IEEE Robotics & Automation Magazine - June 2011 - 71

Design Principles
The architecture and implementation of RoboEarth is
guided by a number of design principles, centered around
the idea of allowing robots to reuse and expand each
other's knowledge. To facilitate reuse of data, RoboEarth
supports and leverages existing standards. The database is
made available via standard Internet protocols and is based
on an open-source cloud architecture to allow others to set
up their own instance of RoboEarth, resulting in a truly
distributed network. The code generated by the RoboEarth
Consortium will be released under an open-source license,
and will provide well-documented, standardized interfaces. Finally, RoboEarth stores semantic information
encoded in the World Wide Web Consortium (W3C)standardized Web Ontology Language (OWL [44]) using
typed links and uniform resource identifiers (URIs) based
on the principles of linked data [42].
Architecture
RoboEarth is implemented based on a three-layered architecture (Figure 1). The core of this architecture is a server
layer that holds the RoboEarth database [Figure 1(a), the
"Architecture: Database" section]. It stores a global world
model, including reusable information on objects (e.g.,
images, point clouds, and models), environments (e.g.,

Databases
Objects, Environments, Action Recipes

Services

(a)

Action Execution

Environment Modeling
Semantic Mapping

Action and Situation
Recognition and Labeling

Generic
Components

Learning
(b)

Skill Abstraction Layer

Robot
Specific

Proof of Concept
This article describes RoboEarth, a
worldwide, open-source platform
that allows any robot with a network
connection to generate, share, and
reuse data. The work described here
is a first step by the RoboEarth Consortium [43] toward delivering a
proof of concept that
l RoboEarth
greatly speeds up
robot learning and adaptation in
complex tasks
l robots using RoboEarth can execute tasks that were not explicitly
planned for at design time.
This proof of concept includes a
distributed database that stores reusable data for objects, maps, and
tasks. As part of its proof of concept,
the RoboEarth Consortium will also

implement a series of demonstrators centered around the
hospital scenario outlined in the "RoboEarth: Example
Scenario" section.

Server

object locations or a different hospital room) mean that the
downloaded information may not be sufficient to allow
robot B to reperform a previously successful task, this information can nevertheless provide a useful starting point-a
prior. Recognized objects, such as the bed, can now provide
rich occupancy information even for areas not directly
observed by robot B. Detailed object models (e.g., of a bottle)
can increase the speed and reliability of robot B's interactions. Task descriptions of previously successful actions
(e.g., driving around the bed) can provide guidance on how
robot B may be able to successfully perform its task.
This and other prior information (e.g., the previous location of the bottle, and the bed is a likely place to find the
patient) can guide the second robot's search and execution
strategy. In addition, as the two robots continue to perform
their tasks and pool their data, the quality of prior information will improve and begin to reveal the underlying patterns
and correlations about the robots and their environment.
Although many of the requirements for this scenario are
challenging research questions, we believe that the availability of such prior information is a necessary condition for the
robots to operate in more complex, unstructured environments. The benefits of storing, sharing, and reusing information are not restricted to tomorrow's mobile service
robots. Today, thousands of robotic systems solve the same
essential problems over and over again.
Ultimately, the nuanced and complicated nature of
human environments cannot be summarized within a
limited set of specifications but will require robots to systematically share data and build on
each other's experience. We believe
that a World Wide Web for robots
will allow the robots to achieve successful performance in increasingly
complex tasks and environments.

JUNE 2011

IEEE ROBOTICS & AUTOMATION MAGAZINE

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - June 2011