Robust 3D Scan Point Classification using Associative Markov Networks

In this paper we present an efficient technique to learn associative Markov networks (AMNs) for the segmentation of 3D scan data. Our technique is an extension of the work recently presented by Anguelov et al. (2005), in which AMNs are applied and the learning is done using max-margin optimization. In this paper we show that by adaptively reducing the training data, the training process can be performed much more efficiently while still achieving good classification results. The reduction is obtained by utilizing kd-trees and pruning them appropriately. Our algorithm does not require any additional parameters and yields an abstraction of the training data. In experiments with real data collected from a mobile outdoor robot we demonstrate that our approach yields accurate segmentations

Authors

• Rudolph Triebel
• Kristian Kersting
• Wolfram Burgard

Keywords

• Robustness
• Markov random fields
• Mobile robots
• Training data
• Labeling
• Computer science
• Laser modes
• Piecewise linear approximation
• Supervised learning
• Maximum likelihood estimation
• 3D Scanning
• Robust Classification
• Markov Random Field
• Point Scanning
• Mobile Robot
• Typical Features
• Training Dataset
• Learning Task
• Data Reduction
• Undirected
• Normal Vector
• Class Labels
• Weight Vector
• Learning Phase
• Quadratic Programming
• Leaf Node
• Subtree
• Partition Function
• Neighboring Points
• Building Walls
• Two-dimensional Scanning
• Original Scan
• Bayes Classifier
• Edge Potential
• Laser Ranging
• Higher Level Of Abstraction