Robot soccer game is an interesting emerging domain for multiple cooperative robotic system. This paper discusses the detailed design of a simulator, and describes the architecture of soccer server and client in detai...Robot soccer game is an interesting emerging domain for multiple cooperative robotic system. This paper discusses the detailed design of a simulator, and describes the architecture of soccer server and client in detail. This simulator is sufficiently flexible and robust for the users to develop strategies for a simulated competition and to test algorithms of intelligent robotics.展开更多
The fast paced nature of robotic soccer necessitates real time sensing coupled with quick decision making and behaving. The robot must have high response rate, exact motion ability, and must robust enough to confront ...The fast paced nature of robotic soccer necessitates real time sensing coupled with quick decision making and behaving. The robot must have high response rate, exact motion ability, and must robust enough to confront interfere during drastic match. But during the match, we find that the robot usually do not act exactly as the commands from host computer. In this paper, we analyze the reason and present a method that uses BP neural network to output robotic velocity directly instead of conventional path plan strategy, to reduce the error between actual motion and ideal plan.展开更多
The decision making system of robot soccer is a kind of knowledge system. A Six step Reasoning Model is established by formalizing its expert knowledge and decision making process. Furthermore, many other models can b...The decision making system of robot soccer is a kind of knowledge system. A Six step Reasoning Model is established by formalizing its expert knowledge and decision making process. Furthermore, many other models can be considered as mutation and evolution of the Six step Reasoning Model.展开更多
Introduces of robot soccer’s competition software of Harbin Institute of Technology (HIT), the concept of running range and the method of calculating the running range for both the opponent and our teammates accordin...Introduces of robot soccer’s competition software of Harbin Institute of Technology (HIT), the concept of running range and the method of calculating the running range for both the opponent and our teammates according to the distances between the ball and robot soccers, and therefore the method of calculating the angle that the robot passes or shoots the ball according to the running ranges of both sides. And gives the examples of passing the ball when the ball’s position is in the backcourt and shooting the ball when the ball’s position is in the frontcourt.展开更多
The soccer robot system and the effective multi agents cooperation strategy applied to the MASKARO team composed of a set of action controllers, a set of behavior module and a behavior selector are introduced. The act...The soccer robot system and the effective multi agents cooperation strategy applied to the MASKARO team composed of a set of action controllers, a set of behavior module and a behavior selector are introduced. The action is the primitive low level component of the robot control system necessary to move the robots on the playground. Each action controller determines the linear and angular velocity commands of the robots corresponding to its own purpose. The behavior is the high level component of the robot control system composed of necessary action sequences. Each behavior module determines the desired action sequences and action commands corresponding to its own objective. The behavior selector considering the information that comes from the vision system selects the behavior of each robot every sampling time. Thus, the behavior of each robot is changed dynamically. The presented strategy is successfully applied to the MASKARO team and the team is ranked in the first place in the 2000 FIRA Korea Cup K League.展开更多
This paper researches robot soccer action selection based on Q learning .The robot learn to activate particular behavior given their current situation and reward signal. We adopt neural network to implementations ...This paper researches robot soccer action selection based on Q learning .The robot learn to activate particular behavior given their current situation and reward signal. We adopt neural network to implementations of Q learning for their generalization properties and limited computer memory requirements.展开更多
In the RoboSot category, no central station or over head camera is allowed. Therefore several issues such as sensors, mechanisms and CPU boards must be taken into consideration before building an autonomous and intell...In the RoboSot category, no central station or over head camera is allowed. Therefore several issues such as sensors, mechanisms and CPU boards must be taken into consideration before building an autonomous and intelligent system for this category. Each robot must have appropriate sensors and must be able to communicate with other robots to have timely information concerning the ball and robot position in the field. Computing power also becomes and improtant factor in implementing intelligence and autonomy onboard. These issues are discussed. Then the robot structure of the Dreams Come True(DCT) from KAIST is introduced. Each robot has a Pentium Π single board computer, and Ethernet card, and USB camera and an omni directional mobile mechanism.展开更多
A new design of vision based soccer robot using the type TMS320F240 of DSPs for MiroSot series is presented. The DSP used enables cost effective control of DC motor, and features fewer external components, lower syste...A new design of vision based soccer robot using the type TMS320F240 of DSPs for MiroSot series is presented. The DSP used enables cost effective control of DC motor, and features fewer external components, lower system cost and better performances than traditional microcontroller. The hardware architecture of robot is firstly presented in detail, and then the software design is briefly discussed. The control structure of decision making subsystem is illuminated also in this paper. The conclusion and prospect are given at last.展开更多
Summary of typical information fusion systems, synthesis analysis of Robot Soccer’s architecture, recognition of its characteristic and key technique are given. The result is prompted that Robot Soccer can be treated...Summary of typical information fusion systems, synthesis analysis of Robot Soccer’s architecture, recognition of its characteristic and key technique are given. The result is prompted that Robot Soccer can be treated as a platform of the information fusion.展开更多
Discusses the application of artificial neural network for MIROSOT, introduces a layered model of BP network of soccer robot for learning basic behavior and cooperative behavior, and concludes from experimental result...Discusses the application of artificial neural network for MIROSOT, introduces a layered model of BP network of soccer robot for learning basic behavior and cooperative behavior, and concludes from experimental results that the model is effective.展开更多
A soccer robot system (HIT 1) was built to participate in MIROSOT_China99 held in Harbin Institute of Technology. Robot soccer game is a very complex robot application that incorporates real time vision system, robot ...A soccer robot system (HIT 1) was built to participate in MIROSOT_China99 held in Harbin Institute of Technology. Robot soccer game is a very complex robot application that incorporates real time vision system, robot control, wireless communication and control of multiple robots. In the paper, we present the design and the hardware architecture and software architecture of our distributed multiple robot system.展开更多
A new ball passing strategy for robot soccer is proposed in this paper. With introduce of a new algorithm on ball passing, the optimum strategy is confirmed to be more efficient and exact when passing a ball. Question...A new ball passing strategy for robot soccer is proposed in this paper. With introduce of a new algorithm on ball passing, the optimum strategy is confirmed to be more efficient and exact when passing a ball. Questions of role switching in multi-intelligent agent cooperation in robot soccer are described based on Generalized Stochastic Petri-Net （GSPN）. Results of computer simulation have confirmed the feasibility and efficiency of above Petri-net method.展开更多
A dynamic cooperation model of multi-agent is established by combining reinforcement learning with distributed artificial intelligence(DAI),in which the concept of individual optimization loses its meaning because of ...A dynamic cooperation model of multi-agent is established by combining reinforcement learning with distributed artificial intelligence(DAI),in which the concept of individual optimization loses its meaning because of the dependence of repayment on each agent itself and the choice of other agents.Utilizing the idea of DAI,the intellectual unit of each robot and the change of task and environment,each agent can make decisions independently and finish various complicated tasks by communication and reciprocation between each other.The method is superior to other reinforcement learning methods commonly used in the multi-agent system.It can improve the convergence velocity of reinforcement learning,decrease requirements of computer memory,and enhance the capability of computing and logical ratiocinating for agent.The result of a simulated robot soccer match proves that the proposed cooperative strategy is valid.展开更多
Robot soccer competition provides an excellent opportunity for robotics research. We have built a soccer robot system to participate in internal and oversea matches. Firstly, we propose a new learning control scheme a...Robot soccer competition provides an excellent opportunity for robotics research. We have built a soccer robot system to participate in internal and oversea matches. Firstly, we propose a new learning control scheme adaptive PID learning controller. It means to overcome the drawbacks of the conventional PID type control methods. Secondly, we introduce our vision recognition algorithm. It remarkably increases the speed of recognition. Finally, we refer the communication system. We adopt bulletin board system to prevent communication confusion.展开更多
The fast paced nature of robotic soccer necessitates real time sensing coupled with quick behaving and decision making. In the field with real robots, it is important to well perceive the location of ball, team robots...The fast paced nature of robotic soccer necessitates real time sensing coupled with quick behaving and decision making. In the field with real robots, it is important to well perceive the location of ball, team robots and opponent robots through the vision system in real time. In this paper the architecture of global vision system of our small size robotic team and the process of object recognition is described. According to the study on color distribution in different color space and quantitative investigation, a method which uses H (Hue) thresholds as the major thresholds to feature exact and recognize object in real time is presented.展开更多
Presents a strategy for soccer robot path planning using genetic algorithms for which, real number coding method is used, to overcome the defects of binary coding method, and the double crossover operation adopted, to...Presents a strategy for soccer robot path planning using genetic algorithms for which, real number coding method is used, to overcome the defects of binary coding method, and the double crossover operation adopted, to avoid the common defect of early convergence and converge faster than the standard genetic algorithms concludes from simulation results that the method is effective for robot path planning.展开更多
Presents an algorithm which can be used to achieve complete decentralization of Kalman filter algorithm amongst sensing nodes of a multi sensor system, and points out the algorithm can be used for position estimation ...Presents an algorithm which can be used to achieve complete decentralization of Kalman filter algorithm amongst sensing nodes of a multi sensor system, and points out the algorithm can be used for position estimation in Robot Soccer because it does not require any form of central processing facility or centralized communications medium, and illustrates with a simulation example that it is very effective.展开更多
A large sample size is required for Monte Carlo localization （MCL） in multi-robot dynamic environ- ment, because of the ＂kidnapped robot＂ phenomenon, which will locate most of the samples in the regions with small...A large sample size is required for Monte Carlo localization （MCL） in multi-robot dynamic environ- ment, because of the ＂kidnapped robot＂ phenomenon, which will locate most of the samples in the regions with small value of desired posterior density. For this problem the crossover and mutation operators in evolutionary computation are introduced into MCL to make samples move towards the regions where the desired posterior density is large, so that the sample set can represent the density better. The proposed method is termed genetic Monte Carlo localization （GMCL）. Application in robot soccer system shows that GMCL can considerably reduce the required number of samples, and is more precise and robust in dynamic environment.展开更多
In the robot soccer competition platform, the cur- rent confrontation decision-making system suffers from dif- ficulties in optimization and adaptability. Therefore, we pro- pose a new self-adaptive decision-making （...In the robot soccer competition platform, the cur- rent confrontation decision-making system suffers from dif- ficulties in optimization and adaptability. Therefore, we pro- pose a new self-adaptive decision-making （SADM） strategy. SADM compensates for the restrictions of robot physical movement control by updating the task assignment and role assignment module using situation assessment techniques. It designs a self-adaptive role assignment model that assists the soccer robot in adapting to competition situations similar to how humans adapt in real time. Moreover, it also builds an accurate motion model for the robot in order to improve the competition ability of individual robot soccer. Experimental results show that SADM can adapt quickly and positively to new competition situations and has excellent performance in actual competition.展开更多
文摘Robot soccer game is an interesting emerging domain for multiple cooperative robotic system. This paper discusses the detailed design of a simulator, and describes the architecture of soccer server and client in detail. This simulator is sufficiently flexible and robust for the users to develop strategies for a simulated competition and to test algorithms of intelligent robotics.
文摘The fast paced nature of robotic soccer necessitates real time sensing coupled with quick decision making and behaving. The robot must have high response rate, exact motion ability, and must robust enough to confront interfere during drastic match. But during the match, we find that the robot usually do not act exactly as the commands from host computer. In this paper, we analyze the reason and present a method that uses BP neural network to output robotic velocity directly instead of conventional path plan strategy, to reduce the error between actual motion and ideal plan.
文摘The decision making system of robot soccer is a kind of knowledge system. A Six step Reasoning Model is established by formalizing its expert knowledge and decision making process. Furthermore, many other models can be considered as mutation and evolution of the Six step Reasoning Model.
文摘Introduces of robot soccer’s competition software of Harbin Institute of Technology (HIT), the concept of running range and the method of calculating the running range for both the opponent and our teammates according to the distances between the ball and robot soccers, and therefore the method of calculating the angle that the robot passes or shoots the ball according to the running ranges of both sides. And gives the examples of passing the ball when the ball’s position is in the backcourt and shooting the ball when the ball’s position is in the frontcourt.
文摘The soccer robot system and the effective multi agents cooperation strategy applied to the MASKARO team composed of a set of action controllers, a set of behavior module and a behavior selector are introduced. The action is the primitive low level component of the robot control system necessary to move the robots on the playground. Each action controller determines the linear and angular velocity commands of the robots corresponding to its own purpose. The behavior is the high level component of the robot control system composed of necessary action sequences. Each behavior module determines the desired action sequences and action commands corresponding to its own objective. The behavior selector considering the information that comes from the vision system selects the behavior of each robot every sampling time. Thus, the behavior of each robot is changed dynamically. The presented strategy is successfully applied to the MASKARO team and the team is ranked in the first place in the 2000 FIRA Korea Cup K League.
文摘This paper researches robot soccer action selection based on Q learning .The robot learn to activate particular behavior given their current situation and reward signal. We adopt neural network to implementations of Q learning for their generalization properties and limited computer memory requirements.
文摘In the RoboSot category, no central station or over head camera is allowed. Therefore several issues such as sensors, mechanisms and CPU boards must be taken into consideration before building an autonomous and intelligent system for this category. Each robot must have appropriate sensors and must be able to communicate with other robots to have timely information concerning the ball and robot position in the field. Computing power also becomes and improtant factor in implementing intelligence and autonomy onboard. These issues are discussed. Then the robot structure of the Dreams Come True(DCT) from KAIST is introduced. Each robot has a Pentium Π single board computer, and Ethernet card, and USB camera and an omni directional mobile mechanism.
文摘A new design of vision based soccer robot using the type TMS320F240 of DSPs for MiroSot series is presented. The DSP used enables cost effective control of DC motor, and features fewer external components, lower system cost and better performances than traditional microcontroller. The hardware architecture of robot is firstly presented in detail, and then the software design is briefly discussed. The control structure of decision making subsystem is illuminated also in this paper. The conclusion and prospect are given at last.
文摘Summary of typical information fusion systems, synthesis analysis of Robot Soccer’s architecture, recognition of its characteristic and key technique are given. The result is prompted that Robot Soccer can be treated as a platform of the information fusion.
文摘Discusses the application of artificial neural network for MIROSOT, introduces a layered model of BP network of soccer robot for learning basic behavior and cooperative behavior, and concludes from experimental results that the model is effective.
基金Supported by the High Technology Research and Developmeent Program of China
文摘A soccer robot system (HIT 1) was built to participate in MIROSOT_China99 held in Harbin Institute of Technology. Robot soccer game is a very complex robot application that incorporates real time vision system, robot control, wireless communication and control of multiple robots. In the paper, we present the design and the hardware architecture and software architecture of our distributed multiple robot system.
文摘A new ball passing strategy for robot soccer is proposed in this paper. With introduce of a new algorithm on ball passing, the optimum strategy is confirmed to be more efficient and exact when passing a ball. Questions of role switching in multi-intelligent agent cooperation in robot soccer are described based on Generalized Stochastic Petri-Net （GSPN）. Results of computer simulation have confirmed the feasibility and efficiency of above Petri-net method.
文摘A dynamic cooperation model of multi-agent is established by combining reinforcement learning with distributed artificial intelligence(DAI),in which the concept of individual optimization loses its meaning because of the dependence of repayment on each agent itself and the choice of other agents.Utilizing the idea of DAI,the intellectual unit of each robot and the change of task and environment,each agent can make decisions independently and finish various complicated tasks by communication and reciprocation between each other.The method is superior to other reinforcement learning methods commonly used in the multi-agent system.It can improve the convergence velocity of reinforcement learning,decrease requirements of computer memory,and enhance the capability of computing and logical ratiocinating for agent.The result of a simulated robot soccer match proves that the proposed cooperative strategy is valid.
文摘Robot soccer competition provides an excellent opportunity for robotics research. We have built a soccer robot system to participate in internal and oversea matches. Firstly, we propose a new learning control scheme adaptive PID learning controller. It means to overcome the drawbacks of the conventional PID type control methods. Secondly, we introduce our vision recognition algorithm. It remarkably increases the speed of recognition. Finally, we refer the communication system. We adopt bulletin board system to prevent communication confusion.
文摘The fast paced nature of robotic soccer necessitates real time sensing coupled with quick behaving and decision making. In the field with real robots, it is important to well perceive the location of ball, team robots and opponent robots through the vision system in real time. In this paper the architecture of global vision system of our small size robotic team and the process of object recognition is described. According to the study on color distribution in different color space and quantitative investigation, a method which uses H (Hue) thresholds as the major thresholds to feature exact and recognize object in real time is presented.
文摘Presents a strategy for soccer robot path planning using genetic algorithms for which, real number coding method is used, to overcome the defects of binary coding method, and the double crossover operation adopted, to avoid the common defect of early convergence and converge faster than the standard genetic algorithms concludes from simulation results that the method is effective for robot path planning.
文摘Presents an algorithm which can be used to achieve complete decentralization of Kalman filter algorithm amongst sensing nodes of a multi sensor system, and points out the algorithm can be used for position estimation in Robot Soccer because it does not require any form of central processing facility or centralized communications medium, and illustrates with a simulation example that it is very effective.
文摘A large sample size is required for Monte Carlo localization （MCL） in multi-robot dynamic environ- ment, because of the ＂kidnapped robot＂ phenomenon, which will locate most of the samples in the regions with small value of desired posterior density. For this problem the crossover and mutation operators in evolutionary computation are introduced into MCL to make samples move towards the regions where the desired posterior density is large, so that the sample set can represent the density better. The proposed method is termed genetic Monte Carlo localization （GMCL）. Application in robot soccer system shows that GMCL can considerably reduce the required number of samples, and is more precise and robust in dynamic environment.
文摘In the robot soccer competition platform, the cur- rent confrontation decision-making system suffers from dif- ficulties in optimization and adaptability. Therefore, we pro- pose a new self-adaptive decision-making （SADM） strategy. SADM compensates for the restrictions of robot physical movement control by updating the task assignment and role assignment module using situation assessment techniques. It designs a self-adaptive role assignment model that assists the soccer robot in adapting to competition situations similar to how humans adapt in real time. Moreover, it also builds an accurate motion model for the robot in order to improve the competition ability of individual robot soccer. Experimental results show that SADM can adapt quickly and positively to new competition situations and has excellent performance in actual competition.
