Goal-Driven Autonomy (GDA)

Goal-Driven Autonomy

Project Description

Goal-Driven Autonomy is a new conceptual framework for intelligent agents that are empowered to modify and pre-empt their own goals in response to changes in the environment. We created and formalized this concept in collaboration with the Naval Research Laboratory and the Defense Advanced Research Projects Agency for a DARPA seedling effort, which has since led to a workshop at the 24th conference on Artificial Intelligence (AAAI-10) and a number of research systems and projects.

In support of this effort, we created the artificially intelligent goal-driven agent Autonomous Response to Unexpected Events (ARTUE), which modifies its own goals when surprised by unexpected changes in the environment. This makes it especially robust to surprising circumstances.

Accomplishments

We successfully published two papers on the ARTUE system at AAAI-10, a prestigious conference with an acceptance rate of 26.9%. Our experimental results showed that the performance of our HTN planner, the first extended to plan in the context of temporal processes and events, statistically outperformed non-HTN planning on a range of benchmark problems.

On a broader scope, we also showed that ARTUE responds robustly to surprising events in a series of difficult scenarios in a naval anti-submarine warfare domain, and showed that ARTUE's ability to formulate and manage goals was instrumental in achieving high performance.

Publications

Klenk, M., Molineaux, M., & Aha, D.W. (in press). Goal-driven autonomy for responding to unexpected events in strategy simulations. To appear in Computational Intelligence.

Molineaux, M., Kuter, U., & Klenk, M. (in press). DiscoverHistory: Understanding the Past in Planning and Execution. To appear in Proceedings of the Eleventh International Conference on Autonomous Agents and Multi-Agent Systems. Valencia, Spain.

Klenk, M., Aha, D.W., & Molineaux, M. (2011). Making the case for transfer: Case-based transfer learning. AI Magazine, 32(1), 54-69.

Molineaux, M., Kuter, U., & Klenk, M. (2011). What Just Happened? Explaining the Past in Planning and Execution. In T. Roth-Berghofer, N. Tintarev, & D.B. Leake (Eds.) Explanation-Aware Computing: Papers from the IJCAI Workshop. Barcelona, Spain.

Molineaux, M., Aha, D.W., & Kuter, U. (2011). Learning event models that explain anomalies. In T. Roth-Berghofer, N. Tintarev, & D.B. Leake (Eds.) Explanation-Aware Computing: Papers from the IJCAI Workshop. Barcelona, Spain.

Molineaux, M., Klenk, M., & Aha, D.W. (2010). Goal-driven autonomy in a Navy strategy simulation. To appear in Proceedings of the Twenty-Fourth AAAI Conference on Artificial Intelligence. Atlanta, GA: AAAI Press.

Molineaux, M., Klenk, M., & Aha, D.W. (2010). Planning in Dynamic Environments: Extending HTNs with Nonlinear Continuous Effects. To appear in Proceedings of the Twenty-Fourth AAAI Conference on Artificial Intelligence. Atlanta, GA: AAAI Press.

Molineaux, M. (2010). Design and Evaluation of a Goal-Directed Autonomous Agent. To appear in Goal-Directed Autonomy: Papers from the AAAI Workshop. Atlanta, GA: AAAI Press.

Auslander, B., Molineaux, M., Aha, D.W., Munro, A., & Pizzini, Q. (2009). Towards research on goal reasoning with the TAO Sandbox. (Technical Report AIC-09-155). Washington, DC: Naval Research Laboratory, Navy Center for Applied Research on Artificial Intelligence.