Marker-Assisted Vehicle Positioning to Correct Visual-SLAM Drift
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In environments with varying backgrounds, such as factories and warehouses, it is challenging to accurately determine the current position of a moving body using a pre-created map, as errors in position calculation using Visual-SLAM methods cumulatively increase over time due to relative positioning without a stable reference.
Innovation Solution
A positioning apparatus that uses an image capturing system to extract feature points and markers, correcting relative positions and attitudes based on absolute positions and attitudes, ensuring accurate measurement by verifying the reliability of recognized markers for precise positioning and attitude calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If Visual-SLAM method is used to measure position of moving body, then positioning can be achieved without wireless transmitters, but measurement precision deteriorates due to cumulative errors over time
Solution Approach 1:
The patent introduces markers as intermediary objects placed in the environment to serve as reference points for position correction. The image capturing apparatus detects these markers to obtain absolute position information, which then corrects the cumulative relative position errors from Visual-SLAM. This mediator approach allows the system to maintain both the simplicity of Visual-SLAM and the accuracy of absolute reference systems.
Solution Approach 2:
The patent implements a feedback mechanism where the system continuously compares the relative position calculated by Visual-SLAM with the absolute position derived from marker detection. The difference (error) is fed back to correct the position estimate, preventing cumulative drift. This closed-loop control ensures long-term positioning accuracy while maintaining the autonomous operation of Visual-SLAM.
2Measurement precision
If markers are used to correct absolute position, then measurement precision improves, but device complexity increases due to marker placement requirements
Solution Approach 1:
The patent transforms the marker from a complex physical object into a simple visual pattern (AR marker) that can be easily captured by standard image cameras. By changing the marker's representation from a physical reference object to a two-dimensional visual code, the system reduces the complexity of marker placement while maintaining positioning accuracy. The markers can be printed on existing surfaces without requiring specialized hardware.
Solution Approach 2:
The patent uses visual copies (images) of markers instead of physical reference objects. The image capturing apparatus captures visual information of markers, and the system processes these image copies to determine position. This approach eliminates the need for complex physical marker structures and allows multiple markers to be placed easily on various surfaces throughout the environment.
Data Source
Figure 1
Figure 2~3
Figure 4~6
AI summary
A first calculator calculates relative position and attitude of a vehicle based on images captured by an image capturing apparatus on the vehicle. A second calculator extracts one marker from a captured image, calculates absolute position and attitude of the vehicle based on a position and an attitude of the one extracted marker, and calculates an angle of a direction of the vehicle 1 with respect to a normal line to a plane of the one marker 4 based on a position and an attitude of the vehicle 1 in a marker coordinate system. A corrector generates a corrected position and attitude, not using the absolute position and attitude calculated based on the marker for which the angle is not larger than a first threshold, but using the absolute position and attitude calculated based on the marker for which the angle is larger than the first threshold.