Stable Feature Point Extraction for Autonomous Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing self-position estimation techniques for autonomous movement, such as automatic driving and robot travel control, face instability due to the use of all feature points from scenic images, which requires long-time observation and complicates the extraction of stable feature points.
Innovation Solution
An information processing apparatus that extracts stable feature points by calculating context feature amounts with preset conditions and selecting feature points with a product of context feature amounts equal to or greater than a threshold, without relying on long-time observation, using a processor and memory to facilitate this process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If all feature points from scenic images are used for self-position estimation, then the quantity of feature points increases, but the stability of estimation deteriorates due to fluctuations and occlusions
Solution Approach 1:
The patent extracts only the stable feature points from the set of all feature points by calculating context feature amounts and comparing them against thresholds. This selective extraction removes unstable feature points that would otherwise degrade estimation reliability, while maintaining a sufficient quantity of reliable feature points for accurate self-position estimation.
2Reliability
If long-time observation is used to identify stable feature points, then the stability of feature points improves, but the time required for processing increases
Solution Approach 1:
The patent performs preliminary calculation of context feature amounts for each feature point based on preset conditions (such as position in shot space, movement amount, and appearance probability). This preliminary analysis allows the system to identify stable feature points in real-time without requiring long-time observation, thus maintaining stability while reducing processing time.
3Measurement precision
If context feature amounts are calculated with multiple preset conditions, then the precision of feature-point selection improves, but the device complexity increases
Solution Approach 1:
The patent segments the evaluation of feature point stability into multiple independent context feature amounts, each corresponding to a specific preset condition (position, movement, appearance). By dividing the complex evaluation into separate calculable components, the system achieves high selection precision while keeping the computational process manageable and structured.
Data Source
AI summary
An information processing apparatus includes: a memory; and a processor coupled to the memory and configured to: acquire feature points from a shot image; calculate, from each of the acquired feature points, with preset conditions, a plurality of context feature amounts that has a success-or-failure vector of a single-attribute; and extract, from among the acquired feature points, feature points of which a product of the plurality of context feature amounts is equal to or greater than a preset threshold.