Loop Closure Confidence Scoring for SLAM Robustness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Pose graph SLAM systems face challenges in maintaining accuracy and robustness due to high rates of false-positive loop closures, especially when using less informative sensors, which can lead to degraded localization and map consistency.
Innovation Solution
A method that generates confidence scores for loop closures through pairwise consistency tests and creates an augmented graph by retaining or deleting loop closures based on these scores, while also injecting artificial loop closures to reinforce consistent ones, thereby enhancing the robustness of the system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If loop closure detection is performed using less informative sensors, then the system can operate in emerging applications, but the number of false-positive loop closures increases significantly
Solution Approach 1:
The patent segments loop closures into two categories: geometric loop closures (based on spatial consistency of nodes) and non-geometric loop closures (based on sensor observations). This segmentation allows the system to handle different types of loop closures differently, improving robustness when using less informative sensors by relying more on geometric consistency which is less susceptible to false positives.
Solution Approach 2:
The patent introduces a consistency check mechanism as an intermediary between loop closure detection and graph optimization. This intermediary verifies whether loop closures are geometrically consistent before incorporating them into the pose graph, acting as a filter that reduces false-positive loop closures while maintaining system adaptability to various sensor types.
2Ease of manufacture
If existing back-end optimisers are used, then the system is simple to implement, but it fails when the proportion of false-positive loop closures exceeds a certain threshold
Solution Approach 1:
The patent performs preliminary consistency checks on loop closures before they are used in graph optimization. By pre-verifying geometric consistency and categorizing loop closures, the system prepares the data in advance to ensure that only reliable loop closures are fed to the optimizer, preventing failure when false-positive rates are high while maintaining implementation simplicity.
Solution Approach 2:
The patent changes the parameters used for loop closure validation by introducing geometric consistency checks and confidence score thresholds. These parameter changes enable the system to distinguish between reliable and unreliable loop closures, improving optimization robustness without significantly complicating the implementation of the back-end optimizer.
3Quantity of substance
If all detected loop closures are included in the pose graph, then the graph is dense and provides good coverage, but false-positive loop closures degrade localization accuracy
Solution Approach 1:
The patent applies local quality by differentiating the treatment of loop closures based on their geometric consistency and confidence scores. High-confidence, geometrically-consistent loop closures are retained to maintain graph density, while low-confidence or inconsistent loop closures are removed to preserve localization accuracy. This selective retention ensures each part of the graph has appropriate quality characteristics.
Solution Approach 2:
The patent creates a simplified representation of loop closure relationships through geometric consistency checks, copying the essential spatial relationships into a normalized format. This copying process allows the system to maintain graph density by preserving valid spatial constraints while filtering out false-positive loop closures that would degrade accuracy.
Data Source
AI summary
A computer-implemented method 1000 of constructing a model of the motion of a mobile device, wherein the method comprises using a sensor of the device to obtain 1002 positional data providing an estimated pose of the mobile device, generating an initial graph 1004 based upon the positional data from the sensor, nodes of which graph provide a series of possible poses of the device, and edges of which graph represent odometry and/or loop closure constraints; processing the graph to estimate 1006 confidence scores for each loop closure by performing pairwise consistency tests between each loop closure and a set of other loop closures; and generating an augmented graph from the initial graph by retaining or deleting 1008 each loop closure based upon the confidence scores.