ADAPT is a robotic cognitive architecture, based on the
RS language developed by Damian Lyons of Fordham University. RS is a language
for programming concurrent, distributed sensory-motor schemas, and provides the
high-level abstractions for ADAPT. The cognitive substrate of ADAPT is the Soar
cognitive architecture.
The ADAPT project has been funded by the National Science
Foundation and the Department of Energy, and is a collaboration among Prof. Benjamin
of Pace University, Prof. Damian Lyons of Fordham University and Prof. Deryle Lonsdale
of Brigham Young University.
Here is an overview paper on ADAPT and
a paper on the natural language system.
Check out some video clips.
Tom Achtemichuk designed and implemented
the active vision system, which compares visual percepts with predictions
generated from the CrystalSpace model. The differences between the actual
vision input and the predicted input drives the proposal of Soar operators
to attend to the differences and model them.
Oz Michaeli is integrating Soar with
Ogre3D, an open-source video game platform, which is the basis
for the robot's internal world model. Ogre3D gives the robot
a dynamic world model with sophisticated 3D graphics and a physics
engine. Check out Ogre3D.
We are working on VMSoar, which is an autonomous agent for
network security analysis, which monitors a network and learns to project its
future behavior. This agent can hack autonomously into NT machines and is learning
to hack into XP.
Here is an overview paper on VMSoar.
We are also collaborating with BBN Technologies in a DARPA-funded project
to develop an intelligent agent to
defend against internal cybersecurity threats.
Here is an overview paper on the BBN project.
Bob Follek has implemented an architecture
for an autonomous poker player. His SoarBot
uses Soar
to play poker with the poker server set up by the University of Alberta Computer Poker Research Group.
Bob can be reached at bob@codeblitz.com
