RoboCup 2002 Publications

From RoboCup Wiki
Jump to: navigation, search
Author Title Year Journal/Proceedings Reftype DOI/URL
Asada, M. & Kaminka, G.A.

An Overview of RoboCup 2002 Fukuoka/Busan

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 1-7 inproceedings


Abstract: The sixth Robot World Cup Competition and Conference (RoboCup 2002) Fukuoka/Busan took place between June 19th and 25th in Fukuoka: competitions were held at Fukuoka Dome Baseball Stadium from June 19th to 23rd, 2002, followed by the International RoboCup Symposium on June 24th and 25th, 2002.
Bentivegna, D.C. & Atkeson, C.G.

A Framework for Learning from Observation Using Primitives

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 263-270 inproceedings


Abstract: This paper describes a method to learn task primitives from observation. A framework has been developed that allows an agent to use observed data to initially learn a predefined set of task primitives and the conditions under which they are used. A method is also included for the agent to increase its performance while operating in the environment. Data that is collected while a human performs a task is parsed into small parts of the task called primitives. Modules are created for each primitive that encode the movements required during the performance of the primitive, and when and where the primitives are performed.
Birk, A. & Kenn, H.

A Rescue Robot Control Architecture Ensuring Safe Semi-Autonomous Operation

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 254-262 inproceedings


Abstract: The rescue robots developed at the International University Bremen (IUB) are semi-autonomous mobile robots providing streams of video and other essential data via wireless connections to human operated basestations, supplemented by various basic and optional behaviors on board of the robots. Due to the limitations of wireless connections and the complexity of rescue operations, the full operation of a robot can not be constantly supervised by a human operator, i.e., the robots have to be semi-autonomous. This paper describes how the main challenge of safe operation under semi-autonomous control can in general be solved. The key elements are a special software architecture and a scheduling framework that ensure Quality of Service (QoS) and Fail-Safe Guarantees (FSG) despite the unpredictable performance of standard Internet/Intranet-technologies, especially when wireless components are involved.
Bonarini, A.

Medium Size League: 2002 Assessment and Achievements

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 460-468 inproceedings


Abstract: Robots in the Robocup Middle Size League (MSL) have dimensions comparable with robots that could be used in other real world applications. MSL provides a framework to test these robots in a challenging environment where actions should be decided in real-time. In many cases, the achievements shown in this competition are important also for real world applications, and have been exported there. Some other results put in evidence what can be done by focusing on specific issues with the aim of producing more interesting and effective entertainment by autonomous robots.
Browning, B.

RoboCup 2002 Small-Size League Review

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 453-459 inproceedings


Abstract: The RoboCup challenge, to build soccer playing robots able to compete against the best human soccer players, is a goal well beyond our current levels of robot technology. Our current competition structure, with its various leagues, aims to step towards this goal by focusing on different aspects of the RoboCup problem. In the small-size league, global perception allows us to focus primarily on single and multi-robot control and multi-robot teamwork.
Bruce, J. & Veloso, M.M.

Real-Time Randomized Path Planning for Robot Navigation

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 288-295 inproceedings


Abstract: Mobile robots often find themselves in a situation where they must find a trajectory to another position in their environment, subject to constraints posed by obstacles and the robot’s capabilities. This poses the problem of planning a path through a continuous domain. Several approaches have been used to address this problem each with some limitations, including state discretizations, planning efficiency, and lack of interleaved execution. Rapidly-exploring random trees (RRTs) are a recently developed algorithm on which fast continuous domain path planners can be based. In this work, we build a path planning system based on RRTs that interleaves planning and execution, first evaluating it in simulation and then applying it to physical robots. Our algorithm, ERRT (execution extended RRT), introduces two novel extensions of previous RRT work, the waypoint cache and adaptive cost search, which improve replanning efficiency and the quality of generated paths. ERRT is successfully applied to a multi-robot system. Results demonstrate that ERRT is improves efficiency and performs competitively with existing heuristic and reactive real-time path planning approaches. ERRT has shown to offer a major step with great potential for path planning in challenging continuous, highly dynamic domains.
This research was sponsored by Grants Nos. DABT63-99-1-0013, F30602-98-2-0135 and F30602-97-2-0250. The information in this publication does not necessarily reflect the position of the funding agencies and no official endorsement should be inferred.
Carpenter, P., Riley, P., Veloso, M.M. & Kaminka, G.A.

Integration of Advice in an Action-Selection Architecture

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 195-205 inproceedings


Abstract: The introduction of a coach competition in the RoboCup-2001 simulation league raised many questions concerning the development of a “coachable” team. This paper addresses the issues of dealing with conflicting advice and knowing when to listen to advice. An action-selection architecture is proposed to support the integration of advice into an agent’s set of beliefs. The results from the coach competition are discussed and provide a basis for experiments. Results are provided to support the claim that the architecture is well-suited for such a task.
Christaller, T.

Lessons Learned from Fukuoka 2002 Humanoid League

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 485-488 inproceedings


Abstract: We would like to trigger developments towards fully autonomous self-build humanoid robots. Therefore we took so-called performance factors for the different dimensions with regard to autonomy (external power cord, computer outside robot, remote control). Each were to be 1.2 and if more then one is applicable then they are multiplied (1.2, 1.44, 1.728, 2.0736). These factors were either used as penalty factor (e.g. in the walking the time was multiplied by them) or as handicap (in penalty kicking the score was divided by them). I think that they are working quite well (with regard to the above stated intention) and will certainly prefer the more autonomous robots but will also allow for semi-autonomous ones if their performance is much better then that of the autonomous ones. No changes needed.
Dahm, I. & Ziegler, J.

Adaptive Methods to Improve Self-Localization in Robot Soccer

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 393-408 inproceedings


Abstract: This paper shows adaptive strategies to improve the reliability and performance of self-localization in robot soccer with legged robots. Adaptiveness is the common feature of the presented algorithms and has proved essential to enhance the quality of localization by a new classification technique, essential to increase the confidence level of internal information about the environment by extracting reliability information and by communicating them via parameterizable acoustic communication, and essential to circumvent manual implementations of walking patterns by evolving them automatically.
Damas, B.D., Lima, P.U. & Custódio, L.M.

A Modified Potential Fields Method for Robot Navigation Applied to Dribbling in Robotic Soccer

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 65-77 inproceedings


Abstract: This paper describes a modified potential fields method for robot navigation, especially suited for unicycle-type non-holonomic mobile robots. The potential field is modified so as to enhance the relevance of obstacles in the direction of the robot motion. The relative weight assigned to front and side obstacles can be modified by the adjustment of one physically interpretable parameter. The resulting angular speed and linear acceleration of the robot can be expressed as functions of the linear speed, distance and relative orientation to the obstacles. For soccer robots, moving to a desired posture with and without the ball are relevant issues. To enable a soccer robot to dribble a ball, i.e., to move while avoiding obstacles and pushing the ball without losing it, under severe restrictions to ball holding capabilities, a further constraint among the angular speed, linear speed and linear acceleration is introduced. This dribbling behavior has been used successfully in the robots of the RoboCup Middle-Size League ISocRob team.
Douret, J., Benosman, R., Bouzar, S. & Devars, J.

Localization of Robots in F180 League Using Projective Geometry

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 312-318 inproceedings


Abstract: The F180 RoboCup league relies on a single camera mounted on top of the field. It is of great importance to use an adapted calibration method to locate robots. Most of the methods used are developped for specific application where 3D is required. This paper presents a new calibration method specially developped for the F180 league geometry, allowing the determination of the camera pose parameters and the correction of the parallax in the image due to different heights of observed robots. This method needs one calibration plane that also could be used for correcting optical distortions introduced by the lens.
Endo, K., Yamasaki, F., Maeno, T. & Kitano, H.

Co-Evolution of Morphology and Controller for Biped Humanoid Robot

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 327-341 inproceedings


Abstract: In this paper, we present a method for co-evolving structures and control circuits of bi-ped humanoid robots. Currently, bi-ped walking humanoid robots are designed manually on trial-and-error basis. Although certain control theory exists, such as zero moment point (ZMP) compensation, these theories does not constrain design space of humanoid robot morphology or detailed control. Thus, engineers has to design control program for apriori designed morphology, neither of them shown to be optimal within a large design space. We propose evolutionary approaches that enables: (1) automated design of control program for a given humanoid morphology, and (2) co-evolution of morphology and control. An evolved controller has been applied to a humanoid PINO, and attained more stable walking than human designed controller. Co-evolution was achieved in a precision dynamics simulator, and discovered unexpected optimal solutions. This indicate that a complex design task of bi-ped humanoid can be performed automatically using evolution-based approach, thus varieties of humanoid robots can be design in speedy manner. This is a major importance to the emerging robotics industries.
Fujita, M.

Sony Four Legged Robot League at RoboCup 2002

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 469-476 inproceedings


Abstract: We report research activities in the Sony Four Legged Robot League at RoboCup 2002. 19 teams including 3 new teams participated in the league. We revised some rules and setup specifications such as a larger field and 4 robots for each team. In addition wireless LAN system were employed for inter-robot communication. These revisions encouraged the participants to develop team play behaviors.
Fukase, T., Kobayashi, Y., Ueda, R., Kawabe, T. & Arai, T.

Real-Time Decision Making Under Uncertainty of Self-Localization Results

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 375-383 inproceedings


Abstract: In this paper, we present a real-time decision making method for a quadruped robot whose sensor and locomotion have large errors. We make a State-Action Map by off-line planning considering the uncertainty of the robot’s location with Dynamic Programming (DP). Using this map, the robot can immediately decide optimal action that minimizes the time to reach a target state at any state. The number of observation is also minimized. We compress this map for implementation with Vector Quantization (VQ). Using the differences of the values between the optimal action and others as distortion measure of VQ minimizes the total loss of optimality.
Golubovic, D. & Hu, H.

An Interactive Software Environment for Gait Generation and Control Design of Sony Legged Robots

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 279-287 inproceedings


Abstract: This paper presents a modular approach to the development of an interactive software environment for gait generation and control design of Sony legged robots. A number of modules have been developed for monitoring robot states, gait generation, control design and image processing. A dynamic model of the leg and wheel-like motion are proposed to combine both wheeled and legged properties to produce smooth quadruped motion and high flexibility. Experimental results are presented to show the feasibility of the system
Hanek, R., Schmitt, T., Buck, S. & Beetz, M.

Towards RoboCup Without Color Labeling

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 179-194 inproceedings


Abstract: Object recognition and localization methods in RoboCup work on color segmented camera images. Unfortunately, color labeling can be applied to object recognition tasks only in very restricted environments, where different kinds of objects have different colors. To overcome these limitations we propose an algorithm named the Contracting Curve Density (CCD) algorithm for fitting parametric curves to image data. The method neither assumes object specific color distributions, nor specific edge profiles, nor does it need threshold parameters. Hence, no training phase is needed. In order to separate adjacent regions we use local criteria which are based on local image statistics. We apply the method to the problem of localizing the ball and show that the CCD algorithm reliably localizes the ball even in the presence of heavily changing illumination, strong clutter, specularity, partial occlusion, and texture.
Hardt, M. & von Stryk, O.

The Role of Motion Dynamics in the Design, Control and Stability of Bipedal and Quadrupedal Robots, The

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 206-223 inproceedings


Abstract: Fundamental principles and recent methods for investigating the nonlinear dynamics of legged robot motions with respect to control, stability and design are discussed. One of them is the still challenging problem of producing dynamically stable gaits. The generation of fast walking or running motions require methods and algorithms adept at handling the nonlinear dynamical effects and stability issues which arise. Reduced, recursive multibody algorithms, a numerical optimal control package, and new stability and energy performance indices are presented which are well-suited for this purpose. Difficulties and open problems are discussed along with numerical investigations into the proposed gait generation scheme. Our analysis considers both biped and quadrupedal gaits with particular reference to the problems arising in soccer-playing tasks encountered at the RoboCup where our team, the Darmstadt Dribbling Dackels, participates as part of the German Team in the Sony Legged Robot League.
Hibino, S., Kodama, Y., Nagasaka, Y., Takahashi, T., Murakami, K. & Naruse, T.

Fast Image Processing and Flexible Path Generation System for RoboCup Small Size League

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 53-64 inproceedings


Abstract: Two key features of successful multi-robot systems in RoboCup are robustness of a vision system and optimization of feedback control in order to reach the goal point and generate the action under the team strategy. This paper proposes two new methods. One is a fast image processing method, which is coped with the spatial variance of color parameters in the field, to extract the positions of robots and ball in 1/30 sec. The separation problem in the interlaced format image is solved. Another one is a path generation method in which the robot approaches the goal by changing its direction convergently. With these two algorithms, the real time processing system is realized by generating a stable path under a low quality input image.
Kaminka, G.A., Fidanboylu, M., Chang, A. & Veloso, M.M.

Learning the Sequential Coordinated Behavior of Teams from Observations

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 111-125 inproceedings


Abstract: The area of agent modeling deals with the task of observing other agents and modeling their behavior, in order to predict their future behavior, coordinate with them, assist them, or counter their actions. Typically, agent modeling techniques assume the availability of a plan- or behavior-library, which encodes the full repertoire of expected observed behavior. However, recent applications areas of agent modeling raise challenges to the assumption of such a library, as agent modeling systems are increasingly used in open and/or adversarial settings, where the behavioral repertoire of the observed agents is unknown at design time. This paper focuses on the challenge of the unsupervised autonomous learning of the sequential behaviors of agents, from observations of their behavior. The techniques we present translate observations of the dynamic, complex, continuous multi-variate world state into a time-series of recognized atomic behaviors. This time-series is then analyzed to find repeating subsequences characterizing each team. We compare two alternative approaches to extracting such characteristic sequences, based on frequency counts and statistical dependencies. Our results indicate that both techniques are able to extract meaningful sequences, and do significantly better than random predictions. However, the statistical dependency approach is able to correctly reject sequences that are frequent, but are due to random co-occurrence of behaviors, rather than to a true sequential dependency between them.
Kleiner, A., Dietl, M. & Nebel, B.

Towards a Life-Long Learning Soccer Agent

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 126-134 inproceedings


Abstract: One problem in robotic soccer (and in robotics in general) is to adapt skills and the overall behavior to a changing environment and to hardware improvements. We applied hierarchical reinforcement learning in an SMDP framework learning on all levels simultaneously. As our experiments show, learning simultaneously on the skill level and on the skill selection level is advantageous since it allows for a smooth adaption to a changing environment. Furthermore, the skills we trained turn also out to be quite competitive when run on the real robotic players of the players of our CS Freiburg team.
This work has been partially supported by Deutsche Forschungsgemeinschaft (DFG) and by SICK AG.
Kok, J., de Boer, R., Vlassis, N. & Groen, F.C.A.

Towards an Optimal Scoring Policy for Simulated Soccer Agents

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 296-303 inproceedings


Abstract: This paper describes the scoring policy which is used by the agents of the UvA Trilearn simulation team. In a given situation this policy enables an agent to determine the best shooting point in the goal, together with an associated probability of scoring when the ball is shot to this point. Our policy is based on an approximate method for learning the relevant statistics of the ball motion which can be regarded as a geometrically constrained continuous-time Markov process.
Meyer, J., Adolph, R., Stephan, D., Daniel, A., Seekamp, M., Weinert, V. & Visser, U.

Decision-Making and Tactical Behavior with Potential Fields

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 304-311 inproceedings


Abstract: Using potential-fields is a seldomly used technique in RoboCup scenarios. The existing approaches mainly concentrate on world state representation on single actions such as a kick. In this paper we will show how to apply potential fields to assist fast and precise decisions in an easy and intuitive way. We go beyond the existing approaches in using potential fields to determine all possible player actions, basic and advanced tactics an also general player behaviors. To ensure fast computing we mainly use basic mathematical computation for potential field related calculations. This gives us the advantage of both determining and understanding player actions. Therefore, integrating future features such as a complex online coach and progressive localization methods will be easier. We implemented the approach in our team Bremen University Goal Seekers (BUGS) and tested it in numerous games against other simulation league teams. The results show that CPU-time of making a decision per team has been decreased significantly. This is a crucial improvement for calculations in time-critical environments.
Montgomery, J.D. & Mackworth, A.K.

Adaptive Synchronisation for a RoboCup Agent

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 135-149 inproceedings


Abstract: We describe an algorithm that adaptively synchronises an agent with its environment enabling maximal deliberation time and improved action success rates. The method balances its reliance upon noisy evidence with internal representations, making it robust to interaction faults caused by both communication and timing. The notion of action correctness is developed and used to analyse the new method as well as two special cases: Internal and External synchronisation. Action correctness is determined online by a novel action accounting procedure that determines the outcome of commanded actions. In conjunction, these elements provide online analysis of agent activity, action confirmation for model prediction, and a coarse measure of the agent’s coherence with the environment that is used to adapt its performance.
Nair, R., Tambe, M. & Marsella, S.

Team Formation for Reformation in Multiagent Domains Like RoboCupRescue

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 150-161 inproceedings


Abstract: Team formation, i.e., allocating agents to roles within a team or subteams of a team, and the reorganization of a team upon team member failure or arrival of new tasks are critical aspects of teamwork. They are very important issues in RoboCupRescue where many tasks need to be done jointly. While empirical comparisons (e.g., in a competition setting as in RoboCup) are useful, we need a quantitative analysis beyond the competition — to understand the strengths and limitations of different approaches, and their tradeoffs as we scale up the domain or change domain properties. To this end, we need to provide complexity-optimality tradeoffs, which have been lacking not only in RoboCup but in the multiagent field in general.
To alleviate these difficulties, this paper presents R-COM-MTDP, a formal model based on decentralized communicating POMDPs, where agents explicitly take on and change roles to (re)form teams. R-COM-MTDP significantly extends an earlier COM-MTDP model, by introducing roles and local states to better model domains like RoboCupRescue where agents can take on different roles and each agent has a local state consisting of the objects in its vicinity. R-COM-MTDP tells us where the problem is highly intractable (NEXP-complete) and where it can be tractable (P-complete), and thus understand where algorithms may need to tradeoff optimality and where they could strive for near optimal behaviors. R-COM-MTDP model could enable comparison of various team formation and reformation strategies — including the strategies used by our own teams that came in the top three in 2001 — in the RoboCup Rescue domain and beyond.
Noda, I.

Hidden Markov Modeling of Multi-Agent Systems and Its Learning Method

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 94-110 inproceedings


Abstract: Two frameworks of hidden Markov modeling for multi-agent systems and its learning procedure are proposed. Although a couple of variations of HMMs have been proposed to model agents and their interactions, these models have not handled changes of environments, so that it is hard to simulate behaviors of agents that act in dynamic environments like soccer. The proposed frameworks enables HMMs to represent environments directly inside of state transitions. I first propose a model that handles the dynamics of the environments in the same state transition of the agent itself. In this model, the derived learning procedure can segment the environments according to the tasks and behaviors the agent is performing. I also investigate a more structured model in which the dynamics of the environments and agents are treated as separated state transitions and coupled each other. For this model, in order to reduce the number of parameters, I introduce “symmetricity” among agents. Furthermore, I discuss relation between reducing dependency in transitions and assumption of cooperative behaviors among multiple agents.
Obst, O.

Simulation League - League Summary

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 443-452 inproceedings


Abstract: In the simulation league the RoboCup soccer server provides a standard platform for simulated soccer teams to play against each other over a local network. Each team connects 11 player programs and possibly a coach client to the server, which simulates the 2D soccer field and distributes the sensory information to the clients. Besides the team clients the RoboCup soccer monitor or other visualization and debug tools can be connected as a client to the server to provide 2D or 3D visual information or information like game statistics and analysis for the spectators.
Ohta, M.

Direct Reward and Indirect Reward in Multi-Agent Reinforcement Learning

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 359-366 inproceedings


Abstract: When we apply reinforcement learning onto multi-agent environment, credit assignment problem will occur, because it is sometimes difficult to define which agents are the real contributors. If we praise all agents, when a group of cooperative agents get reward, some agents which did not contribute it will also reinforce their policies. On the other hand, if we praise obvious contributors only, indirect contribution will not be reinforced. For the first step to reduce this dilemma, we propose a classification of reward, and then investigate the feature of it. We treat a positioning task on SoccerServer for the experiments. The empirical results show that direct reward takes effect faster and helps obtaining individuality. On the contrary, indirect reward takes effect slower, but agents tend to form a group and obtain another effective positioning.
Okhotsimsky, D.E., Pavlovsky, V.E., Touganov, A.N., Plakhov, A.G., Pavlovsky, V.V., Stepanov, S.S. & Zaslavsky, A.Y.

Robosoccer-RU Open Simulation League: Principles and Algorithms

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 271-278 inproceedings


Abstract: In this paper we outline basic principles and competition rules used in a Computer soccer Simulation contest held in Eastern European countries (Robosoccer-RU League). The programming environment of this tournament (“Virtual Soccer” Software package) is described, as well as base algorithms that are implemented for powering team agents. A comparison is given between the reviewed approach and the one used in the RoboCup Simulation League, and directions for future convergence are drafted.
Prokopenko, M. & Wang, P.

Relating the Entropy of Joint Beliefs to Multi-Agent Coordination

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 367-374 inproceedings


Abstract: Current approaches to activity coordination in multi-agent systems (teams) range from strictly top down (plan-based coordination) to purely emergent (reactive coordination), with many hybrid variants, each having its specific advantages and disadvantages. It appears to be extremely difficult to rigorously compare various hybrid approaches to multi-agent coordination (and communication), given the lack of a generic semantics or some guidelines. In this paper, we studied some intuitive inter-agent communication policies and characterised them in terms of generic information-theoretic properties. In particular, the relative entropy of joint beliefs was suggested as an indicator of teams coordination potential. Our novel behaviour-based agent architecture (based on the Deep Behaviour Projection framework) enabled consistent reasoning about belief change, including beliefs about other agents. This allowed us to examine some of the identified communication policies empirically. The obtained results confirmed that there are certain interesting invariants – in particular, a change in team coordination (and overall performance) was shown to be within the boundaries indicated by the relative information entropy.
Restelli, M., Sorrenti, D.G. & Marchese, F.M.

MUREA: A MUlti-Resolution Evidence Accumulation Method for Robot Localization in Known Environments

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 351-358 inproceedings


Abstract: We present MUREA (MUlti-Resolution Evidence Accumulation): a mobile robot localization method for known 2D environments. It is an evidence accumulation method where the complexity is reduced by means of a multi-resolution scheme. The added value of the contribution, in the authors opinion, are 1) the method per sé; 2) the capability of the system to accept both raw sensor data as well as independently generated localization estimates; 3) the capability of the system to give out a (less) accurate estimate whenever asked to do so (e.g. before its regular completion), which could be called any-time localization.
Riley, P.

MPADES: Middleware for Parallel Agent Discrete Event Simulation

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 162-178 inproceedings


Abstract: Simulations are an excellent tool for studying artificial intelligence. However, the simulation technology in use by and designed for the artificial intelligence community often fails to take advantage of much of the work by the larger simulation community to produce stable, repeatable, and efficient simulations. We present the new system Middleware for Parallel Agent Discrete Event Simulation (MPADES) as a simulation substrate for the artificial intelligence community. MPADES focuses on the agent as a fundamental simulation component. The “thinking time” of an agent is tracked and reflected in the results of the agents’ actions. MPADES supports and manages the distribution of agents across machines while being robust to variations in network performance and machine load. We present the system in detail and give experimental results for a simple world model and set of agents. MPADES is not tied to any particular simulation, and is a powerful new tool for creating simulations for the study of artificial intelligence.
Röfer, T.

An Architecture for a National RoboCup Team

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 417-425 inproceedings


Abstract: This paper describes the architecture used by the GermanTeam 2002 in the Sony Legged Robot League. It focuses on the special needs of a national team, i.e. a “team of teams” from different universities in one country that compete against each other in national contests, but that will jointly line up at the international RoboCup championship. In addition, the tools developed by the GermanTeam will be presented, e.g. the first 3-D simulation used in the Sony Legged Robot League.
Schmitt, T., Hanek, R., Buck, S. & Beetz, M.

Probabilistic Vision-Based Opponent Tracking in Robot Soccer

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 426-434 inproceedings


Abstract: Good soccer players must keep their eyes on their opponents in order to make the right plays and moves. The same holds for soccer robots, too. In this paper, we apply probabilistic multiple object tracking to the continual estimation of the positions of opponent players in autonomous robot soccer. We extend MHT [3], an existing tracking algorithm, to handle multiple mobile sensors with uncertain positions, discuss the specification of probabilistic models needed by the algorithm, and describe the required vision-interpretation algorithms. The tracking algorithm enables robots to estimate the positions and motions of fast moving robots both accurately and robustly. We have applied the multiple object tracking algorithm throughout the RoboCup 2001 world championship. Empirical results show the applicability of multiple hypotheses tracking to vision-based opponent tracking and demonstrates the advantages for crowded environments.
Sekimori, D., Usui, T., Masutani, Y. & Miyasaki, F.

Evaluation of Self-Localization Performance for a Local Vision Robot in the Small Size League

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 41-52 inproceedings


Abstract: The main purpose of this paper is to examine the upper limits of self-localization ability using a local-vision system in the RoboCup Small Size League. Using an omni-directional vision system on a mobile robot, we originally developed a self-localization method based on imaging of the floor region. However, we could not explore its full potential because of a quantization error in the images. Therefore, we developed a self-localization method that allowed for better estimates of values than by individual methods, by integrating omni-directional vision and dead reckoning with the Kalman filter. This paper describes the algorithms and experimental results with an actual robot. In addition, we examine error recovery when the robot position changes due to collision with other objects and human intervention in the RoboCup competition.
Sklar, E.

RoboCupJunior 2002: The State of the League

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 489-495 inproceedings


Abstract: The RoboCupJunior division of RoboCup has just completed its third year of international participation and is growing rapidly in size and popularity. This paper describes the state of the league and looks closely at three components: participants, challenge events and educational value. We discuss the technical and educational progress of the league, identify problems and outline plans for future directions.
Sklar, E., Eguchi, A. & Johnson, J.

RoboCupJunior: Learning with Educational Robotics

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 238-253 inproceedings


Abstract: The RoboCupJunior division of RoboCup is now entering its third year of international participation and is growing rapidly in size and popularity. This paper first outlines the history of the Junior league, since it was demonstrated in Paris at RoboCup 1998, and describes how it has evolved into the international sensation it is today. While the popularity of the event is self-evident, we are working to identify and quantify the educational benefits of the initiative. The remainder of the paper focuses on describing our efforts to encapsulate these qualities, highlighting results from a pilot study conducted at RoboCupJunior 2000 and presenting new data from a subsequent study of RoboCupJunior 2001.
Spaan, M.T.J. & Groen, F.C.A.

Team Coordination Among Robotic Soccer Players

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 409-416 inproceedings


Abstract: We present an approach for coordinating a team of soccer playing robots, used by Clockwork Orange in the RoboCup middle-size league. It is based on the idea of dynamically distributing roles among the team members and adds the notion of a global team strategy (attack, defend and intercept). Utility functions are used for estimating how well suited a robot is for a certain role. They are not only based on the time the robot expects to need to reach the ball but also on the robot’s position in the field. Empirical results from the RoboCup 2001 tournament are presented demonstrating the value of extending role distribution with a team strategy.
Stone, P.

Multiagent Competitions and Research: Lessons from RoboCup and TAC

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 224-237 inproceedings


Abstract: This paper compares and contrasts two recent series of competitions in which multiple agents compete directly against one another: the robot soccer world cup (RoboCup) and the trading agent competition (TAC). Both of these competitions have attracted large numbers of competitors and have motivated important research results in artificial intelligence. Based on extensive personal experiences, both as a participant and as an organizer, this paper reflects upon and characterizes both the benefits and the hazards of competitions with respect to academic research.
Stroupe, A.W., Sikorski, K. & Balch, T.

Constraint-Based Landmark Localization

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 8-24 inproceedings


Abstract: We present an approach to the landmark-based robot localization problem for environments, such as RoboCup middle-size soccer, that provide limited or low-quality information for localization. This approach allows use of different types of measurements on potential landmarks in order to increase landmark availability. Some sensors or landmarks might provide only range (such as field walls) or only bearing measurements (such as goals). The approach makes use of inexpensive sensors (color vision) using fast, simple updates robust to low landmark visibility and high noise. This localization method has been demonstrated in laboratory experiments and RoboCup 2001. Experimental analysis of the relative benefits of the approach is provided.
Tadokoro, S.

RoboCupRescue Robot League

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 482-484 inproceedings


Abstract: Ten teams from five countries participated in the RoboCupRescue RobotLeague in 2002 as shown in Table 1. Most robots are remotely teleoperated and have limited autonomy. Because of the complexity of the problem, fully autonomous robots cannot be practical. Adjusted autonomy, shared autonomy, and autonomy for human interface are suitable to apply AI to the real disaster problems.
Takahashi, T.

RoboCupRescue Simulation League

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 477-481 inproceedings


Abstract: This paper overviews all results of RoboCupRescue simulation league at 2002.
RoboCupRescue simulation has a lot in common with RoboCupSoccer simulation. It handles distributed, multiagent domains and agents do their tasks with limited communication and sensing abilities. The distinctions between rescue and soccer are scales of domain, multiple hierarchies in agents and interactions with various disaster simulations [1]. The agents are firefighters, police workers, ambulance workers and their control centers.
Takahashi, Y., Edazawa, K. & Asada, M.

Behavior Acquisition Based on Multi-Module Learning System in Multi-Agent Environment

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 435-442 inproceedings


Abstract: The conventional reinforcement learning approaches have difficulties to handle the policy alternation of the opponents because it may cause dynamic changes of state transition probabilities of which stability is necessary for the learning to converge. This paper presents a method of multi-module reinforcement learning in a multiagent environment, by which the learning agent can adapt itself to the policy changes of the opponents. We show a preliminary result of a simple soccer situation in the context of RoboCup.
Tuyls, K., Maes, S. & Manderick, B.

Reinforcement Learning in Large State Spaces

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 319-326 inproceedings


Abstract: Large state spaces and incomplete information are two problems that stand out in learning in multi-agent systems. In this paper we tackle them both by using a combination of decision trees and Bayesian networks (BNs) to model the environment and the Q-function. Simulated robotic soccer is used as a testbed, since there agents are faced with both large state spaces and incomplete information. The long-term goal of this research is to define generic techniques that allow agents to learn in large-scaled multi-agent systems.
Utz, H., Neubeck, A., Mayer, G. & Kraetzschmar, G.

Improving Vision-Based Self-Localization

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 25-40 inproceedings


Abstract: After removing the walls around the field, vision-based localization has become an even more interesting approach for robotic soccer. The paper discusses how removal of the wall affects the localization task in RoboCup, both for vision-based and non-visual approaches, and argues that vision-based Monte Carlo localization based on landmark features seems to cope well with the changed field setup. An innovative approach for landmark feature detection for vision-based Monte Carlo Localization is presented. Experimental results indicate that the approach is robust and reliable.
Visser, U. & Weland, H.-G.

Using Online Learning to Analyze the Opponent’s Behavior

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 78-93 inproceedings


Abstract: Analyzing opponent teams has been established within the simulation league for a number of years. However, most of the analyzing methods are only available off-line. Last year we introduced a new idea which uses a time series-based decision tree induction to generate rules on-line. This paper follows that idea and introduces the approach in detail. We implemented this approach as a library function and are therefore able to use on-line coaches of various teams in order to test the method. The tests are based on two ‘models’: (a) the behavior of a goalkeeper, and (b) the pass behavior of the opponent players. The approach generates propositional rules (first rules after 1000 cycles) which have to be pruned and interpreted in order to use this new knowledge for one’s own team. We discuss the outcome of the tests in detail and conclude that on-line learning despite of the lack of time is not only possible but can become an effective method for one’s own team.
Weigel, T. & Nebel, B.

KiRo - An Autonomous Table Soccer Player

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 384-392 inproceedings


Abstract: This paper presents KiRo – a system capable of playing table soccer on a competitive level and in a fully autonomous way. It can serve a human both as a teammate and an opponent but also allows for matches between two artificial players. KiRo introduces the table soccer game as a new domain for the research in the fields of robotics and artificial intelligence.
Yoshimura, K., Barnes, N., Rönnquist, R. & Sonenberg, L.

Towards Real-Time Strategic Teamwork: A RoboCup Case Study

2003 RoboCup 2002: Robot Soccer World Cup VI, pp. 342-350 inproceedings


Abstract: RooBots competed in the F180 League of the RoboCup 2001 competition in Seattle, USA. In this article, we present an architectural overview of our system involving an integration of an agent-oriented programming framework to support strategic decisions, with various low-level perception and control elements. Our AI Module includes a novel mechanism to facilitate dynamic formation change by an individual agent and we report a preliminary evaluation of the approach drawn from performance in the 2001 competition.