Requirements for Perceptual Systems for Learning of Concepts and Navigation Tasks

The report provides a summary of some of the key issues involved in the design of a robot system for the Home-Tour Scenario. The main emphasis is on the perceptual requirements to enable mapping and annotation of the environment. The components involved in " human augmented mapping " includes: tracking / following of humans, interpretation of human instructions, and geometric mapping & object recognition. For each of the components a brief literature study is provided with a condensed assessment of the different options available for construction of a system. A fundamental component of any mobile system that is to be considered cognitive is spatial mapping of the environment. To be of utility to a user there is a need to generate a " joint " representation between the user and the robot. Such a representation enables the robot to maintain localized and to utilize user provided " labels " to navigate to destinations and recover information / access to places, structures and objects that have been specified by the user. The present report outlines the options available in terms of endowing the system with perceptual competencies that will enable the robot to perform mapping and user based annotation of the environment to generate a common representation of an environment. The report specifically focuses on the different methods that can be used for " human augmented mapping " as part of the Hour Tour Scenario (Key Experiment 1).