Vision Aided Navigation Using Image Registration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Navigation systems with inertial measurement units (IMUs) face errors such as gyroscope scale factor error, bias, and drift, which can accumulate and significantly affect the accuracy of vehicle positioning and orientation over time.
Innovation Solution
A navigation system that combines an image sensor with an IMU and a processing unit to estimate motion, calculate transformations, and apply these to produce predicted images for error compensation, using intensity-based image registration to refine IMU data and correct for errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Duration of action of moving object
If an IMU is used for navigation, then the system can provide continuous position and orientation data, but measurement errors accumulate over time causing significant positioning drift
Solution Approach 1:
The system uses visual feedback from image sequences to correct IMU measurement errors. The visual odometry algorithm continuously compares predicted image positions with actual captured images, generating correction signals that are fed back to compensate for IMU drift and accumulate errors over time.
Solution Approach 2:
The patent introduces visual odometry as an intermediary system that mediates between the IMU and the final navigation solution. The visual odometry algorithm processes image data to estimate motion, then uses this estimation to correct IMU measurements, acting as a bridge that transfers accurate position information to the navigation system.
2Measurement precision
If high-precision IMUs are used to maintain positioning accuracy, then measurement errors are reduced, but system cost and complexity increase
Solution Approach 1:
The system replaces expensive, high-precision IMUs with cheaper, lower-precision units by introducing visual odometry as a corrective mechanism. The visual system acts as a disposable or auxiliary component that compensates for the limitations of the inexpensive IMU, achieving high overall accuracy without requiring costly hardware.
Solution Approach 2:
The patent substitutes mechanical improvements (better IMU hardware) with a computational approach (visual odometry algorithm). Instead of relying on superior mechanical sensors, the system uses image processing and algorithmic correction to achieve the same positioning accuracy, reducing hardware complexity and cost.
3Measurement precision
If visual odometry is used to correct IMU errors, then positioning accuracy is maintained, but processing time and computational load increase
Solution Approach 1:
The system performs partial visual odometry processing by focusing on key image comparisons and motion estimations rather than exhaustive analysis. The algorithm selectively processes image data at critical moments or uses simplified comparison methods, achieving sufficient correction accuracy without requiring complete and exhaustive image processing.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A navigation system comprises an image sensor operable to obtain a first image at a first moment in time and a second image at a second moment in time; an inertial measurement unit (IMU) operable to obtain measurement data corresponding to the first and second moments in time; and a processing unit coupled to the image sensor and the IMU. The processing unit is operable to estimate motion between the first and second moments in time based on the measurement data from the IMU; calculate a plurality of transformations based on the estimated motion; apply each of the plurality of transformations to the first image to produce a plurality of predicted images; compare the second image to each of the plurality of predicted images; select the predicted image from the plurality of predicted images which most closely matches the second image; and compensate for error in the IMU measurement data based on the transformation corresponding to the selected predicted image.