Interoperable RT component for object detection and 3D pose estimation for service robots

Finding objects and tracking their poses are essential functions for service robots, in order to manipulate objects and interact with humans. We present an approach for object detection and 3D pose estimation for autonomous mobile robots, that is suitable for general uses in a modularized robot control system. Our approach extracts local features from the input images, searches for the reference pattern, and then produces the 3D pose in camera coordinate system, using only a single reference image and the 6-DOF pose in it. We have created an RT (Robot Technology) component that can be used in any RT-based system, and developed an algorithm that can extend the range of detection and produce robust pose estimation. For evaluation, we have integrated our vision component in an autonomous robot system with a search-and-grasp task, and tested it with several objects that are found in ordinary domestic environment. We present the details of our approach, the design of our modular component design, and the results of the experiments in this paper.