Research in multi-robot systems has made many advances in the last decade, leading to demonstrations of multi-robot teams solving a variety of challenging problems. However, the majority of this prior work assumes teams are composed of large numbers of homogeneous robots, or swarms. While viewing the team as a swarm offers many advantages, such as analytical modeling and built-in redundancy, these research advances are difficult to extend to heterogeneous teams, in which robots are no longer fully interchangeable. Challenging research questions in heterogeneous teams include: How do we enable sensor sharing for strongly cooperative tasks when robots vary in their sensor and effector capabilities? How do we detect faults in such systems, so that robots can recognize when the strong cooperation is breaking down? How can robot team members learn from these faults and more quickly recover in the future? This talk will present three interrelated approaches we have developed to address the issue of achieving strongly cooperative multi-robot teams when dealing with heterogeneity and faulty systems. First, I will present our approach, called ASyMTRe, which is a general technique enabling sensor-sharing and coalition formation in heterogeneous robot teams. The AsyMTRe approach is based on combining schema building blocks to achieve the required information flow through the system. Since we are interested not only forming coalitions, but also ensuring their reliability, I will next present our approach, called SAFDetection, enabling robots in coalitions to detect faults that occur in their cooperation. The SAFDetection approach is based on modeling the sensor readings of the robot team as a probabilistic state transition diagram, then using this model on-line to detect when problems occur. Once faults occur, we want the robot team to be able to learn from that fault in order to more quickly recover in the future. The third approach I will present, called LeaF, is an adaptive causal model approach that enables robot teams to not only deal with modeled faults, but also to use a case-based reasoning approach to extend their knowledge of faults that occur during cooperation, enabling them to more quickly recover from future faults. Together, these three approaches make important advances towards addressing heterogeneity and faulty systems in achieving strongly cooperative multi-robot teams.
Biography
Dr. Lynne
Parker received her Ph.D. degree in Computer Science from the
Massachusetts Institute of Technology (MIT) in 1994, performing
research on cooperative control algorithms for multi-robot systems in
MIT's Artificial Intelligence Laboratory, with a minor in brain and
cognitive science. Dr. Parker joined the faculty of the Dept. of
Computer Science at The University of Tennessee, Knoxville, as
Associate Professor in 2002, founding the Distributed Intelligence
Laboratory at that time. She also holds an appointment as Adjunct
Distinguished Research and Development Staff Member in the Computer
Science and Mathematics Division at Oak Ridge National Laboratory
(ORNL), where she worked as a full time researcher for several
years. Her current research is in the areas of distributed mobile
robotics, artificial intelligence, sensor networks, machine learning,
embedded systems, and multi-agent systems. Dr. Parker's research
has been supported by NSF, DARPA, ORNL, DOE, JPL, SAIC, Caterpillar,
and HRL.
Dr. Parker received the PECASE Award (U.S. Presidential Early
Career Award for Scientists and Engineers) in 2000, the DOE
Office of Scinece Early Career Scientist Award in 1999, the UT-Battelle
Technical Achievement Award for Significant Research Accomplishments in
2000, and the University of Tennessee Angie Warren Perkins Award for
scholarship, teaching, and contributions to campus life in 2006. She
has published over 80 articles in peer-reviewed literature, including
five edited books on the topic of distributed robotics. She is a
frequent invited speaker at international conferences, workshops, and
universities, having given over 90 invited lectures. She is a Senior
Editor of the IEEE Transactions on Robotics, an Associate Editor of
IEEE Intelligent Systems Magazine, and is on the Editorial Advisory
Board of the International Journal of Advanced Robotic Systems.
Dr. Parker is a senior member of IEEE, and is also a member of Sigma
Xi, AAAI, and ACM.