ROBOTICS

[This page is CSS2 enabled. Your browser might not fully support it]

[http://www.ee.oulu.fi/research/robotics/roseman/]

ROSEMAN - Robots Serving Humans

$RCSfile: index.html,v $ $Revision: 1.9 $ $Date: 2006/02/09 09:59:59 $

ABSTRACT

The purpose of the this project is to develop components for intelligent mobile robot that operates with humans in every day environments. The emphasis of the research is on the following areas: machine vision, control systems, learning, and teleoperation. The motivation for the basic research conducted in this project stems from the future needs of the mobile robot applications. The developed components can also be utilized in different and simpler devices than mobile robot. Tomorrow’s communication devices provide numerous application scenarios as the devices can be used to collect information from the environment, which can further be used to activate various functionalities based on the current state of the environment – just like mobile robot does. Intelligent devices that aid elderly and disabled people in their every day life also provide an interesting application area. During the project a color vision system were developed for tracking multiple colored objects simultaneously. The system can be used, for example, to track a person based on her skin color, which was also demonstrated. In this project a neural network based evolutionary computation techniques were developed for mobile robotics. During the project a method was developed which forms network structures, consisting of laterally interacting neurons, which are capable of controlling a mobile robot.    The feasibility of the approach was demonstrated by evolving a navigation behavior for a real robot. The approach was also tested with non-Markov double-pole-balancing problem in which the method turn out to be the most efficient one among the tested methods. Robot co-operation was also studied in this project. The emphasis was on robot formations – moving in formations and dynamic change of formations.  This research utilized graph theoretical approaches for robot formations and a thesis was written about the subject to the department of mathematics. A localization method that uses color histograms and ICondensation algorithm was developed for mobile robots. The method worked well in environments where spatial locations and their color properties were related. The Samba-architecture was further developed for real robots. The Samba-architecture was previously used in RoboCup simulation league.  In this project the modification of the Samba-architecture was used in the coffee serving system in which a mobile robot serves a cup of coffee, prepared by a robotic arm, for a person. An important part of the coffee serving system was the distributed XML-based state machine architecture developed during the project. The architecture provides a mean to dynamically implement state machines, which utilize available robotic resources (e.g. sensors and actuators), from corresponding XML-descriptions. The state machine system can also be used in other domains than mobile robotics in a straightforward manner.During the project, a voice controlled teleoperation was demonstrated. By using a video conferencing software the user can control a remote mobile robot using simple high level speech commands. The operator can activate high level behaviors via speech and monitor the performance of the robot. This research is a first step towards natural human-robot interaction.

Contact information

janne . haverinen  (at) ee . oulu . fi

[http://www.ee.oulu.fi/research/robotics/roseman/]

[This page is CSS2 enabled. Your browser might not fully support it]