Undistorted LiDAR Scan Alignment for Autonomous Vehicle Ground Truth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for generating ground truth datasets for autonomous vehicles are costly and inefficient, particularly in processing and analyzing real-world data, and existing datasets may not be well generalized across different environments, limiting the effectiveness of autonomous driving algorithms.
Innovation Solution
A method involving sensor calibration and time synchronization for generating ground truth datasets, which includes obtaining undistorted LiDAR scans, aligning points based on pose estimates, transforming reference scans, and generating LiDAR static-scene point clouds, along with sparse image point correspondences refinement, to create a robust dataset for motion planning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If traditional methods are used to process and analyze real-world data for autonomous vehicles, then comprehensive datasets can be obtained, but the cost of processing and analyzing these data becomes extremely high
Solution Approach 1:
The patent extracts only the essential and useful information from real-world data by identifying static points in the environment and generating ground truth datasets selectively, rather than processing all collected data. This extraction approach reduces the volume of data requiring expensive processing while maintaining the quality and utility of the dataset for autonomous vehicle development.
2Productivity
If existing datasets are used for autonomous driving algorithm development, then immediate testing can be performed, but the datasets may not be well generalized to different environments
Solution Approach 1:
The patent creates ground truth datasets with universal applicability by capturing static environmental points that remain consistent across different environments and conditions. The methodology generates datasets that can serve multiple functions and be applied to various autonomous driving scenarios, improving generalization while maintaining efficient development cycles.
3Measurement precision
If comprehensive sensor calibration and data collection procedures are implemented, then accurate ground truth datasets can be generated, but the complexity of the process increases
Solution Approach 1:
The patent segments the complex calibration and data collection process into distinct, manageable steps: identifying static points in the environment, matching points between frames, generating ground truth datasets, and validating results. This segmentation reduces procedural complexity while maintaining measurement precision by focusing on essential operations rather than comprehensive processing.
4Reliability
If extensive data collection and processing procedures are used, then robust ground truth datasets can be generated, but the time and resources required increase significantly
Solution Approach 1:
The patent performs preliminary actions by pre-identifying static points in the environment and pre-matching points between frames before generating the final ground truth dataset. This preliminary processing reduces the time and resources required during actual data collection while ensuring robustness, as the foundational work is already completed and validated.
Data Source
AI summary
A method of generating a ground truth dataset for motion planning of a vehicle is disclosed. The method includes: obtaining undistorted LiDAR scans; identifying, for a pair of undistorted LiDAR scans, points belonging to a static object in an environment; aligning the close points based on pose estimates; and transforming a reference scan that is close in time to a target undistorted LiDAR scan so as to align the reference scan with the target undistorted LiDAR scan.