Dr. Damian Lyons Associate Professor Department Chair Department of Computer & Information Sciences 320 John Mulcahy Hall 441 East Fordham Road, Bronx NY 10458 Email: dlyons(at)cis.fordham.edu Phone: 718-817-4485 Dr. Lyons's Homepage Dr. Lyons's Short CV

Bio

Dr. Damian M. Lyons is an Associate Professor of Computer Science at Fordham University. Prior to this, he worked for 15 years as a researcher and research program manager in the area of Computer Vision and Robotics.  He completed his undergraduate education in Math, Engineering and Computer Science at Trinity College, University of Dublin in Ireland. He earned his Ph.D. in Computer Science from the University of Massachusetts at Amherst for research on a formal model of computation for sensory-based robotics (1986).
    He worked for many years as a researcher in the US branch of the corporate research laboratories of Philips Electronics, the European Semiconductor and Consumer Electronics giant. His work there included research in representing and analyzing robot action plans, integrating reactive/behavior-based & deliberative approaches to action planning, multimodal user interfaces, and automated video surveillance. Dr. Lyons served as project leader for Philips' research activities in Automated Video Surveillance, and later as Department head for the Video and Display Processing research department, responsible for technical leadership and funding for this diverse group. Dr. Lyons is currently the Director of Fordham’s Computer Vision and Robotics Lab, and Associate Chair for Graduate Studies in the Computer & Information Science Department.
    From 1990 through 1995, Dr. Lyons served as chair of the IEEE Robotics & Automation technical committee on assembly and task planning. He has served on numerous conference program committees, has published over 80 technical papers in conferences, journals and books, and holds 8 patents. Dr. Lyons is a member of ACM and IEEE.

Research

• the integration of planning and reaction in robot systems
• automated video surveillance, and
• vision-based human-machine interfaces.

I am involved in two pieces of research in my role as Directory of the Computer Vision & Robotics Laboratory at Fordham:

• Performance Guarantees for Emergent Behavior in Mobile Robot Systems

The objective of this research program is to develop the necessary advances in theory and software to build robot systems that can be deployed in critical environments in a safe, effective and reliablefashion. The approach in this proposal is based on understanding what ar ethe computational characteristics of the behavior-based approach to robotics. Unique characteristics include: the necessity for some form of sensory-motorstructure such as Schemas, the necessity for asynchronous methods of composing concurrent behaviors, and the conclusion that even simple behavior-based systems exhibit complex behavior when acting in a complex environment.
    We are developing a succinct formalism that captures these issues, and especially addresses the issue of modeling the complex environment. A process-based environment model is being developed, allowing a shared vocabulary (processes) between robot controller and environment model. It employs the Port Automata model as an operational semantics, and a CSP like selection of process composition operators for process description. The emergent behaviors of a robot interacting with an active and dynamic environment can be modelled and explored with this approach.

• Multi-Sensor Fusion and Target-Tracking for Automated Surveillance

Target tracking in CCTV (Close-Circuit TV) surveillance involves determining which portions of an image in a CCTV video sequence corresponds to which surveillance target - where targets are typically people. The standard approaches developed for point target tracking, e.g.,MHT and JPDAF, have been applied to visual target tracking with some success. However, the information from a video sequence, even from a single camera, is much richer than from the point tracking applications with which multi-target methods originated. For this reason a crucial problem becomes the integration of multiple sensory cues, especially in cases where some cues can be misleading some of the time.
    We are studing sensory fusion modules based on the ”Rankand Fuse” (RAF) method, (doc) which is a general, efficient and easily scalable approach to sensory fusion. The RAF method considers both the spatial and temporal results from the sensory system, and offers an elegant method to provide top-down feedback from the application system to the fusion process, so that task context can be used to select a sensory fusion method appropriate for the specific task.

Selected Publications

• D.M. Lyons (2006). "Combinatorial fusion Criteria for Real-Time Tracking", International Conference on Information Fusion.

• D.M. Lyons (2006). "Combining Multiple Scoring systems for Video Target Tracking Based on Rank-Score Function Variation", Symposium on the intergace of statistics, computing science, and applications (Interface 2006).

• D.M. Lyons (2006). "Feature selection for real-time tracking, Multisensor, Multisource Information Fusion: Architectures, Algorithms, and Applications 2006", SPIE Defense and Security Symposium.

• D.M. Lyons (2006). "Combinatorial Fusion Criteria for Real-Time Tracking", International conference on Advanced Information Networking and Application (AINA 2006).

• D.M. Lyons (2006). "Developing a Cognitive Architecture to be Embedded in the Physical World", Behavior Representation in Modeling and Simulation (BRIMS2006).

• D.M. Lyons (2006). "Obstacle Avoidance using Predictive Vision Based on a dynamic 3D World Model", SPIE conference on Intelligent Robots and Computer Vision.

• D.M. Lyons (2006). "Embodying a Cognitive Model in a Mobile Robot", SPIE conference on Intelligent Robots and Computer Vision.

• D.M. Lyons and R. Arkin (2004). "Towards Performance Guarantees for Emergent Behavior", IEEE International Conference on Robotics and Automation, New Orleans, LA.  [pdf]

• J. Drysdale and D. Lyons (2004). "Learning Image-based Landmarks for Wayfinding using a Neural Network", Artificial Neural Networks in Engineering, St. Louis, MO.  [pdf]

• D.P. Benjamin, D. Lonsdale, and D. Lyons (2004). "Designing a Robot Cognitive Architecture with Concurrency and Active Perception", AAAI Fall Symposium on Cognitive Science and Robotics, Washington DC, October 2004.

• D.P. Benjamin, D. Lonsdale, and D. Lyons (2004). "Integrating Perception, Language and Problem Solving in a Cognitive Agent for a Mobile Robot", Third International Joint Conference on Intelligent Agents and Multiagent Systems, NY, July 2004.  [doc]

• D.P. Benjamin, D. Lyons, and D. Lonsdale (2004). "ADAPT: A Cognitive Architecture for Robotics", 2004 International Conference on Cognitive Modeling, Pittsburgh, PA, July 2004.  [pdf]

• D.F. Hsu, D.M. Lyons, C. Usandivaras and F. Montero (2003). "RAF: A Dynamic and Efficient Approach to Fusion for Multitarget Tracking in CCTV Surveillance", IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems, Tokyo, Japan, July 29-August 1, 2003.  [doc]