1. Introduction

We are in the early days of RoboCup [Kitano95IJCAI], with half a century to go before we can “…build a team of robot soccer players, which can beat a human world cup champion team” [RoboCup97]. The challenge posed by the goal is enormous and inspires hundreds of researchers yearly throughout the world to engage themselves and their students in RoboCup. RoboCup has been used as a research challenge in parallel with a usage for educational purposes, and to stimulate the interest of the public for robotics and artificial intelligence(AI). Each year since 1997, researchers from different countries have gathered to play the world cup. The event has drawn an increasing amount of interest from the public, as robotics is still not commonplace.

The intention of this manual [1] is to guide the developers of simulated league teams in the beginning steps, and also serve as a reference manual for the experienced users.

1.1. Background

Mackworth [Mackworth93] introduced the idea of using soccer-playing robots in research. Unfortunately, the idea did not get the proper response until the idea was further developed and adapted by Kitano, Asada, and Kuniyoshi, when proposing a Japanese research programme, called Robot J-League [2]. During the autumn of 1993, several American researchers took interest in the Robot J-League, and it thereafter changed name to the Robot World Cup Initiative or RoboCup for short. RoboCup is sometimes referred to as the RoboCup challenge or the RoboCup domain.

In 1995, Kitano et al. [Kitano95IJCAI] proposed the first Robot World Cup Soccer Games and Conferences to take place in 1997. The aim of RoboCup was to present a new standard problem for AI and robotics, somewhat jokingly described as the life of AI after Deep Blue [3] . RoboCup differs from previous research in AI by focusing on a distributed solution instead of a centralised solution, and by challenging researchers from not only traditionally AI-related fields, but also researchers in the areas of robotics, sociology, real-time mission critical systems, etc.

To co-ordinate the efforts of all researchers, the RoboCup Federation was formed. The goal of RoboCup Federation is to promote RoboCup, for example by annually arranging the world cup tournament. Members of the RoboCup Federation are all active researchers in the field, and represent a number of universities and major companies. As the body of researchers is quite large and widespread, local committees are formed to promote RoboCup-related events in their geographical area.

1.2. The Goals of RoboCup

The RoboCup Federation has set goals and a timetable for the research. Setting goals and a timetable are means of pushing the state-of-the-art further, in conjunction with formalised test-beds. In resemblance with John F. Kennedy’s national goal of “landing a man on the moon and returning him safely to earth” ([4], p. 8276), the main accomplishment was not to land a man on the moon and returning him safely, but the overall technological advancement. Therefore, the most important goal of RoboCup is to advance the overall technological level of society, and as a more pragmatic goal to achieve the following:

By mid-21st century, a team of fully autonomous humanoid robot soccer players shall win the soccer game, comply with the official rule of the FIFA 4 , against the winner of the most recent World Cup.

There will be several technological advancements, even if the goal of the robotic soccer team is not reached, starting with Team-Partitioned, Opaque-Transition Reinforcement Learning (TPOT-RL) [Stone98] which has found application in the domain of packet routing in computer networks. TPOT-RL is a distributed learning method in domains where “agents have limited information about environmental state transitions” ([Stone98], p. 22).

In most RoboCup leagues, the teams consist of either robots or programs that cooperate in order to defeat the opponent team. RoboCup Rescue and the commentator exhibition diverge from the other RoboCup leagues. The goal of defeating an opponent would raise ethical issues in RoboCup Rescue, since we cannot assign comparable utilities to human lives and buildings. Hence, the focus in RoboCup Rescue is on the co-operative efforts between heterogeneous agents. In the commentator exhibition, the goal is to observe and comment.

Besides the commentator exhibition and RoboCup Rescue, the main body of the RoboCup challenge consists of several leagues for soccer playing. However, as this manual is about the simulated league we will only focus on it.

1.2.1. Simulated League

The RoboCup simulator league is based on the RoboCup simulator called the soccer server [Noda97RoboCup97], a physical soccer simulation system. All games are visualised by displaying the field of the simulator by the soccer monitor on a computer screen. The soccer server is written to support competition among multiple virtual soccer players in an uncertain multi-agent environment, with real-time demands as well as semi-structured conditions. One of the advantages of the soccer server is the abstraction made, which relieves the researchers from having to handle robot problems such as object recognition [9], communications, and hardware issues, e.g., how to make a robot move. The abstraction enables researchers to focus on higher level concepts such as cooperation and learning.

Since the soccer server provides a challenging environment, i.e., the intentions of the players cannot mechanically be deduced, there is a need for a referee when playing a match. The included artificial referee is only partially implemented and can detect trivial situations, e.g., when a team scores. However, there are several hard-to-detect situations in the soccer server, e.g., deadlocks, which brings the need for a human referee. All participating teams are also obliged to play according to a gentlemen’s agreement, e.g., not to use loopholes.

Since the first version of the soccer server was completed in 1995, there have been four world cups and one pre-world cup event, not to mention all other RoboCup-related events. The 1996 pre-RoboCup event [PreRoboCup96] was held in Osaka, with only seven entrants in the competition which ended with a Japanese victory by the team Ogalets from Tokyo University. In Nagoya the following year, the first formal competition was held in conjunction with the IJCAI’97 conference. The competition had 29 teams participating, and the winner was AT Humboldt [Burkhard97]. The RoboCup world cup of 1998 was played in conjunction with the human world cup in Paris, and the winning team was CMUnited98 [CMUnited98]. During the world cup, media was heavily covering the event, as it was public in a museum in the suburbs of Paris. The year after, the world cup was held in conjunction with IJCAI’99 in Stockholm, and the winners (once again) were CMUnited99 [CMUnited99]. An unchanged version of the champion team must participate, as a benchmark, in the next world cup. The benchmarking teams have always been able to win their group, but only in 2000 did the benchmark team advance further than the first game after group play.

1.2.2. What is the Soccerserver

Soccer Server is a system that enables autonomous agents consisting of programs written in various languages to play a match of soccer (association football) against each other.

A match is carried out in a client/server style: A server, soccerserver, provides a virtual field and simulates all movements of a ball and players. Each client controls movements of one player. Communication between the server and each client is done via UDP/IP sockets. Therefore users can use any kind of programing systems that have UDP/IP facilities.

The soccerserver consists of 2 programs, soccerserver and soccermonitor. Soccer Server is a server program that simulates movements of a ball and players, communicates with clients, and controls a game according to rules. Soccermonitor is a program that displays the virtual field from the soccerserver on the monitor using the X window system. A number of soccermonitor programs can connect with one soccerserver, so we can display field-windows on multiple displays.

A client connects with soccerserver by an UDP socket. Using the socket, the client sends commands to control a player of the client and receives information from sensors of the player. In other words, a client program is a brain of the player: The client receives visual and auditory sensor information from the server, and sends control-commands to the server.

Each client can control only one player 56 . So a team consists of the same number of clients as players. Communications between the clients must be done via soccerserver using say and hear protocols. (See section Player Command Protocol.) One of the purposes of soccerserver is evaluation of multi-agent systems, in which efficiency of communication between agents is one of the criteria. Users must realize control of multiple clients by such restricted communication.

1.3. History

In this section we will first describe the history of the soccerserver and thereafter the history of the RoboCup Simulation League. To end the section we will also describe the history of the manual effort.

1.3.1. History of the Soccer Server

The first, preliminary, original system of soccerserver was written in September of 1993 by Itsuki Noda, ETL. This system was built as a library module for demonstration of a programming language called MWP, a kind of Prolog system that has multi-threads and high level program manipulation. The module was a closed system and displayed a field on a character display, that is VT100.

The first version (version 0) of the client-server style server was written in July of 1994 on a LISP system. The server shows the field on an X window system, but each player was shown in an alphabet character. It used the TCP/IP protocol for connections with clients. This LISP version of soccerserver became the original style of the current soccerserver. Therefore, the current soccerserver uses S-expressions for the protocol between clients and the server.

The LISP version of soccerserver was re-written in C++ in August of 1995 (version 1). This version was announced at the IJCAI workshop on Entertainment and AI/Alife held in Montreal, Canada, August 1995.

The development of version 2 started January of 1996 in order to provide the official server of preRoboCup-96 held at Osaka, Japan, November 1996. From this version, the system is divided into two modules, soccerserver and soccerdisplay (currently, soccermonitor). Moreover, the feature of coach mode was introduced into the system. These two features enabled researchers on machine learning to execute games automatically. Peter Stone at Carnegie Mellon University joined the decision-making process for the development of the soccerserver at this stage. For example, he created the configuration files that were used at preRoboCup-96.

After preRoboCup-96, the development of the official server for the first RoboCup, RoboCup-97 held at Nagoya, Japan, August 1997, started immediately, and the version 3 was announced in February of 1997. Simon Ch’ng at RMIT joined decisions of regulations of soccerserver from this stage. The following features were added into the new version:

  • logplayer

  • information about movement of seen objects in visual information

  • capacity of hearing messages

The development of version 4 started after RoboCup-97, and announced November 1997. From this version, the regulations are discussed on the mailing list organized by Gal Kaminka. As a result, many contributers joined the development. Version 4 had the following new features:

  • more realistic stamina model

  • goalie

  • handling offside rule

  • disabling players for evaluation

  • facing direction of players in visual information

  • sense body command

Version 4 was used in JapanOpen 98, RoboCup98 and Pacific Rim Series 98.

Version 5 was used in JapanOpen 99, and will also be used in RoboCup99 in Stockholm during the summer of 1999.

In Melbourne 2000, version 6 was used, and for the world cup in 2001 version 7 will be used.

1.3.2. History of the RoboCup Simulation League

The RoboCup simulation league has had 26 main official events: Starting with a preRoboCup96 in 1996 event, from 1997 onward an official world championship tournament was held each year (from RoboCup97 to RoboCup 2021). Research results have been reported extensively in the proceedings of the workshops and conferences associated with these competitions. In this section, we focus mainly on the competitions themselves.


preRoboCup96 was the first robotic soccer competition of any sort. It was held on November 5–7, 1996 in Osaka, Japan [5]. In conjunction with the IROS-96 conference, preRoboCup96 was meant as an informal, small-scale competition to test the RoboCup soccerserver in preparation for RoboCup97. 5 of the 7 entrants were from the Tokyo region. The other 2 were from Ch’ng at RMIT and Stone and Veloso from CMU. The winning teams were entered by:

  1. Ogawara (Tokyo University)

  2. Sekine (Tokyo Institute of Technology)

  3. Inoue (Waseda University)

  4. Stone and Veloso (Carnegie Mellon University)

In this tournament, team strategies were generally quite straightforward. Most of the teams kept players in fixed locations, only moving them towards the ball when it was nearby.


The RoboCup97 simulator competition was the first formal simulated robotic soccer competition. It was held on August 23–29, 1997 in Nagoya, Japan in conjunction with the IJCAI-97 conference [6]. With 29 teams entering from all around the world, it was a very successful tournament. The winning teams were entered by:

  1. Burkhard et al. (Humboldt University)

  2. Andou (Tokyo Institute of Technology)

  3. Tambe et al. (ISI/University of Southern California)

  4. Stone and Veloso (Carnegie Mellon University)

In this competition, the champion team exhibited clearly superior low-level skills. One of its main advantages in this regard was its ability to kick the ball harder than any other team. Its players did so by kicking the ball around themselves, continually increasing its velocity so that it ended up moving towards the goal faster than was imagined possible. Since the soccerserver did not (at that time) enforce a maximum ball speed, a property that was changed immediately after the competition, the ball could move arbitrarily fast, making it almost impossible to stop. With this advantage at the low-level behavior level, no team, regardless of how strategically sophisticated, was able to defeat the eventual champion.

At RoboCup97, the RoboCup scientific challenge award was introduced. Its purpose is to recognize scientific research results regardless of performance in the competitions. The 1997 award went to Sean Luke [10] of the University of Maryland “for demonstrating the utility of evolutionary approach by co-evolving soccer teams in the simulator league.”


The second international RoboCup championship, RoboCup-98, was held on July 2–9, 1998 in Paris, France [1]. It was held in conjunction with the ICMAS-98 conference.

The winning teams were entered by:

  1. Stone et al. (Carnegie Mellon University)

  2. Burkhard et al. (Humboldt University)

  3. Corten and Rondema (University of Amsterdam)

  4. Tambe et al. (ISI/University of Southern California)

Unlike in the previous year’s competition, there was no team that exhibited a clear superiority in terms of low-level agent skills. Games among the top three teams were all quite closely contested with the differences being most noticeable at the strategic, team levels.

One interesting result at this competition was that the previous year’s champion team competed with minimal modifications and finished roughly in the middle of the final standings. Thus, there was evidence that as a whole, the field of entries was much stronger than during the previous year: roughly half the teams could beat the previous champion.

The 1998 scientific challenge award was shared by Electro Technical Laboratory (ETL), Sony Computer Science Laboratories, Inc., and German Research Center for Artificial Intelligence GmbH (DFKI) for “development of fully automatic commentator systems for RoboCup simulator league.”

To encourage the transfer of results from RoboCup to the scientific community at large, RoboCup98 was the first to host the Multi-Agent Scientific Evaluation Session. 13 different teams participated in the session, in which their adaptability to loss of team-members was evaluated comparatively. Each team was played against the same fixed opponent (the 1997 winner, AT Humboldt’97) four half-games under official RoboCup rules. The first half-game (phase A) served as a base-line. In the other three half- games (phases B-D), 3 players were disabled incrementally: A randomly chosen player, a player chosen by the representative of the fixed opponent to maximize “damage” to the evaluated team, and the goalie. The idea is that a more adaptive team would be able to respond better to these.

Very early on, even during the session itself, it became clear that while in fact most participants agreed intuitively with the evaluation protocol, it wasn’t clear how to quantitatively, or even qualitatively, analyse the data. The most obvious measure of the goal-difference at the end of each half may not be sufficient: some teams seem to do better with less players, some do worse. Performance, as measured by the goal-difference, really varied not only from team to team, but also for the same team between phases. The evaluation methodology itself and analysis of the results became open research problems in themselves. To facilitate this line of research, the data from the evaluation was made public at: http://www.isi.edu/~galk/Eval/


The third international RoboCup championship, RoboCup-99, was held in late July and early August, 1999 in Stockholm, Sweden [3]. It was held in conjunction with the IJCAI-99 conference.


The fourth international RoboCup championship, RoboCup 2000, was held in early September, 2000 in Melbourne, Australia [16]. It was held in conjunction with the PRICAI-2000 conference.

RoboCup 2004

The eigth international RoboCup championship, RoboCup 2004, was held in Lisbon, Portugal. It was accompanied by the RoboCup 2004 Symposium, held at the Instituto Superior Tecnico and was co-located with the 5th IFAC/EURON International Symposium on Intelligent Autonomous Vehicles (IAV 2004).

The main novelty in the Soccer Simulation League in 2004 was the introduction of the 3D soccer simulator, where players are spheres in a three-dimensional environment with a full physical model. This sub-competition was the spawning point for the Soccer Simulation 3D League in later years.

The winning teams in the Soccer Simulation 2D competition, for which 24 teams were qualified, were:

  1. STEP (ElectroPult Plant Company, Russia)

  2. Brainstormers (University of Osnabrueck, Germany)

  3. Mersad (Allameh Helli High School, Iran)

The winning teams in the coach competition were:

  1. MRL (Azad University of Qazvin, Iran)

  2. FC Portugal (Universities of Porto and Aveiro, Portugal)

  3. Caspian (Iran University of Science and Technology, Iran)

The winning teams in the 3D competition were:

  1. Aria (Amirkabir University of Technology, Iran)

  2. AT-Humboldt (Humboldt University Berlin, Germany)

  3. UTUtd 2004 (University of Tehran, Iran)

RoboCup 2005

The tenth international RoboCup championship, RoboCup 2005, was held in July 2005 in Osaka, Japan [16]. It was accompanied by the RoboCup Symposium. Since, for the first time, the 3D sub-league of soccer simulation had its own tournament, the number of teams that were maximally allowed to qualify for the Soccer Simulation 2D competitions at RoboCup 2005 was reduced to 16 (though a 17th team was permitted for reasons of the qualifying procedure).

The winning teams in the Soccer Simulation 2D competition were:

  1. Brainstormers (University of Osnabrueck, Germany)

  2. WrightEagle (University of Science and Technology of China, China)

  3. TokyoTech SFC (Tokyo Institute of Technology, Japan)

An interesting observation, quite similar to the related remark for RoboCup98, could be made in this year: Last year’s champion (STEP, Russia) entered the competition without any modifications made to their team and finished the tournament on rank 4.

1.3.3. History of the Soccer Manual Effort

The first versions of the manual were written by Itsuki Noda, while developing the soccerserver, and around version 3.00 there were several requests on an updated manual, to better correspond to the server as well as enable newcomers to more easily participate in the RoboCup World Cup Initiative. In the fall of 1998 Peter Stone initiated the Soccer Manual Effort, over which Johan Kummeneje took responsibility to organize and as a result the Soccer Server Manual version 4.0 was released on the 1st of November 1998.

In 1999, the manual for the soccerserver version 5.0 was released. Unfortunately the manual lost part of its pace, and there was no release of the manual for soccerserver version 6.0.

Since 1999, the soccerserver has changed major version to 7 and is continuously developed. Therefore the Soccer Manual Effort has developed a new version, which resulted in a PDF version of the Soccer Manual (available on Sourceforge) that has been the main reference document for many years.

In 2009 and 2010 (soccerserver versions 12 and 14), significant changes were introduced to the way the soccerserver simulates soccer, including a changed tackle model and a sideward dash model to mention just a few. The corresponding changes of those times were, unfortunately, not incorporated into the existing soccerserver manual, but were reflected only in the NEWS text file as part of the soccerserver software package.

In 2019, a joint effort was started to migrate the existing Latex-based soccerserver manual to the Github-hosted version that is based on reStructured text and that you are reading here.

1.4. About This Manual

This manual is the joint effort of the authors from a diverse range of universities and organizations, which build upon the original work of Itsuki Noda. If there are errors, inconsistencies, or oddities, please notify johank@dsv.su.se or fruit@uni-koblenz.de with the location of the error and a suggestion of how it should be corrected.

We are always looking for anyone who has an idea on how to improve the manual, as well as proofread or (re)write a section of the manual. If you have any ideas, or feel that you can contribute with anything to the SoccerServer Manual Effort. .. please mail johank@dsv.su.se or fruit@uni-koblenz.de.

1.5. Reader’s Guide to the Manual

The thesis is written for a wide range of readers, and therefore the chapters are not equally important to all readers. We shortly describe the remaining chapters to give an overview of the thesis.

Chapter 2 introduces the concepts of the simulated league and will help the newcomer to get to terms with the different parts.

Chapter 3 helps the beginners to start compiling and running the software.

Chapter 4 describes the soccerserver.

Chapter 5 describes the soccermonitor.

Chapter 6 describes the soccerclient and how to create one.

Chapter 7 describes the coachclient.

Chapter 8 suggests some further reading.