Robust Localization Using Context in Omnidirectional Imaging

This work presents the concept to recover and utilize the visual context in panoramic images. Omnidirectional imaging has become recently an efficient basis for robot navigation. The proposed Bayesian reasoning over local image appearances enables to reject false hypotheses which do not fit the structural constraints in corresponding feature trajectories. The methodology is proved with real image data from an office robot to dramatically increase the localization performance in the presence of severe occlusion effects, particularly in noisy environments, and to recover rotational information on the fly.

Authors

• Lucas Paletta
• Simone Frintrop
• Joachim Hertzberg

Keywords

• Robustness
• Bayesian methods
• Mobile robots
• Working environment noise
• Solid modeling
• Sonar navigation
• Biological system modeling
• Robot sensing systems
• Robot vision systems
• Cameras
• Omnidirectional Images
• Performance In The Presence
• Noisy Environments
• Robot Navigation
• Occlusion Effect
• Panoramic Images
• Severe Occlusion
• Bayesian Reasoning
• Field Of View
• Autonomic System
• Local System
• Object Recognition
• Training Images
• Visual Input
• Spatial Context
• Similar Appearance
• Local Cues
• Input Noise
• Local Sector
• Eigenspace
• Single Sector
• Bayesian Context
• Decision Fusion
• Navigation Performance