SLAM Map Update Using Common Object 3D Data Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Outdoor SLAM systems face challenges in tracking features due to dynamic elements like pedestrians, vehicles, trees, and clouds, as well as the infinite expanse of the skyline, which reduces the availability of trackable features.
Innovation Solution
An electronic device equipped with a SLAM module, a camera circuit, and a processor that constructs a SLAM map by capturing environmental images, extracting feature points, aligning common object 3D data with the images, and updating the SLAM map accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional SLAM systems are used outdoors, then the system can operate in outdoor environments, but the tracking accuracy deteriorates due to dynamic elements and limited trackable features
Solution Approach 1:
The patent segments the outdoor environment into static background elements and dynamic foreground objects. By identifying and excluding dynamic elements (pedestrians, vehicles, moving trees) from feature tracking, the system maintains reliable tracking using only static features. This segmentation approach resolves the contradiction by isolating harmful dynamic factors from the tracking process.
Solution Approach 2:
The system performs preliminary classification of environmental features to identify static versus dynamic elements before tracking begins. By pre-processing the environment to distinguish trackable static features from non-trackable dynamic objects, the system establishes a reliable foundation for accurate tracking despite the presence of harmful dynamic factors.
2Measurement precision
If feature tracking is performed in outdoor environments, then the system can locate the device, but the availability of trackable features deteriorates due to infinite skyline and lack of end walls
Solution Approach 1:
The patent changes the parameters of feature selection by adapting to outdoor-specific characteristics. Instead of relying on indoor-like vertical structures (end walls), the system identifies and utilizes alternative static features appropriate for outdoor environments such as distant landmarks, building rooftops, and fixed infrastructure elements. This parameter adaptation resolves the contradiction by finding sufficient trackable features despite the infinite skyline condition.
Solution Approach 2:
The system introduces an intermediary classification layer that mediates between the unlimited outdoor space and the limited tracking capabilities. This intermediary process identifies and selects suitable static features from the complex outdoor scene, effectively bridging the gap between the vast environment and the device's tracking requirements.
3Reliability
If the system processes all features in the environmental image, then comprehensive tracking is achieved, but the processing complexity increases due to dynamic elements and variations
Solution Approach 1:
The patent extracts and separates dynamic elements from the feature set before processing. By removing pedestrians, vehicles, and moving objects from consideration, the system reduces processing complexity while maintaining tracking robustness through focus on stable static features. This extraction approach resolves the contradiction by eliminating unnecessary processing of harmful dynamic factors.
Solution Approach 2:
The system performs partial processing by selectively tracking only static features rather than all detected features. This partial action approach reduces computational complexity while maintaining sufficient tracking robustness, as it processes only the essential static elements needed for reliable device localization in outdoor environments.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An electronic device is disclosed. The electronic device includes a memory, a camera circuit, and a processor. The memory is configured to store a physical map and a database. The database includes several predefined common object images corresponding several common object 3D data. The camera circuit is configured to capture an environmental image. The processor is configured to: obtain a first common object image from the environmental image; extract several feature points from the environmental image; adjust several first feature points of the feature points when a first common object 3D data of the several common object 3D data is aligned to the first common object image; and update the SLAM map according to the plurality of feature points of the environmental image.