Pole-Like Object Detection from Sparse 3D Point Cloud Slices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pole-like object detection methods in autonomous vehicles face challenges with sparse point cloud data, leading to inaccurate computation of point features and failure to detect pertinent structures, resulting in incorrect exclusion of relevant data points.
Innovation Solution
The proposed solution involves identifying horizontal slices of point cloud data, generating 2D occupancy images, detecting vertical clusters, and modeling them as cylinders using machine learning algorithms to improve detection accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional point cloud processing methods are used, then detection coverage is limited, but detection accuracy deteriorates due to sparse data points
Solution Approach 1:
The patent merges multiple sparse point cloud data points that belong to the same vertical pole-like structure into a single consolidated cluster. By detecting vertical relationships across multiple horizontal slices and combining points that align vertically, the system transforms sparse individual points into dense vertical clusters, thereby improving detection accuracy without requiring increased data point density in the original sensor measurements.
Solution Approach 2:
The patent transitions from processing point cloud data in three-dimensional space to generating two-dimensional occupancy images by projecting points onto horizontal slices. This dimensional transformation allows the system to detect vertical clusters by analyzing patterns across multiple 2D slices, effectively using the vertical dimension as an additional analysis axis to overcome sparsity in the original 3D point cloud.
2Measurement precision
If all point cloud data is processed in detail, then detection accuracy improves, but processing time increases
Solution Approach 1:
The patent segments the point cloud data into multiple horizontal slices at different heights, processing each slice independently to generate 2D occupancy images. This segmentation allows parallel processing of different spatial regions and enables the system to focus computational resources only on slices that contain relevant vertical cluster patterns, significantly reducing overall processing time while maintaining detection accuracy.
Solution Approach 2:
The patent extracts only the essential vertical cluster features from the complete point cloud data by analyzing occupancy patterns across horizontal slices. Instead of processing all raw point cloud data in detail, the system extracts vertical relationships and cluster formations, discarding redundant information and focusing computation on the critical features needed for pole-like object detection.
3Reliability
If complete point cloud data is stored, then detection reliability improves, but storage requirements increase
Solution Approach 1:
The patent creates simplified 2D occupancy image representations as copies of the original 3D point cloud data. These occupancy images capture the essential vertical cluster patterns needed for detection while requiring significantly less storage space. The system maintains detection reliability by preserving the vertical relationship information in the 2D projections, eliminating the need to store complete high-resolution point cloud data.
Solution Approach 2:
The patent discards redundant point cloud data points that do not contribute to vertical cluster formation while recovering and preserving only the essential vertical relationship information. By filtering out horizontal noise and retaining only vertically-aligned points across slices, the system reduces storage requirements while maintaining the critical features needed for reliable pole-like object detection.
Data Source
AI summary
Embodiments include apparatus and methods for automatic detection of pole-like objects for a location at a region of a roadway and automatic localization based on the detected pole-like objects. Pole-like objects are modeled as cylinders and the models are generated based on detected vertical clusters of point cloud data associated to corresponding regions along the region of the roadway. The modeled pole-like objects are stored in a database and associated with the region of the roadway. Sensor data from a user located at the region of the roadway is received. The pole-like object model is accessed and compared to the received sensor data. Based on the comparison, localization of the user located at the region of the roadway is performed.