Heterogeneous Map Element Selection for Accurate SLAM Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing SLAM algorithms struggle to effectively utilize heterogeneous sensor data from different types of sensors, leading to inconsistencies in map representation of the same location, particularly when the environment has changed, as different sensors capture sensor data from the same type, the technical problem is to determine which map information to use for localization.
Innovation Solution
A mapping device that receives sensor data and selects map elements based on capturing time, type, and information difference metrics to update a heterogenous map, ensuring accurate localization and mapping using pairs of corresponding elements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If multiple heterogeneous map elements from different sensors are used to represent the same location, then the map contains more information and is more comprehensive, but determining which map information to use for localization becomes complex and inconsistent
Solution Approach 1:
The patent applies parameter changes by introducing an information difference metric that quantifies the divergence between heterogeneous map elements. This metric transforms the qualitative problem of choosing between different sensor representations into a quantitative decision based on measurable parameters including capturing time, sensor type, and information difference, thereby resolving the complexity of localization decisions while preserving comprehensive map information
Solution Approach 2:
The patent implements preliminary action by pre-calculating and storing the information difference metric for each pair of corresponding map elements from different sensors. This preprocessing step allows the localization algorithm to quickly retrieve and compare pre-computed metrics rather than performing complex comparisons in real-time, reducing computational complexity while maintaining information completeness
2Adaptability or versatility
If image-based and structure-based map elements are both stored in the heterogenous map, then the map can support multiple sensor types, but the computational complexity increases when selecting which elements to use
Solution Approach 1:
The patent handles sensor type compatibility by incorporating sensor type as a discrete parameter in the information difference metric calculation. This allows the system to systematically compare and select between image-based and structure-based map elements based on the querying device's sensor type, maintaining versatility while managing selection complexity through parameterized comparison
Solution Approach 2:
The patent implements universality by creating a unified heterogeneous map structure that can store and manage both image-based and structure-based map elements using a common framework. The information difference metric serves as a universal comparison mechanism that works across different sensor types, allowing the same map structure to serve multiple sensor modalities without requiring separate processing paths
3Reliability
If map elements from different capturing times are compared, then the map can be updated with newer information, but inconsistencies arise when the environment has changed
Solution Approach 1:
The patent addresses environmental changes by incorporating capturing time as a dynamic parameter in the information difference metric. When the environment has changed, map elements captured at different times will exhibit higher information difference values, allowing the system to identify and handle such cases appropriately. This time-aware approach maintains localization accuracy by accounting for environmental dynamics while preserving valuable historical information
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A mapping device (140) for supporting localization and mapping using a heterogenous map is provided. The mapping device is operative to receive a request (311) pertaining to localization using current sensor data which is image- based sensor data or structure-based sensor data, access (313) image-based map elements and structure-based map elements, each map element being associated with a capturing time indicating a time of capturing the sensor data based on which it was derived, identify (314) pairs of corresponding map elements, each pair comprising an image-based map element and a structure-based map element representing features at the same real-world location, and for each pair of corresponding map elements: determine (315) an information difference metric between the image-based map element and the structure-based map element, and select (316) one of the map elements of the pair of corresponding map elements for updating the heterogenous map based on a comparison of the respective capturing times, a type of the current sensor data, and the information difference metric.