2D Radar Object Detection via Cross-1D Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current radar systems face challenges in accurately detecting large objects at close range and suffer from high false alarm rates due to clutter interference, particularly in automotive applications, where traditional one-dimensional and two-dimensional detection methods are inefficient in terms of computational cost and memory access.
Innovation Solution
A 2D cross object detection method is introduced, which performs an initial one-dimensional analysis along the range axis to identify candidate objects, followed by a cross one-dimensional analysis along the Doppler axis at each candidate location, and applies a reduced detection threshold in the neighborhood of each object to enhance detection of large objects, thereby reducing false alarms and improving detection efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional two-dimensional detection methods are used, then detection accuracy is improved, but computational cost and memory access increase significantly
Solution Approach 1:
The patent segments the detection process into two separate one-dimensional analyses: first along the range axis to identify candidate objects, then along the Doppler axis at each candidate location. This segmentation reduces the computational complexity from O(N×M) in traditional 2D methods to O(N+M×K) where K is the number of candidates, while maintaining detection accuracy by performing cross-validation at identified locations.
Solution Approach 2:
The patent transitions from traditional simultaneous two-dimensional analysis to sequential one-dimensional analysis by adding a temporal dimension to the detection process. The first 1D analysis identifies candidates in the range dimension, then the second 1D analysis validates them in the Doppler dimension, effectively transforming a complex 2D problem into two simpler 1D problems solved in sequence.
2Speed
If traditional detection methods are used, then processing speed is maintained, but false alarm rates increase due to clutter interference
Solution Approach 1:
The patent extracts and removes clutter interference by performing the first 1D analysis along the range axis to identify candidate objects, then performing the second 1D analysis along the Doppler axis only at those candidate locations. This extraction approach isolates true objects from clutter by requiring confirmation in both dimensions, significantly reducing false alarms while maintaining processing speed through efficient candidate filtering.
3Measurement precision
If reduced detection threshold is applied in neighborhood of objects, then detection of large objects at close range is improved, but false alarms may increase
Solution Approach 1:
The patent applies local quality by using a reduced detection threshold specifically in the neighborhood of identified candidate objects, rather than applying a uniform threshold across the entire data matrix. This localized threshold reduction enhances detection of large objects at close range that may have lower signal strength, while the requirement for cross-validation in both dimensions prevents spurious false alarms from the lowered threshold.
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 significantly reduces false alarm rates while maintaining low computational costs and efficient memory access, effectively detecting large objects at close range with improved accuracy in automotive radar systems.
Implementation Method 1
If there is an object in the path of the radio wave, it will reflect some of the electromagnetic energy, and the radio wave will bounce back to the radar device.
Implementation Method 2
Radio waves move through the air at a constant speed (the speed of light), so the radar device can calculate how far away the object is based on how long it takes the radio signal to return.
Implementation Method 3
Radar can also be used to measure the speed of an object, due to a phenomenon called Doppler shift. When the object is moving away from the radar system, the second segment of the signal has to travel a greater distance to reach the car than the first segment of the signal. This has the effect of 'stretching out' the wave, or lowering its frequency. If the object is moving toward the radar system, the second segment of the wave travels a shorter distance than the first segment before being reflected. As a result, the peaks and valleys of the wave get squeezed together and the frequency increases.
Data Source
AI summary
A set of radar data is organized as a two dimensional (2D) array with multiple lines of data. A 2D cross point analysis is performed on each set of data to detect objects in range of the radar system. A set of candidate objects is identified by performing an initial one dimensional (1D) analysis on a line of data along a first axis of the set if data to determine a candidate location of each candidate object along the first axis; however, the set of candidate objects may include false objects. The set of candidate objects is pruned by performing a cross 1D analysis of the data along a second axis of the set of data at a position corresponding to each candidate location along the first axis to select a set of most likely candidate objects from the set of candidate objects.