From Autonomous Multirobot Systems

Contents 1 Autonomous Multi-Robot Systems Spring 2012 1.1 Administrative 1.2 Topics 1.3 Main Components of Course 1.4 Grading Components & Expectations 1.5 Resources 1.6 Prerequisite 1.7 Final Preparation 1.8 Schedule 1.9 Projects 1.10 Content We May Use 1.11 Previous Assignments We May Use 1.12 Evaluation 1.13 Resources

Autonomous Multi-Robot Systems Spring 2012

Administrative

• web:http://www.intelligentappliances.org/multirobot/
• piazza http://piazza.com/class#spring2012/cs76318803mrs4632
• Professor: Tucker Balch CCB 317
• TA: Brian Hrolenok
• BioSim help: Hai Shang
• Tuesday/Thursday 3:00-4:30 Klaus 2443
• T-square site: TBD
• Link to 2008 course, 2010 course

Topics

Hierarchical robot control; Behavior-based robot control; Sensors and perception; Navigation and path planning; Localization; Reinforcement learning; Communication; Diversity.

Students should have strong programming skills.

Main Components of Course

First third, Basics:

• Basics of behavior-based control for robots
• Basics of multi-robot systems
• Project: Simple multi-robot system, ASCII soccer (C programming language)

Second third, Research:

• Survey of important literature
• Each student chooses a group of papers on a multi-agent or multi-robot topic
• Each student leads a 30 minute discussion of their topic

Last third, Build Multi-Robot Systems:

• Several projects using BioSim, Java based multi agent simulation system (Java)
• Key component is experimental evaluation of the performance of these systems.

Grading Components & Expectations

• Basic projects: 30%
• Presentation: 30% (re-do allowed)
• Multi-robot projects: 30%
• Quizzes: 10% (easy questions about the content of presentations)

If you take the course P/F you only need to do the presentation and quiz component.

Resources

Some papers and pointers:

• motor schema formulations
• motor schemas for formation control
• Voting paper by Rosenblatt
• Voting paper by USC folks
• Q-learning on wikipedia
• Q-learning paper by Kaelbling, Littman, Moore

Software we'll be using:

• ascii soccer
• MASON multiagent toolkit
• MASON instructions

Prerequisite

Strong programming skills. We will be using Unix, C, Java.

Final Preparation

• Arkin paper on motor schemas
• motor schema formulations
• article about diversity (sections 1, 2, & 3)
• paper on Reinforcement Learning (sections 1.1, 1.2, 3.1, 4.2)

Schedule

January
Tuesday	January 10	Class overview vid1.mov vid2.mov vid3 vid4
Thursday	January 12	Sense Think Act cycle multirobot lecture 1	First project discussed (ASCII Soccer)
Tuesday	January 17	Class overview (repeat for France) More on ascii soccer
Thursday	January 19	ascii soccer workshop
Tuesday	January 24	Navigation Part 1: Planning.

Projects

• 2012_Project_1, Part 1 is due Monday January 23 at 11:55PM
• 2012_Project_1, Part 2 is due Friday January 27 at 11:55PM

Content We May Use

• multirobot lecture 0
• multirobot lecture 1
• multirobot lecture 2
• Diversity Part 1
• Diversity Part 2
• article about diversity (sections 1, 2, 3, 4, & 7)
• Behavior Based Control Part 1
• motor schema formulations
• Behavior Based Control Part 2
  
• Introduction to MASON,
• Reinforcement Learning: media:RL-slides.pdf
• RRTs
• Intro to multi robot / multi objective visit tasks. PRIM Algorithm details
• Ants & ERRTs
• Paper on communication (with sections to read)
• Reinforcement Learning paper
• MASON assignment
• Shang Hai MASON slides
• midterm (take home)

Previous Assignments We May Use

• 2010Fall Project 1 Part 1
• 2010Fall Project 1 Part 2
• 2010Fall Project 2
• 2010Fall Project 3
• 2010Fall Project 4
• 2010Fall Project 5
• 2010Fall Project 6

Evaluation

Grading will be based on projects and tests as follows as follows:

• Midterm exam: 10%
• Final exam: 10%
• Programming assignments: 70%
• Paper presentation: 10%

Resources