Monocular Visual SLAM Initialization Using Inertial Odometry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing VSLAM systems face high failure rates and initialization delays due to lack of initial data and adverse robot motion, especially in feature-based Monocular VSLAM, which are sensitive to image quality and require complex heuristic algorithms and frequent parameter tuning.
Innovation Solution
The system uses an Inertial Measurement Unit (IMU) to detect friendly motion, triggering the initialization process only when such motion is detected, and employs a simpler, deterministic approach that does not rely on RANSAC-based Fundamental and Essential matrix fitting, with Global Bundle Adjustment used after successful VO initialization to generate an initial map.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If VSLAM systems use complex heuristic algorithms and RANSAC-based Fundamental and Essential matrix fitting for initialization, then they may achieve better robustness to some extent, but they result in high computation complexity, random initialization results, and require frequent parameter tuning
Solution Approach 1:
The patent extracts and removes the complex RANSAC-based Fundamental and Essential matrix fitting procedures from the initialization pipeline. Instead, it uses a simplified approach that directly leverages IMU data to compute relative poses and initialize the map, eliminating unnecessary computational steps while maintaining robustness.
Solution Approach 2:
The patent replaces the visual-only RANSAC-based geometric fitting mechanism with an inertial-mechanical based approach. By using IMU data to directly compute relative poses and guide initialization, the system substitutes complex visual feature matching and RANSAC iterations with more direct inertial measurement-based pose estimation.
2Productivity
If VSLAM systems perform initialization without screening robot motion, then they can start immediately, but they suffer from high failure rates due to unfriendly motions such as shaking and turning
Solution Approach 1:
The patent performs preliminary screening of robot motion using IMU data before initiating the VSLAM initialization process. By detecting friendly motion patterns in advance, the system ensures that initialization only starts when conditions are favorable, preventing initialization failures due to shaking or turning while maintaining quick startup when motion is appropriate.
3Ease of manufacture
If feature-based Monocular VSLAM uses only a single camera to reconstruct 3D scenes, then it reduces hardware cost, but it makes initialization more severe and more sensitive to image quality and environment
Solution Approach 1:
The patent introduces IMU data as an intermediary to bridge the limitations of monocular vision. The inertial measurements provide additional constraints and information that compensate for the lack of depth information from a single camera, making initialization more robust and less sensitive to image quality and environmental features.
Data Source
AI summary
A method and system for capturing, by a camera a sequence of frames at respective locations within a portion of an environment; capturing, by an inertial measurement unit, a sequence of inertial odometry data corresponding to the sequence of frames at the respective locations; storing in a queue a data record includes information extracted from processing the respective frame and information from the inertial measurement unit; in accordance with a determination that the sequence of inertial odometry data satisfies a first criterion: calculating a first relative pose between the first frame and the second frame; and in accordance with a determination that a difference between the first relative pose and the information extracted from processing the respective frame satisfy a first threshold: generating an initial map of the portion of the environment based on the first data record and the second data record.