Vision Aided Navigation Feature Tracking via Localized Image Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Vision-based navigation systems face challenges in adapting to dynamic environmental conditions, such as changes in lighting or feature obscuration, which can lead to loss of tracked features and reduced navigation accuracy.
Innovation Solution
The system employs a dynamic localized parameter adjuster that applies targeted image enhancement techniques to specific regions of the image where tracked features are expected to appear, based on predicted locations and vehicle kinematics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If vision-based sensors are calibrated for predefined environmental conditions, then navigation accuracy is maintained under those conditions, but navigation performance degrades when environmental conditions change dynamically
Solution Approach 1:
The system dynamically adjusts image processing parameters based on real-time environmental conditions and predicted feature locations. The parameter adjuster modifies extraction parameters adaptively as the vehicle operates through changing environments, transitioning from static calibration to dynamic adaptation.
Solution Approach 2:
The system changes image processing parameters (such as feature extraction thresholds, contrast enhancement levels, and detection sensitivity) based on predicted environmental conditions and actual feature detection results, allowing the navigation system to maintain accuracy across varying environmental conditions.
2Reliability
If the system processes entire image frames to maintain feature tracking, then feature tracking reliability is improved, but processing time and computational resources increase
Solution Approach 1:
The system divides the image processing task by focusing computational resources only on bounded regions around predicted feature locations rather than processing entire image frames. This segmentation maintains tracking reliability while significantly reducing processing time and computational load.
Solution Approach 2:
The system applies enhanced processing quality (detailed feature extraction and analysis) only to local regions where features are expected to appear, rather than uniformly processing the entire image. This localized approach maintains tracking accuracy for critical features while reducing overall processing requirements.
3Loss of information
If features are lost due to environmental changes, then navigation data quality decreases, but system complexity increases if comprehensive fallback mechanisms are implemented
Solution Approach 1:
The system performs preliminary adjustments to image parameters based on predicted environmental conditions and vehicle kinematics before features are completely lost. This proactive approach maintains feature detectability and navigation data quality without requiring complex reactive fallback mechanisms.
Solution Approach 2:
The system uses feedback from feature detection results and environmental condition predictions to continuously adjust image processing parameters. This closed-loop approach maintains navigation data quality by adapting to changing conditions in real-time without requiring pre-programmed fallback procedures for every possible scenario.
Data Source
Figure 1
Figure 1A
Figure 2A
AI summary
In one embodiment, a vision aided navigation system comprises: at least one image sensor configured to produce image frames of a surrounding environment; a feature extractor configured to extract at least one image feature from a first image frame; a navigation filter configured to output a navigation solution based navigation data from a navigation device and changes in position of the image feature in the images; a feature tracker to receive the image frames and predict a location of the image feature in a subsequent image frame; a dynamic localized parameter adjuster to adjust at least one image parameter of the subsequent image frame; and wherein the feature tracker is configured so that when the image feature cannot be identified in the subsequent image frame within a bounded region around the predicted location, the dynamic localized parameter adjuster adjusts the at least one image parameter within the bounded region.