Radar Detection Bin Sampling for Uniform Parallel Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Farthest point sampling (FPS) in radar systems is computationally inefficient due to the need for numerous distance calculations, which overwhelms the processing power of computers, especially those without coprocessors, and limits parallel processing capabilities.
Innovation Solution
Assign bin identifiers (BIDs) based on distance and Doppler values to radar detections, allowing for unbiased sampling and grouping without the need for extensive distance calculations, enabling uniform sampling and parallel processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If farthest point sampling (FPS) is used to sample radar detections, then uniform sampling coverage is achieved, but computational workload increases due to numerous distance calculations
Solution Approach 1:
The patent segments the radar detection space into discrete bins based on distance and Doppler value ranges. Each bin acts as an independent unit for sampling, replacing the continuous space approach of FPS. This segmentation allows parallel processing of bins while maintaining uniform coverage, as each bin contributes samples proportionally to its detection density.
Solution Approach 2:
The patent changes the sampling parameter from Euclidean distance (used in FPS) to bin identifiers based on distance and Doppler ranges. This parameter transformation converts the complex distance calculation problem into a simpler bin assignment and selection process, reducing computational complexity while preserving sampling uniformity across the radar scene.
2Measurement precision
If farthest point sampling (FPS) is used to sample radar detections, then uniform sampling coverage is achieved, but device complexity increases due to requirements for coprocessors
Solution Approach 1:
By segmenting the detection space into bins and assigning each detection to a specific bin based on distance and Doppler ranges, the patent enables independent processing of each bin. This segmentation allows standard processors to handle sampling without requiring specialized coprocessors, as each bin can be processed independently using simple comparison and selection operations.
Solution Approach 2:
The patent replaces expensive, complex distance calculation operations with simpler bin assignment operations. Instead of computing Euclidean distances between all pairs of points (which requires powerful hardware), the system uses pre-defined bin ranges and simple comparisons to assign detections to bins and select samples, enabling operation on less powerful, more accessible hardware.
3Measurement precision
If farthest point sampling (FPS) is used to sample radar detections, then uniform sampling coverage is achieved, but parallel processing capability is limited
Solution Approach 1:
The patent divides the radar detection space into multiple independent bins, where each bin can be processed in parallel. Since bin assignment and sampling decisions for one bin do not depend on other bins, the system can simultaneously process multiple bins using multi-core processors or parallel computing architectures, dramatically improving processing throughput while maintaining uniform sampling coverage.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A technique of identifying features of objects in a physical environment of a set of radar sensors includes accessing a plurality of bin identifiers assigned to respective radar detections made by the set of radar sensors, the bin identifiers specifying respective bins. A bin has at least (i) a first dimension representing a range of distance values and (ii) a second dimension representing a range of Doppler values. The technique further includes selecting radar detections based on the bin identifiers, the selecting producing a plurality of samples, and processing the plurality of samples to determine one or more features of the objects.