PhD student in the CMU Portugal Dual Degree Program in Computer Science, between Carnegie Mellon University and Instituto Superior Técnico.

My research focused on extending planning approaches with Markov Decision Processes to account for alternative configurations of the environment. Specifically, the goal is to allow the agent to plan over possible modifications to its environment, in order to maximize its reward function.

Evaluated in multiple simulation domains, and in human-robot interaction scenarios. In the latter, we successfully allowed a robot to learn a task, reason about its limitations, and request the assistance of a human user if necessary, effectively trading-off the benefits arising from such assistance with the cost of disturbing the user.

Teaching Assistant of the graduate course 15-887 Planning, Execution and Learning.

Topics covered included classical planning, probabilistic path planning, planning under uncertainty, and learning in planning.