Range-Doppler Radar Detection for Small Stationary Objects

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional radar systems face challenges in detecting and classifying smaller stationary objects due to their sparse point cloud representations, leading to slower response times and decreased safety in autonomous and semiautonomous driving applications.

Innovation Solution

The use of low-level radar data in the form of range-Doppler maps, which retains all information from raw electromagnetic signals, is preprocessed to filter out non-stationary bins, and then fed into an end-to-end deep convolutional detection and classification network for accurate object classification and bounding box estimation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional radar systems use point cloud representations for object detection, then the system structure is simple, but smaller stationary objects cannot be detected accurately leading to slower response times

Engineering Contradiction:
Improvedetection accuracyVSAvoidresponse time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent transforms radar data from traditional point cloud representation to range-Doppler spectrum representation, adding spectral dimension information. This dimensional transformation enables better differentiation of stationary objects from clutter by utilizing Doppler frequency characteristics, thereby improving detection accuracy and response time for small stationary objects.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Solution Approach 2:

The patent applies preprocessing operations including range-Doppler transformation and stationary bin identification before object detection. By performing these preliminary actions to filter and organize the data, the system prepares optimized input for the neural network, enabling faster and more accurate detection of stationary objects without requiring complex post-processing.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If conventional radar systems process raw electromagnetic signals through multiple transformation steps, then detection capability improves, but computational complexity increases

Engineering Contradiction:
Improveobject classification accuracyVSAvoidprocessing system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary processing stage that performs range-Doppler transformation and stationary bin identification on raw electromagnetic signals. This intermediary step creates a simplified intermediate representation that retains essential object characteristics while removing redundant information, making subsequent neural network processing more efficient and accurate without requiring overly complex system architecture.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent extracts only the relevant stationary object information from raw radar signals by identifying stationary Doppler bins and filtering out non-stationary components. This extraction process removes unnecessary data and computational burden while preserving the critical features needed for accurate stationary object detection and classification.

Inventive Principle:
Principle #2Taking out (Extraction)

3Measurement precision

If radar systems filter out non-stationary range-Doppler bins to focus on stationary objects, then detection precision for stationary objects improves, but information from moving objects is lost

Engineering Contradiction:
Improvestationary object detection precisionVSAvoidmoving object data
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent segments the radar signal processing into distinct pathways: one for stationary objects (using stationary bin identification from range-Doppler maps) and another for moving objects (using traditional point cloud processing). This segmentation allows each pathway to be optimized for its specific target type, improving stationary object detection precision while preserving moving object information through separate processing channels.

Inventive Principle:
Principle #1Segmentation

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 enables more timely and accurate detection and classification of stationary objects, improving the computational efficiency and safety of driving systems by providing object data quicker to other vehicle systems.

Implementation Method 1

Raw electromagnetic signals reflected off stationary objects and received by a radar system

Methodology Applied
Scientific EffectElectromagnetic radiation reflection: Reflection

Implementation Method 2

transforming the raw electromagnetic signals into low-level electromagnetic spectrum data to identify whether a set of objects of the one more objects are stationary objects

Methodology Applied
Scientific EffectFourier transformation:

Implementation Method 3

estimating, by a deep convolutional detection and classification network and based on the low-level electromagnetic spectrum data, a detection and classification of the stationary objects

Methodology Applied
Scientific EffectConvolution operation:

Data Source

PatentEP4350385A1Stationary object detection and classification based on low-level radar data
Publication Date: 2024.04.10 APTIV TECHNOLOGIES AG
  • EP4350385A1 patent drawingFigure 1
  • EP4350385A1 patent drawingFigure 2
  • EP4350385A1 patent drawingFigure 3

AI summary

This document describes techniques and systems for stationary object detection and classification based on low-level radar data. Raw electromagnetic signals reflected off stationary objects and received by a radar system may be preprocessed to produce low-level spectrum data in the form of range-Doppler maps that retain all or nearly all the data present in the raw electromagnetic signals. The preprocessing may also filter non-stationary range-Doppler bins. The remaining low-level spectrum data represents stationary objects present in a field-of-view (FOV) of the radar system. The low-level spectrum data representing stationary objects can be fed to an end-to-end deep convolutional detection and classification network that is trained to classify and provide object bounding boxes for the stationary objects. The outputted classifications and bounding boxes related to the stationary objects may be provided to other driving systems to improve their functionality resulting in a safer driving experience.