Patrick Riley, Manuela Veloso, and Gal Kaminka. An empirical study of coaching. In H. Asama, T. Arai, T. Fukuda, and T. Hasegawa, editors, Distributed Autonomous Robotic Systems 5, pages 215--224. SpringerVerlag, 2002.  Home/Search   Document Details and Download   Summary   Related Articles   Check

No context found.

Riley, P., Veloso, M., Kaminka, G.: An empirical study of coaching. In Asama, H., Arai, T., Fukuda, T., Hasegawa, T., eds.: Distributed Autonomous Robotic Systems 5. Springer-Verlag (2002) 215--224

....well as how advice from other agents (e.g. a team captain , others teammates, or even humans) can be used [2] A speci c issue of consideration is whether to blindly follow all advice that is given. For example, the ChaMeleon s online coach, OWL, o ers advice in the form of passing rules [14] [13]. Consider the situation when an agent has a clear shot on goal but one of the coach s passing rules is applicable should the agent follow the advice and pass the ball or should it attempt to score on the open goal In the same situation, a human is able to quickly reason about the choices and ....

....a higher priority. For example, an agent should always choose to shoot when the probability of success is high as opposed to making a coach recommended pass. 4 Coach Advice This section gives an overview of the type of advice given by the OWL coach, which was also created at Carnegie Mellon [14] [13]. There are ve main types of advice. The rst four are given at the beginning the game. The rst gives a home position for each member of the team to de ne a team formation. Next, static marking advice assigns defenders to mark the opponent forwards, based on the expected opponent formation. ....

[Article contains additional citation context not shown here]

Patrick Riley, Manuela Veloso, and Gal Kaminka. An empirical study of coaching. In Distributed Autonomous Robotic Systems 6. Springer-Verlag, 2002. (to appear).

....their behavior, in order to predict their future behavior, coordinate with them, assist them, or counter their actions. For instance, agent modeling techniques have been used in intelligent user interfaces [1,2] in virtual environments for training [3,4] and in coaching and team training [5]. Typically, agent modeling techniques assume the availability of a plan or behavior library, which encodes the complete repertoire of expected observed behavior. The agent modeling techniques typically focus on matching the observations with library prototypes, and drawing inferences based on ....

....In such settings, the behavioral repertoire of observed agents is unknown to the observer. Indeed, such is the case in our application area in robotic soccer. We are developing a coach that is able to improve the performance of a team by analyzing its behavior, and that of the opponent [5]. Since both the coached team and the opponent are unknown at the time of the design, an accurate behavior library describing their expected behavior is unavailable to the coach agent. In particular, a key challenge here is to determine what sequential behaviors best characterize a team, and best ....

Riley, P., Veloso, M., Kaminka, G.: An empirical study of coaching. In: Proceedings of Distributed Autonomous Robotic Systems 6, Springer-Verlag (2002) (to appear).

....One hypothesis that underlies our current work is that in order for a coach to be e#ective coaching multiple di#erent teams of agents, it needs to adapt its advice to peculiarities of the team being coached. Notably, see the results of the coach competition at RoboCup 2001 [3] as discussed in [22, 23] where the ability of a coach to improve a team s performance varied vastly across di#erent teams. The Advisee Information box should take observations about the team being coached and use that information to change the advice being sent. This component is the main focus of this paper. Lastly, ....

P. Riley, M. Veloso, and G. Kaminka. An empirical study of coaching. In H. Asama, T. Arai, T. Fukuda, and T. Hasegawa, editors, Distributed Autonomous Robotic Systems 5, pages 215--224. Springer-Verlag, 2002.

....their behavior, in order to predict their future behavior, coordinate with them, assist them, or counter their actions. For instance, agent modeling techniques have been used in intelligent user interfaces [1,2] in virtual environments for training [3,4] and in coaching and team training [5]. Typically, agent modeling techniques assume the availability of a plan or behavior library, which encodes the complete repertoire of expected observed behavior. The agent As of July 2002, the author is a senior lecturer (assistant professor) at Bar Ilan University, Israel, and an adjunct ....

....In such settings, the behavioral repertoire of observed agents is unknown to the observer. Indeed, such is the case in our application area in robotic soccer. We are developing a coach that is able to improve the performance of a team by analyzing its behavior, and that of the opponent [5]. Since both the coached team and the opponent are unknown at the time of the design, an accurate behavior library describing their expected behavior is unavailable to the coach agent. In particular, a key challenge here is to determine what sequential behaviors best characterize a team, and best ....

Riley, P., Veloso, M., Kaminka, G.: An empirical study of coaching. In: Proceedings of Distributed Autonomous Robotic Systems 6, Springer-Verlag (2002) (to appear).

....well as how advice from other agents (e.g. a team captain , others teammates, or even humans) can be used [2] A specific issue of consideration is whether to blindly follow all advice that is given. For example, the ChaMeleon s online coach, OWL, o#ers advice in the form of passing rules [14] [13]. Consider the situation when an agent has a clear shot on goal but one of the coach s passing rules is applicable should the agent follow the advice and pass the ball or should it attempt to score on the open goal In the same situation, a human is able to quickly reason about the choices and ....

....a higher priority. For example, an agent should always choose to shoot when the probability of success is high as opposed to making a coach recommended pass. 4 Coach Advice This section gives an overview of the type of advice given by the OWL coach, which was also created at Carnegie Mellon [14] [13]. There are five main types of advice. The first four are given at the beginning the game. The first gives a home position for each member of the team to define a team formation. Next, static marking advice assigns defenders to mark the opponent forwards, based on the expected opponent formation. ....

[Article contains additional citation context not shown here]

Patrick Riley, Manuela Veloso, and Gal Kaminka. An empirical study of coaching. In Distributed Autonomous Robotic Systems 6. Springer-Verlag, 2002. (to appear).

....designed by one group on the team of another. We participated in the coach competition, which consisted a single game in each test case. This section reports on our thorough empirical evaluation of the teams involved in the competition. An analysis of the techniques our coach uses can be found in [3]. Each experimental condition was run for 30 games and the average score di#erence (as our score minus their score) is reported. All significance values reported are for a two tailed t test. We will use initials to denote the teams. The four teams that understand CLang are: WrightEagle (WE) ....

P. Riley, M. Veloso, and G. Kaminka. An empirical study of coaching. In Distributed Autonomous Robotic Systems 6. Springer-Verlag, 2002. (to appear).

....as how advice from other agents (e.g. a team captain , others teammates, or even humans) can also be used [2] A speci c issue of consideration is whether to blindly follow all advice that is given. For example, the ChaMeleon s online coach, OWL, o ers advice in the form of passing rules [13] [12]. Consider the situation when an agent has a clear 2 Paul Carpenter et al. shot on goal but one of the coach s passing rules is applicable should the agent follow the advice and pass the ball or should it attempt to score on the open goal It is clear that in this situation,the coach should be ....

....a higher priority. For example, an agent should always choose to shoot when the probability of success is high as opposed to making a coach recommended pass. 4 Coach Advice This section gives an overview of the type of advice given by the OWL coach, which was also created at Carnegie Mellon [13] [12]. There are ve main types of advice. The rst four are given at the beginning the game. The rst gives a home position for each member of the team to de ne a team formation. Next, static marking advice assigns defenders to mark the opponent forwards, based on the expected opponent formation. ....

Patrick Riley, Manuela Veloso, and Gal Kaminka. An empirical study of coaching. In Distributed Autonomous Robotic Systems 6. Springer-Verlag, 2002. (to appear).

No context found.

Patrick Riley, Manuela Veloso, and Gal Kaminka. An empirical study of coaching. In H. Asama, T. Arai, T. Fukuda, and T. Hasegawa, editors, Distributed Autonomous Robotic Systems 5, pages 215--224. SpringerVerlag, 2002.

No context found.

Patrick Riley, Manuela Veloso, and Gal Kaminka. An empirical study of coaching. In H. Asama, T. Arai, T. Fukuda, and T. Hasegawa, editors, Distributed Autonomous Robotic Systems 5, pages 215-224. Springer-Verlag, 2002.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC