CMUnited RoboCup-99 Simulator Team

This page presents the CMUnited team entered in the RoboCup-99 simulator league. For more information about our entire projcet (including small and legged robots) please see our RoboSoccer project page.

People

Peter Stone, Patrick Riley, Manuela Veloso

Results

CMUnited-99 is the RoboCup-99 Simulator League World Champion Team!

The tournament included 37 teams from around the world.
Over the course of 8 games, CMUnited-99 outscored its opponents by a combined score of 110-0!
Results of RoboCup-99 are available from the main RoboCup homepage and from the RoboCup-99 page (including logs of all the games).

Accelerators
Summary | Using Optimal Teammate and Opponent Models | Layered Disclosure
Software

Summary

The CMUnited-99 simulator team is based closesly upon CMUnited-98 simulator team. In particular, it uses all of the techniques that are described in detail, with pointers to relevant publications, on the CMUnited-98 simulator team page:
(1) Hierarchical machine learning (Layered learning)
(2) Flexible, adaptive formations (Locker-room agreement)
(3) Single-channel, low-bandwidth communication
(4) Predictive, locally optimal skills (PLOS)
(5) Strategic positioning using attraction and repulsion (SPAR)

CMUnited-99 also improves upon CMUnited-98 in many ways. For example, several of the low-level skills are improved and updated to deal with server changes (such as the introduction of a neck), the use of opponent and teammate models is introduced, and some coordination procedures are improved, in part by taking advantage of the on-line coach. In addition, we introduce a development paradigm called layered disclosure.
These contributions are described in detail in our team description paper that is included in the source code and binary distributions below:
In "RoboCup-99: Robot Soccer World Cup III", M. Veloso, E. Pagello, and H. Kitano (eds.), 2000. Springer Verlag, Berlin.
The CMUnited-99 Champion Simulator Team. (extended version)
Peter Stone, Manuela Veloso, and Patrick Riley

@InProceedings(StoneRileyVeloso2000a, Author="Peter Stone and Patrick Riley and Manuela Veloso", Title="The {CMU}nited-99 Champion Simulator Team", booktitle= "{R}obo{C}up-99: Robot Soccer World Cup III", Editor="Manuela Veloso and Enrico Pagello and Hiroaki Kitano", Publisher="Springer Verlag",address="Berlin",year="2000")

Using Optimal Teammate and Opponent Models

A common challenge for agents in multiagent systems is trying to predict what other agents are going to do in the future. Such knowledge can help an agent determine which of its current action options is most likely to help it achieve its goals. We introduce ideal model-based behavior outcome prediction'' (IMBBOP) which models the results of other agents' future actions in relation to their optimal actions based on an ideal world model. Our technique also includes a method for relaxing this optimality assumption.
CMUnited-99 uses IMBBOP to determine whether to shoot, pass, or dribble when an agent has the ball close to the opponent's goal.
We have written the following paper about IMBBOP.
Defining and Using Ideal Teammate and Opponent Agent Models
Peter Stone, Patrick Riley, and Manuela Veloso.
It will appear in the Twelfth Innovative Applications of AI Conference (IAAI-2000).
@InProceedings(StoneRileyVeloso2000b, Author="Peter Stone and Patrick Riley and Manuela Veloso", Title="Defining and Using Ideal Teammate and Opponent Agent Models", BookTitle="Proceedings of the Twelfth Innovative Applications of AI Conference", Year="2000")

Layered Disclosure

A perennial challenge in creating and using complex autonomous agents is following their choice of actions as the world changes dynamically and understanding why they act as they do. We introduce the concept of layered disclosure by which autonomous agents include in their architecture the foundations necessary to allow a person to probe into the specific reasons for an agent's action. This probing may be done at any level of detail, and either retroactively or while the agent is acting.
Our source code and executable release below includes a test function for generating detailed logfiles for performing layered disclosure, an augmented logplayer for viewing the detailed information, and a sample game with disclosure logfiles.
Seventh International Workshop on Agent Theories, Architectures, and Languages (ATAL-2000)
Layered Disclosure: Revealing Agents' Internals.
Patrick Riley, Peter Stone, and Manuela Veloso

@InProceedings(RileyStoneVeloso2000, Author="Patrick Riley and Peter Stone and Manuela Veloso", Title="Layered Disclosure: Revealing Agents' Internals", BookTitle="The Seventh International Workshop on Agent Theories, Architectures, and Languages (ATAL-2000)", Year="2000")

The Software

You can now download a portion of the CMUnited-99 source code. It is written in C++ and has been tested under Linux and under SunOS.
We made some effort to package the code in such a way that people will be able to learn from and incorporate it. The README file included in the directory describes how to compile and run the code. You should be able to easily produce dribbling, kicking, and ball interception behaviors.
Included with the source code release is a paper describing our control algorithms in detail:
In "RoboCup-99: Robot Soccer World Cup III", M. Veloso, E. Pagello, and H. Kitano (eds.), 2000. Springer Verlag, Berlin.
The CMUnited-99 Champion Simulator Team. (extended version)
Peter Stone, Manuela Veloso, and Patrick Riley
The purpose of the code release is to allow people to get quickly past the low-level implementation details involved in working with the soccer server. Our high-level behaviors, are best described in our various papers (see above).
Please note that the code is released as is, with no support provided. Also, please keep track of what code and ideas of ours you use.
You may also run CMUnited-99 as it ran in Stockholm.
Download the tarred, gzipped executables for Linux or Solaris.
A README file with instructions on how to start the clients is included.
Additional packages available for download are:
(1) Coach modules for creating both on-line and off-line coaches;
(2) An extended logplayer for performing layered disclosure; and
(3) A sample game for demonstrating layered disclosure.

Simulator Team Homepage | RoboSoccer Project Homepage | Computer Science Department | School of Computer Science

	People Peter Stone, Patrick Riley, Manuela Veloso
	Results CMUnited-99 is the RoboCup-99 Simulator League World Champion Team! The tournament included 37 teams from around the world. Over the course of 8 games, CMUnited-99 outscored its opponents by a combined score of 110-0! Results of RoboCup-99 are available from the main RoboCup homepage and from the RoboCup-99 page (including logs of all the games).
	Accelerators Summary \| Using Optimal Teammate and Opponent Models \| Layered Disclosure Software
	Summary The CMUnited-99 simulator team is based closesly upon CMUnited-98 simulator team. In particular, it uses all of the techniques that are described in detail, with pointers to relevant publications, on the CMUnited-98 simulator team page: (1) Hierarchical machine learning (Layered learning) (2) Flexible, adaptive formations (Locker-room agreement) (3) Single-channel, low-bandwidth communication (4) Predictive, locally optimal skills (PLOS) (5) Strategic positioning using attraction and repulsion (SPAR) CMUnited-99 also improves upon CMUnited-98 in many ways. For example, several of the low-level skills are improved and updated to deal with server changes (such as the introduction of a neck), the use of opponent and teammate models is introduced, and some coordination procedures are improved, in part by taking advantage of the on-line coach. In addition, we introduce a development paradigm called layered disclosure. These contributions are described in detail in our team description paper that is included in the source code and binary distributions below: In "RoboCup-99: Robot Soccer World Cup III", M. Veloso, E. Pagello, and H. Kitano (eds.), 2000. Springer Verlag, Berlin. The CMUnited-99 Champion Simulator Team. (extended version) Peter Stone, Manuela Veloso, and Patrick Riley @InProceedings(StoneRileyVeloso2000a, Author="Peter Stone and Patrick Riley and Manuela Veloso", Title="The {CMU}nited-99 Champion Simulator Team", booktitle= "{R}obo{C}up-99: Robot Soccer World Cup III", Editor="Manuela Veloso and Enrico Pagello and Hiroaki Kitano", Publisher="Springer Verlag",address="Berlin",year="2000")
	Using Optimal Teammate and Opponent Models A common challenge for agents in multiagent systems is trying to predict what other agents are going to do in the future. Such knowledge can help an agent determine which of its current action options is most likely to help it achieve its goals. We introduce ideal model-based behavior outcome prediction'' (IMBBOP) which models the results of other agents' future actions in relation to their optimal actions based on an ideal world model. Our technique also includes a method for relaxing this optimality assumption. CMUnited-99 uses IMBBOP to determine whether to shoot, pass, or dribble when an agent has the ball close to the opponent's goal. We have written the following paper about IMBBOP. Defining and Using Ideal Teammate and Opponent Agent Models Peter Stone, Patrick Riley, and Manuela Veloso. It will appear in the Twelfth Innovative Applications of AI Conference (IAAI-2000). @InProceedings(StoneRileyVeloso2000b, Author="Peter Stone and Patrick Riley and Manuela Veloso", Title="Defining and Using Ideal Teammate and Opponent Agent Models", BookTitle="Proceedings of the Twelfth Innovative Applications of AI Conference", Year="2000")
	Layered Disclosure A perennial challenge in creating and using complex autonomous agents is following their choice of actions as the world changes dynamically and understanding why they act as they do. We introduce the concept of layered disclosure by which autonomous agents include in their architecture the foundations necessary to allow a person to probe into the specific reasons for an agent's action. This probing may be done at any level of detail, and either retroactively or while the agent is acting. Our source code and executable release below includes a test function for generating detailed logfiles for performing layered disclosure, an augmented logplayer for viewing the detailed information, and a sample game with disclosure logfiles. Seventh International Workshop on Agent Theories, Architectures, and Languages (ATAL-2000) Layered Disclosure: Revealing Agents' Internals. Patrick Riley, Peter Stone, and Manuela Veloso @InProceedings(RileyStoneVeloso2000, Author="Patrick Riley and Peter Stone and Manuela Veloso", Title="Layered Disclosure: Revealing Agents' Internals", BookTitle="The Seventh International Workshop on Agent Theories, Architectures, and Languages (ATAL-2000)", Year="2000")
	The Software You can now download a portion of the CMUnited-99 source code. It is written in C++ and has been tested under Linux and under SunOS. We made some effort to package the code in such a way that people will be able to learn from and incorporate it. The README file included in the directory describes how to compile and run the code. You should be able to easily produce dribbling, kicking, and ball interception behaviors. Included with the source code release is a paper describing our control algorithms in detail: In "RoboCup-99: Robot Soccer World Cup III", M. Veloso, E. Pagello, and H. Kitano (eds.), 2000. Springer Verlag, Berlin. The CMUnited-99 Champion Simulator Team. (extended version) Peter Stone, Manuela Veloso, and Patrick Riley The purpose of the code release is to allow people to get quickly past the low-level implementation details involved in working with the soccer server. Our high-level behaviors, are best described in our various papers (see above). Please note that the code is released as is, with no support provided. Also, please keep track of what code and ideas of ours you use. You may also run CMUnited-99 as it ran in Stockholm. Download the tarred, gzipped executables for Linux or Solaris. A README file with instructions on how to start the clients is included. Additional packages available for download are: (1) Coach modules for creating both on-line and off-line coaches; (2) An extended logplayer for performing layered disclosure; and (3) A sample game for demonstrating layered disclosure.