Laser Scanner Ego-Motion Estimation via Point Cloud Subsampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mapping systems face challenges in generating robust, real-time maps of terrain surrounding autonomous devices while in motion, due to issues like scan distortions from slow laser scanning rates, extrinsic motion effects, and limitations in GPS/INS techniques, especially in GPS-denied environments.
Innovation Solution
A modularized mapping system incorporating an inertial measurement unit, camera, and laser scanner, with a computing system that processes data from these sensors to perform real-time ego-motion estimation and map registration, using a three-layer computational module structure for bidirectional information flow and dynamic reconfiguration to handle sensor degradation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If SLAM-based mapping approaches are used to estimate ego-motion, then ego-motion estimation can be obtained, but computational latency increases and real-time performance is compromised
Solution Approach 1:
The patent extracts only the essential features needed for ego-motion estimation from the point cloud data, rather than performing complete SLAM mapping. By taking out only the necessary computational elements (feature extraction and ICP registration) and eliminating redundant SLAM processes, the system achieves real-time performance while maintaining estimation accuracy.
Solution Approach 2:
The patent segments the ego-motion estimation process into distinct modules: point cloud preprocessing, feature extraction, ICP registration, and transformation calculation. This segmentation allows each module to be optimized independently and processed in real-time, avoiding the computational burden of unified SLAM approaches.
2Loss of information
If complete SLAM mapping is performed, then comprehensive environmental mapping is achieved, but processing time increases and real-time control is delayed
Solution Approach 1:
The patent extracts only the critical information needed for autonomous navigation and control from the environment, rather than creating complete SLAM maps. By extracting essential geometric features and relationships, the system maintains sufficient mapping information for safe operation while achieving real-time processing speeds.
3Measurement precision
If heavy computational algorithms are used for ego-motion estimation, then estimation accuracy improves, but system responsiveness decreases
Solution Approach 1:
The patent replaces heavy computational SLAM algorithms with a more efficient computational approach based on ICP registration and transformation mathematics. This substitution maintains estimation accuracy while dramatically improving computational speed and system responsiveness, enabling real-time autonomous operation.
Data Source
AI summary
A method comprises accessing a data set comprising a LIDAR acquired point cloud comprising a plurality of points each of which are attributed with at least a geospatial coordinate, sub-sampling at least a portion of the plurality of points to derive a representative sample of the plurality of points and displaying the representative sample of the plurality of points.