LIDAR Point Cloud Alignment via Planar Feature Transform
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in aligning LIDAR-based 3D point clouds with reference point clouds, leading to misalignment and incorrect obstacle detection due to differences in LIDAR device pose and calibration, which can result in false obstacle detection or failure to detect actual obstacles.
Innovation Solution
A method involving a computing device that determines a planar feature in the 3D point cloud and a corresponding planar feature in the reference point cloud, calculates a transform based on their comparison, and applies this transform to align the 3D point cloud with the reference point cloud, facilitating accurate obstacle detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LIDAR data is collected from a vehicle-mounted device with different pose and calibration, then the LIDAR-based 3D point cloud can be obtained for obstacle detection, but misalignment with reference point cloud occurs leading to incorrect obstacle detection
Solution Approach 1:
The patent replaces manual or mechanical alignment methods with an automated computational approach. A computing device automatically determines a transform by comparing planar features between the LIDAR-based point cloud and reference point cloud, then applies this transform to align the data. This substitution of mechanical alignment with computational transformation resolves the contradiction by achieving precise alignment without physical adjustment, thereby improving both measurement precision and detection reliability.
Solution Approach 2:
The patent changes the spatial parameters (position and orientation) of the LIDAR point cloud through a computed transform. By determining the transform based on planar feature comparison and applying it to adjust the point cloud's spatial parameters, the system achieves accurate alignment with the reference point cloud. This parameter transformation resolves the misalignment issue while maintaining the reliability of obstacle detection.
2Reliability
If transform is applied to align 3D point cloud with reference point cloud, then obstacle detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent extracts and utilizes only the planar features from the LIDAR point cloud and reference point cloud to determine the transform. By focusing on these specific geometric features rather than processing the entire point cloud data, the system reduces computational complexity while still achieving accurate alignment and maintaining obstacle detection reliability. This selective feature extraction resolves the contradiction between alignment accuracy and computational burden.
3Measurement precision
If planar feature comparison is used to determine transform, then alignment precision is improved, but processing time increases
Solution Approach 1:
The patent applies partial action by using only the necessary planar features for alignment rather than processing all point cloud data. By focusing on the essential geometric characteristics (planar features) that suffice for determining the transform, the system achieves sufficient alignment precision without the excessive processing time that would result from analyzing the complete point cloud dataset.
Data Source
AI summary
Methods and systems for alignment of light detection and ranging (LIDAR) data are described. In some examples, a computing device of a vehicle may be configured to compare a three-dimensional (3D) point cloud to a reference 3D point cloud to detect obstacles on a road. However, in examples, the 3D point cloud and the reference 3D point cloud may be misaligned. To align the 3D point cloud with the reference 3D point cloud, the computing device may be configured to determine a planar feature in the 3D point cloud of the road and a corresponding planar feature in the reference 3D point cloud. Further, the computing device may be configured to determine, based on comparison of the planar feature to the corresponding planar feature, a transform. The computing device may be configured to apply the transform to align the 3D point cloud with the reference 3D point cloud.