SLAM Loop Closure with Inverted Key Frame Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing Simultaneous Localization and Mapping (SLAM) algorithm tends to miss key frames near inflection points of a curve motion trajectory, leading to large accumulative errors and difficulties in accurate positioning, and it suffers from redundant feature extraction and computation in densely textured scenes due to the use of both point and line features for pose estimation.
Innovation Solution
A closed-loop detecting method using an inverted index-based key frame selection strategy, which supplements missed key frames during curvilinear motion and selectively uses point or line features based on information entropy for pose tracking, reducing redundant feature extraction and computation by introducing an inverted indexing mode and conditional screening of line features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional forward indexing key frame selection is used, then the SLAM algorithm runs efficiently with standard computation, but key frames near inflection points of curve motion trajectory are missed leading to large accumulative errors
Solution Approach 1:
The patent introduces inverted indexing that processes image frames in reverse order (from newest to oldest) in addition to the traditional forward indexing. This bidirectional approach ensures that key frames near inflection points are not missed, as the inverted indexing can detect trajectory changes that forward indexing alone might overlook, thereby improving positioning accuracy and key frame selection reliability
Solution Approach 2:
The patent adds a temporal dimension to key frame selection by implementing both forward and inverted indexing strategies. This creates a multi-dimensional verification system where key frames are selected based on both forward progression and reverse verification, ensuring that inflection points are captured from multiple temporal perspectives, thus improving measurement precision without sacrificing reliability
2Reliability
If both point features and line features are extracted for pose estimation, then enough image features are obtained for reliable results, but redundant features and unnecessary computation occur in densely textured scenes
Solution Approach 1:
The patent applies local quality by adaptively selecting feature types based on scene characteristics. In densely textured scenes where point features provide sufficient information, only point features are extracted. In scenes lacking sufficient point features, line features are added. This localized adaptation of feature extraction strategy maintains pose estimation reliability while improving processing efficiency by avoiding redundant feature extraction
Solution Approach 2:
The patent dynamically changes the feature extraction parameters based on scene texture density. When the scene is determined to be densely textured, the system switches to point-feature-only mode. When the scene lacks sufficient texture, it switches to combined point and line feature mode. This parameter change strategy optimizes the balance between reliability and productivity by matching feature extraction complexity to scene requirements
Data Source
AI summary
A closed-loop detecting method using an inverted index-based key frame selection strategy, storage media and apparatus are provided. The method includes following steps: step I: acquiring image information at a current position, processing the image information to extract corresponding image features therefrom and solve a camera pose; step II: capturing image features successively during movement of a robot, as consecutive image frames, performing, on the consecutive image frames, an indexing, in which the inverted index-based key frame selection strategy is introduced into a key frame selection strategy to supplement key frames which are prone to be missed in a conventional forward indexing during a curvilinear movement of the robot; and step III: performing closed-loop detection and correction of accumulative errors based on image features carried by key frames.