Vehicle Vision System Self-Calibration via Object Position Tracking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing vision systems for motor vehicles face challenges in accurately estimating the extrinsic calibration of imaging devices of different types or wavelength ranges, leading to increased computational costs and risks of false detections due to low precision in camera calibration.

Innovation Solution

The system uses the position of detected objects for calibration estimation, leveraging existing object detectors to track and store object positions over time, allowing for systematic deviation analysis and adjustment of camera calibration parameters, independent of imaging device type or viewpoint, thereby optimizing calibration for object detection applications.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional calibration methods (vanishing point detection, optical flow, eight-point algorithm) are used, then calibration can be performed, but computational cost increases and detection reliability decreases due to larger search areas and false detections

Engineering Contradiction:
Improvecalibration precisionVSAvoidcomputational cost
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system uses detected objects themselves to provide calibration information, eliminating the need for separate calibration procedures or artificial markers. The object detector automatically provides both detection results and calibration data, making the system self-calibrating and reducing computational overhead.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The object detector serves dual purposes: detecting objects for navigation and simultaneously providing calibration data for the imaging system. This multi-functionality eliminates the need for separate calibration procedures and reduces overall system complexity.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Measurement precision

If traditional calibration methods are used, then calibration can be achieved, but detection reliability decreases due to false detections and erroneous associations in larger search areas

Engineering Contradiction:
Improvecalibration precisionVSAvoiddetection reliability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system uses feedback from detected object positions to continuously refine and update calibration parameters. By comparing expected object positions with actual detections and adjusting calibration accordingly, the system maintains high detection reliability while using smaller, more focused search areas.

Inventive Principle:
Principle #23Feedback

3Measurement precision

If calibration is performed using standardized objects (traffic signs, guide posts), then calibration can be achieved, but the system becomes dependent on specific object types and viewpoints

Engineering Contradiction:
Improvecalibration precisionVSAvoidimaging device type independence
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The calibration method works with any detectable object type regardless of imaging device characteristics. The system adapts to different camera types, wavelengths, and viewpoints by using the generic properties of detected objects rather than relying on specific standardized objects, making it universally applicable.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP2662828B1A vision system and method for a motor vehicle
Publication Date: 2020.05.06 VEONEER SWEDEN AB
  • EP2662828B1 patent drawingFigure 1
  • EP2662828B1 patent drawingFigure 2~4

AI summary

A vision system (10) for a motor vehicle comprises an imaging means (11) with a plurality of imaging devices (12.1, 12.2) for acquiring images from a surrounding of a motor vehicle, and a processing means (14) adapted to perform image processing on images from said imaging devices (12.1, 12.2). The processing means comprises an object detection means (20) adapted to detect objects in the surrounding of the motor vehicle based on image data from said imaging means (11), and a calibration means (21) adapted to estimate the extrinsic calibration of at least one of said imaging devices (12.1, 12.2). The calibration means (21) is adapted to perform said estimation using the position of detected objects provided by said object detection means (20).