![]() |
CSCI 599
|
Office : SAL 228 Phone : (213) 740-4520 Email : mataric@cs.usc.edu WWW : http://robotics.usc.edu/~maja Office Hours : by appointment, send email or talk w/ prof. after class
We will cover the development of the field and overview the different approaches to mobile robot control, including planner-based control, reactive-control, behavior-based control, and hybrid control. We will focus on the issues of resolving the well-known conflict between thinking and acting, i.e., high-level deliberation and low-level real-time control. We will cover different approaches and robot control architectures for addressing this issue. We will also study issues in real-time and active sensing as well as robot learning, both of which are integral to mobile robot control.
In the second part of the course we will discuss scaling up mobile robot control to multi-robot systems and swarms of robots. We will relate this problem to multi-agent control and to Distributed AI, and discuss scaling up of centralized, hierarchical, and decentralized architectures for control. Finally, we will address learning in multi-agent and multi-robot systems, and deal with the issues of interference and credit assignment.
Jan 8 : Introduction and overview of the course; intro to the available robot testbeds
Jan 9 :
A lab introduction to the R1 and Geo robots, presented by Gaurav
Sukhatme
and Barry Werger; 1pm, SAL 103.
Jan 13 :
Survey of basic control methods
Jan 15 :
Survey of sensors and effectors; sensor fusion and fault
tolerance
Guest lecturer: Gaurav Sukhatme
Jan 20 :
Reactive systems
Jan 22 :
Reactive systems
Jan 27 :
Hybrid systems
Jan 29 :
Hybrid systems
Feb 3 :
Behavior-based systems
Feb 5 :
Behavior-based systems
Feb 10 :
Reinforcement Learning (RL) in Robotics
Feb 12 :
Supervised Learning in Robotics
Feb 17 :
Reinforcement Learning in Robots/Agents
Feb 19 :
Genetic/Evolutionary Approaches to Robot Learning
Feb 24 :
Genetic/Evolutionary Approaches to Robot Learning
Feb 26 :
Multi-agent control (graphical agents)
Mar 3 :
Summary of control and learning; Quiz
Mar 5 :
Multi-robot control
Mar 10 :
No class: Spring Break
Mar 12 :
No class: Spring Break
Mar 17 :
Multi-agent/robot control
Mar 19 :
Multi-agent control
Mar 24 :
Guest Speaker and Videos
Mar 26 :
Biologically-Inspired Models
Mar 31 :
Multi-agent/robot learning
Apr 2 :
Multi-agent/robot learning
Apr 7 :
Philosophical Overviews
Apr 9 :
Philosophical Overviews
Apr 14 :
Overview of the material
Apr 16 :
No class, project preparation
Apr 21 :
Project presentations and discussion
2-3:30pm, location: SAL 322
Apr 24 :
Project presentations and discussion
4-6pm, location: SAL 322
Apr 28 :
Project presentations and discussion
2-4pm, location: EEB 349
Helpful information on how to present the readings in class can be found here.
1) Information on robot kits that you can purchase is listed below, under Related Web Sites.
2) We are making Geo II, a high-degree-of-freedm legged robot, available for your project use; it is shown below on the left. For more information, look here.
3) We are also providing access to some R1 research robots in exchange for help with hardware development. These robots are shown on the right; for more information, look here. However, the robots are currently undergoing redesign and will require work before being useful.
4) Information on some simulation testbeds is also listed below, under Related Web Sites.
Project topics and platforms must be approved by Prof. Mataric.
A detailed final paper is a required part of the final project; it will be due on the last day of presentations, and it will be made available to any interested students in the class.
Helful information for preparing your project presentation can be found here.
Helful information for preparing your final paper can be found here.
Specifically, final course grades will consist of:
15% quiz 20% in-class discussion participation 30% in-class paper presentations 35% final project implementation and final paper
Johuco Ltd. Four different cheap, versatile, and programmable robot kits by Jonathan Connell.
Information on implementing a neural controller for a mobile robot by using a robot kit, by Jonathan Connell (Johuco, Ltd.)
Information on designing behavior-based robots , tied into a robot kit, by Jonathan Connell (Johuco, Ltd.).
Information on Lynxmotion , a series of inexpensive mobile robot, manipulator and sensor kits .
Information on the Rug Warrior , the robot kit that comes with a book: "Mobile Robots, Inspiration to Implementation" by J. L. Jones and A. M. Flynn, published by A. K. Peters, MA. Rug warrior technical information is also available, including schematics, etc. Rug Warrior kits are quite popular so you can find people who have used them to implement interesting robots. For example see Franz, implemented by Francois Michaud.
For information on obtaining the MIT kits (previously used in the USC CSCI499 course), email fredm@media.mit.edu
Information on KIPR, the KISS Institute for Practical Robotics.
All Electronics, a source of affordable sensors and other components. LA store: (213) 380-8000.
Diversified Enterprises Web page with educational and robot kit information.
The IEEE Computer Society Web page with links to "Stiquito: advanced experiments with a simple and inexpensie robot" textbook and kit by J. Conrad and J. Mills.
Public domain software for the official RoboCup robot soccer simulator. The page is full of useful information. People who definitely do not want to use physical robots can use this simulation to explore issues of multi-agent coordination, cooperation, and learning for the course project.
Public domain software for a Java-based soccer simulation called JavaSoccer, written by Tucker Balch.
A public domain Lisp-based simulation of Valentino Braitenberg's Vehicles.
Look here for info on how to prepare your final project presentation.
Adaptive Biological and Artificial Behavior:
Adaptive Behavior Journal
Simulation of Adaptive Behavior Conference (SAB)
Cognitive Science Journal
Cognitive Science Conference
Learning:
Machine Learning Journal
International Conference on Machine Learning (ICML)
Artificial Intelligence:
Artificial Intelligence Journal
Journal of Artificial Intelligence Research
Journal of Experimental and Theoretical AI
American Association for Artificial Intelligence Conference (AAAI)
International Joint Conference on AI (IJCAI)
Autonomous Agents:
Autonomous Agents Journal
Autonomous Agents Conference (Agents)
Artificial Life:
Artificial Life Journal
Artificial Life Conference (Alife)
European Artificial Life Conference (ECAL)
Computational Neural Network Approaches:
Neural Computation
Neural Information Processing Systems Conference (NIPS)
Date | presenter | presenter | presenter |
1/8 | Maja Mataric |
Barry Werger |
Gaurav Sukhatme |
1/13 | --- | Maja Mataric | --- |
1/15 | --- | Gaurav Sukhatme |
--- |
1/20 | Ilia Ovsiannikov |
Wilko Hein |
Aswath Mohan |
1/22 | Richard Kolodji |
Les Williams |
--- |
1/27 | Brian Ellenberger |
Göksel Dedeoglu |
--- |
1/29 | Tony Mactutis |
Srinivasan Lakshmanan |
--- |
2/3 | Pavlin Radoslavov |
Juan Francisco Lopez |
--- |
2/5 | Brian Ellenberger |
Les Williams |
--- |
2/10 | Tony Mactutis |
Richard Kolodji |
--- |
2/12 | Aswath Mohan |
Aswath Mohan |
Göksel Dedeoglu |
2/17 | Göksel Dedeoglu |
Göksel Dedeoglu |
Srinivasan Lakshmanan |
2/19 | Wilko Hein |
Wilko Hein |
--- |
2/24 | Pavlin Radoslavov |
Sean Morrow |
--- |
2/26 | Sean Morrow |
Tony Mactutis |
--- |
3/3 | Review | Quiz | --- |
3/5 | Pavlin Radoslavov |
Juan Francisco Lopez |
Srinivasan Lakshmanan |
3/17 | Pavlin Radoslavov |
Aswath Mohan |
--- |
3/19 | Les Williams |
Wilko Hein |
--- |
3/24 | Guest | speaker | Videos |
3/26 | Brian Ellenberger |
Juan Francisco Lopez |
--- |
3/31 | Aswath Mohan |
Sean Morrow |
--- |
4/2 | Juan Francisco Lopez |
Tony Mactutis |
Les Williams |
4/7 | Richard Kolodji |
Sean Morrow |
--- |
4/9 | Brian Ellenberger |
Srinivasan Lakshmanan |
--- |
4/14 | Richard Kolodji |
course | review |
4/21 | course Brian Ellenberger |
project Pavlin Radoslavov |
presentations Aswath Mohan |
4/24 | course Göksel Dedeoglu |
project Richard Kolodji |
presentations Tony Mactutis Les Williams |
4/28 | course Wilko Hein |
project Juan Francisco Lopez |
presentations Srinivasan Lakshmanan |