CURRICULUM VITAE

CURRICULUM VITAEMARK ALAN PEOT800 Highgrove Drive mpeot@ Chapel Hill, North Carolina 27516 919.929.3155 as of October 30th, 2007 RESEARCH INTERESTSRecent interests include:•Route and mission planning for autonomous collaborative UAVs•Probabilistic and physics-based planning for UGVs•Simultaneous localization and mapping•Probabilistic expert systems for machine diagnosis•Bayesian vision and super-resolution•Inference and learning algorithms for Bayesian networks•Convergence and accuracy issues for approximate reasoningEDUCATION9/88 – 5/98 Stanford University, Stanford, CAPh.D., Engineering-Economic Systems, June 1998Dissertation: Decision-Theoretic PlanningAdvisor: Professor Ross D. ShachterM.S., Engineering-Economic Systems,June 1989Engineering-Economic Systems is an interdisciplinary program that emphasizes theapplication of the mathematical tools of optimal decision making to problem solving indomains ranging from investment banking and public policy to control theory andartificial intelligence. The core curriculum for EES includes courses in optimization;probability and stochastic processes; dynamic systems and control theory;microeconomics; and decision analysis. In 2000, EES merged with Operations Researchand Industrial Engineering to form the new Stanford program in Management Scienceand Engineering.8/80 – 6/84 Massachusetts Institute of Technology, Cambridge, MAB.S., Electrical Engineering and Computer ScienceThesis: An Upper Bound on Element Efficiency in Infinite Planar ArraysAdvisor: Professor Walter K. KahnPROFESSIONAL EXPERIENCETeledyne Scientific & Imaging, LLC, Research Triangle Park, NC9/06 – Present Principal Scientist, Information SciencesTeledyne Technologies purchased Rockwell Scientific Company on 15 September 2006.• 07: Reflective Autonomous Systems (PI $50 K) (Proprietary)Developed physics-based “deep” planning algorithms for UxVs.• 07: Unmanned Autonomous Collaborative Operations CRADA (PI $500 K,Proprietary) Internally-funded continuation of the AATD Unmanned AutonomousCollaborative Operations. Developing mission and route planning for UxVs.• 07: Radar Terrain Reconstruction (PI $250 K, Proprietary) Led development Bayesianalgorithms to sharpen radar terrain. Sharpen terrain features by 4X over the raw sensormeasurement.Rockwell Scientific Company, LLC, Research Triangle Park, NC8/04 – 9/06 Principal Scientist, Information Sciences6/03 – 8/04 Technical Manager, Decision Sciences East5/98 – 6/03 Principal Scientist reporting to Executive Director, Information Sciences.• Founded the Rockwell Scientific Research Triangle Laboratory, a twelve-person groupperforming research in unmanned systems, planning, computer vision, and computersecurity. Principal investigator for several contract or internal research anddevelopment efforts focusing on the application of artificial intelligence to problems ofmilitary interest.• 07: Weather Radar Terrain Reconstruction (PI $250 K) Rockwell Collins (Proprietary)Developed isotropic and anisotropic terrain reconstruction algorithms for weatherradar, improving radar resolution by up to 4X.• 06: Enhanced Sensor Resolution (PI $250 K) Rockwell Collins (Proprietary)Developed 3D simultaneous localization and mapping algorithms for terrainreconstruction for UGV/UAVs.• 05: Embedded Mission Management (PI $350 K) Rockwell Collins (Proprietary)This effort aims to develop practical probabilistic route planning algorithms(probabilistic roadmap planning and execution-enhanced rapidly-exploring randomtrees) for multi-UAV planning for urban operations.• 05: Enhanced Sensor Resolution (PI $300 K) Rockwell Collins (Proprietary)Developed a patent-pending technique for Bayesian filtering and super-resolutionyielding 4X linear resolution improvements for IR and EO data.• 05: Spatial Plan Editor Phase 1 (PI $500 K) SOCOMSecond of a four phase program to transition the Spatial Plan Editor (nee SOFPlans)developed during the DARPA Active Templates program to operational use. Theobjective of Phase 1 is two fold: (a) replicate native SPE capabilities available inC2PC in the PFPS application and (b) develop a Plan Navigator to integrate multipleplanning applications into a single virtual framework.• 04-07: Unmanned Autonomous Collaborative Operations (PI $3.2M) Army/AATD.Developed planning and control software for implementing multi-UAV behaviors forreconnaissance, surveillance, strike and fault recovery. Used live hardware-in-the-loopsimulation to support planned flight tests.• 04: Spatial Plan Editor Phase 0 (PI $120 K) SOCOMFirst of a four phase program to transition the Spatial Plan Editor (nee SOFPlans)developed during the DARPA Active Templates program to operational use. Theobjective of Phase 0 was to integrate the SPE into the GD-AIS SOFTools editor foroperational evaluation. The resulting tool is currently in daily use at SOCOM.• 04-06: Enhanced Sensor Resolution (PI) Rockwell CollinsProprietary IRAD project developing probabilistically-sound Bayesian approaches tomulti-sensor multi-band fusion and super-resolution. Developed a patent-pendingalgorithm for super-resolution that improves linear image resolution by up to 3X.• 04-05: Embedded Mission Management (PI) Rockwell CollinsPrincipal Investigator. Proprietary IRAD project developing embedded missionplanning software primarily for UAVs. Developed software for• 03-05: DD(X) Ship Resource Manager (PI, $1.8 M) ONRDeveloped a resource manager for all ship board functions, ranging from sensors andweapons to navigation and logistics. The effort is a production effort conducted underthe aegis of the DD(X) engineering development model program, the risk-reductionphase of the DD(X) production effort. DD(X) is first in the Navy’s “family of ships”concept, which will eventually include a guided missile cruiser, a fast attack ship and alittoral combat craft, all using common components.• 98-01: Advanced Multifunction RF System – Resource Manager (PI $800 K) ONRDeveloped a resource allocation tool for a large multifunction phased array system.This prototype demonstrates decision-theoretic resource and frequency allocation ofmultiple RF functions onto a pair (transmit/receive) of phased arrays.• 00-04: Active Templates (co-PI $1.7 M) DARPADeveloped planning tools for special operations. Our major product was SOFPlans, acollaborative tool for integrating flight planning, geographical information, andtracking in a COM plug-in to C2PC, the common-operating picture component ofGCCS. This tool was deployed to Joint Special Operations Command in January 2003and was deployed to SOCOM for Exercise Cobra Gold in June 2003.• 98-04: Effects-Based Operations (PI $450 K) Air Force Research LaboratoryDeveloped software for planning and analyzing effects-based air campaigns. The resultof this effort: the Plan Analysis Component of the Campaign Assessment Tool is slatedfor demonstration at Unified Vision 01 and JEFX 02. The algorithm uses two variantsof sequential importance sampling to rapidly solve difficult temporal inferenceproblems with thousands of variables.• 98-00: Bayesian Information Fusion Toolbox (PI) BoeingProgram to develop dynamic Bayesian network tools for target tracking andidentification. Developed a Bayes net model for multiple hypothesis tracking and IDand a toolkit for constructing ID models from a database.Duke University, Durham, NC6/98 - 5/99 Visiting Assistant Professor, Institute for Statistics and Decision Sciences6/98 – 6/02 Scholar in Residence, Department of Computer Science• Taught “Belief Networks,” a course on graphical models for probability. Coursematerials: /courses/Spring99/sta294/• Member, dissertation committees for Stephen Majercik (Computer Science, Planningunder Uncertainty via Stochastic Satisfiability, graduated Spring 2000), Mark Borsuk(School of Environment, A Graphical Probability Network Model to Support WaterQuality Decision Making for the Neuse River Estuary, North Carolina, graduatedSpring 2001) and Paul Pauca (Computer Science, Automated Development of FastTransforms, graduated Spring 2002)Knowledge Industries, Menlo Park, CA6/92 - 5/99 Member of Board8/96 - 3/97 Acting President6/89 - 4/98 Chief ScientistKnowledge Industries is a Silicon Valley start-up developing probabilistic expert systemsfor capital equipment and medical domains.• Principal investigator for ARPA’s Standard Modules for Decision Analysis project.• Led the development of DXpress, a tool for constructing diagnostic expert systemsusing Bayesian networks.• Developed inference software for Intellipath, a commercial expert system tool forpathology.• Developed or co-developed probabilistic expert systems for many domains, includingcomputer networks (frame relay and CSU/DSU), gas turbines (performance analysisand fault detection), radiology (optimal test selection), sleeping disorders, glaucoma(prediction and diagnosis of scotomas), and nerve compression disorders of the hand.• Developed learning and inference software for the time series analysis.• Authored or co-authored research proposals worth over $2M.Science Center, Rockwell International, Palo Alto, CA5/88 - 6/96 Research Specialist (part time)• Developed or co-developed several AI planning systems including 2 decision-theoretic planners, 3 partial-order planning systems, the first contingent partial-orderplanner and a popular widely-used lisp implementation of GraphPlan.• Developed three new probabilistic reasoning algorithms: the popular likelihoodweighting technique for importance sampling on Bayesian networks, an improvedversion of Pearl’s cutset conditioning algorithm and a technique for solving any influencediagram using nearly any Bayesian network algorithm.Air Force Wright Aeronautical Laboratories, Wright-Patterson AFB, OH7/84 - 7/88 Officer in the Air Force• Managed classified research and development efforts on communicationscountermeasures, electronic warfare signal processing and low-cross section antennas forElectronic Warfare Division.• Managed research on real-time AI and neural networks for the AI Applications Office.Calspan Corporation, Dayton Technical Center, Dayton, OH6/83 - 9/83 InternAnalyzed the effects of terrain on monopulse radar countermeasures. PUBLICATIONSArticlesJournal1.Fusion and Propagation with Multiple Observations in Belief Networks, M. Peot andR. Shachter, Artificial Intelligence, Vol. 48, No. 3, April 1991. [65 citations]Conference Papers2.Y.-L, Chen, M.Peot, J.Lee, V.Sundareswaran, T.Altshuler. Multi-UAV AutonomousCollaborative Behaviors for Convoy Protection. In Proc. SPIE Unmanned SystemsTechnology IX, Vol. 6561: 65611C, April 2007.3.Y.-L Chen, M.Peot, J.Lee, V.Sundareswaran, T.Altshuler. Autonomous Collaborationfor Heterogeneous Multi-UAV Operations. In Proc. AHS International Specialists’Meeting on Unmanned Rotorcraft, January 2007Winner: Best paper in session4.Y.-L Chen, M.Peot, J.Lee, V.Sundareswaran, T.Altshuler. Autonomous CollaborativeBehaviors for Multi-UAV Missions. In Proc. 2006 SPIE Defense Transformationand Network-Centric Systems, Vol. 6249, May 2006.5.M. Peot, A. T. Hawkins, M. Aguilar, and A. Breiholz. Enhanced Sensor Resolution ofDBFM Imagery. 2006 Meeting of the Military Sensing Symposia (MSS), SpecialtyGroup on Passive Sensors. Orlando, Florida, 13-17 February 20066.Y.-L Chen, M.Peot, J.Lee, T.Altshuler. An Extensible Architecture for CollaborativeNetworked Unmanned Air Vehicles Operations. In Proc. SPIEUnmanned/Unattended Sensors and Sensor Networks II, Vol. 5986, September 20057.Y.-L Chen, B.Gregory, M.Easley, M.Peot, J.Lee, T.Altshuler. Market-BasedCollaborations for Autonomous Operations of Unmanned Air Vehicles. In Proc.IEEE Workshop on Distributed Intelligent Systems: Collective Intelligence and Its Applications, pp. 273-278, June 2006.8.M.Peot, T.Altshuler, A.Breiholz, R.Bueker, K.Fertig, A.Hawkins, S.Reddy. PlanningSensing Actions for UAVs in Urban Domains. In Proc. SPIE Unmanned/Unattended Sensors and Sensor Networks II, Vol. 5986, September 2005.9.Autonomous collaborative mission systems (ACMS) for multi-UAV missions. Y.-L.Chen, M. Peot, J. Lee, V. Sundareswaran, and T. Altshuler. Proceedings of SPIE – Volume 5820. Defense Transformation and Network-Centric Systems, Raja Suresh, Editor, May 2005, pp. 152-159. [2 citation]10.A flexible learning algorithm for decision making—Combining integratedclassification paradigm with Bayesian networks. D. Kil, M. Peot, and D. Banard.Recent Developments and Applications in Decision Making S.H. Zanakis, G.Doukidis, C. Zopounidis, eds. To be published by Kluwer Academic Publishers, 2000.11.A Compositional Structured Query Approach to Automated Inference, Y. El Fattahand M. Peot. Proceedings of the Seventh International Conference on Principles of Knowledge Representation and Reasoning (KR2000), 2000.12.Learning from what you don’t observe, M. Peot and R. Shachter, Proceedings of theThirteenth Conference on Uncertainty in Artificial Intelligence, Madison Wisconsin, 1998. [10 citations]13.Geometric Implications of the Naive-Bayes Assumption, M. Peot, Proceedings of theEleventh Conference on Uncertainty in Artificial Intelligence, Portland, Oregon, 1996. [12 citations]14.Suspending Recursion in Causal-link Planning. D. Smith and M. Peot, Proceedingsof The Third International Conference on Artificial Intelligence Planning Systems.Menlo Park: AAAI Press, 1996. [17 citations]15.Conditional Nonlinear Planning, M. Peot and D. Smith, Proceedings of the FirstInternational Conference on AI Planning Systems, College Park, Maryland, 1992.Also appears in Austin Tate, ed. Advanced Planning Technology, Menlo Park: AAAI Press, 1996. [184 citations]Winner of the first-ever ICAPS influential paper award, 2007.16.Postponing Conflicts in Partial-Order Planning, D. Smith and M. Peot, Proceedingsof the Eleventh National Conference on Artificial Intelligence, Washington D.C.1993. [40 citations]17.Threat-Removal Strategies for Partial-Order Planning, M. Peot and D. Smith,Proceedings of the Eleventh National Conference on Artificial Intelligence, Washington D.C., 1993. [51 citations]18.Decision Making Using Probabilistic Inference Methods, R. Shachter and M. Peot,Proceedings of the Eighth Conference on Uncertainty in Artificial Intelligence, Stanford, California, 1992. [53 citations]19.A Critical Look at Knoblock's Hierarchy Mechanism, D. Smith and M. PeotProceedings of the First International Conference on AI Planning Systems, College Park, Maryland, 1992. [12 citations]20.Automated Decision-Analytic Diagnosis of Thermal Performance in Gas Turbines, J.Breese, E. Horvitz, M. Peot, R. Gay, and G. Quentin, Proceedings of the International Gas Turbine and Aeroengine Congress and Exposition, Cologne, Germany, 1992. [12 citations]21.Simulation Approaches to Probabilistic Inference for General Probabilistic Inferenceon Belief Networks, R. Shachter and M. Peot, Workshop on Uncertainty in AI 1989,also appears in Uncertainty in AI 5. [138 citations]Unrefereed Publications22.E. Horvitz and M. Peot. Flexible strategies for computing the value of information indiagnostic systems. Proceedings of the AAAI Fall Symposium on FlexibleComputation in Intelligence Systems, pp. 89-95, Cambridge, Massachusetts, 1996. [2citations]23.P ostponing Simple Conflicts in Nonlinear Planning. D. Smith and M. Peot, In Lansky,A., ed. Foundations of Automatic Planning: The Classical Approach and Beyond,TR SS-93-03, Menlo Park, CA: AAAI Press, 1993. [4 citations]24.M odel Construction in Planning, M. Peot and J. Breese, Workshop on the Knowledge-Based Construction of Probabilistic and Decision Models. AAAI-91, Anaheim,California, 1991. [2 citation]25.A pplication of Simulated Annealing and Genetic Search to a Nonlinear TravelingSalesman Problem. M. Peot, S. Cross, M. Fausett; 1987 AIAA Conference onNavigation, Guidance, and Control. Monterey, California. [1 citation]26.E lectronic Warfare Signal Processing in the 21st Century. Microwave Journal, March1987 and The EW Handbook, 1989. [1 citation]DissertationDecision-Theoretic Planning, Ph.D. Dissertation, Department of Engineering-EconomicSystems, Stanford University, June 1998. [24 citations]Abstract: In recent years, researchers in AI planning have been trying to extend the classical planning paradigm to handle uncertainty and partial goal satisfaction. This dissertation focuses on the development of two partial-order planners: DTPOP, a contingent decision-theoretic planner, and UDTPOP, its non-contingent sibling. Both planners assemble actions with conditional, uncertain effects into plans of high overall utility.UDTPOP uses a new criterion for identifying actions that are relevant to utility and for linking these actions into a partial order plan. The simplified linking strategy of UDTPOP allows us to abandon the multiple support paradigm used in the probabilistic planner Buridan. This same innovation also allows us to develop a strict (and tight) upper bound on the utility of the plan that may be used as an admissible cost function for best-first search. UDTPOP is significantly faster than previous probabilistic planners. An empirical study compares a variant of UDTPOP against Buridan. In nearly every domain, the performance of UDTPOP is superior, sometimes by 2-3 orders of magnitude.The second planner, DTPOP is the first sound and complete contingent partial-order planner. Efficient contingency plans satisfy a “no augury” condition: information used to decide between plan branches must be relevant (in the sense of D-separation) to the utility of the branches under consideration. DTPOP uses this criterion directly to develop tests that can reveal the distribution over unknown, but relevant, events. During planning, DTPOP reverses its direction of search, using regression to establish preconditions and forward planning to identify observations. [21 citations]Undergraduate ThesisAn Upper Bound on Element Efficiency in Infinite Planar Arrays. B.S. Thesis,Department of Electrical Engineering and Computer Science, Massachusetts Institute ofTechnology, June 1984.Abstract: The efficiency of a radiating element in an infinite planar array can be definedto be the ratio of the power radiated to the full power available to the element. Thisefficiency is a simple measure of the mutual coupling effects in a phased array. An idealupper bound to this efficiency, based only on the array geometry, has been described byW. K. Kahn: “Ideal Efficiency of a Radiating Element in an Infinite Array” (1967). Amethod for finding a lower upper bound on this efficiency for elements in an array withconstant matching resistance (non-interconnecting matching networks) is describedherein. This lower upper bound is calculated using the element radiation characteristicsas well as the geometry of the array. The element efficiency is calculated for dipoles ininfinite rectangular arrays with varying dipole orientations and spacings for a fewdifferent impedance matching schemes. Comparison is made between these efficienciesand those of high gain elements and those produced by an ideal element pattern which isfound to maximize the array efficiency.The results from this thesis can be used to demonstrate that element/element coupling andloss is inevitable (and large) in broadband solid state active arrays proposed for severalrecent programs (including DD(X)). [1 citation]Cited ReportsOrdering Clauses in Nonlinear Planning, Technical Report, Rockwell International, PaloAlto Laboratory, 1992. [1 citation]ManuscriptsA Note on the DMIN Strategy, D. Smith and M. Peot. Cited addendum to “Threat-Removal Strategies for Partial-Order Planning,” 1993. [1 citation]Evidential Reasoning using Likelihood Weighting, R. Shachter and M. Peot, 1990. [4citations]PatentsBAYESIAN APPROACH FOR SENSOR SUPER-RESOLUTION, Patent Application11/239,981, 30 September 2005.TUTORIALS, PANELS AND INVITED TALKSUncertainty in Artificial Intelligence, Providence, RI, 1997.Panel Chair: UAI Panel on the Commercialization of Bayesian Reasoning, The ThirteenthConference on Uncertainty in Artificial Intelligence, Providence, RI, 1997.Tutorial: Algorithms for Decision-Making, The Twelfth Conference on Uncertainty inArtificial Intelligence, Portland, OR, 1996.Panelist: UAI Panel on the Application of Bayesian Reasoning, The Twelfth Conferenceon Uncertainty in Artificial IntelligenceTutorial: Diagnosis, (w/ Greg Provan), The Summer Institute on Probability in AI,Corvallis, OR, 1994.Invited Talk: Gas Turbine Diagnosis, The First Annual Normative Systems Workshop,Palos Verdes, CA, 1991.PROFESSIONAL ACTIVITIESReviewing:The Journal of AI ResearchArtificial IntelligenceManagement ScienceProgram Committees:The National Conference on Artificial Intelligence (AAAI). 1996, 1997, 1998, 1999,2000, 2001.The Conference on Uncertainty in Artificial Intelligence (UAI). 1993, 1995, 1996,1997, 1998, 1999, 2000, 2001.The European Conference on Artificial Intelligence, 1998.The International Joint Conference on Artificial Intelligence, 1997.AWARDSstNonlinear Planning, Proceedings AIPS-92 – awarded to the most influential paper out ofall AIPS, EWSP, ECP, and ICAPS papers published before 1997.1989 - 1993 National Science Foundation Graduate Fellowship.1988 - 1989 Noel Eldridge Memorial Fellowship.1988 Air Force Commendation Medal.1987 Junior Engineer of the Year; Association of Old Crows, Kittyhawk Chapter(Electronic warfare fraternity).1984 Eta Kappa Nu Electrical Engineering Honor Society.SOFTWARE1991 - 1998 The DXSolution Series: DXPress, Win-DX, and API-DX (w/Michael Shwe, Alan Wada, Dan Sommerfield, Sanjay Vanvari). The DXSolution Series is KnowledgeIndustrie’s suite of diagnosis software products. Expert systems fielded included theexpert systems for frame relay modems (TTC), office equipment diagnosis (Ricoh), andjet engine performance analysis (Enter/GE).1990 SleepIt: (w/Michael Shwe, Dr. German Nino-Murcia, and CNS, Inc.). SleepIt is a diagnostic expert system for use by general practitioners for the interpretation of resultsfrom a home polysomnogram system. SleepIt diagnosed over 100 different sleepingdisorders using over 200 different symptoms, test results, and historical findings.1989 -1992 Intellipath: (w/K.C. Chang, Michael Shwe, Dr. David Heckerman, Dr. Eric Horvitz).Developed the inferential core for Intellipath, an expert system tool for pathology. Thisinference system was also used for the development of an expert system for gas turbinediagnosis (United), as well as an expert system for the diagnosis of critical reactioncontrol and orbital maneuvering system faults on the Space Shuttle using real-timetelemetry (NASA).Government Software2003-2006 Autonomous Collaborative Mission System (w/Brian Gregory, Mark Anderson, Yi-Liang Chen): Developed a planning and scheduling service for reconnaissance,surveillance and communications tasks for unmanned aerial vehicles. The system used asimple distributed planner that allowed multiple UAVs to plan and collaborate on areasearch missions, point observation missions, convoy protection missions,communications relay missions and combinations of the above. Some features of thesystem included: GPU-based inference to perform fast line of sight calculations,constraint-based scheduling for optimal resource allocation of aircraft to search tasks, andnonlinear numerical constraint propagation using detailed sensor / vehicle /environmental physics to develop optimal sensing plans for a wide variety of objectives.The system ran in a hardware-in-the-loop simulation environment based on the Piccoloautopilot and simulator.2001-2003 Spatial Plan Editor (nee SOFPlans) (w/Craig Anderson and Wallace Kelly): Production planning and tracking tool for special operations. Deployed to SOCOM since 2003.Integrates flight planning (via PFPS), geographic information retrieval, mission trackingand mission planning into Command and Control PC (C2PC), the mapping component ofGCCS.2001-2003 PFPS2C2PC (w/Wallace Kelly): A popular file-translation utility used in SOCOM. 2001-2003 Bayesian Algorithm Toolkit: Developed a toolkit of Bayesian inference and sensitivity algorithms that is part of the AFRL Causal Analysis Tool. CAT is the centerpiece of theAir Force’s Effects Based Operations Advanced Technology Demonstration and is to bedeployed as part of Falconer, the Advanced Technology Air Operations Center.Commercial Prototypes1998 DXInsight: (w/Michael Shwe): A system for automated referral and test sequence optimization for radiology. Derived optimal radiographic test sequences for diagnosingabdominal diseases.1993 DXP: A rule-based expert system for identifying scotomas (patches of decreased sensitivity or blindness) using visual fields collected from a computerized perimetrysystem.1989 Glaucoma: A probabilistic expert system and inference engine for projecting the probability that glaucoma will develop in a patient based on a dynamic Bayesian networkmodel for the evolution of intraocular pressure based on the history of diabetes, eyepressure and angle measurements, myopia, etc.Research Software2007 Isotropic Terrain Reconstruction (w/ Julio Castrillon) Developed a novel (patent-pending) isotropic reconstruction algorithm for terrain using data collected from aconventional weather radar. The algorithm improved weather radar resolution by 4X. 2005 Enhanced Sensor Resolution (w/ M. Aguilar and A. Hawkins): Developed software demonstrating a novel (patent-pending) Bayesian super-resolution algorithm. Thisalgorithm simultaneously (and directly) infers the optimal alignment for multiple imagesalong with the optimal point spread function to model observed blurring. In tests, itimproved IR image linear resolution by over 3X. Developed a preliminary computeralgebra package to automatically derive super-resolution algorithms from a declarativespecification of the sensor model.2005 Embedded Mission Management: Co-developed (with R. Bueker and K. Fertig) a UAV planning system using probabilistic roadmap algorithms. Developed extensive 3Dgraphics, collision detection algorithms and planning algorithms. Developed strategies tooptimally plan sensing missions using symbolic and numeric models for sensor, platformand environment.2005 Autonomous Collaborative Mission System: Authored the manned/unmanned deconfliction components for this system. Designed the top-level message flow for thefull architecture. Authored a linear/discrete constraint reasoning system that was used forscheduling and resource allocation.2003 Fusion Resource Manager: Authored a constraint-based system for dispatching sensing requests to a large constellation of sensing systems. This system was used for Boeing onseveral proposal efforts.1998-1999 CAT 2.0: Constructed a prototype tool for analyzing air campaign plans. Used extensively by the Air Force to demonstrate the new Effects-Based Operations conceptfor planning air campaigns (in 1999, the #2 rated Air Force R+D need).1998-2001 AMRFS RAM Prototype: Authored (w/ W. Kelly) prototype for a real-time Bayesian-network based resource management system for the Advanced Multifunction RF System.Schedules multiple ESM, COM and radar functions in time, space, and frequency tominimize conflicts while maximizing effectiveness. A technology differentiator (keydemonstration, prototype architecture) for three production programs, including DD(X). 1997-1999 DTPOP, UCPOP: Probabilistic planning software (dissertation). DTPOP is a complete contingent partial-order planner. UCPOP is its non-contingent sibling.1995 GraphPlan (port to Common Lisp): Developed a popular public-domain Lisp port of Blum & Furst’s GraphPlan. This software is used in several planning systems includingConditional GraphPlan (Anderson, Smith & Weld), Conformant GraphPlan (Smith &Weld), Sensory GraphPlan (Weld, Anderson & Smith), BSR-GraphPlan (Parker) andgoal-based variants of GraphPlan (Kambhampati, Lambrecht & Parker).1992 CNLP: The first contingent partial-order planner. Featured in Russell and Norvig, Artificial Intelligence: A Modern Approach, 1st Edition, Chapter 13 (called CPOP). 1991-1996 RNLP, ONLP: (w/David Smith) DARPA-funded research software developed by Rockwell to explore operator graph-based planning, threat-resolution techniques,recursion detection, and subgoal selection techniques in partial-order planning.1988-1990 IDEAL: Contributed cutset conditioning and simulation code to IDEAL (Srinivas and Breese), a public-domain tool for exploring influence diagram and Bayesian networkalgorithms.1982 Stix Editor (w/Henry Minsky): VLSI design system developed for Danny Hillis (of connection machine fame). Used virtual grid compaction (Weste) to enforce design rulesand optimize circuit layout.。