PhD Seminar Course on

Automated planning and machine learning

Cagliari, 23-29.06.2011



This activity was made possible by the "Visiting Professors 2010" program of the University of Cagliari, sponsored by the Autonomous Region of Sardinia
Instructor: Prof. Daniel Borrajo, University of Carlos III, Madrid
Duration: 8h
Schedule: Thu 23/6 (15-18), Mon 27/6 (15-18), Wed 29/6 (15-17)
Venue: B1
Topics: [ 3hours + 3hours ]
The goal of the subfield of Artificial Intelligence called Planning&Scheduling is the development of software tools that generate plans of actions for physical  (humans or robots) or software agents. Examples of successful deployment of this technology have been shown on tasks such are: robotics, military operations, or civil emergencies. Very efficient general algorithms exist that work indepedently of the task. We will cover the main approaches currently used in automated planning.

[ 2hours ]
In order to apply automated planning to real-world tasks, there are two major configuration tasks that have to be performed manually:  defining the actions model of the domain (as what actions the robot can execute and what are the effects of those actions), and defining the domain specific knowledge to make the software efficient for that domain (when performing an action is better than performing another one). This is what machine learning techniques can help with. We will cover some of these machine learning approaches applied to automated planning during the seminar.
Assessment: Paper review (assigned by the teacher)
Organizer: Giuliano Armano
Dep. of Electrical and Electronic Engineering
University of Cagliari, Italy
Email: armano@diee.unica.it