Augmenting Appearance-Based Localization and Navigation using Belief Update

Appearance-based localization compares the current image takenfrom a robot's camera to a set of pre-recorded images in order toestimate the current location of the robot. Such techniques oftenmaintain a graph of images, modeling the dynamics of the imagesequence. This graph is used to navigate in the space of images. In this paper we bring a set of techniques together, includingPartially-Observable Markov Decision Processes, hierarchical staterepresentations, visual homing, human-robot interactions, and soforth, into the appearance-based approach. Our approach providesa complete solution to the deployment of a robot in a relativelysmall environment, such as a house, or a work place, allowing therobot to robustly navigate the environment after minimal training. We demonstrate our approach in two environments using a realrobot, showing how after a short training session, the robot is ableto navigate well in the environment.

Authors

• George Chrysanthakopoulos
• Guy Shani MSAI, Microsoft & Ben Gurion University of the Negev

Keywords

• Localization
• Navigation
• Topological SLAM
• Hierarchical models
• POMDP