A tutorial at AAMAS 2016, May 10, 2016, Singapore

Automated Planning

by Roman Barták

Action planning is a typical activity of an autonomous agent dealing with finding a plan – a sequence of actions – that leads to a given goal. The tutorial gives an introduction to classical planning methods. It explains the planning problem and its possible representations, gives a survey of the most widespread planning techniques, namely state-space planning and plan-space planning, an explains some practicalities such as using heuristics and control rules to improve efficiency of planning algorithms.

The tutorial is targeted to researchers designing autonomous agents with planning capabilities. The goal of the tutorial is to explain major planning techniques so the attendees can understand what planning is about and how the planning algorithms work. No prior knowledge of planning is required though some knowledge of search algorithms such as A* and logical representation (first-order logic) are expected.

author

About the Author

Roman Barták works as a full professor and a researcher at Charles University, Prague (Czech Republic), where he leads Constraint Satisfaction and Optimisation Research Group. In 1999-2004 he led research activities of a multinational company Visopt BV, where he was the main architect of the scheduling engine developed by this company. His work focuses on techniques of constraint satisfaction and their application to planning, scheduling, and other areas. Recently, his research focus is on formal modelling of planning and scheduling problems. At Charles University he is teaching (among others) a course “Planning and Scheduling” that will serve as the major source for this tutorial.