Colloquium in Mathematics and Computer Science
LANDMARKS IN CLASSICAL PLANNING AND BEYOND
Mr. Erez Karpas
Sunday, Febuary 8 | 14:00 | Science Building 8, Room 424
Classical planning is the problem of finding a sequence of instantaneous actions from some initial state to a goal, in a deterministic, fully observable world. In classical planning, a landmark is a fact or action which must occur in every plan. Landmarks have proven to be a very useful tool in classical planning, and planners which exploit landmarks have won several tracks in previous International Planning Competitions.
However, if we want a robot to plan its actions in the real world, it must deal with realistic settings, in which actions take time and can be executed in parallel, as well as dealing with partial observability and the risk of actions failing. In this talk, I will review the use of landmarks in classical planning, and present some recent work on landmarks in non-classical planning settings.
Mr. Erez Karpas is a post-doctorial fellow at the Model-based Embedded and Robotic Systems group at MIT’s Computer Science and Artificial Intelligence Lab.
His research interests include autonomous systems, automated planning, heuristic search, and machine learning.
He received his Ph.D. from the Technion - Israel Institute of Technology in 2012, and his M.Sc. and B.Sc. from Ben-Gurion University.