FRAS (Flexible and Resilient Autonomous Systems) is a joint project of our department and Robotic Institute at Carnegie Melon University (Pittsburgh, PA). On top of this international academic cooperation, FRAS is funded by a rather exceptional institution – US Air Force Research Lab, which gives us rich research possibilities and exciting real world applications such as drone fleet route planning.
In this thrilling research we ingeniously combine classical planning and game theory to design new algorithms that will work in adversarial environment such as bad weather conditions or attack of the enemy unit. Later we also apply machine learning techniques on the simulated executions to find out how can we improve the model.
This means that FRAS tackles a number of research areas: Planning, Game Theory, AI and Machine Learning. We cannot see many different reearch projects where you get this kind of interdisciplinarity. Get to know FRAS in more detail and join our team! We aree looking for students who seek a part time job or those who wish to write a Bachelor/Master thesis with research in planning.
Tough challenges of classical planning
Many problems from classical planning are applied in the environment with other, possibly adversarial agents (aka enemies). However, plans found by classical planning algorithms lack the robustness against the actions of other agents – the quality of computed plans can be significantly worse compared to the model. To explicitly reason about other (adversarial) agents, the game-theoretic framework can be used. The scalability of game-theoretic algorithms, however, is limited and often insufficient for real-world problems.
Our solution? Add a little bit of game theory!
We combine classical domain-independent planning algorithms and game-theoretic strategy-generation algorithm where plans form strategies in the game. To evaluate the efficiency of our project, we analyze the trade-off between the quality of the planning algorithm and the robustness of final randomized plans and the computation time. Finally, we analyze different variants of integration of classical planning algorithms into the game-theoretic framework and show that at the cost a minor loss in the robustness of final plans, we can significantly reduce the computation time (yay!). If you want to dig deeper into our methodology, read our new paper we published on this topic.
See you soon,
Lukáš and Pavel