Course Description

Prerequisites. Students taking this course should have knowledge of object-oriented analysis, modeling and design, relational databases, and basic notions of probabilities, as acquired in previous courses, as well as logic and discrete mathematics. Knowledge and understanding of basic artificial intelligence techniques and concepts is also required.

Content: The course provides an introduction to the theory and algorithms for automated planning, with an emphasis on classical planning, and reasoning about actions. Automated planning is a branch of AI that concerns realizing strategies or action sequences to transform a given initial state into a desirable goal state. Reasoning about actions incorporates sophisticated forms of reasoning, such as action sequence executability and future effects (projection), which intelligent agents use to proceed from whatever state they find themselves in into a state that satisfies their goals. 

Objectives. The students will learn the theoretical and algorithmic foundations of automated planning and reasoning about actions and their practical implementation. They will understand the fundamental concepts underlying modern planning algorithms, and they will be equipped to conduct projects in this area.

Intended learning outcomes. After the course, the student will be able to:

• Evaluate and apply a variety of planning techniques for classical planning.

• Explain the practical advantages and disadvantages of different levels of expressivity in planning models.

• Model classical as well as probabilistic planning problems in commonly used domain definition languages.

• Model dynamic domains in Situation Calculus and Golog.

• Solve planning problems with automatic solvers.

• Apply reasoning algorithms to infer knowledge from dynamic domains.

Main Topics. 

• Introduction to the Classical Planning model

• Languages for Classical Planning

• Search Algorithms: Blind and Heuristic

• Domain-Independent Heuristics and Relaxations

• Classical Planning: Complexity, Variations and Extensions

• Getting to Know and Use a Planner 

• Probabilistic Planning: MDP e POMPD 

• Reasoning about actions through Situation Calculus and Golog

• Modeling dynamics of the domain of interest: Precondition Axioms, Successor State Axioms, the Frame Problem

• Regression and Projection

Teaching material. 

[1] Course slides, notes, and additional material available on this site. 

[2] Artificial Intelligence: A Modern Approach, Global Edition, 4th Edition by Stuart Russell, Peter Norvig, Pearson, 2020

[3] A Concise Introduction to Models and Methods for Automated Planning, by Hector Geffner and Blai Bonet, Springer, 2013

[4] Automated Planning: Theory and Practice, by Malik Ghallab, Dana Nau, Paolo Traverso, Elsevier, 2001

[5] An Introduction to the Planning Domain Definition Language, by Patrik Haslum, Nir Lipovetzky, Daniele Magazzeni, Christian Muise, Springer, 2019

[6] Knowledge in Action, by Raymon Reiter, MIT Press, 2001