Natural Terrain Classification using 3-D Ladar Data

Because of the difficulty of interpreting laser data in a meaningful way, safe navigation in vegetated terrain is still a daunting challenge. In this paper, we focus on the segmentation of ladar data using local 3-D point statistics into three classes: clutter to capture grass and tree canopy, linear to capture thin objects like wires or tree branches, and finally surface to capture solid objects like ground terrain surface, rocks or tree trunks. We present the details of the method proposed, the modifications we made to implement it on-board an autonomous ground vehicle. Finally, we present results from field tests using this rover and results produced from different stationary laser sensors.

Authors

• Nicolas Vandapel
• Daniel F. Huber
• Anuj Kapuria
• Martial Hebert

Keywords

• Laser radar
• Statistics
• Surface emitting lasers
• Vegetation mapping
• Laser modes
• Robot sensing systems
• Laser tuning
• Navigation
• Wires
• Solids
• Light Detection And Ranging
• Terrain Classification
• Vegetation
• Field Test
• Tree Branches
• Balance Of System
• Tree Canopy
• Tree Trunks
• Terrain Surface
• Use Of Data
• Final Results
• Training Dataset
• Covariance Matrix
• Classification Results
• Autonomic System
• Expectation Maximization
• Salient Features
• Point Cloud
• Intermediate Results
• Line Scan
• Gaussian Mixture Model
• Linear Structure
• Angular Distance
• Thin Branches
• Tallgrass
• Scan Pattern
• Principal Directions
• Local Neighborhood
• Classification Error Rate
• Laser Ranging