Composite Sensor Calibration Using Geometric Feature Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor calibration techniques fail to accurately combine data from different types of sensors, such as 3D distance sensors and image sensors, used in mobile robots, leading to mismatched coordinate systems and inefficient localization.
Innovation Solution
A composite sensor calibration apparatus and method that extracts planes and feature points from range data and image data, projects these features, and calculates coordinate transformation parameters to align the sensors' coordinate systems, allowing for precise matching of data points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional calibration using a lattice rig is used, then coordinate matching between sensors is attempted, but measurement precision deteriorates due to inability to accurately combine data from different sensor types
Solution Approach 1:
The patent introduces a calibration object with specific geometric features (planes and feature points) as an intermediary between different sensor types. This calibration object serves as a common reference that both the 3D distance sensor and image sensor can observe and measure, enabling accurate coordinate transformation between the two sensor coordinate systems through feature point correspondence.
2Ease of operation
If representative calibration with lattice rig is performed, then some coordinate alignment is achieved, but localization efficiency deteriorates due to mismatched coordinate systems
Solution Approach 1:
The patent performs calibration in advance by establishing coordinate transformation parameters between different sensors using a calibration object. This preliminary calibration ensures that subsequent localization operations can directly use the pre-established coordinate relationships, improving localization efficiency without sacrificing measurement precision.
3Adaptability or versatility
If different sensor types are combined for environmental recognition, then adaptability improves, but coordinate system matching becomes more difficult
Solution Approach 1:
The patent creates a universal calibration approach that works across different sensor types (3D distance sensor and image sensor) by using a common calibration object with geometric features that can be detected by both sensor types. This universal method simplifies the coordinate transformation process despite the diversity of sensor combinations.
Data Source
AI summary
An apparatus and method capable of calculating a coordinate transformation parameter without having to utilize a rig are provided. The apparatus and method extract a first feature point based on a plane of first data, project the first feature point onto second data and then extract a second feature point from a part of the second data onto which the first feature point is projected. Then, calibration is performed based on the extracted feature points. Therefore, it is possible to perform calibration immediately as necessary without having to utilize a separate device such as a rig.