Grants‎ > ‎

Complex Decision Making - ARC LP0882234

Project Description


Following the work incorporating HTN planning into BDI systems, this project is concerned with in solving further problems, in particular in the areas of dynamic plan repair (or re-planning if necessary), more comprehensive planning techniques, and interactions between planning and other aspects of reasoning about goals and their achievement. These issues are extremely important for the kinds of complex autonomous decision making. For example it is certainly important for a UAV (Unmanned Aerial Vehicle) to be able to develop a plan to reach its intended destination. However, after some unplanned for event such as an Air Traffic Control instruction to alter course to avoid other traffic, it is critical that it can repair its plan to take account of the new situation. Failure to do this effectively could cause further conflicts or lead to dangerous outcomes. In doing this repair it must therefore take account of priorities and potential conflicts involving other goals. For example a goal to get back on course as quickly as possible may suggest one course of action, while a goal to avoid some restricted airspace may suggest something different. In addition, the situation that has arisen may require techniques borrowed from first principles planning, rather than the structured HTN style planning which we have currently used. The ability to use general-purposes reasoning and planning capabilities, integrated within the agent infrastructure, will enable complex applications to be built more rapidly, as well as being assured of being more robust. Not every detail needs to be programmed at the outset, as the system can fill the “gaps” automatically according to the circumstances. In addition, the specification of such systems would be more declarative, facilitating maintenance over time.

Industry Partner: Agent Oriented Software Pty. Ltd. (AOS)


Participants 

  • Lin Padgham (from RMIT University) 
  • Andrew Lucas (from AOS)
  • Sebastian Sardina (from RMIT University)

Students involved: Lavindra de Silva and Dhirendra Singh


Publications

  1. Dhirendra Singh, Sebastian Sardina, Lin Padgham, and Geoff James. Integrating learning into a BDI agent for environments with changing dynamics. In Craig Knoblock Toby Walsh and Carles Sierra, editors, Proceedings of the International Joint Conference on Artificial Intelligence (IJCAI), pages 2525-2530, Barcelona, Spain, August 2011. AAAI Press.
  2. Sebastian Sardina and Lin Padgham. A BDI agent programming language with failure recovery, declarative goals, and planning. Autonomous Agents and Multi-Agent Systems, 23(1):18-70, 2011.
  3. Dhirendra Singh, Sebastian Sardina, Lin Padgham, and Stéphane Airiau. Learning context conditions for BDI plan selection. In van der Hoek, Kaminka, Lespérance, Luck, and Sen, editors, Proceedings of Autonomous Agents and Multi-Agent Systems (AAMAS), pages 325-332, Toronto, Canada, May 2010. IFAAMAS.
  4. Giuseppe De Giacomo, Fabio Patrizi, and Sebastian Sardina. Agent programming via planning programs. In van der Hoek, Kaminka, Lespérance, Luck, and Sen, editors, Proceedings of Autonomous Agents and Multi-Agent Systems (AAMAS), pages 491-498, Toronto, Canada, May 2010. IFAAMAS.
  5. Giuseppe De Giacomo, Fabio Patrizi, and Sebastian Sardina. Generalized planning with loops under strong fairness constraints. In Proceedings of Principles of Knowledge Representation and Reasoning (KR), pages 351-361, Toronto, Canada, May 2010.
  6. Sarah Hickmott and Sebastian Sardina. Optimality properties of planning via Petri net unfolding: A formal analysis. In Alfonso Gerevini and Adele Howe, editors, Proceedings of the International Conference on Automated Planning and Scheduling (ICAPS), pages 170-177, Thessaloniki, Greece, September 2009. AAAI Press.
  7. Stavros Vassos, Sebastian Sardina, and Hector Levesque. Progressing basic action theories with non-local effect actions. In Gerhard Lakemeyer, Leora Morgenstern, and Mary-Anne Williams, editors, Proceedings of the Ninth International Symposium on Logical Formalizations of Commonsense Reasoning (CS'09), pages 135-140, Toronto, Canada, June 2009. UTSePress, University of Technology, Sydney, Australia.
  8. Sebastian Sardina and Giuseppe De Giacomo. Composition of ConGolog programs. In Craig Boutilier, editor, Proceedings of the International Joint Conference on Artificial Intelligence (IJCAI), pages 904-910, Pasadena, CA, USA, July 2009. IJCAI.
  9. Lavindra de Silva, Sebastian Sardina, and Lin Padgham. First principles planning in BDI systems. In Proceedings of Autonomous Agents and Multi-Agent Systems (AAMAS), Budapest, Hungary, May 2009. ACM Press.
  10. Sebastian Sardina and Yves Lespérance. GologSpeak: Golog speaks the BDI language. In Proceedings of the Programming Multiagent Systems Languages, Frameworks, Techniques and Tools workshop (PROMAS), Lecture Notes in Computer Science (LNCS), volume 5919. To appear.
  11. Stéphane Airiau, Lin Padham, Sebastian Sardina, and Sandip Sen. Enhancing adaptation in BDI agents using learning techniques. International Journal of Agent Technologies and Systems (IJATS), 1(2):1-18, January 2009.