Ph.D Dissertation Proposal Defense "Anticipation in Dynamic Environments: Knowing What to Monitor" by Zohreh A. Dannenhauer

Ph.D. Committee:  Drs. Michael T. Cox (Co-advisor), Michael Raymer (Co-advisor), Nikolaos Bourbakis, Michelle Cheatham, and Hector Munoz-Avila (Lehigh University)

ABSTRACT:

Autonomy for an intelligent agent involves not only the capacity to achieve the goals but also to recognize problems and to generate new goals. Discrepancies in the world signal that a problem exists that needs to be explained. Then given an explanation of the cause of the discrepancy, the agent can generate a new goal to remove or mitigate the cause. Then the agent generates a plan to accomplish the selected goal. In dynamic environments, external changes may occur that may affect planning decisions and goal choices. An agent should be aware of these unexpected changes, and reason about them to be able to respond to them appropriately. If new  encountered world information affects the plan,  a planner adapts to it through the refinement of plans that are under construction. If the justification for goal selection changes, then the agent should transform the goal. In my thesis, I focus on the relationship between vision, interpretation and planning. My approach is to make vision sensitive to relevant changes in the environment that can affect plans and goals. I will present results with a cognitive architecture in different domains such as blocksworld, logistics and a Baxter humanoid robot to show the effectiveness of the proposed approach.