Radar Object Detection Using FFT Heat Maps and RAD Tensors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional radar sensors over-process raw data using signal filters and clustering beamforming, leading to a loss of detailed information and increased processing latency, which hampers effective object detection.
Innovation Solution
Generate configurable radar data representations such as 1D range heat map tensors, 2D RD heat maps, 2D RA heat maps, or 3D RAD matrices, which are directly input to machine learning models for object detection, reducing signal processing and latency by performing FFTs in real-time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional signal filters and clustering beamforming are used to process radar data, then processing completeness is improved, but information loss increases and processing latency increases
Solution Approach 1:
The patent performs Fast Fourier Transforms (FFTs) in real-time to generate configurable radar data representations (1D range heat maps, 2D RD heat maps, 2D RA heat maps, or 3D RAD matrices) before they are input to machine learning models. This preliminary processing preserves detailed information in the radar data while maintaining processing completeness, avoiding the information loss associated with traditional signal filters and clustering beamforming.
2Reliability
If traditional signal filters and clustering beamforming are used to process radar data, then processing completeness is improved, but processing latency increases
Solution Approach 1:
The patent performs Fast Fourier Transforms (FFTs) in real-time to generate configurable radar data representations before they are input to machine learning models. This preliminary processing approach reduces processing latency compared to traditional methods while maintaining processing completeness, as the transformations are computed efficiently and continuously rather than through multi-stage filtering and beamforming operations.
3Measurement precision
If configurable radar data representations are generated and input to machine learning models, then object detection accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent generates configurable radar data representations (1D range heat maps, 2D RD heat maps, 2D RA heat maps, or 3D RAD matrices) by adjusting parameters such as sensing range, distance resolution, velocity resolution, and angular resolution. These configurable representations provide rich information for machine learning models to achieve high object detection accuracy while allowing flexibility to balance processing complexity based on specific application requirements.
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 preserves detailed information in radar data, enhances object detection accuracy, and reduces processing latency by configuring radar output based on sensing range, distance, velocity, and angular resolution, thereby improving the efficiency of autonomous systems.
Implementation Method 1
Radar sensors transmit electromagnetic wave signals that are reflected by objects in the environment. Radar sensors capture the reflected signals
Implementation Method 2
Radar sensors transmit electromagnetic wave signals that are reflected by objects in the environment
Data Source
AI summary
Provided are methods for generation of representations of radar data. Some methods described include: receiving ADC raw data of a radar sensor of a vehicle; performing range FFT, Doppler FFT, and azimuth FFT on the ADC raw data; generating a 1D range heat map tensor representing the range FFT, a 2D RD heat map tensor representing a combination of the range FFT and the Doppler FFT, a 2D RA heat map tensor representing a combination of the range FFT and the azimuth FFT, or a 3D RAD matrix tensor representing a combination of the range FFT, the Doppler FFT, and the azimuth FFT; and inputting at least one of the 1D range heat map tensor, the 2D RD heat map tensor, the 2D RA heat map tensor, or the 3D RAD matrix tensor, to a machine learning model for detecting objects on a road network around the vehicle.