Robust place recognition for 3D range data based on point features

The problem of place recognition appears in different mobile robot navigation problems including localization, SLAM, or change detection in dynamic environments. Whereas this problem has been studied intensively in the context of robot vision, relatively few approaches are available for three-dimensional range data. In this paper, we present a novel and robust method for place recognition based on range images. Our algorithm matches a given 3D scan against a database using point features and scores potential transformations by comparing significant points in the scans. A further advantage of our approach is that the features allow for a computation of the relative transformations between scans which is relevant for registration processes. Our approach has been implemented and tested on different 3D data sets obtained outdoors. In several experiments we demonstrate the advantages of our approach also in comparison to existing techniques.

Authors

• Bastian Steder
• Giorgio Grisetti
• Wolfram Burgard

Keywords

• Robustness
• Mobile robots
• Motion planning
• Simultaneous localization and mapping
• Robot vision systems
• Image recognition
• Change detection algorithms
• Image databases
• Spatial databases
• Testing
• Feature Points
• Place Recognition
• 3D Scanning
• Recognition Problem
• False Positive
• Confusion Matrix
• Second Derivative
• Value Of Image
• Aerial Images
• Corresponding Points
• Image Point
• Feature Matching
• Elimination Method
• Recall Rate
• Acceptable Threshold
• Image Position
• Coordinate Frame
• Valid Point
• Pixel Position
• Odometry
• Laplacian Of Gaussian
• Advances In Hardware
• Creation Of Images
• SLAM
• loop closing
• point clouds
• range images
• range sensing