My research aims to help people make decisions in complex situations. In applying mathematics and artificial intelligence to practical problems, my aim is to assist, not replace, the human decision maker in planning and combinatorial optimisation problems. Especially when the situation features uncertainty, change, incomplete knowledge, and multiple decisions that could be desirable, the decision maker is best enabled by a greater understanding of the problem: advice, rather than prescription, over the course of action.

Motivated by problems from the real-world, I have interests in planning and scheduling, temporal reasoning, preferences, uncertainty, intelligent agents, constraint programming, operations research (including the integration of CP and AI techniques within OR), applications in office assistance, aerospace, and internet domains, and problem modelling. In constraint programming, my thesis research focused on handling uncertainty in both theory and practice.

I chaired the organizing committee for the AAAI 2007 Spring Symposium on Interaction Challenges for Intelligent Assistants. The symposium brought together practitioners and researchers of artificial intelligence, human-computer interaction, cognitive science, robotics, assistive and agent technologies, and fields that address complex socio-technical systems.

As part of my work on applications of hybrid methods, I maintain a page of results for the still life problem (currently in need of update). I initiated a list of frequently asked questions about the ECLiPSe constraint programming environment, now maintained by the community.

I serve or served on the programme committee of the following events:

• IUI'09: Eleventh International Conference on Intelligent User Interfaces
• ICAPS'08 Doctoral Consortium
• ICAPS'08 Scheduling and Planning Applications woRKshop (SPARK'08)
• ICAPS'08 Workshop on A Reality Check for Planning and Scheduling Under Uncertainty
• AAAI-08: Twenty-Third National Conference on Artificial Intelligence
• AAAI-08 Multidisciplinary Workshop on Advances in Preference Handling
• CHI 2008 Trends: Workshops and Courses: Usable Artificial Intelligence
• IUI'08 Workshop on Recommendation and Collaboration
• ISAIM'08: Tenth International Symposium on Artificial Intelligence and Mathematics
• ICAPS'07 Workshop on Moving Planning and Scheduling Systems into the Real World
• AAAI-07: Twenty-Second National Conference on Artificial Intelligence
• AAAI-07 Workshop on Preference Handling for Artificial Intelligence
• AAAI 2007 Spring Symposium on Interaction Challenges for Intelligent Assistants
• CP'06 Workshop on Preferences and Soft Constraints
• AAAI-06: Twenty-First National Conference on Artificial Intelligence
• ICAPS'06 Workshop on Planning under Uncertainty and Execution Control for Autonomous Systems
• CP'05 Workshop on Constraint Solving under Change and Uncertainty
• CP'05 Workshop on Quantification in Constraint Programming
• ICAPS'05 Workshop on Planning under Uncertainty for Autonomous Systems

I also serve as the coordinator for the AIC Seminar Series. Please email me to join the mailing list that announces the talks.

The following sections group my research by theme; some publications that overlap multiple themes are described accordingly.

[ Planning and scheduling | Temporal reasoning | Preferences | Uncertainty | BDI-agent theory | User-assistive agents | Aerospace applications | Hybrid CP-AI-OR methods ]
 

Constraint reasoning, temporal reasoning, uncertainty, preferences, and user-assistance are manifest in the dual problems of planning a course of action to achieve a set of goals, and arranging the planned steps in a schedule.

My work, joint with colleagues inside and outside SRI, spans personal calendar scheduling, broad-scale resource scheduling, planning and scheduling with preferences and uncertainty, and planning schemes for aerospace applications.

• Berry, P.; Donneau-Golencer, T.; Duong, K.; Gervasio, M.; Peintner, B.; and Yorke-Smith, N. Emma: An Event Management Assistant. ICAPS'08 System Demonstrations, Sydney, Australia, September 2008 (to appear). Demonstration of an adaptive personalized calendar management agent.
   
• Meuleau, N.; Morris, R. A.; and Yorke-Smith, N. A Variable Elimination Approach for Optimal Scheduling with Linear Preferences. CP/ICAPS'08 Joint Workshop on Constraint Satisfaction Techniques for Planning and Scheduling Problems, Sydney, Australia, September 2008 (to appear).
  Develops a tractable elimination function to apply Bucket Elimination to solve temporal CSPs with piecewise linear constraints on temporal preferences over continuous domains.
   
• Berry, P.; Bulka, B.; Peintner, B.; Roberts, M.; and Yorke-Smith, N. Neptune: A Mixed-Initiative Environment for Planning and Scheduling. Proceedings of FLAIRS'08, Coconut Grove, FL, May 2008 (Best Poster).   Abstract | PDF
  Describes the design of a mixed-initiative system for integrated hierarchical planning and scheduling, Neptune.
   
• Berry, P.; Moffitt, M. D.; Peintner, B.; and Yorke-Smith, N. The Design of a User-Centric Scheduling System for Multi-Faceted Real-World Problems. Proceedings of ICAPS'07 Workshop on Moving Planning and Scheduling Systems into the Real World, Providence, RI, September 2007.   Abstract | PDF
  Describes the design of a mixed-initiative scheduling system, Pisces.
   
• Berry, P.; Gervasio, M.; Peintner, B.; and Yorke-Smith, N. A Preference Model for Over-Constrained Meeting Requests. Proceedings of AAAI 2007 Workshop on Preference Handling for Artificial Intelligence, Vancouver, Canada, July 2007.   Abstract | PDF  
  Presents a MAUT model for user meeting scheduling preferences that balances expressiveness with amenability for elicitation, reasoning, and learning.
   
• Berry, P.; Peintner, B.; and Yorke-Smith, N. Bringing the User Back into Scheduling: Two Case Studies of Interaction with Intelligent Scheduling Assistants. Proceedings of AAAI 2007 Spring Symposium on Interaction Challenges for Intelligent Assistants, Stanford, CA, March 2007.   Abstract   ©2007 AAAI
  Speaks for the importance of the user in two scheduling applications.
   
• Berry, P.; Conley, K.; Gervasio, M.; Peintner, B.; Uribe, T.; and Yorke-Smith, N. Deploying a Personalized Time Management Agent. Proceedings of AAMAS'06 Industrial Track, Hakodate, Japan, May 2006.   Abstract | PDF   ©2006 ACM
  Reports on the ongoing practical experience designing, implementing, and deploying PTIME, a personalized agent for time management and meeting scheduling in an open, multi-agent environment.
   
• Berry, P.; Gervasio, M.; Peintner, B.; Uribe, T.; and Yorke-Smith, N. Multi-Criteria Evaluation in User-Centric Distributed Scheduling Agents. Proceedings of AAAI 2006 Spring Symposium on Distributed Plan and Schedule Management, Stanford, CA, March 2006.  Abstract | PDF   ©2006 AAAI
  Positions a distributed scheduling task for personalised calendaring as the co-operation of selfish scheduling agents.
   
• Morris, R. A.; Dungan, J.; Edgington, W.; Williams, J.; Carlson, C.; Fleming, D.; Wood, T.; and Yorke-Smith, N. Coordinated Science Campaign Scheduling for Sensor Webs. Proceedings of i-SAIRAS'05, Munchen, Germany, September 2005.  Abstract | PDF
  Describes the DESOPS software architecture for coordinated planning, scheduling and execution of Earth-orbiting science campaigns.
   
• Guettier, C. and Yorke-Smith, N. Enhancing the Anytime Behaviour of Mixed CSP-Based Planning. Proceedings of ICAPS'05 Workshop on Planning under Uncertainty for Autonomous Systems, Monterey, CA, June 2005.  Abstract | PDF
  Investigates the anytime performance of solving a full observability mixed CSP. Proposes algorithmic enhancements to improve the anytime behaviour w.r.t. plan completeness and plan executability.
   
• Yorke-Smith, N. and Guettier, C. Towards Automatic Robust Planning for the Discrete Commanding of Aerospace Equipment. Proceedings of 2003 IEEE International Symposium on Intelligent Control, Houston, TX, October 2003.  Abstract | PDF  ©2003 IEEE
  Applies the certainty closure framework to constraint-based control of aerospace equipment. Shows that a conditional plan set leads to more robust autonomous behaviour.

My work in temporal reasoning is motivated by the need for expressive frameworks to meet the requirements of real-world problems, particularly in planning.

Temporal Constraint Satisfaction Problems allow for reasoning with events happening over time. Their expressiveness has been extended independently in two directions: to account for uncontrollable events, and, more recently, to account for soft temporal preferences. The motivation for both extensions is from real-life temporal problems; and indeed such problems may well necessitate both preferences and uncertainty. We propose the study of temporal problems with both preferences and uncertainty, and put forward some methods for their resolution. We formally define the Simple Temporal Problem with Preferences and Uncertainty (STPPU). We show that STPPUs can be dynamically controlled with the same complexity as temporal problems with either preferences or uncertainty alone.

This work was joint with Brent Venable and Francesca Rossi.

• Rossi, F.; Venable, K. B.; and Yorke-Smith, N. Uncertainty in Soft Temporal Constraint Problems: A General Framework and Controllability Algorithms for the Fuzzy Case. Journal of Artificial Intelligence Research 27, 617-674, December 2006.  Abstract | PDF
  Presents definitively the STPPU formalism, where quantitative temporal constraints with both preferences and uncertainty can be defined. Showing how three classical notions of controllability (strong, weak, and dynamic), which have been developed for uncertain temporal problems, can be generalized to handle preferences as well, giving algorithms for the case of fuzzy preferences.
   
• Rossi, F.; Venable, K. B.; and Yorke-Smith, N. Controllability of Soft Temporal Constraint Problems. Proceedings of CP'04, Toronto, Canada, September 2004.  Abstract | PDF  ©2004 Springer-Verlag
  Shows how temporal constraint networks with both preferences and uncertainty (STPPUs) can be dynamically controlled, and gives a polynomial execution algorithm.
   
• Rossi, F.; Venable, K. B.; and Yorke-Smith, N. Preferences and Uncertainty in Simple Temporal Problems. Proceedings of CP'03 Workshop on Online Constraint Solving, Kinsale, Ireland, September 2003.  Abstract | PDF
  Develops methods for resolving strong, weak, and some types of dynamic controllability in Simple Temporal Problems with Preferences and Uncertainty.
   
• Yorke-Smith, N.; Venable, K. B.; and Rossi, F. Temporal Reasoning with Preferences and Uncertainty. Proceedings of IJCAI'03, Acapulco, Mexico, August 2003.  Abstract | PDF   ©2003 IJCAI
  Introduces a more expressive model for simple temporal problems, combining existing models for preference and for contingency in a new formalism. Proves that strong and weak controllability for STPPUs have the same complexity as for STPUs.

The STPPU framework considers STPs with preferences and implicitly uniform probabilities on the contingent time-points. The Simple Temporal Problems with Preferences and Probabilities, or STP3, permits arbitrary probability distributions on the occurrence of these time-points.

This work is joint with Lina Khatib, Nico Meuleau, Paul Morris, and Robert Morris.

• Meuleau, N.; Morris, R. A.; and Yorke-Smith, N. A Variable Elimination Approach for Optimal Scheduling with Linear Preferences. CP/ICAPS'08 Joint Workshop on Constraint Satisfaction Techniques for Planning and Scheduling Problems, Sydney, Australia, September 2008 (to appear).
  Develops a tractable elimination function to apply Bucket Elimination to solve temporal CSPs with piecewise linear constraints on temporal preferences over continuous domains. Probabilities and uncertainty are not considered.
   
• Morris, R.; Morris, P.; Khatib, L.; and Yorke-Smith, N. Temporal Planning with Preferences and Probabilities. Proceedings of ICAPS'05 Workshop on Constraint Programming for Planning and Scheduling, Monterey, CA, June 2005.  Abstract | PDF
  Introduces the Simple Temporal Problem with Preferences and Probabilities, and outlines two decision problems with STP3s for planning.

Disjunctive Temporal Problems are able to express relations such as "either before or after" that cannot be expressed with a Simple Temporal Problem. We formally define the Disjunctive Temporal Problem with Uncertainty (DTPU) to extend the DTP to account for uncontrollable events.

This work is joint with Bart Peintner and Brent Venable.

• Peintner, B.; Venable, K. B.; and Yorke-Smith, N. Strong Controllability of Disjunctive Temporal Problems with Uncertainty. Proceedings of CP'07, Providence, RI, September 2007.  Abstract | PDF   ©2007 Springer-Verlag
  Refines the semantics of DTPU constraints and gives the first algorithm to determine Strong Controllability of a DTPU.
   
• Venable, K. B. and Yorke-Smith, N. Disjunctive Temporal Planning with Uncertainty. Proceedings of IJCAI'05, Edinburgh, UK, August 2005.  Abstract | PDF   ©2005 IJCAI
  Introduces the semantics of the DTPU model.

Although one of the most expressive constraint-based temporal frameworks derived from the STP, the Disjunctive Temporal Problem with Preferences (DTPP) is limited to simple, fixed combinations of multiple criteria. The Multi-Criteria Disjunctive Temporal Problem with Preferences (MC-DTPP) models the aggregation of multiple criteria by integrating methods from Multi-Attribute Utility Theory. Solutions to an MC-DTPP can express subtle trade-offs between criteria, and yet an MC-DTPP can be solved with only modest computation cost beyond solving a DTPP.

This work is joint with Mike Moffitt and Bart Peintner.

• Moffitt, M. D.; Peintner, B.; and Yorke-Smith, N. Multi-Criteria Optimization of Temporal Preferences. Proceedings of CP'06 Workshop on Preferences and Soft Constraints, Nantes, France, September 2006.   Abstract | PDF  
  Premiers an extended framework for disjunctive temporal reasoning in the presence of multiple optimization criteria, and two initial algorithms to derive favoured solutions.

Quantitative temporal constraints are an essential requirement for many planning domains. The HTN planning paradigm has proven to be better suited than other approaches to many applications. To date, however, integrating temporal reasoning with HTN planning has been little explored. This paper describes a means to exploit the structure of a HTN plan in performing temporal propagation on an associated Simple Temporal Network. By exploiting the natural restriction on permitted temporal constraints, the time complexity of propagation can be sharply reduced, while completeness of the inference is maintained. Empirical results indicate an order of magnitude improvement on real-world plans.

This work is joint with Hung Bui and Mabry Tyson.

• Bui, H. H.; Tyson, M.; and Yorke-Smith, N. Efficient Message Passing and Propagation of Simple Temporal Constraints: Results on Semi-Structured Networks. CP/ICAPS'08 Joint Workshop on Constraint Satisfaction Techniques for Planning and Scheduling Problems, Sydney, Australia, September 2008 (to appear).
  Reports experimental results on belief propagation for solving semi-structured Simple Temporal Problems over continuous domains.
   
• Bui, H. H.; Tyson, M.; and Yorke-Smith, N. Efficient Message Passing and Propagation of Simple Temporal Constraints. Proceedings of AAAI 2007 Workshop on Spatial and Temporal Reasoning, Vancouver, Canada, July 2007.   Abstract | PDF  
  Investigates belief propagation for solving structured Simple Temporal Problems.
   
• Yorke-Smith, N. Exploiting the Structure of Hierarchical Plans in Temporal Constraint Propagation. Proceedings of AAAI'05, Pittsburgh, PA, July 2005.  Abstract | PDF  ©2005 AAAI
  Describes the sibling-restricted propagation algorithm for HTN temporal inference and its strong performance on real-world plans.

"It can be anytime on Tuesday, but I prefer earlier" -- a hard constraint ("it must be Tuesday") and a soft constraint ("I prefer earlier") indicating a preference. One source of preferences is people: our inclinations, decision-making, and ideals. Preferences result in an ordering (total or partial) over possible solutions.

For my work in temporal reasoning with preferences, please see above.

This work is joint with Bart Peintner and others.

• Peintner, B.; Viappiani, P.; and Yorke-Smith, N. Preferences in Interactive Systems: Technical challenges and case studies. AI Magazine 29(4), Winter 2008 (to appear).  ©2008 AAAI
  Surveys the role preferences have in Interactive Artificial Intelligence systems in both reasoning and interaction with the user.
   
• Berry, P.; Gervasio, M.; Peintner, B.; and Yorke-Smith, N. A Preference Model for Over-Constrained Meeting Requests. Proceedings of AAAI 2007 Workshop on Preference Handling for Artificial Intelligence, Vancouver, Canada, July 2007.   Abstract | PDF  
  Presents a MAUT model for user meeting scheduling preferences that balances expressiveness with amenability for elicitation, reasoning, and learning.
   

Uncertainty is inevitable in many real-world problems. Constraint programming (CP) has proved an effective paradigm to model and solve difficult large-scale combinatorial optimisation problems from many domains. Nonetheless, there has been little work on how to handle uncertainty within CP, yet the need for reliable solutions to real-world problems can only be met when uncertainty is taken into account.

Our interest is in data uncertainty in particular. Uncertainty can arise due to the dynamic and unpredictable nature of the commercial world, but also due to the information available to those modelling the problem. The former is handled by seeking robust solutions: those that hold under the most possible cases, for example, the solution that maximises expectation. For the latter source of uncertainty, however, arising from incomplete and erroneous data, there is little value in a robust solution to an approximation of the problem. Rather, the user's desire is for help to formulate the true problem and then solve it.

We provide the user with reliable information by enclosing the uncertainty with what we do know; thus we guarantee that the true problem is contained in the model. We find a closure, a set of possible solutions to this model; thus we guarantee a reliable basis from which to seek insight into the problem. This is the basis of the certainty closure framework. The practical value of our approach is seen in applications to computer networking and to aerospace planning.

This work is joint with Carmen Gervet, commenced under the CERCLe EPSRC project.

• Yorke-Smith, N. and Gervet, C. Certainty Closure: Reliable Constraint Reasoning with Uncertain Data. ACM Transactions on Computational Logic (to appear).  Abstract | Preprint PDF (597K)
   
• Yorke-Smith, N. and Gervet, C. Closures of Uncertain Constraint Satisfaction Problems. Proceedings of CP'05 Workshop on Quantification in Constraint Programming, Sitges, Spain, October 2005.  Abstract | PDF
  Investigates the characteristics of a relevant solution to an uncertain CSP, according to the nature of the data uncertainty and the outcome sought in the application.
   
• Yorke-Smith, N. and Gervet, C. Uncertain Constraint Optimisation Problems. Proceedings of CP'05 Workshop on Preferences and Soft Constraints, Sitges, Spain, October 2005.  Abstract | PDF
  Develops an extension of the uncertain CSP for optimisation problems with data incompleteness or errors.
   
• Berry, P.; Myers, K.; Uribe, T.; and Yorke-Smith, N. Task Management under Change and Uncertainty: Constraint Solving Experience with the CALO Project. Proceedings of CP'05 Workshop on Constraint Solving under Change and Uncertainty, Sitges, Spain, October 2005.  Abstract | PDF
  Puts forward the challenges to constraint programming, not restricted to change and uncertainty, that arise in task management in the CALO intelligent user assistant project.
   
• Yorke-Smith, N. and Gervet, C. Tight and Tractable Reformulations for Uncertain CSPs. Proceedings of CP'04 Workshop on Modelling and Reformulating Constraint Satisfaction Problems, Toronto, Canada, September 2004.  Abstract | PDF
  Defines two sufficient conditions for constraint classes that guarantee a tight and tractable reformulation of a uncertain CSP, to derive its full closure.
   
• Yorke-Smith, N. Reliable Constraint Reasoning with Uncertain Data. PhD thesis, IC-Parc, Imperial College London, June 2004.  Abstract
  Develops a formal framework and its practical instances, for reliably modelling and solving constraint problems with incomplete and erroneous data.
   
• Yorke-Smith, N. and Guettier, C. Towards Automatic Robust Planning for the Discrete Commanding of Aerospace Equipment. Proceedings of 2003 IEEE International Symposium on Intelligent Control, Houston, TX, October 2003.  Abstract | PDF  ©2003 IEEE
  Applies the certainty closure framework to constraint-based control of aerospace equipment. Shows that a conditional plan set leads to more robust autonomous behaviour.
   
• Yorke-Smith, N. and Gervet, C. Certainty Closure: A Framework for Reliable Constraint Reasoning with Uncertainty. Proceedings of CP'03, Kinsale, Ireland, September 2003.  Abstract | PDF  ©2003 Springer-Verlag
  Defines a comprehensive framework for uncertainty in constraint programming due to incomplete or erroneous data, together with practical resolution forms. Illustrates the framework with two diverse case studies.
   
• Yorke-Smith, N. and Gervet, C. On Constraint Problems with Incomplete or Erroneous Data. Proceedings of CP'02, Ithaca, NY, September 2002.  Abstract | PDF  ©2002 Springer-Verlag
  Summarises the need to tackle data uncertainty in CP, and presents a case study of the certainty closure approach in computer networking.
   
• Yorke-Smith, N. and Gervet, C. Data Uncertainty in Constraint Programming: A Non-Probabilistic Approach. Proceedings of AAAI 2001 Fall Symposium on Using Uncertainty within Computation, Cape Cod, MA, November 2001.  Abstract | PDF
  Demonstrates why non-probabilistic reasoning about data uncertainty is suitable for a diagnosis problem in computer networking. Describes a model of the problem as an uncertain CSP, and gives a transformation-based resolution of the model.

The Belief-Desire-Intention framework is widely adopted as a pragmatic but principled basis for intelligent agents. The limitations of the framework are challenged at both theoretical and implementation levels.

To explore the limitations and develop means to overcome them, I collaborate with researchers inside SRI and outside, particularly at RMIT.

• Thangarajah, J.; Harland, J.; Morley, D.; and Yorke-Smith, N. Suspending and Resuming Tasks in Intelligent Agents. Proceedings of AAMAS'08, Estoril, Portugal, May 2008.   Abstract | PDF   ©2008 IFAAMAS
  Develops a principled approach to suspending and resume tasks in a BDI-based agent, formalized in the CAN agent language.
   
• Thangarajah, J.; Harland, J.; and Yorke-Smith, N. A Soft COP Model for Goal Deliberation in a BDI Agent. Proceedings of CP'07 Workshop on Constraint Modelling and Reformulation, Providence, RI, September 2007.  Abstract | PDF
  Models the BDI goal deliberation process as a soft Constraint Optimization Problem.
   
• Thangarajah, J.; Harland, J.; Morley, D.; and Yorke-Smith, N. Aborting Tasks in BDI Agents. Proceedings of AAMAS'07, Honolulu, HI, May 2007.   Abstract | PDF   ©2007 ACM
  Presents extensions to the CAN agent language to enable a BDI-based agent to reason over goal and plan aborts and fails in a unified way.
   
• Myers, K. and Yorke-Smith, N. Proactive Behavior of a Personal Assistive Agent. Proceedings of AAMAS'07 Workshop on Metareasoning in Agent-Based Systems, Honolulu, HI, May 2007.   Abstract | PDF  
  Presents a BDI-based agent cognition model designed to support proactive assistance, employing a meta-level layer to identify potentially helpful actions and determine when it is appropriate to perform them
   
• Morley, D.; Myers, K.; and Yorke-Smith, N. Continuous Refinement of Agent Resource Estimates. Proceedings of AAMAS'06, Hakodate, Japan, May 2006.   Abstract | PDF   ©2006 ACM
  Allows a BDI-based agent to estimate resource consumption of tasks prior to their adoption for execution, and refine those estimates as execution proceeds.
   
• Myers, K. and Yorke-Smith, N. A Cognitive Framework for Delegation to an Assistive User Agent. Proceedings of AAAI 2005 Fall Symposium on Mixed-Initiative Problem Solving Assistants, Arlington, VA, November 2005.  Abstract | PDF   ©2005 AAAI
  Presents a BDI-based framework for a cognitive agent that acts as an assistant to a human user by perfoming tasks on her behalf.
   

Constraints, temporal reasoning, preferences, and uncertainty, are all manifest is the daily tasks of knowledge-based office professionals such as busy managers. SRI's CALO is an intelligent, adaptive agent designed to assist such a user. One component is CALO's Personalized Time Manager, PTIME.

As part of the reasoning and action aspects of the CALO project, I collaborate with researchers inside and outside SRI.

• Berry, P.; Donneau-Golencer, T.; Duong, K.; Gervasio, M.; Peintner, B.; and Yorke-Smith, N. Emma: An Event Management Assistant. ICAPS'08 System Demonstrations, Sydney, Australia, September 2008 (to appear). Demonstration of an adaptive personalized calendar management agent.
   
• Bui, H. H.; Cesari, F.; Elenius, D.; House, N.; Morley, D.; Myers, K. M.; Natarajan, S.; Saadati, S.; Yeh, E.; and Yorke-Smith, N. CALO Workflow Recognition and Proactive Assistance. AAAI-08 AI Video Competition, Chicago, IL, July 2008.   Abstract | AVI
  Short video shows how a CALO agent provides potentially helpful suggestions according to your work context, including your current desktop activity.
   
• Bui, H. H.; Cesari, F.; Elenius, D.; Morley, D.; Natarajan, S.; Saadati, S.; Yeh, E.; and Yorke-Smith, N. A Context-Aware Personal Desktop Assistant. Proceedings of AAMAS'08 Demonstration Track, Estoril, Portugal, May 2008.   Abstract | PDF   ©2008 IFAAMAS
  Demonstration of a personal assistant agent that provides potentially helpful suggestions according to your work context, including your current desktop activity.
   
• Weber, J. S. and Yorke-Smith, N. Time Management with Adaptive Reminders: A Pair of Ethnographic Studies and Their Design Implications. Working notes of CHI'08 Workshop: Usable Artificial Intelligence, Florence, Italy, April 2008.  Abstract | PDF  
  Reports initial findings from a pair of user studies into how people manage their time, and how they could benefit from an adaptive reminder system.
   
• Berry, P.; Gervasio, M.; Peintner, B.; and Yorke-Smith, N. A Preference Model for Over-Constrained Meeting Requests. Proceedings of AAAI 2007 Workshop on Preference Handling for Artificial Intelligence, Vancouver, Canada, July 2007.   Abstract | PDF  
  Presents a MAUT model for user meeting scheduling preferences that balances expressiveness with amenability for elicitation, reasoning, and learning.
   
• Myers, K. and Yorke-Smith, N. Proactive Behavior of a Personal Assistive Agent. Proceedings of AAMAS'07 Workshop on Metareasoning in Agent-Based Systems, Honolulu, HI, May 2007.   Abstract | PDF  
  Presents a BDI-based agent cognition model designed to support proactive assistance, employing a meta-level layer to identify potentially helpful actions and determine when it is appropriate to perform them
   
• Yorke-Smith, N. (ed). Interaction Challenges for Intelligent Assistants: Papers from the AAAI Spring Symposium. AAAI Technical Report SS-07-04, March 2007. AAAI Press, Menlo Park, CA.   Abstract  
  The symposium asked: what are most useful paradigms, methodologies, and implementations for human interaction with intelligent artificial assistants?
   
• Myers, K. and Yorke-Smith, N. Proactivity in an Intentionally Helpful Personal Assistive Agent. Proceedings of AAAI 2007 Spring Symposium on Intentions in Intelligent Systems, Stanford, CA, March 2007.   Abstract | PDF   ©2007 AAAI
  Argues that personal assistive agents should be able to reason about acting proactively, with care, to help their user.
   
• Berry, P.; Conley, K.; Gervasio, M.; Peintner, B.; Uribe, T.; and Yorke-Smith, N. Deploying a Personalized Time Management Agent. Proceedings of AAMAS'06 Industrial Track, Hakodate, Japan, May 2006.   Abstract | PDF   ©2006 ACM
  Reports on the ongoing practical experience designing, implementing, and deploying PTIME, a personalized agent for time management and meeting scheduling in an open, multi-agent environment.
   
• Berry, P.; Albright, C.; Bowring, E.; Conley, K.; Nitz, K.; Pearce, J.; Peintner, P.; Saadati, S.; Tambe, M.; Uribe, T.; and Yorke-Smith, N. Conflict Negotiation Among Personal Calendar Agents. Proceedings of AAMAS'06 Demonstration Track, Hakodate, Japan, May 2006.   Abstract | PDF   ©2006 ACM
  Demonstration of distributed conflict resolution in the context of personalized meeting scheduling.
   
• Berry, P.; Gervasio, M.; Peintner, B.; Uribe, T.; and Yorke-Smith, N. Multi-Criteria Evaluation in User-Centric Distributed Scheduling Agents. Proceedings of AAAI 2006 Spring Symposium on Distributed Plan and Schedule Management, Stanford, CA, March 2006.  Abstract | PDF   ©2006 AAAI
  Positions a distributed scheduling task for personalised calendaring as the co-operation of selfish scheduling agents.
   
• Myers, K. and Yorke-Smith, N. A Cognitive Framework for Delegation to an Assistive User Agent. Proceedings of AAAI 2005 Fall Symposium on Mixed-Initiative Problem Solving Assistants, Arlington, VA, November 2005.  Abstract | PDF   ©2005 AAAI
  Presents a BDI-based framework for a cognitive agent that acts as an assistant to a human user by perfoming tasks on her behalf.
   
• Berry, P.; Myers, K.; Uribe, T.; and Yorke-Smith, N. Task Management under Change and Uncertainty: Constraint Solving Experience with the CALO Project. Proceedings of CP'05 Workshop on Constraint Solving under Change and Uncertainty, Sitges, Spain, October 2005.  Abstract | PDF
  Puts forward the challenges to constraint programming, not restricted to change and uncertainty, that arise in task management in the CALO intelligent user assistant project.
   
• Berry, P.; Gervasio, M.; Uribe, T.; and Yorke-Smith, N. Mixed-Initiative Issues for a Personalized Time Management Assistant. Proceedings of ICAPS'05 Workshop on Mixed-Initiative Planning And Scheduling, Monterey, CA, June 2005.  Abstract | PDF
  Discusses the collaborative human/agent decision process in the PTIME project.

The aerospace domain is uniquely challenging and uniquely compelling. The scope for artificial intelligence is from fully-automated autonomous systems, through to partially-autonomous systems that exhibit greater fulfillment of mission objectives at lower cost, through to mixed-initiative decision aids for human scientists mission managers. My work is on theoretical foundations and algorithms that supports the range of application scope. See also the research on temporal reasoning.

This work is the result of a number of collaborations.

• Morris, R. A.; Dungan, J.; Edgington, W.; Williams, J.; Carlson, C.; Fleming, D.; Wood, T.; and Yorke-Smith, N. Coordinated Science Campaign Scheduling for Sensor Webs. Proceedings of i-SAIRAS'05, Munchen, Germany, September 2005.  Abstract | PDF
  Describes the DESOPS software architecture for coordinated planning, scheduling and execution of Earth-orbiting science campaigns.
   
• Guettier, C. and Yorke-Smith, N. Enhancing the Anytime Behaviour of Mixed CSP-Based Planning. Proceedings of ICAPS'05 Workshop on Planning under Uncertainty for Autonomous Systems, Monterey, CA, June 2005.  Abstract | PDF
  Investigates the anytime performance of solving a full observability mixed CSP. Proposes algorithmic enhancements to improve the anytime behaviour w.r.t. plan completeness and plan executability.
   
• Yorke-Smith, N. and Guettier, C. Towards Automatic Robust Planning for the Discrete Commanding of Aerospace Equipment. Proceedings of 2003 IEEE International Symposium on Intelligent Control, Houston, TX, October 2003.  Abstract | PDF  ©2003 IEEE
  Applies the certainty closure framework to constraint-based control of aerospace equipment. Shows that a conditional plan set leads to more robust autonomous behaviour.

We have investigated the combination of dynamic symmetry breaking methods with LP-CP hybridisation, and applied our methods to the still life problem. Our results show that combining Symmetry Breaking During Search (SBDS) with LP-CP hybrids outperforms both CP with SBDS and LP-CP without SBDS.

This work was joint with Karen Petrie.

• Petrie, K. E., Smith, B. M., and Yorke-Smith, N. Dynamic Symmetry Breaking in Constraint Programming and Linear Programming Hybrids. Proceedings of STAIRS'04, Valencia, Spain, August 2004.  Abstract | PDF
  Integrates Symmetry Breaking During Search (SBDS) with LP-CP hybrids. Case study on the maximum density still life problem.

The generic hoist scheduling problem is NP-hard and arises from automated manufacturing lines. In recent work using the constraint logic programming (CLP) formalism, a unified model has been developed with the problem description and solution method separated. We provide an improved model and new preprocessing stages where, as before, solutions and proof of optimality are provided by a hybrid CLP-MIP algorithm. The new algorithm is more scalable and robust. We give empirical results for a range of problem classes on benchmark problems from several sources.

This work was joint with Daniel Riera.

• Riera, D. and Yorke-Smith, N. An Improved Hybrid Model for the Generic Hoist Scheduling Problem. Annals of Operations Research 115, 173-191, September 2002.  Abstract | PDF
  Builds on an existing hybrid CP-LP model for multi-hoist, multi-track hoist scheduling problems, to yield more robust computational results without sacrificing expressiveness.