Adaptation and Learning in Autonomous Agents

The ultimate goal of artificial intelligence is to construct a fully autonomous agent that operates in the physical world, much as do humans. True autonomy implies the ability to generate one's own tasks and to decide when one task is more important than another. This in turn requires the ability to adaptively change the relative importance of tasks when the real world violates one's expectations and to learn an improved world model that makes better predictions in the future.

This project explores these issues using Icarus, an architecture for intelligent agents that represents control knowledge at different levels of temporal resolution, uses decision-theoretic criteria to select tasks and actions, and learns the values of these tasks and actions from its experience. We are testing this framework in a simulated traffic domain, where the task is to control the speed and lane of an automobile on a highway among other cars.

Project Personnel

  • Professor Pat Langley

  • Dr. Daniel Shapiro

    • Previous Personnel

  • Dr. Ryutaro Ichise

    • Related Papers

    Ichise, R., Shapiro, D. G., & Langley, P. (in press). Learning hierarchical skills from observation. Proceedings of the Fifth International Conference on Discovery Science.

    Shapiro, D., & Langley, P. (2002). Separating skills from preference: Using learning to program by reward. Proceedings of the Nineteenth International Conference on Machine Learning (pp. 570-577). Sydney: Morgan Kaufmann.

    Shapiro, D., Langley, P., & Shachter, R. (2001). Using background knowledge to speed reinforcement learning in physical agents. Proceedings of the Fifth International Conference on Autonomous Agents (pp. 254-261). Montreal: ACM Press.

    Shapiro, D., & Langley, P. (1999). Controlling physical agents through reactive logic programming. Proceedings of the Third International Conference on Autonomous Agents (pp. 386-387). Seattle: ACM Press.

    Langley, P. (1997). Learning to sense selectively in physical domains. Proceedings of the First International Conference on Autonomous Agents (pp. 217-226). Marina del Rey, CA: ACM Press.


    For more information, please send email to dgs@stanford.edu .


    Return to research overview page.