SLAM Loop Closure Using Hierarchical Spatial Feature Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing SLAM systems face challenges in maintaining accurate and reliable loop closure due to camera calibration errors and limited feature matching accuracy, leading to cumulative errors and inefficient global map optimization.
Innovation Solution
A method involving 3D point set clustering, hierarchical spatial feature matching, and adaptive bundle adjustment (IBA/FBA) to estimate a relative pose between images, optimizing the global map through hierarchical spatial feature matching (HSFM) and hybrid bundle adjustment (HBA).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If feature matching and relative pose calculation are used to optimize global map, then loop closing functionality is achieved, but visual variations are large and optimization time is long
Solution Approach 1:
The patent segments the feature matching process into multiple layers (coarse matching and fine matching). Coarse matching uses clustered 3D point sets to quickly identify potential matches, while fine matching refines these matches. This segmentation reduces the overall optimization time by handling the computationally intensive parts efficiently while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary clustering of 3D point sets before feature matching. By organizing points into clusters in advance, the system prepares data structures that enable faster matching during loop closing operations, reducing the time required for optimization while maintaining reliability.
2Measurement precision
If camera calibration and feature matching are performed, then localization accuracy is achieved, but cumulative errors occur over time
Solution Approach 1:
The patent implements a feedback mechanism where the loop closing module continuously monitors and corrects cumulative drift by identifying loop closures and optimizing the global map. The relative pose estimation between query and search images provides feedback that is used to adjust and correct the global map, preventing error accumulation over time.
Solution Approach 2:
The system performs preliminary stereo matching to generate accurate 3D point sets before feature matching. This preliminary action ensures that the spatial relationships are established with high precision early in the process, reducing the propagation of calibration errors through subsequent matching operations.
3Measurement precision
If 3D point set clustering is performed, then feature matching accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the feature matching process into coarse and fine stages. Coarse matching uses clustered 3D point sets to quickly identify potential matches, reducing the search space. Fine matching then refines these matches with higher precision. This segmentation improves accuracy while managing computational complexity by distributing the workload across different levels of detail.
Solution Approach 2:
The patent applies clustering to only the most critical aspects of feature matching (coarse matching stage) rather than processing all features with full precision. This partial application of complex operations maintains matching accuracy for key features while reducing overall computational complexity.
Data Source
AI summary
A method performed by an electronic device includes: acquiring a search image based on a query image; acquiring first spatial features of the query image and second spatial features of the search image; and estimating a relative pose between the query image and the search image based on the first spatial features and the second spatial features.