Radar Image Evaluation with Reliability-Based Neural Network Fallback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Radar image processing is computationally challenging, leading to increased processing latency or energy consumption, which is undesirable in mobile and vehicle safety applications.
Innovation Solution
A method and apparatus that utilize a trained neural network for rapid position estimation in radar images, falling back to a conventional radar processing pipeline when reliability criteria are not met, and employing continual or federated learning to enhance the neural network model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a conventional radar processing pipeline is used to determine object positions, then measurement precision is improved, but processing time increases and energy consumption increases
Solution Approach 1:
The system dynamically selects between two processing pipelines based on reliability criteria. The neural network pipeline is used when reliability criteria are met (providing low latency), while the conventional pipeline is used as fallback when criteria are not met (ensuring accuracy). This dynamic adaptation resolves the contradiction by adjusting the processing approach based on real-time conditions.
Solution Approach 2:
The invention changes the processing parameters by switching between different algorithms (neural network vs. conventional pipeline) based on reliability criteria. This parameter change allows the system to optimize for either speed or accuracy depending on the situation, resolving the time-precision contradiction.
2Measurement precision
If a conventional radar processing pipeline is used to determine object positions, then measurement precision is improved, but energy consumption increases
Solution Approach 1:
The system dynamically adjusts energy consumption by selecting the neural network pipeline when reliability criteria are met (lower energy usage) and switching to the conventional pipeline only when necessary (higher energy usage). This dynamic approach resolves the contradiction between accuracy and energy consumption.
Solution Approach 2:
The system uses the simpler neural network pipeline partially (when reliability criteria are met) instead of always using the full conventional pipeline. This partial action approach achieves acceptable accuracy with reduced energy consumption, resolving the contradiction.
3Productivity
If a simple processing approach is used, then processing speed is improved, but reliability of position estimation deteriorates
Solution Approach 1:
The system incorporates feedback through reliability criteria evaluation. The neural network output is evaluated against reliability criteria, and based on this feedback, the system decides whether to use the neural network results or fallback to the conventional pipeline. This feedback mechanism ensures reliability while maintaining speed when possible.
Solution Approach 2:
The reliability criteria act as an intermediary between the simple neural network approach and the reliable conventional pipeline. This intermediary evaluates the neural network output and determines whether it meets reliability standards, resolving the contradiction between speed and reliability.
Data Source
AI summary
In an embodiment, a method to evaluate radar images includes providing a first raw radar image and a second raw radar image and determining, whether a reliability criterion is fulfilled. The method further includes using a first coordinate and a second coordinate output by a trained neural network as an estimate of a position of an object if the reliability criterion is fulfilled, the trained neural network using the first raw radar image and the second raw radar image as an input. The method further includes using a third coordinate and a fourth coordinate output by another radar processing pipeline as the estimate of the position of the object if the reliability criterion is not fulfilled, the radar processing pipeline using the first raw radar image and the second raw radar image as an input.