Self-Position Estimation Using Dynamic Object Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In autonomous driving and SLAM technologies, self-position estimation is challenged by dynamic objects not included in map information, leading to low robustness and accuracy issues when their movable portions change form.
Innovation Solution
A self-position estimation device that includes a peripheral object information acquisition unit, a storage unit for map and dynamic object information, and a dynamic object determination unit to identify dynamic objects by collating peripheral object information with fixed and movable portion information, allowing for accurate position estimation by excluding dynamic object portions from the map information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If peripheral object information is collated with map information for self-position estimation, then position estimation can be performed, but position estimation accuracy deteriorates when dynamic objects exist around the moving body
Solution Approach 1:
The patent segments dynamic object information into fixed portion information (which remains collatable with map information) and movable portion information (which is excluded from collation). This segmentation allows the system to maintain position estimation accuracy by using only the fixed portions of dynamic objects that are present in the map, while excluding the movable portions that would cause estimation errors.
Solution Approach 2:
The patent extracts and excludes the movable portion information of dynamic objects from the peripheral object information before collating with map information. By taking out the problematic movable portions that change form and are not in the map, the system prevents these portions from degrading position estimation accuracy while still utilizing the fixed portions for reliable estimation.
2Measurement precision
If dynamic object information is excluded from peripheral object information to improve position estimation, then position estimation accuracy improves, but the ability to handle dynamic objects with changing forms deteriorates
Solution Approach 1:
The patent segments dynamic object information into fixed and movable portions, allowing the system to adaptively handle dynamic objects by identifying and utilizing their fixed portions for collation while excluding their movable portions. This segmentation enables the system to maintain adaptability to various dynamic objects while ensuring accurate position estimation.
Solution Approach 2:
The patent applies different processing qualities to different parts of dynamic objects: fixed portions are processed with collation against map information to maintain adaptability, while movable portions are excluded from collation to ensure accuracy. This local quality approach allows the system to handle the complexity of dynamic objects with changing forms effectively.
3Productivity
If all peripheral object information is used for collation, then comprehensive position estimation is possible, but estimation reliability deteriorates due to inclusion of dynamic object portions
Solution Approach 1:
The patent extracts and removes the movable portion information of dynamic objects from the peripheral object information before collation with map information. This extraction process maintains comprehensive position estimation by utilizing all fixed portions while eliminating the unreliable movable portions that would degrade estimation reliability.
Solution Approach 2:
The patent dynamically processes peripheral object information by identifying and differentially treating fixed and movable portions of dynamic objects. This dynamic approach allows the system to maintain comprehensive estimation coverage while adaptively excluding only the portions that would compromise reliability, rather than excluding all dynamic object information.
Data Source
AI summary
A self-position estimation device includes a peripheral object information acquisition unit acquiring peripheral object information, a storage unit storing map information and dynamic object information, a dynamic object determination unit, and a self-position estimation unit. The dynamic object information includes fixed portion information and movable portion information. The dynamic object determination unit determines whether or not the dynamic object exists around the moving body by collating the peripheral object information with the fixed portion information, when the dynamic object determination unit determines that the dynamic object exists around the moving body, the self-position estimation unit estimates the position of the moving body in the map information by excluding the portion corresponding to the dynamic object from the peripheral object information and collating the map information with the peripheral object information.