SLAM Graph Thinning by Density-Based Node Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing size of SLAM graphs due to constant node and edge additions leads to higher computing demands and potential accuracy loss, necessitating a method to thin the graph while maintaining map accuracy.
Innovation Solution
A method involving the removal of nodes and edges based on scale-invariant density and geometric measures, such as the scale-invariant density measure and n-Group condition, to reduce the SLAM graph size while preserving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If nodes and edges are constantly added to the SLAM graph to improve mapping accuracy and environmental coverage, then the completeness and accuracy of the map is improved, but the computing demand and graph size increase significantly
Solution Approach 1:
The patent extracts and removes redundant nodes from the SLAM graph based on density criteria. Nodes with excessive density (too many nodes in close proximity) are identified and removed, along with their associated edges. This extraction process reduces graph size while preserving essential structural information needed for accurate mapping and localization.
Solution Approach 2:
The patent applies local quality by treating different regions of the SLAM graph differently. Instead of uniformly reducing the graph, it specifically targets high-density local regions for node removal while preserving low-density regions. This localized approach maintains map accuracy in critical areas while reducing complexity in redundant areas.
2Loss of information
If the SLAM graph size increases to improve map detail and coverage, then the environmental representation is improved, but the computational load and processing time increase
Solution Approach 1:
Redundant environmental information represented by dense clusters of nodes is extracted and removed. The patent identifies nodes that provide overlapping or duplicate environmental coverage and removes them, thereby reducing processing time while maintaining the essential environmental representation through the remaining nodes.
3Productivity
If nodes are removed to reduce graph size and computing demand, then the computational efficiency is improved, but the map accuracy may deteriorate
Solution Approach 1:
The patent ensures that node removal is performed selectively in high-density regions where redundancy exists, while preserving nodes in low-density regions that are critical for map accuracy. This localized quality control prevents accuracy deterioration in important areas while achieving computational efficiency gains through targeted node removal in redundant areas.
Solution Approach 2:
The patent employs a feedback mechanism where the density of nodes is continuously monitored and evaluated. Nodes are removed based on density thresholds, and the process can be iterated to maintain optimal graph density. This feedback-based approach ensures that map accuracy is preserved while achieving the desired computational efficiency.
Data Source
Figure 1a
Figure 1b~1c
Figure 2a~2b
AI summary
The invention relates to a method for thinning out a SLAM graph (210) which is used to operate a mobile device (110) and has a multiplicity of nodes (220) and a multiplicity of edges (230, 232, 234) which each end with an end point at a node (220), wherein the SLAM graph (210) is obtained (150), nodes (220) are removed (152) from the SLAM graph (210) and an updated SLAM graph (212) is then output (156), wherein the removal (152) of a node (220) comprises: determining (160) that node of the multiplicity of nodes (220) with the highest scale-invariant density at other nodes (220) around this node (220), and removing (162) the determined node (220) with the highest density from the SLAM graph (210).