Feature Point Integration for Visual SLAM Stability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional visual SLAM systems face instability and poor positioning accuracy, especially in environments with severe variations such as high light contrast and corners, leading to frequent position loss and prolonged re-localization times.
Innovation Solution
A feature point integration positioning system that combines machine vision and deep learning to generate and integrate first and second feature points, enhancing positioning stability and accuracy by compensating for limitations in machine vision detection with deep learning capabilities, particularly in environments with high light contrast and variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional visual SLAM is used, then the system is simple and cost-effective, but the positioning stability deteriorates leading to frequent position loss
Solution Approach 1:
The patent combines conventional machine vision detection with deep learning detection to create an integrated feature point detection system. The machine vision detecting unit generates first feature points while the deep learning detecting unit generates second feature points, and these are integrated by the integrating unit to produce comprehensive feature point information that maintains positioning stability in varying environments.
Solution Approach 2:
The system uses a composite approach by integrating two different detection methodologies (machine vision and deep learning) similar to how composite materials combine different properties. This hybrid detection system leverages the strengths of both methods to achieve reliable positioning without sacrificing system simplicity.
2Productivity
If conventional visual SLAM is used, then the system operates quickly, but the positioning accuracy deteriorates in environments with severe variations
Solution Approach 1:
The patent merges machine vision detection and deep learning detection to simultaneously achieve fast processing and high accuracy. The machine vision unit provides rapid feature point detection while the deep learning unit enhances accuracy in challenging environments, and their integrated results maintain both speed and precision.
Solution Approach 2:
The system applies different detection qualities to different situations: machine vision provides standard-speed detection for normal conditions, while deep learning provides enhanced accuracy for severe environment variations. The integrating unit combines these local quality differences to achieve overall high performance.
3Ease of manufacture
If conventional visual SLAM is used, then the system is easy to implement, but it loses position in situations with high light contrast and corners
Solution Approach 1:
The patent integrates machine vision detection and deep learning detection to compensate for each other's weaknesses. When machine vision fails in high light contrast or corner situations, the deep learning detecting unit generates alternative feature points, ensuring position retention without significantly complicating implementation.
4Use of energy by moving object
If conventional visual SLAM is used, then the system requires minimal resources, but it spends too much time to find the original position after losing position
Solution Approach 1:
The system combines machine vision and deep learning detection to maintain continuous, reliable feature point tracking. This prevents position loss in the first place, eliminating the need for time-consuming re-localization procedures and reducing overall resource consumption by avoiding repeated positioning attempts.
Data Source
AI summary
A feature point integration positioning system includes a moving object, an image input source, an analyzing module and a positioning module. The image input source is disposed at the moving object to shoot an environment for obtaining a sequential image dataset. The analyzing module includes a machine vision detecting unit configured to generate a plurality of first feature points in each of the images based on each of the images, a deep learning detecting unit configured to generate a plurality of second feature points in each of the images based on each of the images, and an integrating unit configured to integrate the first feature points and the second feature points in each of the images into a plurality of integrated feature points in each of the images. The positioning module confirms a position of the moving object relative to the environment at each of the time points.