by M Tenorth and M Beetz
Abstract:
This document describes the current state of implementation of the RoboEarth representation language. This language is designed for two main purposes. First, it should allow to represent all information a robot needs to perform a reasonably complex task. This includes information about (1) Plans, which consist of the actions a task is composed of, ordering constraints among them, monitoring and failure handling, as well as action parameters like objects, locations, grasp types; (2) Objects, especially types, dimensions, states, and other properties, but also locations of specific objects a robot has detected, and object models that can be used for recognition; and the (3) Environment, including maps for self-localization as well as poses of objects like pieces of furniture. The second main task of the RoboEarth language is to allow a robot to decide on its own if a certain piece of information is useful to it. That means, a robot must be able to check if an action description contains a plan for the action it would like to do, if it meets all requirements to perform this action, and if it has the sensors needed to use an object recognition model. Using the semantic descriptions in the RoboEarth language, a robot can perform the checks using logical inference.
Reference:
Deliverable D5.2: The RoboEarth Language – Language Specification (M Tenorth and M Beetz), Technical report, FP7-ICT-248942 RoboEarth, 2010.
Bibtex Entry:
@techreport{tenorth_deliverable_2010,
author = {M Tenorth and M Beetz},
title = {Deliverable D5.2: The {RoboEarth} Language – Language Specification},
institution = {{FP7-ICT-248942} {RoboEarth}},
year = {2010},
number = {D5.2},
abstract = {This document describes the current state of implementation of the
{RoboEarth} representation language. This language is designed for
two main purposes. First, it should allow to represent all information
a robot needs to perform a reasonably complex task. This includes
information about (1) Plans, which consist of the actions a task
is composed of, ordering constraints among them, monitoring and failure
handling, as well as action parameters like objects, locations, grasp
types; (2) Objects, especially types, dimensions, states, and other
properties, but also locations of specific objects a robot has detected,
and object models that can be used for recognition; and the (3) Environment,
including maps for self-localization as well as poses of objects
like pieces of furniture. The second main task of the {RoboEarth}
language is to allow a robot to decide on its own if a certain piece
of information is useful to it. That means, a robot must be able
to check if an action description contains a plan for the action
it would like to do, if it meets all requirements to perform this
action, and if it has the sensors needed to use an object recognition
model. Using the semantic descriptions in the {RoboEarth} language,
a robot can perform the checks using logical inference.},
}