Radar Object Detection Using Doppler Zero Slice Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Radar systems in vehicles face issues with short-range leakage (SRL) leading to false positives and unused sensing data, affecting object detection accuracy in close-range scenarios.
Innovation Solution
A method and system that extract doppler zero slice data from radar range and doppler data, maintain predictions based on previous frames, calculate standard deviation using Mahalanobis distances, and update covariance matrices to mitigate SRL effects, providing cleaned range doppler data for accurate object detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If radar systems are used for object detection in vehicles, then detection capability is improved, but short-range leakage causes false positives and reduces reliability in short-range areas
Solution Approach 1:
The patent segments the range-Doppler data by extracting only the doppler zero slice at low-end distance ranges, separating the problematic short-range data from the rest of the radar data for specialized processing and prediction
Solution Approach 2:
The system performs preliminary action by maintaining a prediction of doppler zero slice data based on previous frames before making object detection decisions, allowing comparison between predicted and actual values to filter false positives
2Device complexity
If standard object detection methods are used, then processing simplicity is maintained, but false positives increase due to SRL disturbances
Solution Approach 1:
The patent implements feedback by maintaining a prediction of doppler zero slice data from previous frames and comparing it with current frame data, using the prediction error and standard deviation to dynamically adjust detection thresholds and filter false positives
Solution Approach 2:
The system applies dynamics by updating the prediction and covariance matrices adaptively based on the prediction error and Mahalanobis distances, allowing the detection algorithm to dynamically adjust to changing environmental conditions while maintaining reliability
3Reliability
If doppler zero slice extraction and prediction methods are applied, then false positives are reduced, but processing complexity increases
Solution Approach 1:
The patent extracts only the doppler zero slice at low-end distance ranges from the full range-Doppler data, focusing processing resources on the specific area where SRL causes false positives while ignoring other areas where standard detection suffices
Solution Approach 2:
The system changes parameters by using Mahalanobis distances and covariance matrices to characterize the prediction error distribution, transforming the detection problem into a statistical framework that improves reliability without requiring excessive computational resources
Data Source
AI summary
Method and systems for object detection using a radar module are disclosed. Frames of range and doppler data are received from a radar module at sample time intervals. Doppler zero slice data is extracted from a current frame of the range and doppler data. A prediction of doppler zero slice data is maintained. The prediction of doppler zero slice data is based at least partly on doppler zero slice data from a previous frame of range and doppler data. Standard deviation data is determined based at least partly on prediction error data. The prediction error data relates to a difference between the prediction of doppler zero slice data and the doppler zero slice data. An object detection output is determined based on a comparison of the standard deviation data and an object detection threshold.