Time-Situated Reasoning within Tight Deadlines and Realistic Space and Computation Bounds

Book One

Volume

Proceedings of the AAAI Conference on Artificial Intelligence, 12

Student Abstracts

We develop an effective representational and inferential framework for fully deadline-coupled, time-situated problem solving. Our effort is to model an agent in a tight and rigid deadline situation, in need of successfully formulating and executing a deadline-feasible plan of action as the world around the agent continues to change. We highlight the severe time-pressure under which the agent must operate with a paradigmatic problem scenario : Nell and Dudley and the railroad tracks. Nell is tied to the railroad tracks as a train approaches. Dudley, our agent, must formulate a plan to save her and carry it out before the oncoming train reaches her. He must deliberate (plan) in order to decide this, yet as he does so, the train draws nearer to Nell.