Inter-classifier feedback for human-robot interaction in a domestic setting

Abstract

For a mobile robot that interacts with humans such as a home assistant or a tour guide robot, activities such as locating objects, following specific people, and distinguishing among different people are fundamental, yet challenging robotic vision tasks. For complex object recognition and tracking tasks such as person recognition and tracking, we use the method of inter-classifier feedback to track both uniquely identifying characteristics of a person (e.g. face), and more frequently visible, but perhaps less uniquely identifying characteristics (e.g. shirt). The inter-classifier feedback enables merging multiple, heterogeneous sub-classifiers designed to track and associate different characteristics of a person being tracked. These heterogeneous sub-classifiers give feedback to each other by identifying additional online training data for one another, thus improving the performance of the overall tracking system. We implement the tracking system on a Segway base that successfully performed aforementioned activities to a second place finish in the RoboCup@Home 2007 competition. The main contribution is a complete description and analysis of the robot system and its implemented algorithms.