LiDAR Free Space Generation With Multi-Modal Noise Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional LiDAR signal processing systems require fast processors to handle large amounts of point cloud data in real time and suffer from noise components that necessitate additional noise removal devices, compromising accuracy.
Innovation Solution
A LiDAR free space data generator and signal processing method employing a multi-modal noise filtering scheme that filters noise at both grid and point levels, optimizing noise determination through a point filter module, 2D grid map generation, and multi-modal noise filter module.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional 2D pointwise point processing scheme is used, then measurement precision is improved, but device complexity increases due to requirement of fast processor and separate noise removal device
Solution Approach 1:
The patent combines the noise filtering function with the existing grid map generation module. The grid map generation module performs both occupancy determination and noise filtering in a unified process, eliminating the need for separate noise removal devices while maintaining measurement precision through multi-modal noise filtering at grid and point levels.
Solution Approach 2:
The grid map generation module is designed to perform multiple functions: it generates occupancy grids from point cloud data, determines free space, and simultaneously filters noise through multi-modal filtering schemes. This multi-functional approach reduces device complexity while preserving the accuracy benefits of pointwise processing.
2Manufacturing precision
If pointwise point processing is used, then manufacturing precision is improved, but productivity decreases due to large amount of data processing requirements
Solution Approach 1:
The patent segments the point cloud data processing into a two-stage approach: first converting points to occupancy grids (reducing data volume), then performing noise filtering on the grid structure. This segmentation enables efficient real-time processing while maintaining the precision of point-level analysis through subsequent point-level noise filtering on representative points.
Solution Approach 2:
The patent transforms 3D point cloud data into 2D occupancy grids, changing the dimensional representation to reduce computational complexity. This dimensionality reduction enables faster processing while the multi-modal noise filtering scheme ensures that accuracy is maintained by filtering noise at both grid and point levels.
3Productivity
If conventional processing without noise filtering is used, then productivity is improved, but measurement precision deteriorates due to noise components
Solution Approach 1:
The patent performs noise filtering as a preliminary step during grid map generation, before final output is produced. By integrating multi-modal noise filtering into the grid generation process, the system eliminates noise components early in the processing pipeline, ensuring high measurement precision without requiring additional processing steps that would reduce productivity.
Data Source
AI summary
An embodiment LiDAR free space data generator includes a point filter module configured to select input points in a region of interest of an autonomous parking system from a three-dimensional (3D) point cloud obtained from a LiDAR, a two-dimensional (2D) grid map generation module configured to perform 2D grid-wise point processing on the selected input points to generate a first occupied grid feature map, and a multi-modal noise filter module configured to perform grid-level noise filtering, grid-level downsampling, or point-level noise filtering on the first occupied grid feature map to generate first LiDAR free space data.