Free Space Estimation Using Camera-LiDAR Occupancy Grids
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for organizing 3D LIDAR point cloud data into 2D depth maps and surface normal maps do not effectively integrate camera image data to estimate free space for autonomous vehicle navigation, lacking a systematic approach to associate sensor data with machine learning models for accurate obstacle detection.
Innovation Solution
A system that receives camera image data and point cloud data, applies a machine learning model for semantic classification, transforms point cloud data into image and robot coordinate systems, and estimates free space by associating points with probabilities based on noise, reliability, and spatial associations, using a grid mapping approach to differentiate obstructed and non-obstructed areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If 3D LIDAR point cloud data is organized into 2D depth maps and surface normal maps, then data processing efficiency is improved, but the ability to accurately estimate free space for autonomous navigation deteriorates
Solution Approach 1:
The system segments the environment into discrete grid cells, where each cell is independently classified as occupied or free based on point cloud data. This segmentation allows efficient processing while maintaining accurate free space estimation by evaluating each cell's occupancy probability separately rather than processing the entire point cloud as a single structure.
Solution Approach 2:
The patent transforms the 3D point cloud data into a 2D grid map representation where the vertical dimension is aggregated into occupancy probabilities. This dimensional transformation enables efficient 2D processing while preserving 3D spatial information through probability values, resolving the contradiction between processing efficiency and estimation accuracy.
2Measurement precision
If camera image data and point cloud data are integrated with machine learning models, then obstacle detection precision is improved, but system complexity increases
Solution Approach 1:
The system merges camera image data and LIDAR point cloud data into a unified probabilistic occupancy grid. By combining multiple sensor types and their respective data streams into a single integrated representation, the system achieves improved obstacle detection precision while managing complexity through unified processing rather than separate parallel systems.
Solution Approach 2:
The patent changes the parameter representation from raw sensor data to occupancy probabilities. By transforming point cloud coordinates and camera images into probability values for each grid cell, the system enables precise obstacle detection through probabilistic reasoning while simplifying the integration of multiple sensor types with different data formats.
3Measurement precision
If point cloud data is transformed into multiple coordinate systems, then spatial association accuracy is improved, but computational time increases
Solution Approach 1:
The system performs preliminary coordinate transformations by pre-defining the relationship between camera and LIDAR coordinate systems. By establishing transformation matrices in advance and applying them to convert point cloud data into the camera coordinate system before processing, the system achieves accurate spatial association while reducing computational time through pre-prepared transformation parameters.
Data Source
AI summary
A system and method for estimating free space including applying a machine learning model to camera images of a navigation area, where the navigation area is broken into cells, synchronizing point cloud data from the navigation area with the processed camera images, and associating probabilities that the cell is occupied and object classifications of objects that could occupy the cells with cells in the navigation area based on sensor data, sensor noise, and the machine learning model.