Cesar Bandera, Ph.D.
EducationCert., Harvard School of Management, Executive Development, 1996.
Ph.D., University at Buffalo, Electrical and Computer Engineering, 1990.
M.S., University at Buffalo, Electrical Engineering, 1984.
Expertise and Research Interests
My technical background is in active perception, which combines real-time computer vision and other sensor modalities with machine learning
and behavioral control. I am interested in all aspects of active vision, including algorithms for signal processing and control, sensor VLSI,
and multiprocessing architectures. I am also interested in offsetting the limitations of each of these components through the purposive i
ntegration into operational systems.
My experience in active perception comes largely from my work in foveal vision, which exploits in the machine setting the multiacuity properties prevalent in vertebrate vision. Foveal vision is intended for autonomous platforms with diverse behaviors competing for limited perception resources, such as navigation, search, recognition, and manipulation. It supports a more refined allocation of perceptual resources, but also requires a tighter coordination of all the aspects of active vision.
While at AT&T Labs, which coincided with the peak of broadband multimedia investments by the telecommunications industry, I became involved with rich media. Since then, I have been active in the field of pervasive rich media. This field endeavors to provide spatiotemporally coordinated multimodal streams to an audience with diverse demographics, player platforms and channel access (e.g., broadband-connected PCs, wireless PDAs, set-top boxes).
Pervasive rich media builds upon object oriented multimodal architectures such as MPEG-4 (the complex profiles, not just a video codec or vector graphics animation) to automate the on-demand tailoring of content and to cost-effectively support audience diversity. I am also involved in extending ratified standards so as to support the many de-facto standards in use today.
Academe Accomplishments
Dissertations Supported (Committee Member, Technical Direction and Funding)
1. 'Reinforcement Learning Gaze Control,' Kit Shan Choy, University of Buffalo Department of Computer Science and Engineering, 2000.
2. 'Issues in the Design of Pipelined VLSI Circuits for DSP Applications,' Dipankar Talukdar, University of Buffalo Department of Electrical and Computer Engineering, 1997.
3. 'Multiresolution Object Recognition usingNeural Networks,' Susan Shiqiong Young, University of Buffalo Department of Electrical and Computer Engineering, 1995
4. 'Representing and Learning Routine Activities,' Henry Hexmoor, University of Buffalo Department of Computer Science, 1995.
5. 'A Computational Model of Color Perception and Color Naming,' Johan M. Lammens, University of Buffalo Department of Computer Science, 1994.
Selected Invited Lectures and Courses Taught
· 'Computer Vision,' Univ. of Buffalo; School of Medicine, Faculty of Applied Sciences (1995, 1996, 1997)
· 'Advanced Topics in Active Vision,' Cornell (1998), Brock University Distinguished Lecturer Series, Canada (1996), University of Malaga, Spain (1995, 1996, 1997, 2000 - 2004)
· 'Introduction to Hierarchical Foveal Vision,' Brock University Distinguished Lecturer Series, Canada (1995), University of Malaga, Spain (1994), Dipartimento di Informatica Sistematica e Telematica of the University of Genova, Italy (1994), Scuola Superiore Santa Anna Advanced RoboticsTechnology and Systems (ARTS) Facility, Pisa, Italy (1994)
· 'Electronic Circuits,' Univ. of Buffalo Dept. of Electrical and Computer Engineering (1982, 1983)
Industry Accomplishments
In 1990, I formed a research department at Amherst Systems dedicated to the development and application of active vision. This work yielded operational platforms with algorithms for video understanding and automaton behavior control, matching multiprocessor architectures, and smart VLSI imaging sensors (imagers with monolithic signal processing). I had profit/loss responsibility, and was able to secure external funding for all R&D (over twenty customer grants and contracts) while exceeding growth and profit estimates.
The technology is now in use in several government programs including NASA's Extravehicular Activity Helper/Retreiver robot, and target tracking systems at the Naval Air Warfare Center. The department is affiliated with NASA's Jet Propulsion Laboratory, SUNY Buffalo, and the University of Malaga.
To date, this active vision research has yielded six Ph.D. and four M.S. degrees, several patents, highest distinction in the Air Force Small Business Innovative Research Accomplishments Report to the U.S. Congress, a Small Business of the Year nomination from Rome Lab, and the 1999 NASA Space Act award from Johnson Space Center.
In 2001, I formed a research department at Manhattan-based Sorceron (now BanDeMar) dedicated to the synthesis and delivery of object-oriented rich media. As CTO, I am responsible for the company’s offerings in its target government markets: e-learning, scientific outreach, and surveillance/security. Through the National Science Foundation’s SBIR/STTR program, I have been able to bootstrap Creneaux’s e-learning operations without equity investment.
Boards
· Rutgers Center for Advanced Information Processing (CAIP) Industry Advisory Board
· InfoValue Technical Advisory Board
· Face2Face (Bell Labs spin-off) Board of Advisors
Reviewer
· National Science Foundation
· Autonomous Robots Journal, Machine Learning Journal (Kluwer Academic Publishers)
· State University of New York Multidisciplinary Pilot Project Program
Memberships
· Association for Computing Machinery
· Institute of Electrical and Electronics Engineers
· International Society for Optical Engineering
Honors, Awards, Patents
· 1999, Space Act Award, NASA, Contribution to Space Exploration
· 1993, Highest Distinction, Small Business Innovative Research Accomplishments Report to the U.S. Congress, US Air Force, Excellence in Research Commercialization
· 1992, Small Business of the Year Nomination, US Air Force, Excellence in Research
· CMOS foveal image sensor chip, Patent Number: 6455831, 2002
Funding Secured
· DARPA: Biological techniques for machine vision, $75,000, Oct 1991 to Apr 1991.
· National Aeronautics and Space Administration (NASA): Rich media engine for hyperspectral collaborative workspaces, $70,000, Nov 1, 2001 to Apr 30, 2002.
· National Science Foundation (NSF): Education on demand for technique training Phase II, $442,000, Mar 15, 2003 to Apr 30, 2005.
· DARPA: Attention in active vision, $75,000, Mar 1992 to Sep 1992.
· National Science Foundation (NSF): Education on demand for technique training Phase I, $100,000, Jan 1, 2002 to Jun 30, 2002.
· US Air Force Wright Laboratory: Reinforcement learning for visual attention, $75,000, Jan 1, 1995 to Jun 30, 1995.
· National Science Foundation (NSF): Foveal gaze control algorithms, $75,000, Jan 1994 to Jun 1994.
· National Science Foundation (NSF): Research Commercialization Award, $197,365, 2004 to 2005.
· National Institutes of Health (NIH): Educating Schools and Families with ToxRAP over MPEG-4, $100,000, 2004 to 2005.
· Air Force Office of Scientific Research (AFOSR): Second generation smart foveal imager, $500,000, 1999 to 2001.
· National Science Foundation (NSF): Microminiature 2-D spherical pointing actuator, $300,000, 1997 to 1999.
· Office of Naval Research: Ground vehicle foveal cueing system development, $750,000, 1997 to 1999.
· National Aeronautics and Space Administration (NASA): Foveal cueing sensor development, $500,000, 1997 to 1999.
· US Air Force Research Laboratory, Eglin AFB: Development of reconfigurable foveal cueing system, $70,000, 1997 to 1997.
· Office of Naval Research: Foveal sensor development, $650,000, 1996 to 1998.
· US Army Research Laboratory: Hierarchical foveal algorithm development for ATR, $650,000, 1995 to 1997.
· National Aeronautics and Space Administration (NASA): Extravehicular helper-retriever robot development, $500,000, 1995 to 1997.
· US Air Force Wright Laboratory: Multispectral scene synthesis for ATR, $750,000, 1993 to 1995.
· US Air Force Wright Laboratory: Integration of the Tactics and Equipment Evaluator with C3I signal simulator, $300,000, 1990 to 1991.
· Office of Naval Research: Active pursuit algorithms for foveal vision, $500,000, 1990 to 1992.
· US Air Force Rome Laboratory: Non Von Neumann software life cycle hypertext database, $500,000, 1989 to 1991.
· US Air Force Wright Laboratory: Real-time Ada SUPPRESSOR study, $300,000, 1988 to 1990.
· US Air Force Wright Laboratory: Tactics and Equipment Evaluator development, $650,000, 1986 to 1988.
· National Science Foundation (NSF), 9362112, Foveal Machine Vision for Mobile Robots Using Agent Based Gaze Control, $64,996 (Estimated), April 1, 1994-January 31, 1995 (Estimated)
· 9560519, SBIR Phase I: Spherical Pointing Actuator, $74,542 (Estimated), May 1, 1996-October 31, 1996 (Estimated)
Selected Publications
· C. Bandera, M. Marsico, Wireless Just-in-Time Training of Mobile First Responders. Proceedings of the 2006 SPIE Defense and Security Symposium. 20 April 2006.
· C. Bandera, M. Marsico, Rich media streaming for just-in-time training of first responders. Proceedings of the 2005 SPIE Defense and Security Symposium. 28 Mar 2005.
· C. Bandera, B. McMullen, R. Beck, Object-Oriented E-Education for PK-16 Students and Teachers. Proceedings of the 2004 International Educause Conference. 18 Oct 2004.
· C. Bandera, Learner Diversity and Content Personalization with MPEG-4. 2004 National Science Foundation Design, Service and Manufacturing Grantees and Research Conference. 3 Jan 2004.
· C. Bandera, B. McMullen, R. Beck, E-Learning Objects: The Value of SCORM and MPEG-7 Packaging for Digital Media Assets. Proceedings of the 2003 International Educause Conference. 3 Nov 2003.
· C. Bandera, SCORM-Compliant Interactive Video. Proceedings of the 2002 IITSEC. 2 Dec 2002.
· B. McMullen, C. Bandera, R. Beck, Object Oriented Multimedia Content; Server Benefits versus Authoring Costs. Proceedings of the 2002 International Educause Conference. 1 Oct 2002.
· C. Bandera, D. Stack, B. Pain, C. Wrigley, Real-time reconfigurable foveal target acquisition and tracking system. Proceedings of the SPIE AeroSense Symposium on Acquisition, Tracking, and Pointing XIII. April 1999.
· H. Hexmoor, C. Bandera, Architectural issues for integration of sensing and acting modalities. Proceedings of the IEEE International Symposium on Intelligent Control. 14 Sep 1998.
· S. Xia, R. Sridhar, P. Scott, C. Bandera, An All CMOS Foveal Image Sensor Chip. Proceedings of the 11th Annual IEEE International ASIC Conference. 13 Sep 1998.
· D. McKee, C. Bandera, Multistage foveal target detection system. Proceedings of the SPIE AeroSense '98 Symposium on Signal Processing, Sensor Fusion, and Target Recognition VII. 13 Apr 1998.
· S. Young, P. Scott, C. Bandera, Foveal automatic target recognition using a multiresolution neural network. IEEE Transactions on Image Processing. 7(8):1122-1135. 1998.
· F. Du, C. Bandera, Parallel programming for real-time image processing using computing agents. Proceedings of the International Conference on Parallel and Distributed Processing Techniques and Applications. 3(Las Vegas, Nevada):1505-1514. 30 Jun 1997.
· D. McKee, C. Bandera, Performance of a foveal target detection system. Proceedings of the SPIE AeroSense '97 Symposium on Signal Processing, Sensor Fusion, and Target Recognition VI. April 1997.
· P. Scott, S. Young, C. Bandera, Foveal automatic target recognition using neural networks. Proceedings of the International Conference on Image Processing. 16 Sep 1996.
· C. Bandera, F. Vico, J. Bravo, M. Harmon, L. Baird, Residual Q-learning applied to visual attention. Proceedings of the 13th International Conference on Machine Learning. 3 Jul 1996.
· F. Du, C. Bandera, Attentive control for model-based object recognition: a preliminary study. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 18 Jun 1996.
· F. Du, A. Izatt, C. Bandera, An MIMD computing platform for hierarchical foveal machine vision system. Proceedings of the IEEE Conf on Computer Vision and Pattern Recognition. 18 Jun 1996.
· S. Ghosal, D. McKee, C. Bandera, Model-based automatic target recognition using hierarchical foveal machine vision. Proceedings of the SPIE AeroSense '96 Symposium on Signal Processing, Sensor Fusion, and Target Recognition V. April 1996.
· C. Bandera, F. Vico, Stable color perception for deformable material characterization. Proceedings of the 1996 Workshop on Automation Robotics for Visco-Elastic Materials. 26 Mar 1996.
· S. Ghosal, A. Izatt, C. Bandera, Retinotopic processing for active foveal vision. Proceedings of the 2nd Asian Conference on Computer Vision. 5 Dec 1995.
· F. Du, C. Bandera, A. Izatt, A supporting environment for parallel algorithm development of a parallel image processing engine. Proceedings of the International Conference on Parallel and Distributed Processing Techniques and Applications. 3 Nov 1995.
· C. Bandera, F. Vico, A multiacuity connectionist model for local speed estimation. Proceedings of the International Workshop on Artificial Neural Networks. June 1995.
· C. Bandera, Attention in Active Foveal Vision. Defense Technical Information Center. Technical Report(ADB-182078L):March 1994.
· C. Bandera, Foveal FPA with Monolithic Early Vision Processing and Gaze Control. Proceedings of the DOD Smart Staring IRFPA Workshop. July 1993.
· P. Scott, S. Young, C. Bandera, Hierarchical Top-Down Classification Based on Multiresolution Skeleton. Proceedings of the IS&T/SPIE Conference on Electronic Imaging Science and Technology. 31 Jan 1993.
· C. Bandera, P. Scott, Machine Vision for Active Pursuit: The Foveal Alternative. Journal of Electronic Defense. 16(10):October 1991.
· C. Bandera, P. Scott, Hierarchical Multiresolution Data Structures and Algorithms for Foveal Vision Systems. Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics. November 1990.
· C. Bandera, P. Scott, Foveal Machine Vision Systems. Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics. November 1989.