Visual Inertial Odometry Localization With Optical Flow Drift Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing localization systems for moving devices, such as unmanned vehicles, face challenges in accurately determining position and orientation in GPS-denied environments due to drift issues and the need for photo-consistency in traditional visual inertial odometry methods.
Innovation Solution
A camera-based localization system that fuses visual and inertial measurement unit (IMU) data using a fast visual inertial odometry method, which estimates linear and angular velocities through optical flow and refinement via quadratic optimization, and reduces drift by utilizing key frames, allowing for accurate pose estimation without relying on photo-consistency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional visual inertial odometry methods are used, then localization can be achieved, but drift issues occur and photo-consistency requirements increase system complexity
Solution Approach 1:
The patent segments the localization process into distinct modules: optical flow computation, angular/linear velocity estimation, and pose calculation. By separating these functions, the system avoids the need for photo-consistency checks while maintaining localization accuracy, thus reducing system complexity.
Solution Approach 2:
The patent replaces the traditional mechanical approach of photo-consistency verification with an optical flow-based method. Instead of comparing full image sequences to ensure consistency, the system uses optical flow fields to estimate motion, simplifying the localization process while maintaining precision.
2Productivity
If fast visual inertial odometry is used, then processing speed improves, but drift accumulation increases
Solution Approach 1:
The patent implements a feedback mechanism where the estimated pose from fast visual inertial odometry is continuously refined using optical flow information. The system uses the estimated angular and linear velocities to predict motion, then corrects these estimates based on optical flow measurements, preventing drift accumulation while maintaining high processing speed.
Solution Approach 2:
The patent performs preliminary estimation of angular and linear velocities using IMU data and optical flow before final pose calculation. This preliminary action allows the system to quickly obtain initial pose estimates and then refine them, achieving both speed and accuracy without drift accumulation.
3Reliability
If optical flow and velocity estimation are separated, then localization robustness improves, but computational steps increase
Solution Approach 1:
The patent merges the optical flow computation and velocity estimation into a unified framework where both are derived from the same optical flow field and IMU measurements. This integration reduces the number of independent computational steps while maintaining the robustness benefits of separating optical flow and velocity estimation.
Data Source
AI summary
A camera-based localization system is provided. The camera-based localization system may assist an unmanned vehicle to continue its operation in a GPS-denied environment with minimal increase in vehicular cost and payload. In one aspect, a method, a computer-readable medium, and an apparatus for localization via visual inertial odometry are provided. The apparatus may construct an optical flow based on feature points across a first video frame and a second video frame captured by a camera of the apparatus. The apparatus may refine the angular velocity and the linear velocity corresponding to the second video frame via solving a quadratic optimization problem constructed based on the optical flow, the initial values of the angular velocity and the linear velocity corresponding to the second video frame. The apparatus may estimate the pose of the apparatus based on the refined angular velocity and the refined linear velocity.