RoboEarth - TU Eindhoven

Today's robots are programmed in such a way that they rely, almost completely, on feedback. Based on the consequences of an action performed by the robot, it tries to make the necessary adjustments. Each has to be programmed in advance and every time a robot has to perform a task the same process has to start all over again. A solution to this problem could be to store the robot's knowledge regarding the actions and the environment. Then, any robot that has access to the stored knowledge can benefit
from it resulting in an improved action. Such a feed forward strategy prevents reinventing the wheel over and over again.

Recently, a consortium consisting of Eindhoven University of Technology (TU/e), Philips Applied Technology, Universität Stuttgart (USTUTT), Eidgenössische Technische Hochschule Zürich (ETHZ), Universidad de Zaragoza (UNIZAR), and Technische Universität München (TUM) started a project called RoboEarth. The goal is to successfully design and implement a world-wide web-style database called RoboEarth. RoboEarth should receive and store knowledge obtained by a robot and supply it to any
robot that can benefit from it, independent of the robot's hardware and configuration. Apart from this database, an interface between the database and the systems referring to
the database, and a language that can be used for communication have to be developed. As a proof of principle, a service robot for patient companion in a hospital environment will be built as a demonstrator. This service robot will, e.g., serve a drink or open a screw cap bottle.

Category:

Education

Tags:

• TU/e
• RoboEarth
• roboearth
• robots
• amigo
• self-learning
• robot
• TU
• Eindhoven
• human-machine
• interaction
• electronics
• holland