300 likes | 478 Views
Large-Scale 3D Terrain Modeling. David L. Page Mongi A. Abidi, Andreas F. Koschan Sophie Voisin, Sreenivas Rangan, Brad Grinstead, Wei Hao, Muharrem Mercimek Imaging, Robotics, & Intelligent Systems Laboratory The University of Tennessee March 23, 2004. Outline. 3D Terrain Modeling
E N D
Large-Scale 3D Terrain Modeling David L. Page Mongi A. Abidi, Andreas F. Koschan Sophie Voisin, Sreenivas Rangan, Brad Grinstead, Wei Hao, Muharrem Mercimek Imaging, Robotics, & Intelligent Systems Laboratory The University of Tennessee March 23, 2004
Outline • 3D Terrain Modeling • UTK mobile terrain scanning system • Simulation needs and Army benefit • Scanning system pipeline • “Knoxville Proving Grounds” • Research problems
UTK Mobile Terrain Scanning System Multi-sensor data collection system for road surface. Video Camera GPS Receiver GPS Base Station 3-Axis IMU and Computer 3D Range Sensor
3 1 4 7 5 2 8 6 Data Acquisition Actual Path 1 – Riegl LMS-Z210 Laser Range Scanner 2 – SICK LMS 220 Laser Range Scanner 3 – JVC GR-HD1 High Definition Camcorder 4 – Leica GPS500 D-RTK Global Positioning System 5 – XSens MT9 Inertial Measurement Unit 6 – CPU for acquiring SICK, GPS, and IMU data 7 – CPU for acquiring Riegl data 8 – Power system Scanned Path Modular System Mounted here on a push cart. Geo-referenced geometric 3D model of an area near IRIS West in Knoxville.
3D View of Terrain (Jump to 3D Viewer)
Outline • 3D Terrain Modeling • UTK mobile terrain scanning system • Simulation needs and Army benefit • Scanning system pipeline • “Knoxville Proving Grounds” • Research problems • Static scanning
Simulation Needs for Terrain Modeling Why needed, in general? • Visualization • Typical terrains only available in 30x30 m2 grids • Probably sufficient with bump mapping • System analysis • Requires high-resolution terrains! • Multi-body dynamics • Linear analysis, PSD • Time series analysis • Requires high-resolution terrains! • Multi-body dynamics • Motion stands Bump Mapping Discussions with Dr. Al Reid
Scanning 3D terrains is a significant enhancement over traditional towed-cart profiling, cart dynamics, 1D profile, etc. Real terrain modeling overcomes potential limitations of linearity, stationarity, and normality assumptions, particularly associated with PSD (Chaika & Gorsich 2004). Research in 3D processing (tools!) addresses relevant issues in… data reduction (Al Reid), terrain analysis (3D EMD), interpolation, etc. Benefit to U.S. Army
Profilometers • Four (4) wheel trailer • Drawn by a tow vehicle • Front axle free to rotate about yaw axis (other constrained) • Linkage to draw bar of tow vehicle • Rear axle free to rotate about roll axis (other constrained) • No compliant suspension components between axles and frame • Inertial gyroscope measures pitch and roll angle • Ultrasonic measurement between axle and terrain (always points down) • Shaft encoder every 0.1 in. of travel • Data acquisitions every 3 inches Towed Trailer Profilometer Highly correlated sensor data (GPS, IMU, Range) = Correction for vehicle dynamics UTK 3D Terrain Modeling
120-360 profiles over a 2-8 m swath (3D surface) vs. 1 profile (1D signal) Correlated data vs. trailer dynamics Agile path vs. linear path (?) Comparison to Profilometer 3D vs. 1D Path Overlaid on Aerial View Zoom View 2 m wide x 8 m length Path is 300 m length +/- 0.5 cm resolution Video Data of Zoom Notice Cracks in Pavement
Outline • 3D Terrain Modeling • UTK mobile terrain scanning system • Simulation needs and Army benefit • Scanning system pipeline • “Knoxville Proving Grounds” • Research problems
System Block Diagram 3D range sensors Position and orientation sensors Visual Thermal IVP RIEGL SICK Leica -GPS Xsens IMU Sony Indigo Range Profiles Video Sequence 3D Position and Orientation Multi-sensor Alignment Inter-profile Alignment Multi-modal Data Integration Multi-sensor Visualization
UTK IRIS Lab 3D Sensors Genex 3D CAM IVP RANGER SC-386 SICK LMS200 Time-of-flight Sheet-of-light triangulation-based system Structured-light stereo system Principle of operation X Laser Camera x’ x c’ c S1 S2 S1 and S2 are two sensors. S1 3D Rendering
Statistical Modeling of Sensors Yaw Measurements Roll Measurements Pitch Measurements Standard Deviation = 0.0336 Standard Deviation = 0.0338 Standard Deviation = 0.0492 Extensive GPS and IMU error characterization and modeling.
Outline • 3D Terrain Modeling • UTK mobile terrain scanning system • Simulation needs and Army benefit • Scanning system pipeline • “Knoxville Proving Grounds” • Research problems
“Knoxville Proving Grounds” Visualization tool built to be able to visualize “z” measurements Blue Line is the GPS Path for the loops that we collected. Cornerstone Drive, off Lovell Road, I-40 Exit #374 Knoxville, Tennessee, Knoxville Each loop a length of 1.1 mile, Total distance covered on scanning that day = 2.2 miles ( 2 times) = 4.4 miles of the same data. The color of the GPS path encodes the height of the terrain. Over 4 miles = ~2 GB of data
Data Collection Automated correction for varying speeds and dynamics of platform.
17 m 17 m 0 m 0 m Elevation Change of Terrain Pathways – Loop scanning Full length scanning
High Accuracy 3D Terrain Full Data ~10 km Zoom ~1 km Zoom ~10 m Aerial View
Triangulated Terrain Mesh The entire stretch, 1.8 meters
Campus Loop Y Latitude and Longitude Measurements from the Leica DGPS Raw Point Cloud
Outline • 3D Terrain Modeling • UTK mobile terrain scanning system • Simulation needs and Army benefit • Scanning system pipeline • “Knoxville Proving Grounds” • Research problems
Interprofile Registration Problem Vehicle (Scanning) Direction GPS curve sampled at 10 Hz. IMU data @ 100 Hz Video recorded at 30 frames/sec Range Profiles @ 30 Hz 4m wide SICK 2000 Hz and 50cms wide IVP Raw Data
Data Interpolation Correct for non-uniform data collection with terrain modeling.
R, T Pose Localization Video Sequence Feature Matching Pose From Motion Oriented Tracks Filtering RANSAC Filtering GPS drop-outs under certain conditions. Improve overall localization accuracy.
Data Reduction Noise Removal Initial Model Multiresolution Analysis and Denoising Adaptive Simplification
Statistical Modeling of Terrain Dataset from near IRIS West The total length of the patch: 20 meters with inter-profile spcaing around 1 cm. Reconstructed 3D profile from the statistical model Mean Longitudinal profile The 3D terrain was generated using our system mounted on a van. The profile is non-linear and non-stationary but all the IMF’s taken separately are linear and stationary, which means the PSD of the IMF’s model the data better than the PSD of the profile alone. Empirical mode decomposition of the terrain sample shown above. EMD implementation : Modified Brad’s functions
Camera Calibration Image Rectification Dense Matching Disparity Estimation Triangulation & Visualization Temporal-Based StereoTire-Soil Terrain Modeling Calibration Pipeline of 3D Reconstruction Test Setup Disparity Map Input
3D Model of Military Tire Model Integration (+/- 0.5 mm) Tire 150 cm dia., 30 cm width Registration (18 Sections, 7 Views) Final Model
17 m 0 m 17 m z (m) 0 m y (m) x (m) Questions? Pathways – Loop scanning