Ground Vehicle VINS with Odometry Constraints for Scale-Aware Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional vision-aided inertial navigation systems (VINS) face challenges in accurately determining position and orientation due to unobservable directions, particularly scale, when navigating in environments with restricted motion, leading to inaccurate state estimates and increased localization errors.
Innovation Solution
Integrating odometry data and motion manifold information to generate geometric constraints, and using a sliding window filter to compute state estimates, thereby accounting for unobservable directions and reducing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional VINS uses only camera and IMU data, then the system structure remains simple, but localization accuracy deteriorates due to unobservable scale information
Solution Approach 1:
The patent combines multiple sensing modalities (camera, IMU, and odometry unit) into a unified VINS framework. The odometry unit provides complementary scale information that compensates for the unobservable scale in pure visual-inertial systems, thereby improving localization accuracy without requiring complex additional hardware beyond standard sensor fusion architecture.
Solution Approach 2:
The odometry unit serves multiple functions: it provides scale information for metric localization, constrains the state estimation in GPS-denied environments, and works synergistically with the camera and IMU data. This multi-functional component enables the system to achieve accurate localization across diverse motion scenarios without requiring separate specialized systems.
2Measurement precision
If VINS maintains states for all observed features, then positioning accuracy improves, but computational complexity increases
Solution Approach 1:
The patent segments the feature set into two distinct categories: features whose states are maintained in the state vector, and features used solely for computing constraints. This segmentation allows the system to process all observed features for positioning accuracy while maintaining a compact state vector that excludes redundant feature position states, thereby reducing computational complexity in the estimation algorithm.
Solution Approach 2:
The patent extracts feature position information from the state vector and represents it through constraints instead. By taking out the explicit feature position states from the state vector and encoding their information through geometric constraints between poses, the system maintains positioning accuracy while significantly reducing the dimensionality of the state space that requires active estimation and updating.
3Productivity
If VINS uses sliding window filter with feature constraints, then real-time performance improves, but accuracy in restricted motion environments deteriorates
Solution Approach 1:
The patent merges odometry constraints with visual and inertial measurements in the sliding window filter framework. This combination provides additional observability for scale and position in restricted motion environments, compensating for the limitations of visual-inertial data alone while maintaining real-time performance through the efficient sliding window optimization structure.
Solution Approach 2:
The system incorporates feedback from odometry measurements into the state estimation process. The odometry-derived constraints provide feedback on scale and position that corrects drift accumulation in the visual-inertial integration, particularly during restricted motion sequences where visual features alone provide insufficient observability, thereby maintaining accuracy without sacrificing real-time performance.
Data Source
AI summary
A vision-aided inertial navigation system (VINS) comprises an image source for producing image data along a trajectory. The VINS further comprises an inertial measurement unit (IMU) configured to produce IMU data indicative of motion of the VINS and an odometry unit configured to produce odometry data. The VINS further comprises a processor configured to compute, based on the image data, the IMU data, and the odometry data, state estimates for a position and orientation of the VINS for poses of the VINS along the trajectory. The processor maintains a state vector having states for a position and orientation of the VINS and positions within the environment for observed features for a sliding window of poses. The processor applies a sliding window filter to compute, based on the odometry data, constraints between the poses within the sliding window and compute, based on the constraints, the state estimates.