of Subarctic Forests Philippe Babin, Philippe Dandurand, Vladimr - PowerPoint PPT Presentation

12th FSR conference, Tokyo, 2019 Large-scale 3D Mapping of Subarctic Forests Philippe Babin, Philippe Dandurand, Vladimír Kubelka, Philippe Giguère and François Pomerleau

Subarctic Boreal Forest: Research Opportunity 2/27

Naive approach Our approach 3/27

Applications 4/27

Challenge of Field Tests Snow Fall Path obstacles Uneven Path Local Wildlife 5/27

Mapping of Subarctic Boreal Forest - Challenges  Unstructured environment → hard to map  Cold temperatures → noisy sensor  Few visual features due to snow → bad for vision based approaches 6/27

Related Work Williams et al., 2009 Paton et al., 2016 7/27

Contributions  Large-scale mapping of difficult environments  Novel fusion of IMU and GNSS measurement inside of ICP  Generated maps are crisp and without long term drifts  Introduced optimization to scale to large map 8/27

Dataset Environment 4.1 km of forest path 9/27

10/27

Data Acquisition Platform GNSS station (RTK)  RS-16 lidar  MTI-30 IMU  10h of battery life  11/27

Iterative Closest Point (ICP) T ? 12/27

Iterative Closest Point (ICP) T init ICP T 13/27

Iterative Closest Point (ICP) 14/27

Legend Point cloud Sensor Fusion Pose Covariance Pose/Covariance Classical approach Our approach Lidar ICP Lidar pose Covariance [1] IMU Penalty-ICP IMU SLAM GNSS GNSS map pose map pose 15/27 [1] D. Landry, F. Pomerleau, and P. Giguère. CELLO-3D: Estimating the Covariance of ICP in the Real World. In ICRA, 2019

Map of lake Dataset 350m ICP with penalty ICP no penalty Prior 16/27

ICP with penalty ICP no penalty 17/27

Prior ICP With penalties ICP Without Penalties Crispiness locally consistent 18/27

Map of forest Dataset 500m ICP with penalty ICP no penalty Prior 19/27

Map of forest Dataset 500m ICP with penalty ICP no penalty Prior 20/27

Map of skidoo Dataset 670m ICP with penalty ICP no penalty Prior 21/27

Map of skidoo Dataset 670m ICP with penalty ICP no penalty Prior 22/27

Full Map with Penalty 23/27

Performance improvements 24/27

Future work 25/27

Future work – Project SNOW 26/27

Questions? 27/27

28

Performance improvements 29/26

Results: effect of penalties 30

Penalty-ICP • Leverage ICP’s minimizer for sensor fusion • Add penalty term based on GNSS and IMU estimate • Introduced a point to Gaussian cost function • Minimize Mahalanobis distance instead of the Euclidian distance 31

Full Map 32