Autonomous Mobility Grid Mapping for Point Cloud Noise Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing micro mobility vehicles face challenges in autonomously navigating without high-precision maps, particularly due to noise in three-dimensional point cloud data that affects the accuracy of obstacle detection and trajectory planning.
Innovation Solution
A moving object system that generates a grid map by mapping three-dimensional point cloud data to a horizontal plane, determining the existence of objects based on point density, and applying noise removal processing to reduce the influence of noise in the data, thereby enhancing obstacle detection and trajectory planning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If three-dimensional point cloud data is projected onto a horizontal plane to generate a grid map, then obstacle detection capability is improved, but noise from outlier point clouds and white noise reduces measurement precision
Solution Approach 1:
The patent segments the point cloud data processing by dividing the grid map into multiple height layers. Each layer is processed independently to identify objects at different heights, allowing selective noise removal while preserving valid obstacle information. This segmentation enables the system to handle outlier point clouds and white noise more effectively by treating different spatial regions separately.
Solution Approach 2:
The patent extracts and removes noise point clouds from the three-dimensional point cloud data before generating the grid map. By identifying and eliminating outlier point clouds and white noise through statistical analysis and filtering algorithms, the system preserves only the valid obstacle information, thereby improving measurement precision while maintaining obstacle detection capability.
2Measurement precision
If conventional contraction and expansion processing is applied to a binary two-dimensional map for noise removal, then some noise is reduced, but the processing does not adequately address the characteristics of point cloud noise
Solution Approach 1:
The patent applies local quality by implementing noise removal processing that adapts to the specific characteristics of point cloud data in different regions. The system analyzes the density and distribution of point clouds locally and applies appropriate filtering thresholds, rather than using uniform contraction and expansion processing. This allows the system to effectively remove noise while preserving valid obstacle information with varying densities.
Solution Approach 2:
The patent changes the processing parameters from conventional binary map contraction/expansion to point cloud-specific parameters such as point density thresholds, height layer divisions, and statistical noise criteria. By adjusting these parameters based on the actual characteristics of the point cloud data, the system achieves superior noise removal effectiveness while maintaining adaptability to different scanning conditions and environments.
3Adaptability or versatility
If micro mobility vehicles operate without high-precision maps, then adaptability to various movement regions is improved, but autonomous navigation becomes more difficult due to noise in environmental perception
Solution Approach 1:
The patent transitions from two-dimensional binary map processing to three-dimensional point cloud processing by introducing height as an additional dimension. This dimensional change allows the system to perceive the environment more accurately and distinguish obstacles from noise more effectively, thereby reducing autonomous navigation difficulty while maintaining adaptability to various movement regions including roadways and sidewalks.
Solution Approach 2:
The patent replaces the mechanical approach of conventional binary map processing with a more sophisticated system that processes three-dimensional point cloud data. By substituting the simple binary representation with rich three-dimensional spatial information and applying advanced noise removal algorithms, the system achieves better autonomous navigation performance without requiring high-precision pre-maps, thus maintaining adaptability to diverse environments.
Data Source
AI summary
A moving object acquires three-dimensional point cloud data in surroundings of the moving object; and generates a grid map based on the three-dimensional point cloud data, the moving object maps point clouds of the three-dimensional point cloud data to a grid on a horizontal plane parallel to a road surface on which the moving object travels and generates the grid map in which each grid indicates whether a three-dimensional object having a predetermined height or a step exists. The moving object determines, for each grid of the grid map, whether the three-dimensional object exists based on a point density when a point cloud corresponding to each grid of the three-dimensional point cloud data is projected on the horizontal plane.