UROP Project

Explaining an Autonomous Agent's Path Plans using Structured Question-Answer Interaction

As artificial intelligence becomes more prevalent, it has become critical to explain AI behavior and decisions to humans. The emerging field of Explainable Path Planning (XAIP) focuses specifically on explaining autonomous agents, like robots and drones, which use AI to plan their own paths and actions. One simple way to describe these systems is through Markov decision processes (MDPs). Here, the agent moves from state to state by choosing from a set of possible actions, taking one action per time step. Our current research focuses on building a method to explain the choices that the agent makes. To help the explanation process, we would like human users to be able to ask the agent specific questions about its plan. Thus, we are looking for a UROP researcher to design and program a question-asking user interface. This interface will be able to ask the agent plan-specific questions and make the questions "agent-understandable" within our current framework.

Task

The UROP student researcher will be asked to do the following, with the help of the supervisor: (1) read the current project paper(s) and a small amount of background material on the theoretical framework for the project, (2) design the question-asking interface, (3) implement and test the interface with the existing simulation, (4) present the final work at the end of the program

Requirements

-Currently enrolled in a bachelor program in engineering, computer science, mathematics, or other relevant area -Comfortable programming in a high-level programming language (e.g., Python) -Interest in optimization, autonomous systems, and/or control -Ready to problem-solve and work independently (with supervisor support)