Feature Map Management via Stationary Dynamic Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Maintaining and managing feature point maps in virtual environments is challenging due to the complexity of updating feature points from multiple sources and the need to differentiate between stationary and dynamic objects.
Innovation Solution
A method for managing feature maps that involves receiving feature points from multiple recording devices, determining whether to update existing feature points based on characteristics such as time and object classification, and selectively updating or adding feature points to the map, ensuring that only relevant and accurate data is stored.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feature points from multiple recording devices are continuously updated in the feature map, then the relevance and accuracy of the feature map is improved, but the complexity of managing and differentiating between stationary and dynamic objects increases
Solution Approach 1:
The patent applies dynamics by classifying feature points as either stationary or dynamic based on their temporal characteristics. Stationary feature points are identified by comparing feature points across multiple time points, and only updates to stationary feature points are accepted. This dynamic classification system resolves the contradiction by automatically adapting the update behavior based on the nature of each feature point, improving accuracy while managing complexity through automated differentiation.
Solution Approach 2:
The patent changes the parameter of time by comparing timestamps of feature points from different recording devices and time periods. By analyzing temporal parameters and identifying whether feature points represent stationary objects (consistent across time) or dynamic objects (changing over time), the system selectively updates the feature map. This parameter-based differentiation resolves the contradiction by using time as a discriminative feature to manage update complexity.
2Reliability
If all feature points are updated frequently to maintain current accuracy, then the relevance of the feature map is improved, but the computational resources and time required for processing increase
Solution Approach 1:
The patent applies partial action by selectively updating only stationary feature points in the feature map while leaving dynamic feature points unchanged. Instead of updating all feature points from multiple recording devices, the system performs partial updates based on the stationary/dynamic classification. This resolves the contradiction by reducing the amount of processing required (saving time) while maintaining reliability for the critical stationary features that form the stable structure of the environment.
Solution Approach 2:
The patent extracts and separates stationary feature points from dynamic feature points through temporal comparison. By taking out only the stationary feature points for update processing and excluding dynamic ones, the system reduces computational overhead. This extraction approach resolves the contradiction by isolating the essential updates needed for map reliability while avoiding unnecessary processing of transient dynamic elements.
3Quantity of substance
If feature points from different time periods are integrated into a single feature map, then the completeness of the feature map is improved, but the stability of the feature map deteriorates due to changes in dynamic objects
Solution Approach 1:
The patent applies dynamics by implementing a temporal classification mechanism that distinguishes between stationary and dynamic feature points based on their behavior across different time periods. By dynamically identifying which feature points represent stable structures versus transient objects, the system integrates feature points from different times selectively. This resolves the contradiction by maintaining completeness through integration while preserving stability by excluding dynamic elements that would otherwise destabilize the feature map composition.
Data Source
AI summary
This disclosure relates to managing a feature point map. The managing can include adding feature points to the feature point map using feature points captured by multiple devices and/or by a single device at different times. The resulting feature point map may be referred to as a global feature point map, storing feature points from multiple feature point maps. The global feature point map may be stored in a global feature point database accessible by a server so that the server may provide the global feature point map to devices requesting feature point maps for particular locations. In such examples, the global feature point map may allow a user to localize in locations without having to scan for feature points in the locations. In this manner, the user can localize in locations in which the user has never previously visited.