Welcome to ADE!
ADE is the Agent Development Environment
conceived and developed by Matthias
Scheutz with the help of several collaborators (most notably
James Kramer and Paul
Schermerhorn)
over the last twelve years.
ADE
is a multi-agent system (MAS) middleware intended for the design,
implementation, and testing of
distributed agent architectures with special emphasis on complex
robotic systems and human-robot interaction (HRI).
Historically, ADE started as a set of JAVA classes to communicate
with
simple robots over a serial interface in 1997. These JAVA
classes
were extended in 1999 at the University
of
Notre Dame under the name of AgeS
to interface various robots (including the Trilobot and
various
MobileRobots Pioneer
robots). The universal agent architecture
framework APOC
developed in the early 2000s led to a systematic
reorganization of AgeS, with particular attention to the needs of
complex agent architectures for robots. Further restructering
of
APOC/AgeS in the 2004-2006 time frame resulted in an extensive code
based that consisted of both infrastructure components as well as
implemented components with specific functionalities (e.g., audio
and
vision processing components, planning components, reasoning
components, natural language components, etc.). The
infastructure
was renamed into ADE and
made
available both as alpha (in 2004) and beta versions (in 2006) on sourceforge.
In recent years, ADE has
been demonstrated on a variety of additional robots (including the Willow
Garage PR2, the Xitome MDS, the Segway
RPM, the Aldebaran
Nao,
as well as various purpose-built robots such as the Indiana
University Golfcart).
ADE is currently used by
several research labs in North America and Europe, most notably by
the Human-Robot Interaction
Laboratory
at Tufts University
(USA),
for developing complex robotic architectures with natural language
capabilities for human-robot interaction (click here
for a demo of the DIARC architecture implemented in ADE). It
has
also been used both on Unix-based platforms (like Linux, Solaris,
and
MacOSX) and Windows-based platforms (like Windows XP, Windows2000,
and
Windows Vista), although support is only provided for Linux.
While ADE has been
developed
with a focus on complex robots, it has been used in a variety of
other,
non-robotic applications, including live audio and video streaming
in
distributed systems, intelligent computerized survey taking,
multi-modal interactions with smart devices, etc.
ADE is unique among
robotic
development environments (RDE) in that it integrates detailed
access
control and security mechanisms into the MAS middleware, thus
allowing
for the secure execution of distributed architecture components on
possibly untrusted hosts. ADE also allows for the automatic
configuration and startup of distributed architectural components,
provides system-wide logging capabilities and multi-session
graphical user interfaces for administrating and monitoring a
distributed ADE
system. Moreover, ADE uniquely integrates automatic
mechanisms
for fault detection and systematic error recovery, which can be
accessed and tightly integrated into the agent control
architecture
implemented in ADE. Finally, ADE provides standard
interfaces to
common RDEs like Player/Stage, ROS, Carmen, and MARIE. For
publications on ADE describing its various features, click here.
To browse the javadocs of ADE and several ADE components
developed
as part of the robotic DIARC architecture, click here.
For a simple demonstration of ADE and an architecture to play
with,
consider the four programming assignments given as part of the
behavior-based
robotics class at Indiana University Bloomington: toy1,
toy2, toy3, and toy4.
The current 1.0 gamma version of ADE from January 2010 (available
here
on sourceforge for download) has been extensively tested in a
variety of robotic
applications. There are still some known bugs that are currently
being ironed out
and features lacking that are currently being implemented (most
notably on the GUI side).
Last modified by Matthias Scheutz on October 29, 2011