Mobile Robot Global Localization Using Submap Feature Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing global localization methods for mobile robots face challenges due to sensor noise and map building errors, leading to mismatches between 2D occupancy grid maps and real spaces, which affect the accuracy of robot positioning.
Innovation Solution
A method that divides a global map into query submap images, calculates histogram values for geometric features, reflection symmetry scores, and similarity scores to determine the most similar submap image, enabling robust global localization based on geometric and structural feature information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a 2D occupancy grid map is used for global localization, then the method can reuse algorithms from computer vision, but mismatches occur between the map and real space due to sensor noise or map building errors
Solution Approach 1:
The patent divides the global map into multiple submaps, each representing a local area. This segmentation allows the system to process and compare smaller map regions independently, reducing the impact of noise and errors in any single area while maintaining the ability to reuse computer vision algorithms for each submap comparison.
2Measurement precision
If feature point extraction methods like SIFT or SURF are used, then object recognition capability is improved, but the method does not address mismatches caused by sensor noise and map building errors
Solution Approach 1:
The patent combines multiple feature extraction methods (geometric features from histogram analysis and structural features from reflection symmetry detection) to create a more robust localization system. This merging of complementary approaches addresses the limitations of using any single method, particularly improving robustness to sensor noise and map building errors while maintaining feature recognition accuracy.
Data Source
AI summary
A method for global localization of a mobile robot is disclosed. The method compares a histogram value, obtained by quantifying geometric and structural features of each query submap image divided from a global map image, with geometric and structural features of submap images stored in a database to select a submap image which is the most similar to a query submap image, and performs the global localization of the mobile robot, based on coordinate information included in the selected submap image.