Road Network Occlusion Detection in 3D Map Point Clouds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Gaps in 3D maps due to LIDAR occlusions, such as temporary objects blocking the view of roadside objects, lead to incomplete data sets and false negatives, affecting localization algorithms in autonomous driving and navigation applications.
Innovation Solution
A method for detecting occlusions from point cloud data using ray tracing and connected component analysis to identify hidden spaces, allowing for the identification of occluded regions and supplementing data collection to enhance map accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If LIDAR scanning is performed to capture roadside objects, then 3D map data is collected, but occlusions by temporary objects cause false negatives and incomplete data
Solution Approach 1:
The patent introduces an intermediary occlusion detection system that uses virtual camera ray tracing to identify blocked regions. This intermediary process analyzes the relationship between LIDAR sensor positions, detected objects, and expected roadside object locations to detect occlusions caused by temporary objects, thereby preventing false negatives in the 3D map data collection process
Solution Approach 2:
The system performs preliminary occlusion detection by generating virtual camera images and tracing rays through the point cloud data before finalizing the 3D map. This preliminary analysis identifies occluded regions in advance, allowing the system to flag potential false negatives and request additional data collection passes to ensure complete and accurate mapping
2Measurement precision
If data collection is performed without occlusion detection, then data collection is faster, but false negatives reduce map accuracy
Solution Approach 1:
The occlusion detection system performs self-service by automatically analyzing the collected point cloud data to identify occluded regions. The virtual camera ray tracing process autonomously determines which roadside objects may be blocked and flags these areas for potential re-scanning, eliminating the need for manual quality review while maintaining high map accuracy
Solution Approach 2:
The system implements feedback by using occlusion detection results to guide subsequent data collection efforts. When occlusions are detected, the system provides feedback to request additional LIDAR scanning in the affected areas, creating a closed-loop process that continuously improves map accuracy while optimizing data collection efficiency through targeted re-scanning
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present embodiments provide for automatically detecting the location and severity of occluded regions within input data. A grid representation of a scene is generated from a data set, characterizing spaces of the grid representation as free, occupied, and hidden/occluded. The grid is bounded, and a connected component analysis is performed on the hidden space to identify the occluded regions.