.
2Measurement precision
If AI/ML model updates are transmitted frequently, then object recognition accuracy is improved, but signaling overhead increases
Solution Approach 1:
The patent extracts only the essential model update parameters and configuration information from the complete AI/ML model, transmitting only these critical updates through the wireless interface. This extraction approach maintains recognition accuracy while minimizing the amount of data that needs to be transmitted.
Solution Approach 2:
The patent changes the parameter representation by transmitting model update parameters in a compressed or optimized format, and by selectively updating only those parameters that have changed significantly. This reduces the signaling overhead while preserving the essential information needed for accurate object recognition.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables improved object recognition and prediction for UAVs by updating AI/ML models, leading to enhanced flight characteristics, better task completion, and reduced likelihood of damage due to improved accuracy.
Implementation Method 1
transmitting an indication of micro-Doppler measurements associated with a first artificial intelligence or machine learning (AI/ML) model
Data Source
AI summary
Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a wireless communication device may transmit an indication of micro-Doppler measurements associated with a first artificial intelligence or machine learning (AI/ML) model. The wireless communication device may receive, in association with transmitting the indication of the micro-Doppler measurements, an indication of a second AI/ML model that is an update of the first AI/ML model. The wireless communication device may transmit, in connection with using the second AI/ML model, an indication associated with object recognition. Numerous other aspects are described.