IEEE Robotics & Automation Magazine - June 2011 - 36

Guiding Principles for Service-Oriented Robots
In many cases, the challenges of robotic services considerably overlap with the provisions offered by SOC, specifically in the area of Web services. As such, there are a
number of guiding principles that can translate from the
SOC world to enable the design of service-oriented interfaces for robotics.

Service Broker

Human
Editing

WIKI

.srv Files
.msg Files

ROS
Run-Time
Environment
f (x )
Service
Requester
(ROS
Environment)

Customized
RPC
Communication

Service
Provider
(ROS
Environment)

Figure 2. Service-oriented computing with ROS.

based on these standards. As such, applications developed
by a diverse set of vendors can still be integrated. Both
ROS and RDS have customized messaging formats that
exclusively facilitate operations within their respective
environments. A feature of SOC paradigms is the use of
machine-interpretable, markup languages for representing
interfaces and messages. Both ROS and RDS use text files
and event-based specifications that are tightly coupled
with their respective operating environments. Moreover,
ROS and RDS use human-generated repositories (such as
WIKIs) to describe the software available within their
respective frameworks. Figure 2 illustrates the ROS environment considering the tenets of SOC.

Relevant Service-Based Robot Capabilities
Even with the existing service-oriented development environments, such as ROS and RDS, discovering and integrating existing software modules is a challenge. This discovery
process involves identifying relevant modules (i.e., services)
that can be integrated into an overall structure that achieves
global task objectives. In this work, we have identified a
number of core functions which, when accessed from distributed locations over the Internet, could increase the ease
of designing a robot application. This taxonomy is illustrated in the unified modeling language (UML) class diagram in Figure 3, which shows the potential capabilities
related to direct robot services and those related to the
robot's client. In this example, data from both the robot's
sensors and its learning modules are used to control actuators and clients can view robot output or provide teleoperative input through a graphical user interface.
Based on this taxonomy, the authors have developed
robotic learning module to control the actuation of a
robotic arm based on using SOC concepts (which is further
discussed in the "Case Study: Instrumenting ROS-Based
Services with Traditional WWW Resources" section). The
overall application, as depicted in Figure 4, leverages three
primary robotic service modules: image analysis, a human
graphical user interface, and a memory module service.
The image analysis module receives data from a camera
sensor that then produces images associated with the location of objects and the robotic arm in context of its actions.

Communicates with

Robot Services

External Actor Client Services
Is a

Is a

Human-Control Graphical Interface
Sensing
Takes Data from

Computation

Action

Is a
Display

Teleoperative Control Module

Has a
Camera

Is a
Thermometer

Microphone

Learning Module

Image Processing Module

1
1..*
Controls
1
1
Wheel

Has Many

IEEE ROBOTICS & AUTOMATION MAGAZINE

Has a

*
Rules (Event-Cond-Action) Image Conversion

Figure 3. Toward robot capabilities as reusable Web-accessible services.

Joystick

Motor

Arm

Is a

Haptic

JUNE 2011

Brain Interface

Image Analysis

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