Radar Object Detection Using Neural Network Signal Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional radar-based surroundings detection systems require substantial signal processing to obtain meaningful information from time signals, leading to computationally intensive and inefficient object detection in driver assistance and automated driving applications.
Innovation Solution
A method for training a radar-based object detection system using a training data set that represents surroundings as a point cloud, point clusters, or reflectance grids, leveraging artificial intelligence and neural networks to simplify data processing and reduce information loss, with pre-processing techniques like two-dimensional fast Fourier transforms and simulation-based data generation to enhance accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional signal processing methods are used to obtain meaningful information from radar time signals, then object detection can be achieved, but the computational complexity and processing time increase substantially
Solution Approach 1:
The patent replaces conventional mechanical signal processing methods (FFT, clustering algorithms, point cloud generation) with a neural network-based AI system. The neural network is trained to directly process raw radar time signals and output object detection results, eliminating the need for complex intermediate processing steps while maintaining or improving detection accuracy.
Solution Approach 2:
The patent changes the processing approach from deterministic signal processing parameters (frequency bins, range gates, velocity cells) to learned parameters within a neural network. The network learns optimal feature extraction and object identification patterns from training data, adapting to different scenarios without requiring manual parameter tuning or complex processing pipelines.
2Loss of information
If conventional radar signal processing is applied to generate surroundings maps, then object information can be extracted, but processing time and computational resources are substantially consumed
Solution Approach 1:
The patent performs preliminary action by pre-training a neural network model on extensive radar data before deployment. The network learns to directly map raw time signals to object characteristics during the training phase, so that during actual operation, detection occurs in real-time without requiring lengthy processing steps. This shifts computational burden from runtime to training time.
Data Source
AI summary
A method for training a radar-based object detection. The method includes: creating a training data set that includes radar data of a radar sensor or of a plurality of radar sensors, the radar data representing a map of surroundings of the radar sensor or of the plurality of radar sensors; training a radar-based object detection based on the created training data set for generating an output representation of the surroundings of the radar sensor.