Automotive Radar Object Height Estimation from Elevation Spectrum Shape
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the height of objects using automotive radar sensors are limited by low elevation resolution, reliance on accurate sensor mounting and pitch angle data, and loss of useful information in peak detection, leading to inaccurate obstacle height estimation.
Innovation Solution
A method and device that utilize the shape of the elevation spectrum, rather than peak positions, to determine object height, incorporating machine learning algorithms to enhance accuracy and reliability, and are invariant to sensor mounting and pitch angle variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If peak detection method is used to determine object height, then the process is simple, but the measurement precision is low due to low elevation resolution
Solution Approach 1:
The patent transitions from one-dimensional peak detection to two-dimensional spectrum shape analysis by utilizing the entire elevation spectrum profile instead of single peak positions. This dimensional expansion allows capturing more information about object height while compensating for the limited elevation resolution of automotive radar sensors.
Solution Approach 2:
The patent introduces spectrum shape features as an intermediary between the raw elevation spectrum and the final height estimation. These features serve as mediators that transform the low-resolution spectral data into more accurate height information by capturing characteristic patterns rather than relying on precise peak localization.
2Productivity
If standard peak detection is used, then computation is fast, but useful information is lost leading to unreliable height estimation
Solution Approach 1:
The patent extracts specific shape features from the elevation spectrum that are most relevant for height estimation, separating these useful characteristics from the rest of the spectral data. This extraction process retains critical information while discarding redundant elements, maintaining computational efficiency.
Solution Approach 2:
The patent changes the parameters being analyzed from peak position coordinates to spectrum shape descriptors. This parameter transformation allows the system to utilize the full information content of the elevation spectrum while maintaining a computationally efficient approach through feature-based analysis.
3Ease of operation
If peak position is used for height determination, then the method is straightforward, but the measurement is highly dependent on accurate sensor mounting and pitch angle data
Solution Approach 1:
The patent inverts the traditional approach by not directly using peak positions to determine height. Instead, it uses spectrum shape features that are inherently more robust to mounting variations, effectively working backwards from the limitations of peak detection to a more reliable methodology.
Solution Approach 2:
The spectrum shape features serve themselves by being inherently invariant to certain transformations and less sensitive to mounting inaccuracies. This self-service characteristic reduces the system's dependence on precise sensor calibration and external correction data.
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
Provides more accurate and reliable height estimation of objects, enabling better obstacle detection and adaptive driving assistance systems.
Implementation Method 1
Modern vehicles typically have, among other sensors such as ultrasound sensors, camera sensors, and light ranging and detection (LIDAR) sensors, one or more radar sensors which can be used to monitor and acquire sensor data of the environment
Implementation Method 2
The multiple antenna elements distributed in vertical direction may be in addition to the antenna elements distributed horizontally
Data Source
AI summary
This document describes systems and techniques for determining a height of an object in a surrounding of a vehicle. In a first aspect, the systems and techniques include acquiring radar data for each of a plurality of vertically distributed antenna elements of a radar antenna. In additional aspects, the systems and techniques include estimating an elevation spectrum from the acquired radar data, extracting one or more features representative of the shape of the estimated elevation spectrum, and determining the height of the object using the extracted one or more features.