Sparse Surface Adjustment
description
Transcript of Sparse Surface Adjustment
![Page 1: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/1.jpg)
April 22, 2023
Sparse Surface AdjustmentM. Ruhnke, R. Kümmerle, G. Grisetti, W. Burgard
![Page 2: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/2.jpg)
Metric 3D Models
►Essential for tasks like:►Object recognition►Manipulation
![Page 3: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/3.jpg)
Metric 3D Models
►Essential for tasks like:►Object recognition►Manipulation
►Key challenges in model acquisition with mobile robots►Errors in pose estimate►Measurement errors
![Page 4: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/4.jpg)
Model Creation
►Optimize the sensor poses►Registration / SLAM
►Reduce the impact of measurement errors►Use optimal sensor distance►Local noise reduction techniques (Moving Least
Squares, Statistical Outlier Removal, …)►Pose information is mostly not considered
►Sensor pose gives information about normal direction and range of the measurement
![Page 5: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/5.jpg)
Sparse Surface Adjustment
► Goal: Jointly optimize robot poses and surface points positions
► Surface Model► Model of measurement uncertainties► Data association: find corresponding points ► Utilize sparse graph optimizer framework g2o
![Page 6: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/6.jpg)
Surface Model
►Range measurements sample surfaces
►Assumption: Piecewise regular surfaces
►Surface sample►3D Position►covariance ►normal (local neighborhood)
![Page 7: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/7.jpg)
►Range measurement►Sensor specific Covariance
►Dependent on range and incidence angle►Gaussian error distributions
Sensor Model
front view side view~ 3.5m
~ 0.7m
Kinect RGB-D
![Page 8: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/8.jpg)
Data Association
►Normal shooting as data association heuristic
►Assign surfaces samples of different observations
►Covariance►Large error weight in
direction of the normal ►Small weight for errors
in tangential direction
![Page 9: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/9.jpg)
Optimization
►Iteratively:►Optimize system with g2o ►Re-compute:
►surface point characteristics (covariance, normals)►data association
![Page 10: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/10.jpg)
SSA 2D: Intel Dataset
![Page 11: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/11.jpg)
Object 5mm Resolution
![Page 12: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/12.jpg)
AASS Loop Dataset*
SLAM result (input)
SSA result
*Courtesy of Martin Magnusson, AASS, Örebro, Sweden
![Page 13: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/13.jpg)
SSA 3D: Example Cup
![Page 14: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/14.jpg)
Example: Scan Refinement
►SSA refines scans based on more certain nearby measurements
after optimizationraw scan
![Page 15: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/15.jpg)
Comparison SSA / MLS
►Moving Least Squares (MLS)►Local smoothing method►No correction of robot poses
►Sparse Surface Adjustment (SSA)►Robot pose correction & smoother surfaces
SSA result
MLS result
![Page 16: Sparse Surface Adjustment](https://reader035.fdocuments.net/reader035/viewer/2022062301/56815b9c550346895dc9a268/html5/thumbnails/16.jpg)
Summary SSA
►Iterative refinement of ►Sensor poses►Surface points positions
►Considering range & sensor dependent uncertainties
►Re-computation of data association►Uses PCL and FLANN