Vision-Based Zero Velocity Update for GPS-Denied Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In GPS-denied navigation scenarios, monocular cameras with inertial measurement units (IMUs) face challenges in providing accurate position and velocity estimations due to signal drift, especially with low-cost IMUs struggling to reliably detect zero velocity or attitude rate updates in noisy sensor data.
Innovation Solution
A system that integrates a digital camera capable of capturing frames at 16 hertz or faster, with image processing functionality to extract features and determine zero velocity or attitude rate updates by comparing frames, using methods like Scale-Invariant Feature Transform (SIFT) or Speeded Up Robust Features (SURF) to identify stationary features and trigger updates, thereby reducing estimation errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If monocular camera with IMU is used for GPS-denied navigation, then position and velocity estimation is provided, but signal drift occurs due to inadequate depth information from monocular camera
Solution Approach 1:
The patent introduces an image processing system as an intermediary between the monocular camera and IMU that extracts depth information through feature matching and correlation analysis. This intermediary processes image frames to generate depth estimates that calibrate the IMU, resolving the depth information deficiency of monocular cameras while maintaining system integration.
Solution Approach 2:
The patent replaces the traditional mechanical inertial navigation system with a vision-based calibration approach. Instead of relying solely on IMU measurements which drift over time, the system uses optical flow and feature tracking from the monocular camera to provide continuous position and velocity updates, substituting mechanical inertial sensing with optical measurement methods.
2Device complexity
If low cost IMU is used, then device complexity is reduced, but reliability of zero velocity update detection deteriorates due to noisy sensor data
Solution Approach 1:
The patent introduces image processing functionality as an intermediary that verifies and validates IMU measurements. The system processes image frames to detect stationary features and confirms zero velocity events through visual evidence, providing a reliability check on low-cost IMU data without requiring expensive high-precision sensors.
Solution Approach 2:
The patent implements feedback loops where image processing results continuously validate and correct IMU measurements. The system monitors feature movement in image frames and provides feedback to confirm or reject zero velocity updates from the IMU, creating a closed-loop verification system that enhances reliability of low-cost inertial sensors.
3Stability of the object's composition
If vision based zero velocity update is implemented, then signal drift is reduced, but device complexity increases due to image processing requirements
Solution Approach 1:
The patent extracts only the essential image processing functionality needed for zero velocity detection, separating it from full computer vision processing. The system extracts key features such as stationary object detection and optical flow estimation without implementing complete scene understanding, reducing computational complexity while maintaining drift correction capability.
Solution Approach 2:
The patent applies partial image processing action by focusing only on specific image regions containing stationary features rather than processing the entire image frame. The system performs correlation analysis on selected regions of interest, reducing computational load while sufficient to detect zero velocity events and correct drift.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method comprises receiving a first frame from at least one imaging device, receiving a second frame from the at least one imaging device, analyzing at least a portion of the first frame and at least a portion of the second frame, and indicating when at least one of a zero velocity update and a zero attitude update should be performed based on at least in part on the analysis of the at least a portion of the first frame and the at least a portion of the second frame. The first frame is captured at a first vantage point and the second frame is captured at a second vantage point.