Page - 63 - in Proceedings of the OAGM&ARW Joint Workshop - Vision, Automation and Robotics
Image of the Page - 63 -
Text of the Page - 63 -
Fig. 2. Overall software architecture with links to the central refbox, the
team server and the connectedn robots.
of the paper. These skills are provided to the next higher
layer, the mid-layer, via defined messages.
The mid-layer, therefore, can use these skills to perform
more complex tasks such as exploring a zone of the game-
field, get a base from the base station or deliver the product
holding in its gripper. Additionally, a first error detection
and recovery behavior are implemented here, e.g. the system
checks if there is a product in the gripper after the low-level
hassuccessfullygrabbedsomething.Thesecomplex tasksare
again provided via defined protobuf messages to our highest
layer, the team server.
Here a central knowledge-base is held and a game strategy
is derived (see Section IV-B for further details). This central
point enables the system to conclude a global optimal game
strategy for the complete robotic fleet. The global strategy
is derived using a simple planning system which uses a hi-
erarchical task network [10] to properly create the products.
Due to the centralized knowledge base one does not need to
deal with synchronization of knowledge bases of the robot
or distributed planning. Instead a “simpler” approach for
planning can be applied. Fig. 3. Communication between the different layers for one robot using
Google’s protocol buffers.
IV. SELECTED SOFTWARE COMPONENTS
To get an idea of the functional interaction of our robot
system some selected components are presented. First of
all, two low-level modules necessary to detect and identify
a machine are presented in SectionIV-A. Additionally, the
scheduling algorithm (located at the high-level component of
our system) which manages the discovery of the unknown
game-field is presented in Section IV-B.
A. Machine Detection and Identification
To be able to gather information about the unknown envi-
ronment it is necessary for the robot to recognize elements
surrounding it. One important type of these elements is a
modular production system, i.e. the machines capable of
producing the ordered products. To identify the machines
AR-tags are used which are placed at two sides of these
machines.
63
Proceedings of the OAGM&ARW Joint Workshop
Vision, Automation and Robotics
- Title
- Proceedings of the OAGM&ARW Joint Workshop
- Subtitle
- Vision, Automation and Robotics
- Authors
- Peter M. Roth
- Markus Vincze
- Wilfried Kubinger
- Andreas MĂĽller
- Bernhard Blaschitz
- Svorad Stolc
- Publisher
- Verlag der Technischen Universität Graz
- Location
- Wien
- Date
- 2017
- Language
- English
- License
- CC BY 4.0
- ISBN
- 978-3-85125-524-9
- Size
- 21.0 x 29.7 cm
- Pages
- 188
- Keywords
- Tagungsband
- Categories
- International
- Tagungsbände