Math to System Design

3D Environment ReconstructionUsing Modified Color ICPAlgorithm by Fusion of a Camera and a 3D Laser Range FinderThe 2009 IEEE/RSJ International Conference onIntelligent Robots and SystemsOctober 11-15, 2009 St. Louis, USA3D Reconstruction BackgroundBuilding rich 3D maps of environments is an importanttask for mobile robotics, with applications in navigation,virtual reality, medical operation, and telepresence.

Most 3D mapping systems contain three main components:the spatial alignment of continuous data frames; the detection of loop closures; the globally consistent alignment of the complete data sequence.

2Concept explainScale-invariant feature transform (SIFT):Detect and describe local features match in images. Random Sample Consensus (RANSAC):An iterative method to estimate parameters of a mathematical model from a set of observed data which contains outliers.Iterative Closest Point (ICP):employed to minimize the differences between two clouds of points.K-D Treea space-partitioning data structure to organize points in a k-dimensional space for range searching and nearest neighbor searching3Processing of Color ICP

Step of algorithm: SIFT RANSAC Color ICP

4Processing of Color ICP, Cont.SIFTColor ICP1. Create image feature matchCreate 3D EnvironmentGet Higher accurate position information

RANSACFiltrate Poisoned Point and get motion Information Get low accurate position information for the initial estimation of ICP algorithmICP mathematical expression

Euclidean distance

Normal ICP algorithm Introduce

Normal ICP algorithm Introduce, Cont.

Color ICP Algorithm Improve Improve the normal ICP algorithm :Transfer the RGB to YIQ space. Influence of luminance is decreased.Build the IQ 2D histogram using I and Q channel for faster searching.

Color ICP Algorithm Improve, Cont. Improve the normal ICP algorithm :Transfer the RGB to YIQ space. Influence of luminance is decreased.Build the IQ 2D histogram using I and Q channel for faster searching.Consider color distance in nearest neighbor search algorithm.Color dynamic range can compress data size.

Higher Accuracy LocationCompare the rotation error graph of SIFT and Color ICP + SIFT algorithm

Compare the rotation error graph of ICP and Color ICP algorithm

10Time ComplexityComparison of searching time to find the closest points

11UAV System StructureInertial Measurement Unit algorithmData FusionDynamics Model Robust ControllerLower Layer VisionAlgorithmMotionInformationSimultaneous localization and mapping(SLAM)(Robot Operating System) Path PlanningAttitude InformationMotionInformationPosition control InformationCloud Point InformationInitial Motion InformationPosition and Environment Information10 millisecondLayer

100 millisecondLayersecondLayerHigh accuracy attitude informationAnd low accuracy position informationHigher accuracy position information12ReferencesD.Lowe, "Distinctive Image Features from Scale-Invariant Keypoints", January 5, 2004T.Lindeberg, "Scale-space theory: A basic tool for analyzing structures at different scales"Jon Louis Bentley, "Multidimensional Binary Search Trees Used for Associative Searching, Stanford University 1975 ACM Student AwardP. Henry, M. Krainin, E. Herbst, X. Ren, and D. Fox. , RGB-D Mapping: Using Depth Cameras for Dense 3D Modeling of Indoor Environments, Proc. of International Symposium on Experimental Robotics (ISER), 2010Lu F and Milios E Globally consistent range scan alignment for environment mapping ,Autonomous Robots 4: 333349, 1997133D Environment ReconstructionThank You!14Random Sample Consensus (RANSAC)

KD-Tree Introduction

Best-Bin-First(BBF) algorithmxyxy