COLLOQUIUM
Department of Computer Science and Engineering 
University of South Carolina
 
Unmanned Air Vehicle: A Team Approach 

Régis Vincent
SRI International

Date: May 15, 2002 (Wednesday)
Time:  1:30-2:30PM
Place: Swearingen 1A03 (Faculty Lounge)



Abstract not Available



Régis Vincent is an active member of the multiagent community. He
recently joined SRI International, where he is a Computer Scientist.
Dr. Vincent received his Ph.D. from the University of Nice, France in
1997. His research interests are real-time artificial intelligence,
from scheduling to planning to negotiation.

Since joining SRI, Dr. Vincent has been working on monitoring and
adaptation for physical systems, and on the Unmanned Combat Air
Vehicle (UCAV) project.  The goal of the project is to demonstrate the
collaboration and adaptation of a team of UCAVs and UGVs (Unmanned
Ground Vehicles) to achieve tactical missions.  Dr. Vincent and the
project team are developing a new multi-level adaptation architecture
to monitor in real-time the progress of the team towards its different
goals. Each team member's jobs are adapted to compensate for
unexpected events (e.g. failures, new missions, and new objectives).
Dr. Vincent designed and implemented the UCAV simulator.

Before joining SRI, Dr. Vincent was Senior Research Scientist at the
University of Massachusetts in the Multi-Agent System Lab under the
direction of Professor Victor Lesser. During this period, he was
project leader of the DARPA ANTS project, where he supervised and
designed a real-time solution for target tracking using radar sensors,
including algorithms for real-time allocation that save resources by
responding to targets of opportunity.  Dr. Vincent was a contributor
on the DARPA SAFER (Self-Adaptive Software) project, where a fleet of
robots was sent into a building to map, evaluate threats and report
information. He helped design a new robot and the control architecture
that allows the easy integration of new type of sensors.

Dr. Vincent also worked on agent adaptation in the survivability
project, which developed distributed detection and diagnosis
algorithms to be used for recognizing and explaining the cause of
unacceptable performance of a distributed, multi-agent system. The
explanation is used to reorganize processing across agents, thus
improving survivability when there are software errors, hardware
malfunctions or hostile attacks.