SS 2001 Vorlesung: Principles of AI Planning

Vorlesung: Principles of AI Planning

Lectures on Monday from 11am to 1pm in room SR -02/017 (building 52).

April 23: Introduction; Different forms of planning; Forms of uncertainty: nondeterminism, partial observability; State variables and operators; Conditional effects
April 30: Deterministic planning; Progression and regression; Forward and backward planners; Search algorithms; Distance heuristics
May 7: Invariants; Representation of transition relations in propositional logic; Planning as satisfiability
May 14: An overview of recent planners;Matrix representation of actions;$n$-step reachability as matrix multiplication; Representation of Boolean functions as Binary Decision Diagrams; Matrix operations with BDDs; Planning with BDDs
May 21: fully observable nondeterministic planning; strong planning; strong cyclic planning; probabilities; construction of program-like branching plans
May 28: partially-observable planning; belief space; search in the belief space; complexity classes
June 4: holiday, day off, no lecture (Pfingstmontag = Whitmonday)
June 11: complexity classes; computational complexity of planning
June 18: computational complexity of planning; Markov decision processes; discounted rewards; value iteration
June 25: policy iteration; solution by linear programming; Reinforcement learning: bandit problems, Q-learning
July 2: job-shop scheduling
July 9: overview of applications of planning and scheduling, summary
July 9: examination