SLAM Navigation Map Closure Using Multi-Field Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional simultaneous localization and mapping (SLAM) algorithms in movable devices often fail to generate a closed profile map due to detection errors, leading to incorrect self-position recognition when returning to a starting point.
Innovation Solution
A movable device equipped with a range finder and processor unit that detects reflected light in multiple non-overlapping fields of view, generates previous and compensation data, and performs SLAM operations to create and compare maps, generating a control signal for navigation based on pre-determined conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a conventional SLAM algorithm is used for navigation, then the device can perform automatic movement and target recognition, but detection errors cause the profile map to fail to form a closed curve, leading to incorrect position recognition
Solution Approach 1:
The system performs preliminary actions by capturing images at multiple predetermined positions (first position, second position, third position) before the final navigation decision. These preliminary images are used to generate preliminary maps that are compared and verified for consistency, ensuring accurate position recognition before the device commits to a navigation path.
Solution Approach 2:
The system implements feedback by comparing the preliminary map generated from the second position with the reference map generated from the first position. When the comparison shows they match within a predetermined threshold, the system confirms accurate position recognition. This feedback loop allows the device to verify its position and correct any detection errors before final navigation.
2Reliability
If the device returns to the vicinity of the starting point, then it should obtain a closed profile map, but detection errors cause the map to be non-closed, preventing correct self-position recognition
Solution Approach 1:
The system captures preliminary images at multiple predetermined positions (first, second, and third positions) before final navigation. These preliminary actions create multiple preliminary maps that can be compared for consistency, ensuring that the profile map forms a closed curve even when detection errors occur during normal operation.
Solution Approach 2:
The system performs excessive action by capturing more images than the minimum single image would require. Instead of relying on a single image at the return position, it captures images at multiple predetermined positions and compares multiple preliminary maps, providing redundant information that compensates for detection errors and ensures map closure.
3Measurement precision
If multiple fields of view are used for detection, then navigation accuracy is improved, but the device complexity increases due to multiple detection operations and map comparisons
Solution Approach 1:
The system segments the detection process into distinct predetermined positions (first position, second position, third position) with specific fields of view. Each position captures images of specific features (first feature, second feature, third feature), dividing the complex navigation task into manageable segments that can be processed and compared independently.
Solution Approach 2:
The system applies local quality by assigning different fields of view to different predetermined positions based on their specific navigation needs. The first field of view captures the first feature at the first position, the second field of view captures the second feature at the second position, and so on, optimizing each local detection operation for its specific purpose rather than using a uniform approach.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Ensures accurate self-position recognition and navigation by comparing map data and adjusting navigation paths, improving the device's ability to return to a specific location accurately.
Implementation Method 1
The range finder detects reflected light in the area. At the first time point, the range finder detects reflected light in a first field of view to generate first previous data. At the second time point, the range finder detects reflected light in the first field of view to generate second previous data.
Data Source
AI summary
A movable device is provided. When the movable device is at a specific position, a range finder detects reflected light in a first field of view to generate first previous data. When the movable device returns to the vicinity of the specific position, the range finder detects reflected light in the first field of view to generate second previous data. The range finder detects reflected light in a second field of view to generate compensation data. The processor unit performs a SLAM operation on the first and second previous data to generate first and second previous maps, respectively. When a probability that the second previous map does not conform to the first previous map is greater than a threshold value, the processor unit performs the SLAM operation on specific data and the compensation data to control the moving path of the movable device.