List of Figures

• An overview of the method.
• Pose constraints in the plane.
• Uncertain pose constraints in the plane.
• Figure-ground ambiguity in the interpretation of two objects.
• A table passing in front of a door.
• Detected landmarks in an image.
• A cross-section of the density function.
• Output of the landmark detector.
• The training process.
• Landmark prototypes and corresponding eigenlandmarks.
• A typical landmark set.
• The initial images and landmark candidates.
• Tracked landmarks and eigenlandmarks built from the bootstrap image.
• Results of adding Figure  to the database.
• Results of adding Figure  to the database.
• The final set of a) prototypes and b) principal components for a traversal of the environment depicted in part in Figure .
• Landmark-prototype matches for a single image.
• The recovery operation.
• Convergence properties for a single training set.
• Position estimate for a single test image.
• Merged ATs.
• A set of filtered predictions.
• Scene I.
• The set of tracked landmarks extracted from Scene I.
• Position estimates and corresponding ground truth for twenty random samples from Scene I.
• Parameter variation results for Scene I.
• Appearance-based estimation error for Scene I.
• Individual appearance-based estimation results for Scene I.
• Estimation results for edge-based estimation.
• Estimation results for edge-based estimation using only appearance.
• Scene II.
• Scene II pose estimates for 100 test cases, and .
• Scene II pose estimates for 100 test cases, and .
• Scene III.
• The Nomad 200.
• Results for Scene III.
• Scene IV.
• The RWI with mounted camera.
• The set of pose estimates obtained for Scene IV.
• Altered Scene IV
• Results from altered Scene IV.
• The consistency measure plotted as a function of orientation.