Feature Quality Filtering for Efficient Environment Mapping

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The density and size of localization data collected by mobile devices for SLAM and AR applications can slow down object identification and impact storage resources, as existing methods do not effectively manage data quality and redundancy.

Innovation Solution

Assigning quality values to features based on consistency, observations, and dynamic characteristics, and reducing localization data by removing low-quality features, non-visual sensor data, and geometrically compressing features, while maintaining high-quality mapping capabilities.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If localization data is collected densely for SLAM and AR applications, then mapping accuracy and object identification capability are improved, but processing speed decreases and storage resources are consumed

Engineering Contradiction:
Improvemapping accuracyVSAvoidprocessing speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent applies local quality by assigning different quality values to different features based on their individual characteristics (e.g., stability, observability, geometric properties). High-quality features that contribute more to mapping accuracy are retained, while low-quality features are removed. This selective retention maintains mapping precision while reducing the overall data volume that requires processing.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent changes the parameter of feature quality by introducing a quality metric that evaluates each feature's contribution to SLAM performance. By filtering features based on this quality parameter, the system reduces data density while preserving the most informative features, thereby maintaining accuracy while improving processing speed.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If localization data is collected densely for SLAM and AR applications, then mapping accuracy and object identification capability are improved, but storage resources are consumed

Engineering Contradiction:
Improvemapping accuracyVSAvoidstorage resources
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent evaluates each feature's local quality based on characteristics such as stability, observability, and geometric properties. Features with high quality values are retained for storage, while low-quality features are discarded. This selective approach maintains the accuracy required for reliable mapping while significantly reducing the total quantity of stored localization data.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent discards low-quality features that contribute minimally to mapping accuracy, thereby reducing storage requirements. The system recovers and retains only the high-quality features that are essential for maintaining SLAM and AR functionality, achieving efficient storage utilization without sacrificing performance.

Inventive Principle:
Principle #34Discarding and recovering

3Reliability

If all collected localization data is retained for processing, then comprehensive environmental mapping is achieved, but processing overhead and computational load increase

Engineering Contradiction:
Improvemapping completenessVSAvoidprocessing overhead
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent assigns quality values to features based on their individual characteristics, identifying which features provide the most reliable information for environmental mapping. By processing only high-quality features, the system maintains mapping completeness while reducing the computational load and energy consumption associated with processing all collected data.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent applies partial action by selectively processing only the necessary subset of localization data (high-quality features) rather than all collected data. This approach achieves sufficient mapping reliability without the excessive computational overhead of processing every data point, optimizing the balance between completeness and resource usage.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3335153B1Managing feature data for environment mapping on an electronic device
Publication Date: 2022.10.05 GOOGLE LLC
  • EP3335153B1 patent drawingFigure 1
  • EP3335153B1 patent drawingFigure 2~3
  • EP3335153B1 patent drawingFigure 4~5

AI summary

An electronic device [100] reduces localization data [230] based on feature characteristics identified from the data [225]. Based on the feature characteristics, a quality value can be assigned to each identified feature [343], indicating the likelihood that the data associated with the feature will be useful in mapping a local environment of the electronic device. The localization data is reduced by removing data associated with features have a low quality value, and the reduced localization data is used to map the local environment of the device [235] by locating features identified from the reduced localization data in a frame of reference for the electronic device.