Radar People Counting Using Dual Range-Doppler Neural Pipelines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing people counting methods based on camera images face accuracy issues due to varying lighting conditions and privacy concerns, while radar-based systems have limitations in distinguishing multiple individuals close together.
Innovation Solution
A computer-implemented method using radar measurement datasets to create range-Doppler measurement maps, separating macro-Doppler and micro-Doppler features, and employing a neural network algorithm with separate data processing pipelines for accurate people counting, independent of lighting conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If camera-based people counting is used, then people counting functionality is provided, but accuracy varies with lighting conditions
Solution Approach 1:
The patent replaces camera-based optical detection with radar-based electromagnetic wave detection. The radar system transmits electromagnetic waves and receives reflections to detect persons, completely eliminating dependence on visible light conditions. This substitution of detection mechanism resolves the contradiction by making people counting accurate regardless of lighting intensity or conditions.
Solution Approach 2:
The patent changes the detection parameter from optical reflection (camera) to electromagnetic wave reflection (radar). By operating in the radio frequency domain rather than the visible spectrum, the system achieves people counting accuracy that is invariant to lighting conditions, as radar waves penetrate darkness, smoke, and other optical obstructions.
2Illumination intensity
If radar-based people counting is used, then lighting condition independence is achieved, but difficulty in distinguishing multiple individuals close together occurs
Solution Approach 1:
The patent segments the radar detection process into multiple independent analysis dimensions: range dimension (distance from radar), velocity dimension (Doppler shift indicating motion), and angular dimension (direction of arrival). By analyzing persons across these segmented dimensions simultaneously, the system can resolve and distinguish multiple individuals even when they are close together, as each person occupies a unique position in the multi-dimensional detection space.
Solution Approach 2:
The patent transitions from two-dimensional camera imaging to three-dimensional radar detection by adding the velocity dimension through Doppler effect measurement. This dimensional enhancement provides an additional degree of freedom for distinguishing persons, allowing the system to separate individuals who may be spatially close but have different motion characteristics, thereby resolving the difficulty of detecting close individuals.
3Loss of information
If camera-based detection is used, then visual information is obtained, but privacy concerns arise
Solution Approach 1:
The patent substitutes camera-based visual detection with radar-based electromagnetic detection. Radar systems capture motion and position information through radio wave reflections rather than optical images, thereby acquiring necessary detection information while avoiding the creation of visual imagery that could reveal sensitive personal details and trigger privacy concerns.
Solution Approach 2:
The patent extracts only the essential detection parameters (range, velocity, angular position) from the radar signals while deliberately excluding visual information. By taking out only the minimum necessary data for people counting and removing all visual content, the system maintains detection functionality while eliminating privacy-invasive information.
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
Enables accurate counting of multiple individuals, including those close together, with improved robustness against variations in radar sensor pose and illumination, while preserving privacy.
Implementation Method 1
a radar measurement dataset that is obtained by a radar measurement of a scene
Implementation Method 2
The 1st range-Doppler measurement map is indicative of macro-Doppler features of one or more persons included in the scene. The 2nd range-Doppler measurement map is indicative of micro-Doppler features of the one or more persons included in the scene
Data Source
AI summary
In accordance with an embodiment, a method includes estimating a people count of one or more persons included in the scene based on a first range-Doppler measurement map and the second range-Doppler measurement map derived from a radar measurement dataset. Estimating the people count includes inputting the first range-Doppler measurement map into a first data processing pipeline of a neural network algorithm, and inputting the second range-Doppler measurement map into a second data processing pipeline of the neural network algorithm. The first data processing pipeline and the second data processing pipeline includes range-Doppler convolutional layers implementing two-dimensional convolutions along the range dimension and the Doppler dimension, and the neural network algorithm includes an output section for processing a combination of a first output of the first data processing pipeline and a second output of the second data processing pipeline in a regression block.