Mobile Robot Self-Localization Using 3D Geometric Feature Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing self-localization methods for mobile bodies in environments with large amounts of three-dimensional point group data suffer from prolonged processing times.
Innovation Solution
A self-localization device that estimates self-location using geometric features such as planes, straight lines, and spheres, reducing the number of combinations required by selecting and matching geometric features rather than three-dimensional points, and evaluating coincidence between current and past maps using sphere groups.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the mobile body estimates self-location by matching a plurality of three-dimensional points, then the estimation accuracy is improved, but the processing time becomes long
Solution Approach 1:
The patent extracts only the essential geometric features (planes, straight lines, spheres) from the complete set of three-dimensional points. By taking out only the necessary geometric elements rather than processing all points, the system maintains self-location estimation accuracy while significantly reducing the data volume and processing time required for matching current and past maps.
Solution Approach 2:
The patent creates simplified geometric feature representations (copies) of the three-dimensional point data. Instead of directly matching raw three-dimensional points, the system generates geometric feature models (planes, lines, spheres) that replicate the essential spatial structure, enabling faster comparison and matching operations while preserving the necessary accuracy for self-location estimation.
2Measurement precision
If the number of three-dimensional points used for matching is increased, then the estimation accuracy is improved, but the computational complexity increases
Solution Approach 1:
The patent segments the complex task of matching three-dimensional points into distinct geometric feature extraction and matching stages. By dividing the processing into separate steps (extracting planes, straight lines, and spheres independently), the system reduces computational complexity while maintaining accuracy, as each geometric feature type can be processed using specialized algorithms rather than general point-cloud matching.
Solution Approach 2:
The patent changes the parameters of the data representation from raw three-dimensional point coordinates to geometric feature parameters (plane equations, line equations, sphere equations). This parameter transformation reduces the dimensionality and complexity of the data, making computational operations more efficient while preserving the essential geometric information needed for accurate self-location estimation.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
[Problem] To provide a self-localization device that can reduce the processing time even when estimating the self-location in an environment where the number of pieces of three-dimensional point group data is large. [Solving Means] A self-localization device 2 includes an estimation unit 5 that estimates a self-location of a mobile robot 1. The estimation unit 5 includes: a map generation unit 51 that generates a map of surroundings of the mobile robot 1 based on three-dimensional point group data detected by a detection unit 4; a geometric feature extraction unit 53 that, based on the map, extracts geometric features from a current map and extracts geometric features from a past map; a self-location calculation unit 54 that selects the geometric features extracted by the geometric feature extraction unit 53 as sets of geometric features and calculates self-locations in the past map based on the set of geometric features; and a self-location evaluation unit 55 that evaluates a degree of coincidence between the current map and the past map for each set of geometric features based on the self-locations calculated by the self-location calculation unit 54 and selects a self-location with a high degree of coincidence.