Dr Karen L Myers

Biography and Interests

Dr. Karen Myers is Director of the Intelligent Mixed-initiative Planning and Control Technologies (IMPACT) program within the AI Center at SRI International. She is also an SRI Principal Scientist. Dr. Myers joined SRI in 1991 after completing a Ph.D. in computer science at Stanford University. Her research interests include the areas of reactive control, multiagent systems, automated planning, advisable technologies, and mixed-initiative problem-solving. Her work in these areas spans the range of basic research, technology development, and applications building. Dr. Myers currently serves on the Executive Council for ICAPS and the AIJ Editorial Board. She recently completed terms on the Executive Council for AAAI and the Editorial Board for JAIR.

Current Projects

Integrated Learning The Integrated Learning project is focused on developing machine learning technology that would enable a system to learn general planning knowledge from user demonstrations of processes.

Past Projects

	Advisable Planners The Advisable Planners project sought to make AI planning technology more accessible and controllable through the metaphor of advisability. User-provided advice specifies characteristics for both the desired solution and the problem-solving process to be employed during plan generation. Such advice is specified in a high-level language that is natural and intuitive for users, then operationalized into constraints that direct the underlying planning technology.
	Automated Capture of Design Rationale This project developed nondisruptive techniques for automatically acquiring rationale information for the detailed design process. The project produced a Rationale Construction Framework (RCF) system that monitors designer interactions with a CAD tool to produce a rich process history. This history is then structured and interpreted relative to a background theory of `design metaphors', thus enabling summarization and explanation of key elements of the design process.
	Cognitive Assistant that Learns and Organizes As part of DARPA’s Personalized Assistant that Learns (PAL) program, SRI and team members are working on developing a next-generation "Cognitive Agent that Learns and Organizes" (CALO).
	Coordination of Distributed Activities (CODA) The CODA system provides targeted information dissemination among distributed planners as a way of improving team coordination. In CODA, an individual planner declares interest in different types of plan changes that could impact his local plan development. As a team of distributed users develop plans with a plan authoring tool, their activities are monitored; changes that match declared interests are forwarded automatically to the person who declared interest in them.
	FASTUS A system for extracting information from free text.
	Inferring Intent of Attackers CAPRE uses plan recognition techniques to automatically determining the intent behind a cluster of security alerts. This allows us to prioritize and explain alert clusters to users.
	Integrated Battle Command Rolling Start DARPA’s Integrated Battle Command program (IBC) aims to support the commander’s intuition, judgment, and creativity using flexible, intelligent decision aids. Unlike some previous efforts, IBC will focus on interactive methods that enable humans to guide the search for solutions.
	JFACC Continuous Planning and Execution The main result of this project was the development of the Continuous Planning and Execution Framework (CPEF), which provides plan generation and replanning capabilities for situated agents in highly dynamic environments. Within CPEF, plans are treated as dynamic, open-ended artifacts that evolve in response to an ever-changing environment. In particular, plans must be updated in response to new information and requirements in a timely fashion to ensure their relevance and viability.
	Mixed-Initiative Planning and Scheduling for Science Missions We are developing mixed-initiative planning and scheduling technology that will enable NASA scientists to construct high quality mission plans that achieve as many of their science goals as possible, while satisfying all operations constraints. Our research will explore fundamental issues in how to reason about decisions and preferences from multiple sources, how to specify and utilize user preferences, and how to automate trade-off analysis. (With Ari Jonsson and John Bresina from NASA Ames)
	Multiagent Planning Architecture MPA is an open planning architecture that facilitates incorporation of new plan-related technologies, capitalizing on the benefits of distributed computing for efficiency and robustness. MPA provides protocols to support the sharing of knowledge and capabilities among agents involved in cooperative problem solving. MPA has been demonstrated in the air campaign planning domain, and was used as the infrastructure for the flagship demonstration of the DARPA Planning Initiative.
	Plan Authoring System based on Sketches, Advice, and Templates (PASSAT) PASSAT is a user-centric plan-authoring system grounded in the concepts of plan sketches, advice, and templates. PASSAT enables users to quickly develop plans that draw upon past experience encoded in templates, but that are customized to their individual preferences of a given user. The PASSAT core consists of an interactive plan authoring capability; tools for task management, constraint reasoning, plan sketching and causal reasoning provide provide complementary automated capabilities.
	Tactical Mission Planning SRI, in collaboration with NASA Ames, is developing mixed-initiative planning technology to support scientists in constructing tactical mission plans for future planetary exploration missions.
	Taskable Reactive Agent Communities (TRAC) The TRAC project developed mixed-initiative technology that enables flexible tasking and direction of agents by a user. Within TRAC, a user assigns tasks to agents along with guidance that imposes boundaries on agent behavior. During execution, the user manages agent activities in accord with a level of involvement that suits his individual needs. In essence, our work can be viewed as providing a form of process management technology that enables ready human control of agent communities.
	The JFACC Planner/Scheduler The main accomplishment on the project was the development of an integrated planning and scheduling capability that supports generation of tightly linked air operations plans and schedules, as well as their adaptation in response to changing tasks and resource availability. This effort built on existing planning (CPEF, from SRI) and scheduling (ACS, from CMU) technologies that provide core generation and repair techniques. (Joint work with Dr. Stephen F. Smith)

Software

	Plan Authoring System based on Sketches, Advice and Templates PASSAT is a user-centric plan-authoring system grounded in the concepts of plan sketches, advice, and templates.
	SPARK: SRI Procedural Agent Realization Kit SPARK is a Belief-Desire-Intention style agent framework grounded in a model of procedural reasoning.

Patents

Patent Pending.

Patent Pending.

Publications

The following are in reverse chronological order of publication.
Showing most recent 5 out of 78 [View All]

• Garvey, T. and Gervasio, M. and Lee, T. and Myers, K. and Angiolillo, C. and Gaston, M. and Knittel, J. and Kolojejchick, J. Learning by Demonstration to Support Military Planning and Decision Making, in Proceedings of the Twenty-first Conference on Innovative Applications of Artificial Intelligence (IAAI-09), AAAI Press, July 2009.  [PDF, Details]

• Yorke-Smith, N., Saadati, S., Myers, K. and Morley, D. Like an Intuitive and Courteous Butler: A Proactive Personal Agent for Task Management, in Proceedings of the Eighth International Joint Conference on Autonomous Agents and Multi Agent Systems (AAMAS’09), Budapest, Hungary, May 2009.  [Details]

• Gervasio, M. and Myers, K., and desJardins, M. and Yaman, F. Question Asking to Inform Preference Learning: A Case Study, in AAAI Spring Symposium on Agents that Learn from Human Teachers, March 2009.  [PDF, Details]

• Peintner, B. and Dinger, J. and Rodriguez, A. and Myers, K. . Task Assistant: Personalized Task Management for Military Environments, in Twenty-first Conference on Innovative Applications of Artificial Intelligence (IAAI-09) , 2009.  [PDF, Details]

• Gervasio, M. and Myers, K. L. Question Asking to Inform Procedure Learning, in Proceedings of the AAAI-08 Workshop "Metareasoning: Thinking about Thinking", July 2008.  [PDF, Details]