Appearance-based localization for mobile robots using digital zoom and visual compass

This paper describes a localization system for mobile robots moving in dynamic indoor environments, which uses probabilistic integration of visual appearance and odometry information. The approach is based on a novel image matching algorithm for appearancebased place recognition that integrates digital zooming, to extend the area of application, and a visual compass. Ambiguous information used for recognizing places is resolved with multiple hypothesis tracking and a selection procedure inspired by Markov localization. This enables the system to deal with perceptual aliasing or absence of reliable sensor data. It has been implemented on a robot operating in an office scenario and the robustness of the approach demonstrated experimentally.

• Type: Journal
• Year: 2008
• Where: RAS
• Authors: Nicola Bellotto, Kevin Burn, E. Fletcher, Stefan Wermter
• Keywords: Appearancebased Place Recognition, Dynamic Indoor Environments, RAS 2008, Reliable Sensor Data