RoboCup 2007 Publications

From RoboCup Wiki
Jump to: navigation, search
Author Title Year Journal/Proceedings Reftype DOI/URL
Abbott, R.G.

Behavioral Cloning for Simulator Validation

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 329-336 inproceedings


Abstract: Behavioral cloning is an established technique for creating agent behaviors by replicating patterns of behavior observed in humans or other agents. For pragmatic reasons, behavioral cloning has usually been implemented and tested in simulation environments using a single nonexpert subject. In this paper, we capture behaviors for a team of subject matter experts engaged in real competition (a soccer tournament) rather than participating in a study. From this data set, we create software agents that clone the observed human tactics. We place the agents in a simulation to determine whether increased behavioral realism results in higher performance within the simulation and argue that the transferability of real-world tactics is an important metric for simulator validation. Other applications for validated agents include automated agent behavior, factor analysis for team performance, and evaluation of real team tactics in hypothetical scenarios such as fantasy tournaments.
Abdi, M.J., Analoui, M., Aghabeigi, B., Rafiee, E. & Tabatabaee, S.M.S.

Evolutionary Design of a Fuzzy Rule Base for Solving the Goal-Shooting Problem in the RoboCup 3D Soccer Simulation League

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 320-328 inproceedings


Abstract: Most of the problems in the RoboCup soccer domain suffer from the noisy perceptions, noisy actions, and continuous state space. To cope with these problems, using Fuzzy logic can be a proper choice, due to its capabilities of inferring and approximate reasoning under uncertainty. However, designing the entire rule base of a Fuzzy rule base system (FRBS) by an expert is a boring and time consuming task and sometimes the performance of the designed Fuzzy system is far from the optimum, especially in cases that the available knowledge of the system is not enough. In this paper, a rule learning method based on the iterative rule learning (IRL) approach is proposed to generate the entire rule base of an FRBS with the help of genetic algorithms (GAs). The advantage of our proposed method compared to similar approaches in the literature is that our algorithm does not need any training set, which is difficult to collect in many cases; cases like most of the problems existing in the RoboCup soccer domain. As a test case, the goal-shooting problem in the RoboCup 3D soccer simulation league is chosen to be solved using this approach. Simulation tests reveal that with applying the rule learning method proposed in this paper on the goal-shooting problem, not only can a rule base with good performance in goal-shooting skill be obtained, but also the number of rules in the rule base can be decreased by using the general rules in constructing the rule base.
Aghazadeh, O., Sharbafi, M.A. & Haghighat, A.T.

Implementing Parametric Reinforcement Learning in Robocup Rescue Simulation

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 409-416 inproceedings


Abstract: Decision making in complex, multi agent and dynamic environments such as Rescue Simulation is a challenging problem in Artificial Intelligence. Uncertainty, noisy input data and stochastic behavior which is a common difficulty of real time environment makes decision making more complicated in such environments. Our approach to solve the bottleneck of dynamicity and variety of conditions in such situations is reinforcement learning. Classic reinforcement learning methods usually work with state and action value functions and temporal difference updates. Using function approximation is an alternative method to hold state and action value functions directly. Many Reinforcement learning methods in continuous action and state spaces implement function approximation and TD updates such as TD, LSTD, iLSTD, etc. A new approach to online reinforcement learning in continuous action or state spaces is presented in this paper which doesn’t work with TD updates. We have named it Parametric Reinforcement Learning. This method is utilized in Robocup Rescue Simulation / Police Force agent’s decision making process and the perfect results of this utilization have been shown in this paper. Our simulation results show that this method increases the speed of learning and simplicity of use. It has also very low memory usage and very low costing computation time.
Ahmadi, M. & Stone, P.

Instance-Based Action Models for Fast Action Planning

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 1-16 inproceedings


Abstract: Two main challenges of robot action planning in real domains are uncertain action effects and dynamic environments. In this paper, an instance-based action model is learned empirically by robots trying actions in the environment. Modeling the action planning problem as a Markov decision process, the action model is used to build the transition function. In static environments, standard value iteration techniques are used for computing the optimal policy. In dynamic environments, an algorithm is proposed for fast replanning, which updates a subset of state-action values computed for the static environment. As a test-bed, the goal scoring task in the RoboCup 4-legged scenario is used. The algorithms are validated in the problem of planning kicks for scoring goals in the presence of opponent robots. The experimental results both in simulation and on real robots show that the instance-based action model boosts performance over using parametric models as done previously, and also incremental replanning significantly improves over original off-line planning.
Akiyama, H. & Noda, I.

Multi-agent Positioning Mechanism in the Dynamic Environment

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 377-384 inproceedings


Abstract: In this paper, we propose a novel agent positioning mechanism for the dynamic environments. In many problems of the real-world multi-agent/robot domain, a position of each agent is an important factor to affect agents’ performance. Because the real-world problem is generally dynamic, a suitable positions for each agent should be determined according to the current status of the environment. We formalize this issue as a map from a focal point like a ball position in a soccer field to a desirable positioning of each player agent, and propose a method to approximate the map using Delaunay Triangulation. This method is simple, fast and accurate, so that it can be implemented for real-time and scalable problems like RoboCup Soccer. The performance of the method is evaluated in RoboCup Soccer Simulation environment compared with other function approximation method like Normalized Gaussian Network. The result of the evaluation tells us that the proposal method is robust to uneven sample distribution so that we can easily to maintain the mapping.
Arenas, M., del Solar, J.R. & Verschae, R.

Detection of AIBO and Humanoid Robots Using Cascades of Boosted Classifiers

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 449-456 inproceedings


Abstract: In the present article a framework for the robust detection of mobile robots using nested cascades of boosted classifiers is proposed. The boosted classifiers are trained using Adaboost and domain-partitioning weak hypothesis. The most interesting aspect of this framework is its capability of building robot detection systems with high accuracy in dynamical environments (RoboCup scenario), which achieve, at the same time, high processing and training speed. Using the proposed framework we have built robust AIBO and humanoid robot detectors, which are analyzed and evaluated using real-world video sequences. This research was partially supported by FONDECYT (Chile) under Project Number 1061158.
Atkinson, J. & Castro, C.

Crossed-Line Segmentation for Low-Level Vision

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 472-479 inproceedings


Abstract: This work describes a new segmentation method for robotic soccer applications. The approach called crossed-line segmentation is based on the combination of region classification and a border detector which meet homogeneity criteria of medians. Experiments suggest that the method outperforms traditional procedure in terms of smoothing and segmentation accuracy. Furthermore, existing noise in the images is also observed to be reduced without missing the objects’ borders.
Atkinson, J. & Rojas, D.

Generating Dynamic Formation Strategies Based on Human Experience and Game Conditions

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 159-170 inproceedings


Abstract: In this paper, a new approach to automatically generating game strategies based on the game conditions is presented. A game policy is defined and applied by a human coach who establishes the attitude of the team for defending or attacking. A simple neural net model is applied using current and previous game experience to classify the game’s parameters so that the new game conditions can be determined so that a robotic team can modify its strategy on the fly. Results of the implemented model for a robotic soccer team are discussed.
Beck, D., Ferrein, A. & Lakemeyer, G.

A Simulation Environment for Middle-Size Robots with Multi-level Abstraction

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 136-147 inproceedings


Abstract: Larger fields in the Middle-size league as well as the effort to build mixed teams from different universities require a simulation environment which is capable to physically correctly simulate the robots and the environment. A standardized simulation environment has not yet been proposed for this league. In this paper we present our simulation environment, which is based on the Gazebo system. We show how typical Middle-size robots with features like omni-drives and omni-directional cameras can be modeled with relative ease. In particular, the control software for the real robots can be used with few changes, thus facilitating the transfer of results obtained in simulation back to the robots. We address some technical issues such as adapting time-triggered events in the robot control software to the simulation, and we introduce the concept of multi-level abstractions. The latter allows switching between faithful but computionally expensive sensor models and abstract but cheap approximations. These abstractions are needed especially when simulating whole teams of robots. This work was partially supported by the German Science Foundation (DFG) in the Priority Program 1125, Cooperating Teams of Mobile Robots in Dynamic Environments and by the NRW Ministry of Education and Research (MSWF). Further support by the Bonn-Aachen International Center for Information Technology (B-IT) is gratefully acknowledged.
Celiberto, L.A., Ribeiro, C.H.C., Costa, A.H.R. & Bianchi, R.A.C.

Heuristic Reinforcement Learning Applied to RoboCup Simulation Agents

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 220-227 inproceedings


Abstract: This paper describes the design and implementation of robotic agents for the RoboCup Simulation 2D category that learns using a recently proposed Heuristic Reinforcement Learning algorithm, the Heuristically Accelerated Q–Learning (HAQL). This algorithm allows the use of heuristics to speed up the well-known Reinforcement Learning algorithm Q–Learning. A heuristic function that influences the choice of the actions characterizes the HAQL algorithm. A set of empirical evaluations was conducted in the RoboCup 2D Simulator, and experimental results show that even very simple heuristics enhances significantly the performance of the agents.
Cesa, S.L., Farinelli, A., Iocchi, L., Nardi, D., Sbarigia, M. & Zaratti, M.

Semi-autonomous Coordinated Exploration in Rescue Scenarios

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 286-293 inproceedings


Abstract: In this paper we study different coordination strategies for a group of robots involved in a search and rescue task. The system integrates all the necessary components to realise the basic behaviours of robotic platforms. Coordination is based on iterative dynamic task assignment. Tasks are interesting points to reach, and the coordination algorithm finds at each time step the optimal assignment of robots to tasks. We realised both a completely autonomous exploration strategy and a strategy that involves a human operator. The human operator is able to control the robots at different levels: giving priority points for exploration to the team of robots, giving navigation goal points to team of robots, and directly tele-operating a single robot. For building a consistent global map, we implemented a centralised coordinated SLAM approach that integrates readings from all robots. The system has been tested both in the UsarSim simulation environment and on robotic platforms.
Cherubini, A., Giannone, F. & Iocchi, L.

Layered Learning for a Soccer Legged Robot Helped with a 3D Simulator

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 385-392 inproceedings


Abstract: Mobile robots can benefit from machine learning approaches for improving their behaviors in performing complex activities. In recent years, these techniques have been used to find optimal parameter sets for many behaviors. In particular, layered learning has been proposed to improve learning rate in robot learning tasks. In this paper, we consider a layered learning approach for learning optimal parameters of basic control routines, behaviours and strategy selection. We compare three different methods in the different layers: genetic algorithm, Nelder-Mead, and policy gradient. Moreover, we study how to use a 3D simulator for speeding up robot learning. The results of our experimental work on AIBO robots are useful not only to state differences and similarities between different robot learning approaches used within the layered learning framework, but also to evaluate a more effective learning methodology that makes use of a simulator.
Chonnaparamutt, W. & Birk, A.

A Fuzzy Controller for Autonomous Negotiation of Stairs by a Mobile Robot with Adjustable Tracks

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 196-207 inproceedings


Abstract: Tracked mobile robots with adjustable support tracks or flippers are popular promising solutions for negotiating rough terrain and 3D obstacles. Though many according robot bases are in principle physically capable of climbing stairs, it is a non-trivial control-task for a remote tele-operator, especially when the user can not directly see the robot like in search and rescue scenarios. To limit training requirements and to ease the cognitive load on operators, respectively to enable fully autonomous rescue robots, we developed a fuzzy controller for this task, which adjusts the drive forces and the posture of the flipper. The design of the controller is guided by observing the strategies of a trained user when tele-operating a robot with unlimited visual information. In doing so, an Open Dynamics Engine (ODE) simulation of our robot is used where the full set of all physical parameters is accessible for analysis. Based on this data, it is shown in several experiments that the controller is not only capable of climbing stairs but that it does so in a more efficient manner than the human user who served as training model.
Colombini, E.L., da Silva Simöes, A., Martins, A.C.G. & Matsuura, J.P.

A Framework for Learning in Humanoid Simulated Robots

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 345-352 inproceedings


Abstract: One of the most important characteristics of intelligent activity is the ability to change behaviour according to many forms of feedback. Through learning an agent can interact with its environment to improve its performance over time. However, most of the techniques known that involves learning are time expensive, i.e., once the agent is supposed to learn over time by experimentation, the task has to be executed many times. Hence, high fidelity simulators can save a lot of time. In this context, this paper describes the framework designed to allow a team of real RoboNova-I humanoids robots to be simulated under USARSim environment. Details about the complete process of modeling and programming the robot are given, as well as the learning methodology proposed to improve robot’s performance. Due to the use of a high fidelity model, the learning algorithms can be widely explored in simulation before adapted to real robots.
Cunha, B., Azevedo, J., Lau, N. & Almeida, L.

Obtaining the Inverse Distance Map from a Non-SVP Hyperbolic Catadioptric Robotic Vision System

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 417-424 inproceedings


Abstract: The use of single viewpoint catadioptric vision systems is a common approach in mobile robotics, despite the constraints imposed by those systems. A general solution to calculate the robot centered distances map on non-SVP catadioptric setups, exploring a back-propagation ray-tracing approach and the mathematical properties of the mirror surface is discussed in this paper. Results from this technique applied in the robots of the CAMBADA team (Cooperative Autonomous Mobile Robots with Advanced Distributed Architecture) are presented, showing the effectiveness of the solution.
Fathzadeh, R., Mokhtari, V. & Kangavari, M.R.

Opponent Provocation and Behavior Classification: A Machine Learning Approach

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 540-547 inproceedings


Abstract: Opponent Modeling is one of the most attractive and practical arenas in Multi Agent System (MAS) for predicting and identifying the future behaviors of opponent. This paper introduces a novel approach using rule based expert system towards opponent modeling in RoboCup Soccer Coach Simulation. In this scene, an autonomous coach agent is able to identify the patterns of the opponent by analyzing the opponent’s past games and advising own players. For this purpose, the main goal of our research comprises two complementary parts: (a) developing a 3-tier learning architecture for classifying opponent behaviors. To achieve this objective, sequential events of the game are identified using environmental data. Then the patterns of the opponent are predicted using statistical calculations. Eventually, by comparing the opponent patterns with the rest of team’s behavior, a model of the opponent is constructed. (b) designing a rule based expert system containing provocation strategies to expedite detection of opponent patterns. These items mentioned are used by coach, to model the opponent and generate an appropriate strategy to play against the opponent. This structure is tested in RoboCup Soccer Coach Simulation and MRLCoach was the champion at RoboCup 2006 in Germany.
Friedmann, M., Petersen, K. & von Stryk, O.

Tailored Real-Time Simulation for Teams of Humanoid Robots

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 425-432 inproceedings


Abstract: Developing and testing the key modules of autonomous humanoid robots (e.g., for vision, localization, and behavior control) in software-in-the-loop (SIL) experiments, requires real-time simulation of the main motion and sensing properties. These include humanoid robot kinematics and dynamics, the interaction with the environment, and sensor simulation. To deal with an increasing number of robots per team the simulation algorithms must be very efficient. In this paper, the simulator framework MuRoSimF (Multi-Robot-Simulation-Framework) is presented which allows the flexible and transparent integration of different simulation algorithms with the same robot model. These include several algorithms for simulation of humanoid robot motion kinematics and dynamics (with O(n) runtime complexity), collision handling, and camera simulation including lens distortion. A simulator for teams of humanoid robots based on MuRoSimF is presented. A unique feature of this simulator is the scalability of the level of detail and complexity which can be chosen individually for each simulated robot and tailored to the requirements of a specific SIL test. Performance measurements are given for real-time simulation on a moderate laptop computer of up to six humanoid robots with 21 degrees of freedom, each equipped with an articulated camera.
Furbach, U., Murray, J., Schmidsberger, F. & Stolzenburg, F.

Model Checking Hybrid Multiagent Systems for the RoboCup

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 262-269 inproceedings


Abstract: This paper shows how multiagent systems can be modeled by a combination of UML statecharts and hybrid automata. This allows formal system specification on different levels of abstraction on the one hand, and expressing real-time system behavior with continuous variables on the other hand. It is shown how multi-robot systems can be modeled by hybrid and hierarchical state machines and how model checking techniques for hybrid automata can be applied. An enhanced synchronization concept is introduced that allows synchronization taking time and avoids state explosion to a certain extent. This research is supported by the grants Fu 263/8 and Sto 421/2 from the German research foundation DFG within the special priority program 1125 on Cooperating Teams of Mobile Robots in Dynamic Environments.
Grasemann, U., Stronger, D. & Stone, P.

A Neural Network-Based Approach to Robot Motion Control

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 480-487 inproceedings


Abstract: The joint controllers used in robots like the Sony Aibo are designed for the task of moving the joints of the robot to a given position. However, they are not well suited to the problem of making a robot move through a desired trajectory at speeds close to the physical capabilities of the robot, and in many cases, they cannot be bypassed easily. In this paper, we propose an approach that models both the robot’s joints and its built-in controllers as a single system that is in turn controlled by a neural network. The neural network controls the entire trajectory of a robot instead of just its static position. We implement and evaluate our approach on a Sony Aibo ERS-7.
Guerrero, P. & del Solar, J.R.

Improving Robot Self-localization Using Landmarks’ Poses Tracking and Odometry Error Estimation

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 148-158 inproceedings


Abstract: In this article the classical self-localization approach is improved by estimating, independently from the robot’s pose, the robot’s odometric error and the landmarks’ poses. This allows using, in addition to fixed landmarks, dynamic landmarks such as temporally local objects (mobile objects) and spatially local objects (view-dependent objects or textures), for estimating the odometric error, and therefore improving the robot’s localization. Moreover, the estimation or tracking of the fixed-landmarks’ poses allows the robot to accomplish successfully certain tasks, even when having high uncertainty in its localization estimation (e.g. determining the goal position in a soccer environment without directly seeing the goal and with high localization uncertainty). Furthermore, the estimation of the fixed-landmarks’ pose allows having global measures of the robot’s localization accuracy, by comparing the real map, given by the real (a priori known) position of the fixed-landmarks, with the estimated map, given by the estimated position of these landmarks. Based on this new approach we propose an improved self-localization system for AIBO robots playing in a RoboCup soccer environment, where the odometric error estimation is implemented using Particle Filters, and the robot’s and landmarks’ poses are estimated using Extended Kalman Filters. Preliminary results of the system’s operation are presented. This research was partially supported by FONDECYT (Chile) under Project Number 1061158.
Guerrero, P., del Solar, J.R. & Díaz, G.

Probabilistic Decision Making in Robot Soccer

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 29-40 inproceedings


Abstract: Decision making is an important issue in robot soccer, which has not been investigated deeply enough by the RoboCup research community. This paper proposes a probabilistic approach to decision making. The proposed methodology is based on the maximization of a game situation score function, which generalizes the concept of accomplishing different game objectives as: passing, scoring a goal, clearing the ball, etc. The methodology includes a quantitative method for evaluating the game situation score. Experimental results in a high-level strategy simulator, which runs our four-legged code in simulated AIBOs’ robots, show a noticeable improvement in the scoring effectiveness achieved by a team that uses the proposed approach for making decisions. This research was partially supported by FONDECYT (Chile) under Project Number 1061158.
Guerrero, P., del Solar, J.R., Fredes, J. & Palma-Amestoy, R.

Automatic On-Line Color Calibration Using Class-Relative Color Spaces

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 246-253 inproceedings


Abstract: In this article we present an automatic on-line color calibration system that makes extensive use of the spatial relationships between color classes in the color space. First, we introduce the definition of class-relative color spaces, where classes are represented in terms of their spatial relation to a base color class. Then, using class-relative color spaces, the system is able to remap classes from the already trained ones, which gives a starting point for training the remaining classes. The color-calibrating system also uses a feedback from the detected objects using the remapped (or partially trained) classes. As a result, the system is able to generate a complete color look-up table from scratch, and to adapt quickly to severe lighting condition changes. A particularity of our system is that it does not need to solve the natural ambiguity in color classes’ intersections, but it is able to keep and use it during color segmentation using the concept of soft-colors. This research was partially supported by FONDECYT (Chile) under Project Number 1061158.
Göhring, D.

Cooperative Object Localization Using Line-Based Percept Communication

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 53-64 inproceedings


Abstract: In this paper we present a novel approach to estimate the position of objects tracked by a team of robots. Moving objects are commonly modeled in an egocentric frame of reference, because this is sufficient for most robot tasks as following an object, and it is independent of the robots localization within its environment. But for multiple robots, to communicate and to cooperate the robots have to agree on an allocentric frame of reference. Instead of transforming egocentric models into allocentric ones by using self localization information, we will show how relations between different objects within the same camera image can be used as a basis for estimating an object’s position. The spacial relation of objects with respect to stationary objects yields several advantages: a) Errors in feature detections are correlated. The error of relative positions of objects within a single camera frame is comparably small. b) The information is independent of robot localization and odometry. c) Object relations can help to detect inconsistent sensor data. We present experimental evidence that shows how two non-localized robots are capable to infer the position of an object by communication on a RoboCup Four-Legged soccer field.
Hayashi, Y. & Fujiyoshi, H.

Mean-Shift-Based Color Tracking in Illuminance Change

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 302-311 inproceedings


Abstract: The mean-shift algorithm is an efficient technique for tracking 2D blobs through an image. Although it is important to adapt the mean-shift kernel to handle changes in illumination for robot vision at outdoor site, there is presently no clean mechanism for doing this. This paper presents a novel approach for color tracking that is robust to illumination changes for robot vision. We use two interleaved mean-shift procedures to track the spatial location and illumination intensity of a blob in an image. We demonstrate that our method enables efficient real-time tracking of the multiple color blobs against changes in illumination, where the illuminace ranges from 58 to 1,300 lx.
He, H. & Chen, X.

A Model-Based Approach to Calculating and Calibrating the Odometry for Quadruped Robots

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 337-344 inproceedings


Abstract: This paper presents a model-based odometry calculation and calibration method (MBO) for quadruped robots. Instead of establishing the direct relation between target and actual speeds as previous methods did, MBO sets up a “parametric physical model” incorporating various properties of the robot and environment such as friction and inertia, through optimization with locomotor data. Based on this optimized model, one can compute the loci of robot legs’ movement by forward kinematics and finally obtain odometric readings by analyzing the loci. Experiments on Sony AIBO ERS-7 robots demonstrate that the odometry error of MBO is generally 50% less than the existing methods. In addition, the calibration complexity is low. This work is supported by the NSFC 60275024 and the 973 program 2003CB317000.
Hein, D., Hild, M. & Berger, R.

Evolution of Biped Walking Using Neural Oscillators and Physical Simulation

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 433-440 inproceedings


Abstract: Controlling a biped robot with a high degree of freedom to achieve stable movement patterns is still an open and complex problem, in particular within the RoboCup community. Thus, the development of control mechanisms for biped locomotion have become an important field of research. In this paper we introduce a model-free approach of biped motion generation, which specifies target angles for all driven joints and is based on a neural oscillator. It is potentially capable to control any servo motor driven biped robot, in particular those with a high degree of freedom, and requires only the identification of the robot’s physical constants in order to provide an adequate simulation. The approach was implemented and successfully tested within a physical simulation of our target system - the 19-DoF Bioloid robot. The crucial task of identifying and optimizing appropriate parameter sets for this method was tackled using evolutionary algorithms. We could show, that the presented approach is applicable in generating walking patterns for the simulated biped robot. The work demonstrates, how the important parameters may be identified and optimized when applying evolutionary algorithms. Several so evolved controllers were capable of generating a robust biped walking behavior with relatively high walking speeds, even without using sensory information. In addition we present first results of laboratory experiments, where some of the evolved motions were tried to transfer to real hardware.
Henderson, N., King, R. & Middleton, R.H.

An Application of Gaussian Mixtures: Colour Segmenting for the Four Legged League Using HSI Colour Space

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 254-261 inproceedings


Abstract: In the colour coded environment of the RoboCup 4 Legged League it is crucial to extract as much colour information as possible from an image without error. To do this requires hours of manual YUV pixel mapping and testing to ensure robustness under all possible lighting conditions. The YUV colour space is a very convenient standard for transmission of video data, but for colour classification and segmentation it suffers from being non-intuitive and sensitive to changes in lighting. Alternatively, colour classification principles can be applied in an HSI colour space; one of the convenient characteristics of the HSI colour space is that the hue value, H, represents the colour wavelength information. From this concept it is easier to separate and label colour regions in an automated process as the theoretical hue and colour wavelength relationship is known. By fitting a Gaussian model using mixtures to HSI histograms we can generate boundaries of colour classes in HSI colour space.
Jüngel, M., Mellmann, H. & Spranger, M.

Improving Vision-Based Distance Measurements Using Reference Objects

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 90-100 inproceedings


Abstract: Robots perceiving their environment using cameras usually need a good representation of how the camera is aligned to the body and how the camera is rotated relative to the ground. This is especially important for bearing-based distance measurements. In this paper we show how to use reference objects to improve vision-based distance measurements to objects of unknown size. Several methods for different kinds of reference objects are introduced. These are objects of known size (like a ball), objects extending over the horizon (like goals and beacons), and objects with known shape on the ground (like field lines). We give a detailed description how to determine the rotation of the robot’s camera relative to the ground, provide an error-estimation for all methods and describe the experiments we performed on an Aibo robot.
Jüngel, M. & Risler, M.

Self-localization Using Odometry and Horizontal Bearings to Landmarks

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 308-317 inproceedings


Abstract: On the way to the big goal - the game against the human world champion on a real soccer field - the configuration of the soccer fields in RoboCup has changed during the last years. There are two main modification trends: The fields get larger and the number of artificial landmarks around the fields decreases. The result is that a lot of the methods for self-localization developed during the last years do not work in the new scenarios without modifications. This holds especially for robots with a limited range of view as the probability for a robot to detect a landmark inside its viewing angle is significantly lower than on the old fields. On the other hand the robots have more space to play and do not collide as often as on the small fields. Thus the robots have a better idea of the courses they cover (odometry has higher reliability). This paper shows a method for self-localization that is based on bearings to horizontal landmarks and the knowledge about the robots movement between the observation of the features.
Kalyanakrishnan, S., Stone, P. & Liu, Y.

Model-Based Reinforcement Learning in a Complex Domain

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 171-183 inproceedings


Abstract: Reinforcement learning is a paradigm under which an agent seeks to improve its policy by making learning updates based on the experiences it gathers through interaction with the environment. Model-free algorithms perform updates solely bas ed on observed experiences. By contrast, model-based algorithms learn a model of the environment that effectively simulates its dynamics. The model may be used to simulate experiences or to plan into the future, potentially expediting the learning process. This paper presents a model-based reinforcement learning approach for Keepaway, a complex, continuous, stochastic, multiagent subtask of RoboCup simulated soccer. First, we propose the design of an environmental model that is partly learned based on the agent’s experiences. This model is then coupled with the reinforcement learning algorithm to learn an action selection policy. We evaluate our method through empirical comparisons with model-free approaches that have been previously applied successfully to this task. Results demonstrate significant gains in the learning speed and asymptotic performance of our method. We also show that the learned model can be used effectively as part of a planning-based approach with a hand-coded policy.
Kashanipour, A., Kashanipour, A.R., Milani, N.S., Akhlaghi, P. & Boukani, K.K.

Robust Color Classification Using Fuzzy Reasoning and Genetic Algorithms in RoboCup Soccer Leagues

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 548-555 inproceedings


Abstract: Color segmentation is typically the first step of vision processing for a robot operating in a color coded environment, like RoboCup soccer, and many object recognition modules rely on that, in this paper we present a method for color segmentation that is based on fuzzy logic. Fuzzy sets are defined on the H, S and L components of the HSL color space and provide a fuzzy logic model that aims to follow the human intuition of color classification. The membership functions used for the fuzzy inference are optimized by genetic algorithms. The method requires the setting of only a few parameters and has been proved to be very robust to noise and light variations, allowing for setting parameters only once. The approach has been implemented on MRL middle size robots, and successfully experimented in the numbers of the friendly matches of the Middle size in the 2006’s games.
Kolberg, E., Reich, Y. & Levin, I.

Design of Design Methodology for Autonomous Robots

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 528-539 inproceedings


Abstract: We present a methodology for deriving design methodology for autonomous robots. We designed this methodology in the context of a robotics course in high schools. The motivation for designing this new methodology was improving the robots’ robustness and reliability and preparing students for becoming better designers. The new methodology proved to be highly successful in designing top quality robots. In the methodology design, we explored and adapted design methods to the specific designers, the nature of the product, the environment, the product needs, and the design context goals. At the end of this comprehensive design, we selected a synergetic integration of six methods to compose the methodology for this product context: conceptual design, fault tolerant design, atomic requirements, using fuzzy logic for the control of robotics systems, creative thinking method, and microprogramming design.
Kyrylov, V. & Razykov, S.

Pareto-Optimal Offensive Player Positioning in Simulated Soccer

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 228-237 inproceedings


Abstract: The ability by the simulated soccer player to make rational decisions about moving without ball is a critical factor of success. Here we limit our scope to the offensive situation, i.e. when the ball is controlled by own team, and propose a systematic method for determining the optimal player position. Existing methods for accomplishing this task do not systematically balance risks and rewards, as they are not Pareto optimal by design. This may result in overlooking good opportunities. One more shortcoming of these methods is over simplifications in predicting the situation on the field, which may lead to performance loss. We propose two new ideas to address these issues. Experiments demonstrate that this results in a substantial increase in the team performance.
Liu, Z., Zhao, M., Shi, Z. & Xu, W.

Multi-robot Cooperative Localization through Collaborative Visual Object Tracking

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 41-52 inproceedings


Abstract: In this paper we present an approach for a team of robots to cooperatively improve their self localization through collaboratively tracking a moving object. At first, we use a Bayes net model to describe the multi-robot self localization and object tracking problem. Then, by exploring the independencies between different parts of the joint state space of the complex system, we show how the posterior estimation of the joint state can be factorized and the moving object can serve as a bridge for information exchange between the robots for realizing cooperative localization. Based on this, a particle filtering method for the joint state estimation problem is proposed. And, finally, in order to improve computational efficiency and achieve real-time implementation, we present a method for decoupling and distributing the joint state estimation onto different robots. The approach has been implemented on our four-legged AIBO robots and tested through different scenarios in RoboCup domain showing that the performance of localization can indeed be improved significantly.
Loncomilla, P. & del Solar, J.R.

Robust Object Recognition Using Wide Baseline Matching for RoboCup Applications

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 441-448 inproceedings


Abstract: As the RoboCup leagues evolve, higher requirements (e.g. object recognition skills) are imposed over the robot vision systems, which cannot be fulfilled using simple mechanisms as pure color segmentation or visual sonar. In this context the main objective of this article is to propose a robust object recognition system, based on the wide-baseline matching between a reference image (object model) and a test image where the object is searched. The wide baseline matching is implemented using local interest points and invariant descriptors. The proposed object recognition system is validated in two real-world tasks, recognition of objects in the RoboCup @Home league, and detection of robots in the humanoid league. This research was funded by Millennium Nucleus Center for Web Research, Grant P04-067-F, Chile.
Mayer, N.M., Boedecker, J., Masui, K., Ogino, M. & Asada, M.

HMDP: A New Protocol for Motion Pattern Generation Towards Behavior Abstraction

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 184-195 inproceedings


Abstract: The control of more than 20 degrees of freedom in real-time is one challenge of humanoid robotics. The control architecture of an autonomous humanoid robot often consists of two parts, namely a real-time part that has direct access to the motors or RC servos, and a non-real-time part, that controls the higher-level behaviors and sensory information processing such as vision and touch. As a result motion patterns are developed separately from the other parts of the robots behavior. In research, particularly when including developmental processes, it is often necessary that the design or the evolution of motion patterns is integrated in the overall development of the robot’s behavior. This is indeed one of the main principles of the embodied intelligence paradigm. The main aim of this work is to define a flexible way of describing motion patterns that can be passed to the motion controller which in turn executes them in real-time. As a result, the Harmonic Motion Description Protocol (HMDP) is presented. It allows the motions to be described as vectors of coefficients of harmonic motion splines. The motion splines are expressed as human-readable ASCII strings that can be passed as a motion stream. Flexibility is achieved by implementing the principle of superposition of several motion patterns. In this way also closed loop control is achievable in principle. Moreover, the HMDP can be implemented into the (deleted for blind review) project of the 3D soccer simulation league as a standard way to communicate motion patterns between the agent and the simulation interface and/or real humanoid robots.
Nakanishi, R., Murakami, K. & Naruse, T.

Dynamic Positioning Method Based on Dominant Region Diagram to Realize Successful Cooperative Play

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 488-495 inproceedings


Abstract: In this paper, we propose a new technique to compute, in real time, the positions of robots in a cooperative play such as the pass-and-shoot play. To evaluate the positioning of the robot, we use the Dominant Region (DR) diagram, which is a kind of a Voronoi diagram. In the DR diagram, the soccer field is divided into regions, each of which shows an area that a robot can reach faster than the other robots. This division is based on the time of arrival while the division by the Voronoi diagram is based on the distance of arrival. Though the DR diagram plays a primary role in the positioning of the robots, it has a serious problem of taking much computation time. To overcome this problem, we show an approximate calculation procedure to obtain the DR diagram, which realizes the real time computation, i.e. a computation within a frame time. Applying the approximate dominant diagram to the positioning of the robots for the pass play, we show, by the simulation study,
• the DR diagram can be calculated in real time,
• an appropriate position for the pass play can be obtained.
Narita, R., Murakami, K. & Naruse, T.

Strategic Layout of Multi-cameras Based on a Minimum Risk Criterion

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 361-368 inproceedings


Abstract: This paper proposes a method to allocate multiple cameras to a better or the best positions. In RoboCup Small Size League(SSL), two or more cameras are used, and we have to decide the layout of them at the venue. This paper gives a criterion which minimizes the risk, for example, the occlusion of a ball by robots, and solves it by using Fletcher-Reeves conjugate gradient algorithm. Experimental result shows the effectiveness of the proposed method.
Noma, K., Takahashi, Y. & Asada, M.

Cooperative/Competitive Behavior Acquisition Based on State Value Estimation of Others

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 101-112 inproceedings


Abstract: The existing reinforcement learning approaches have been suffering from the curse of dimension problem when they are applied to multiagent dynamic environments. One of the typical examples is a case of RoboCup competition since other agents and their behaviors easily cause state and action space explosion. This paper presents a method of hierarchical modular learning in a multiagent environment by which the learning agent can acquire cooperative behaviors with its teammates and competitive ones against its opponents. The key ideas to resolve the issue are as follows. First, a two-layer hierarchical system with multi learning modules is adopted to reduce the size of the state and action spaces. The state space of the top layer consists of the state values from the lower level, and the macro actions are used to reduce the size of the action space. Second, the state of the other to what extent it is close to its own goal is estimated by observation and used as a state value in the top layer state space to realize the cooperative/competitive behaviors. The method is applied to 4 (defence team) on 5 (offence team) game task, and the learning agent successfully acquired the teamwork plays (pass and shoot) within much shorter learning time (30 times quicker than the earlier work).
Ogino, M., Toyama, H., Fuke, S., Mayer, N.M., Watanabe, A. & Asada, M.

Compliance Control for Biped Walking on Rough Terrain

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 556-563 inproceedings


Abstract: In this paper, we propose a control system that changes the compliance based on the walking speed to stabilize biped walking on rough terrain. The proposed system changes walking modes depends on its walking speed. In the downhill terrain, when the walking speed increases, the stiffness of the ankle in the support phase is controlled so as to brake the increased speed. In the uphill terrain, when the walking speed decreases, the stiffness of the waist joint is controlled and the desired trajectory for the supported leg is shifted so as not to falls down backward. To validate the efficiency of the proposed system, the stability of walking with the proposed system is examined in the two dimensional dynamics simulation. It is shown that the robot with the proposed system can walk in the more variable rough terrain and with the broader walking speed than without changing the stiffness of the joints.
Otake, K., Murakami, K. & Naruse, T.

Precise Extraction of Partially Occluded Objects by Using HLAC Features and SVM

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 17-28 inproceedings


Abstract: In the RoboCup competition, robot soccer game, ball and robots are extracted by using color information. If color markers attached on the robot or a ball itself are occluded, especially the occlusion ratio is high, it will be difficult to extract them. This paper proposes a new and high precision method which extracts partially occluded objects based on the statistical features of the pixel and its neighborhoods. Concretely, at first, input image is labeled by using color information and small candidate regions which have similar color to the color markers or the ball are extracted, then each candidate region is classified into partially occluded object or noise by using HLAC features and SVM. We applied our method to the global vision system of RoboCup small size league (SSL) and confirmed that it could extract partially occluded objects, 94.23% for 5 to 8 pixels area and 80.06% for 3 to 4 pixels area, and worked more than 60fps.
Pfingsthorn, M., Slamet, B. & Visser, A.

A Scalable Hybrid Multi-robot SLAM Method for Highly Detailed Maps

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 457-464 inproceedings


Abstract: Recent successful SLAM methods employ hybrid map representations combining the strengths of topological maps and occupancy grids. Such representations often facilitate multi-agent mapping. In this paper, a successful SLAM method is presented, which is inspired by the manifold data structure by Howard et al. This method maintains a graph with sensor observations stored in vertices and pose differences including uncertainty information stored in edges. Through its graph structure, updates are local and can be efficiently communicated to peers. The graph links represent known traversable space, and facilitate tasks like path planning. We demonstrate that our SLAM method produces very detailed maps without sacrificing scalability. The presented method was used by the UvA Rescue Virtual Robots team, which won the Best Mapping Award in the RoboCup Rescue Virtual Robots competition in 2006.
Qining Wang, Yan Huang, G.X. & Wang, L.

Let Robots Play Soccer under More Natural Conditions: Experience-Based Collaborative Localization in Four-Legged League

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 353-360 inproceedings


Abstract: This paper presents an experience-based collaborative approach for a group of autonomous robots to localize in asymmetric, dynamic environments. To help robots play soccer under more natural conditions, we propose a Markov localization based hybrid method with integration of environment experience construction and dynamic reference object based multi-robot localization. By using this method, the robot can estimate and correct its position perception more accurately and effectively among a group of autonomous robots, taking the odometry error and other negative influence into consideration. Satisfactory results are obtained in the RoboCup Four-Legged League environment.
Rajaie, H., Lafrenz, R., Zweigle, O., Käppeler, U.-P., Schreiber, F., Rühr, T., Tamke, A. & Levi, P.

Physical Simulation of the Dynamical Behavior of Three-Wheeled Omni-directional Robots

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 270-277 inproceedings


Abstract: Hardware simulation is a very efficient way for parameter tuning. We developed a Simulink-based simulator for the navigation components of our robotic soccer team. This physical simulation has interfaces to be interconnected with the higher levels of the real control software and is therefore able to perform an overall simulation of single robots.
Ribeiro, F., Moutinho, I., Pereira, N., Oliveira, F., Fernandes, J., Peixoto, N. & Salgado, A.

High Accuracy Navigation in Unknown Environment Using Adaptive Control

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 312-319 inproceedings


Abstract: Aiming to reduce cycle time and improving the accuracy on tracking, a modified adaptive control was developed, which adapts autonomously to changing dynamic parameters. The platform used is based on a robot with a vision based sensory system. Goal and obstacles angles are calculated relatively to robot orientation from image processing software. Autonomous robots are programmed to navigate in unknown and unstructured environments where there are multiple obstacles which can readily change their position. This approach underlies in dynamic attractor and repulsive forces. This theory uses differential equations that produce vector fields to control speed and direction of the robot. This new strategy was compared with existing PID method experimentally and it proved to be more effective in terms of behaviour and time-response. Calibration parameters used in PID control are in this case unnecessary. The experiments were carried out in robot Middle Size League football players built for RoboCup. Target pursuit, namely, ball, goal or any absolute position, was tested. Results showed high tracking accuracy and rapid response to moving targets. This dynamic control system enables a good balance between fast movements and smooth behaviour.
RÖfer, T.

Region-Based Segmentation with Ambiguous Color Classes and 2-D Motion Compensation

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 369-376 inproceedings


Abstract: This paper presents a new approach for color segmentation, in which colors are not only mapped to unambiguous but also to ambiguous color classes. The ambiguous color classes are resolved based on their unambiguous neighbors in the image. In contrast to other approaches, the neighborhood is determined on the level of regions, not on the level of pixels. Thereby, large regions with ambiguous color classes can be resolved. The method is fast enough to run on a Sony AIBO in real time (30 Hz), leaving enough resources for the other tasks that have to be performed to play soccer. In addition, the paper discusses the problem of motion compensation, i.e. reversing the effects of a rolling shutter on the images taken by a moving camera. This work has been funded by the Deutsche Forschungsgemeinschaft in the context of the Schwerpunktprogramm 1125 (Kooperierende Teams mobiler Roboter in dynamischen Umgebungen).
Savage, J., LLarena, A., Carrera, G., Cuellar, S., Esparza, D., Minami, Y. & Peñuelas, U.

ViRbot: A System for the Operation of Mobile Robots

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 512-519 inproceedings


Abstract: This paper describes a robotics architecture, the ViRbot, used to control the operation of service mobile robots. It accomplish the required commands using AI actions planning and reactive behaviors with a description of the working environment. In the ViRbot architecture the actions planner module uses Conceptual Dependency (CD) primitives as the base for representing the problem domain. After a command is spoken to the mobile robot a CD representation of it is generated, a rule based system takes this CD representation, and using the state of the environment generates other subtasks represented by CDs to accomplish the command. By using a good representation of the problem domain through CDs and a rule based system as an inference engine, the operation of the robot becomes a more tractable problem and easier to implement. The ViRbot system was tested in the Robocup@Home [1] category in the Robocup competition at Bremen, Germany in 2006 and in Atlanta in 2007, where our robot TPR8, obtained the third place in this category.
Shahri, A.H., Monfared, A.A. & Elahi, M.

A Deeper Look at 3D Soccer Simulations

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 294-301 inproceedings


Abstract: Developing an intelligent agent requires more than an Integrated Development Environment (IDE). In multi agent environments or systems equipped with artificial intelligence it is often difficult to obtain the function or method which led to a particular behavior that is noticeable from outside. In addition to previous dilemma, the publicity that the RoboCup events get from the media provides an ideal opportunity to show the state of art of these systems during RoboCup World Cup. This paper describes the concept and the implementation of Team Assistant 2006 as the next generation of TA2002. The idea is to provide a tool that is able to assist developers to detect problems of their agents both in single and cooperation mode and also organizers to have better games.TA2006 won the second place in RoboCup 3D development competition 2006.
da Silva Guerra, R., Boedecker, J., Mayer, N., Yanagimachi, S., Hirosawa, Y., Yoshikawa, K., Namekawa, M. & Asada, M.

Introducing Physical Visualization Sub-league

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 496-503 inproceedings


Abstract: This work introduces the new sub-league of the RoboCup Soccer Simulation League, called Physical Visualization. We show how the fundamental collaborative concepts of this new sub-league shift essential research issues from the playing agents themselves to the development of a new versatile research and educational platform. Additionally, we discuss benefits of this new platform in terms of standardization, flexibility and reasonable price. We also try to characterize and discuss the place of this new sub-league within the RoboCup community. Finally, competition formats and roadmaps are presented and discussed.
Silva, H., Almeida, J.M., Lima, L., Martins, A. & Silva, E.P.

A Real Time Vision System for Autonomous Systems: Characterization during a Middle Size Match

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 504-511 inproceedings


Abstract: This paper propose a real-time vision framework for mobile robotics and describes the current implementation. The pipeline structure further reduces latency and allows a paralleled hardware implementation. A dedicated hardware vision sensor was developed in order to take advantage of the proposed architecture. The real-time characteristics and hardware partial implementation, coupled with low energy consumption address typical autonomous systems applications. A characterization of the implemented system in the Robocup scenario, during competition matches, is presented.
Stoye, K. & Elfers, C.

Intuitive Plan Construction and Adaptive Plan Selection

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 278-285 inproceedings


Abstract: Typical tasks of multi agent systems are effective coordination of single agents and their cooperation. Especially in dynamic environments, like the RoboCup soccer domain, the uncertainty of an opponent’s team behavior complicates coordinated team action. This paper presents a novel approach for intuitive multi agent plan construction and adaptive plan selection to attempt these tasks. We introduce a tool designed to represent plans like in tactical playbooks in human soccer which allows easy plan construction, editing and managing. Further we introduce a technique that provides adaptive plan selection in offensive situations by evaluating effectiveness of plans and their actions with statistically interpreted results to improve a team’s style of play. Using experts as a concept for abstracting information about a team’s interaction with another, makes fast accommodated plan selection possible. We briefly describe our software components, examine the performance of our implementation and give an example for rational plan selection in the RoboCup Small Size League.
Taiana, M., Gaspar, J., Nascimento, J., Bernardino, A. & Lima, P.

3D Tracking by Catadioptric Vision Based on Particle Filters

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 77-88 inproceedings


Abstract: This paper presents a robust tracking system for autonomous robots equipped with omnidirectional cameras. The proposed method uses a 3D shape and color-based object model. This allows to tackle difficulties that arise when the tracked object is placed above the ground plane floor. Tracking under these conditions has two major difficulties: first, observation with omnidirectional sensors largely deforms the target’s shape; second, the object of interest embedded in a dynamic scenario may suffer from occlusion, overlap and ambiguities. To surmount these difficulties, we use a 3D particle filter to represent the target’s state space: position and velocity with respect to the robot. To compute the likelihood of each particle the following features are taken into account: i) image color; ii) mismatch between target’s color and background color. We test the accuracy of the algorithm in a RoboCup Middle Size League scenario, both with static and moving targets. This work was supported by Fundação para a Ciência e a Tecnologia (ISR/IST pluriannual funding) through the POS-Conhecimento Program that includes FEDER funds. We would like to thank Dr. Luis Montesano and Dr. Alessio Del Bue for the helpful discussions.
Takahashi, Y., Nowak, W. & Wisspeintner, T.

Adaptive Recognition of Color-Coded Objects in Indoor and Outdoor Environments

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 65-76 inproceedings


Abstract: To achieve robust color perception under varying light conditions in indoor and outdoor environments, we propose a three-step method consisting of adaptive camera parameter control, image segmentation and color classification. A controller for the intrinsic camera para- meters is used to improve color stability in the YUV space. Segmentation is done to detect spatially coherent regions of uniform color belonging to objects in the image. Then, a probabilistic classification method is applied to label the colors by use of a Gaussian color distribution model. Experiments under combination of artificial and natural illuminations indoors and outdoors have been carried out. The results show the feasibility of this approach as well as the problems that occur under these highly diverse light situations. In particular we investigate the application in a RoboCup soccer scenario pointing toward future outdoor use.
Tarizzo, A. & Rella, G.

A Force Sensor Made by Diaphragm Pattern Mounted on a Deformable Circular Plate

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 465-471 inproceedings


Abstract: Measurement of the real distribution of pressure under the foot is an important challenge to obtain an efficacious stability control and a real dynamic walk. This paper is the result of the work of two students in mechanical and mechatronic engineering who have built a force sensor using a diaphragm pattern mounted on a deformable circular plate. The result of this study is a cheaper but accurate force sensor that will be mounted on the humanoid robot “I-2” of Politecnico di Torino. The design of the sensor in centred to analyse the deformation of a circular plate loaded at the center considering various edge conditions. It begin from the structural analyse of the plates considering different loads and edge conditions to obtain a deformation model as near as possible to reality. Final goal will be to obtain an output voltage proportional to the deformation of the plate.
Trieu, M. & Williams, M.-A.

Grounded Representation Driven Robot Motion Design

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 520-527 inproceedings


Abstract: Grounding robot representations is an important problem in Artificial Intelligence. In this paper we show how a new grounding framework guided the development of an improved locomotion engine [3] for the AIBO. The improvements stemmed from higher quality representations that were grounded better than those in the previous system [1]. Since the AIBO is more grounded under the new locomotion engine it makes better decisions and achieves its design goals more efficiently. Furthermore, a well grounded robot offers significant software engineering benefits since its behaviours can be developed, debugged and tested more effectively.
Visser, A., Xingrui-Ji, van Ittersum, M., Jaime, L.A.G. & Stancu, L.A.

Beyond Frontier Exploration

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 113-123 inproceedings


Abstract: This article investigates the prerequisites for a global exploration strategy in an unknown environment on a virtual disaster site. Assume that a robot equipped with a laser range scanner can build a detailed map of a previous unknown environment. The remaining question is how to use this information on this map for further exploration. On a map several interesting locations can be present where the exploration can be continued, referred as exploration frontiers. Typically, a greedy algorithm is used for the decision which frontier to explore next. Such a greedy algorithm only considers interesting locations locally, focused to reduce the movement costs. More sophisticated algorithms also take into account the information that can be gained along each frontier. This shifts the problem to estimate the amount of unexplored area behind the frontiers on the global map. Our algorithm exploits the long range of current laser scanners. Typically, during the previous exploration a small number of laser rays already passed the frontier, but this number is too low to have major impact on the generated map. Yet, the few rays through a frontier can be used to estimate the potential information gain from unexplored area beyond the frontier.
Wagner, T., Bogon, T. & Elfers, C.

Incremental Generation of Abductive Explanations for Tactical Behavior

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 401-408 inproceedings


Abstract: According to the expert literature on (human) soccer, e.g., the tactical behavior of a soccer team should differ significantly with respect to the tactics and strategy of the opponent team. In the offensive phase the attacking team is usually able to actively select an appropriate tactic with limited regard to the opponent strategy. In contrast, in the defensive phase the more passive recognition of tactical patterns of the behavior of the opponent team is crucial for success. In this paper we present a qualitative, formal, abductive approach, based on a uniform representation of soccer tactics that allows to recognize/explain the tactical and strategical behavior of opponent teams based on past (usually incomplete) observations.
Wu, F. & Chen, X.

Solving Large-Scale and Sparse-Reward DEC-POMDPs with Correlation-MDPs

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 208-219 inproceedings


Abstract: Within a group of cooperating agents the decision making of an individual agent depends on the actions of the other agents. A lot of effort has been made to solve this problem with additional assumptions on the communication abilities of agents. However, in some realworld applications, communication is limited and the assumptions are rarely satisfied. An alternative approach newly developed is to employ a correlation device to correlate the agents’ behavior without exchanging information during execution. In this paper, we apply correlation device to large-scale and spare-reward domains. As a basis we use the framework of infinite-horizon DEC-POMDPs which represent policies as joint stochastic finite-state controllers. To solve any problem of this kind, a correlation device is firstly calculated by solving Correlation Markov Decision Processes (Correlation-MDPs) and then used to improve the local controller for each agent. By using this method, we are able to achieve a tradeoff between computational complexity and the quality of the approximation. In addition, we demonstrate that, adversarial problems can be solved by encoding the information of opponents’ behavior in the correlation device. We have successfully implemented the proposed method into our 2D simulated robot soccer team and the performance in RoboCup-2006 was encouraging. This work is supported by the NSFC 60275024 and the 973 programme 2003CB317000.
Wyeth, P. & Wyeth, G.

Robot Building for Preschoolers

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 124-135 inproceedings


Abstract: This paper describes Electronic Blocks, a new robot construction element designed to allow children as young as age three to build and program robotic structures. The Electronic Blocks encapsulate input, output and logic concepts in tangible elements that young children can use to create a wide variety of physical agents. The children are able to determine the behavior of these agents by the choice of blocks and the manner in which they are connected. The Electronic Blocks allow children without any knowledge of mechanical design or computer programming to create and control physically embodied robots. They facilitate the development of technological capability by enabling children to design, construct, explore and evaluate dynamic robotics systems. A study of four and five year-old children using the Electronic Blocks has demonstrated that the interface is well suited to young children. The complexity of the implementation is hidden from the children, leaving the children free to autonomously explore the functionality of the blocks. As a consequence, children are free to move their focus beyond the technology. Instead they are free to focus on the construction process, and to work on goals related to the creation of robotic behaviors and interactions. As a resource for robot building, the blocks have proved to be effective in encouraging children to create robot structures, allowing children to design and program robot behaviors.
Yuan, X. & Yingzi, T.

Rational Passing Decision Based on Region for the Robotic Soccer

2008 RoboCup 2007: Robot Soccer World Cup XI, pp. 238-245 inproceedings


Abstract: For an agent to behave appropriately in an uncertain environment, efficient representation of knowledge and reliable reasoning mechanisms are at the core of design. This paper proposes a novel region based passing scheme for the robotic soccer. The scheme captures qualitative knowledge of soccer in a natural and efficient way. We implemented the rational passing decision based on region(RPDR) in our RoboCup simulation 3D team. Experiments show that our method outperforms the base line method, i.e. position searching approach.