Radar Sensor Calibration via Rotating Corner Reflector

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Autonomous navigation systems for off-road vehicles face challenges in calibrating sensors due to environmental vibrations and misalignment issues, particularly with radar sensors experiencing high noise levels, making it difficult to distinguish between objects and background signals.

Innovation Solution

A calibration method using a highly reflective radar reflector object, such as a corner reflector, is moved in a known manner to provide a detectable signal, and its location is determined using a localization sensor like GNSS, allowing for the calculation and storage of offsets to adjust future object detections, with additional sensors like lidar and camera aiding in correlation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a radar sensor is used for autonomous navigation, then the vehicle can detect objects in rough environments, but the sensor experiences high noise levels making it difficult to distinguish between objects and background signals

Engineering Contradiction:
Improveobject detection reliabilityVSAvoidnoise interference
Core Design Contradiction:
ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The calibration object is rotated at a known frequency to create a periodic signal pattern. This mechanical rotation generates distinct frequency components in the radar return signal that can be easily distinguished from background noise through spectral analysis, thereby improving object detection reliability in noisy environments

Inventive Principle:
Principle #18Mechanical vibration

Solution Approach 2:

The calibration object rotates periodically at a known frequency, creating a periodic signal pattern in the radar data. This periodicity allows the system to distinguish the calibration object from random background noise by looking for signals at the specific rotation frequency, effectively filtering out non-periodic interference

Inventive Principle:
Principle #19Periodic action

2Measurement precision

If sensors are mounted on a vehicle for autonomous navigation, then the system can detect object locations, but vibration and environmental factors cause sensor movement and misalignment

Engineering Contradiction:
Improvesensor location accuracyVSAvoidsensor mounting stability
Core Design Contradiction:
Measurement precisionVSStability of the object's composition

Solution Approach 1:

The calibration process is performed in advance to determine the relative positions and orientations of all sensors before autonomous operation begins. This preliminary calibration establishes baseline offset values that compensate for manufacturing tolerances and initial misalignments, ensuring accurate measurements even if sensors shift slightly during operation

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses the known position and rotation of the calibration object to generate feedback signals that reveal actual sensor positions and orientations. This feedback information is used to calculate offset corrections that are applied to future measurements, continuously compensating for any sensor movement or misalignment

Inventive Principle:
Principle #23Feedback

3Measurement precision

If a calibration system is implemented to adjust sensor locations, then future object detections can be corrected, but the calibration process requires precise known locations and controlled movement

Engineering Contradiction:
Improvedetection accuracyVSAvoidcalibration system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

A dedicated calibration object with known geometric features and a localization sensor serve as intermediaries between the radar sensor and the calibration process. The calibration object acts as a mediator that provides known reference points and motion patterns, simplifying the calibration procedure while achieving high measurement precision without requiring complex calibration equipment

Inventive Principle:
Principle #24Intermediary (Mediator)

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 method enables precise calibration of sensors in various environments, improving the accuracy of obstacle detection and navigation by reducing noise interference and misalignment issues, allowing for repeatable calibration in factory or field settings.

Implementation Method 1

The sensors, such as radar, camera and lidar are mounted at different locations on the vehicle

Methodology Applied
Scientific EffectRadar: Radar

Implementation Method 2

The object is chosen as being a radar reflector (e.g., a cube) with high returned energy in the frequency band of the sensor

Methodology Applied
Scientific EffectReflection: Reflection

Implementation Method 3

The vehicle can have a localization sensor, such as a Global Navigation Satellite System (GNSS)

Methodology Applied
Scientific EffectGlobal Navigation Satellite System:

Implementation Method 4

The object is detected by filtering at one or more frequencies, which includes running a Fourier Transform on a sequence of frames from the sensor to detect peaks at the known frequencies

Methodology Applied
Scientific EffectFourier Transform:

Data Source

PatentEP4332615A1Sensor calibration and localization using known object identification
Publication Date: 2024.03.06 TRIMBLE INC
  • EP4332615A1 patent drawingFigure 1
  • EP4332615A1 patent drawingFigure 2A~2B
  • EP4332615A1 patent drawingFigure 3

AI summary

Disclosed are a method and apparatus for calibrating a sensor. An object is placed in a field of view of the sensor at a known location. The object is moved (e.g., rotated) in a known manner. Data from the sensor is processed to detect an object image moving in the known manner. An apparent location of the object is determined from the object image. The apparent location is compared to the known location of the object to determine an offset. The offset is stored and used to adjust future detected locations of objects.