AGV 3D Localization Using Selective Multi-Plane Point Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AGV localization systems face challenges with occlusions when using 3D scanners due to computational overhead and reliance on single 2D planes, leading to navigation issues in confined spaces with multiple vehicles.
Innovation Solution
Implementing a multi-beam approach in 3D scanning to triangulate position using selective beam angles and a multi-level approach to switch between 2D planes, reducing computational overhead and mitigating occlusions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If 3D point cloud matching is used for localization, then localization accuracy is improved, but computational overhead increases
Solution Approach 1:
The patent segments the 3D point cloud into multiple 2D planes at different heights. Instead of matching the entire 3D point cloud, the system divides it into manageable 2D slices and performs matching on each plane separately, reducing computational complexity while maintaining localization accuracy through multi-plane integration
Solution Approach 2:
The patent transforms the 3D point cloud matching problem into multiple 2D plane matching problems. By projecting 3D points onto 2D planes at different heights and performing 2D matching, the system reduces computational overhead while preserving essential spatial information for accurate localization
2Device complexity
If single 2D plane matching is used for localization, then computational complexity is reduced, but occlusion resistance deteriorates
Solution Approach 1:
The patent introduces the height dimension by creating multiple 2D planes at different vertical levels. This allows the system to maintain simple 2D matching computations while gaining 3D occlusion resistance, as landmarks visible at one height may be occluded at another height but visible from alternative planes
Solution Approach 2:
The patent segments the single 2D plane into multiple 2D planes stacked vertically. Each plane provides an independent view for matching, and the system can switch between planes or combine results, maintaining low computational complexity per plane while achieving robust occlusion handling through multi-plane redundancy
3Reliability
If entire 3D point cloud is matched, then occlusion resistance is improved, but computational overhead increases
Solution Approach 1:
The patent segments the 3D point cloud into multiple 2D planes, transforming one complex 3D matching operation into several simpler 2D matching operations. This segmentation reduces processing time per plane while maintaining occlusion resistance through the collective information from multiple planes
Solution Approach 2:
The patent reduces the 3D matching problem to multiple 2D matching problems by projecting points onto horizontal planes at different heights. This dimensionality reduction significantly decreases computational overhead while preserving occlusion resistance by utilizing vertical separation to avoid occlusions
Data Source
AI summary
A method may include receiving a 3D point cloud of a space, identifying points of the 3D point cloud at selective locations of the 3D point cloud, and comparing the points to a map of the space to localize an AGV within the space. A method may include receiving a 3D point cloud of a space, identifying multiple points of the 3D point cloud at respective beam angles from a sensor, and comparing the multiple points to a map of the space to localize an AGV within the space. A method may include receiving a 3D point cloud of a space, identifying first and second sets of points at respective first and second 2D planes, and comparing the sets of points to a map of the space to localize an AGV within the space. Additional methods and associated systems are also disclosed.