@inproceedings {icaps16-195,
    track    = {​​Robotics Track},
    title    = {Online Learning of Robot Soccer Free Kick Plans Using a Bandit Approach},
    url      = {http://www.aaai.org/ocs/index.php/ICAPS/ICAPS16/paper/view/13180},
    author   = {Juan Pablo Mendoza and  Reid Simmons and  Manuela Veloso},
    abstract = {This paper presents an online learning approach for teams of autonomous soccer robots to select free kick plans. In robot soccer, free kicks present an  opportunity to execute plans with relatively controllable initial conditions. However, the effectiveness of each plan is highly dependent on the adversary, and there are few free kicks during each game, making it necessary to learn online from sparse observations.  To achieve learning, we first greatly reduce the planning space by framing the problem as a contextual multi-armed bandit problem, in which the actions are a set of pre-computed plans, and the state is the position of the free kick on the field. During execution, we model the reward function for different free kicks using Gaussian Processes, and perform online learning using the Upper Confidence Bound algorithm. Results from a physics-based simulation reveal that the robots are capable of adapting to various different realistic opponents to maximize their expected reward during free kicks.},
    keywords = {adversarial action planning in competitive robotic domains,acquisition of planning models for robotics,learning action and task models}
}

{"_id":"4nskt8qkTz4ekNQLX","bibbaseid":"mendoza-simmons-veloso-onlinelearningofrobotsoccerfreekickplansusingabanditapproach","downloads":2,"creationDate":"2016-03-09T03:04:32.868Z","title":"Online Learning of Robot Soccer Free Kick Plans Using a Bandit Approach","author_short":["Mendoza, J. P.","Simmons, R.","Veloso, M."],"year":null,"bibtype":"inproceedings","biburl":"icaps16.icaps-conference.org/papers.bib","bibdata":{"bibtype":"inproceedings","type":"inproceedings","track":"​​Robotics Track","title":"Online Learning of Robot Soccer Free Kick Plans Using a Bandit Approach","url":"http://www.aaai.org/ocs/index.php/ICAPS/ICAPS16/paper/view/13180","author":[{"firstnames":["Juan","Pablo"],"propositions":[],"lastnames":["Mendoza"],"suffixes":[]},{"firstnames":["Reid"],"propositions":[],"lastnames":["Simmons"],"suffixes":[]},{"firstnames":["Manuela"],"propositions":[],"lastnames":["Veloso"],"suffixes":[]}],"abstract":"This paper presents an online learning approach for teams of autonomous soccer robots to select free kick plans. In robot soccer, free kicks present an opportunity to execute plans with relatively controllable initial conditions. However, the effectiveness of each plan is highly dependent on the adversary, and there are few free kicks during each game, making it necessary to learn online from sparse observations. To achieve learning, we first greatly reduce the planning space by framing the problem as a contextual multi-armed bandit problem, in which the actions are a set of pre-computed plans, and the state is the position of the free kick on the field. During execution, we model the reward function for different free kicks using Gaussian Processes, and perform online learning using the Upper Confidence Bound algorithm. Results from a physics-based simulation reveal that the robots are capable of adapting to various different realistic opponents to maximize their expected reward during free kicks.","keywords":"adversarial action planning in competitive robotic domains,acquisition of planning models for robotics,learning action and task models","bibtex":"@inproceedings {icaps16-195,\r\n track = {​​Robotics Track},\r\n title = {Online Learning of Robot Soccer Free Kick Plans Using a Bandit Approach},\r\n url = {http://www.aaai.org/ocs/index.php/ICAPS/ICAPS16/paper/view/13180},\r\n author = {Juan Pablo Mendoza and Reid Simmons and Manuela Veloso},\r\n abstract = {This paper presents an online learning approach for teams of autonomous soccer robots to select free kick plans. In robot soccer, free kicks present an opportunity to execute plans with relatively controllable initial conditions. However, the effectiveness of each plan is highly dependent on the adversary, and there are few free kicks during each game, making it necessary to learn online from sparse observations. To achieve learning, we first greatly reduce the planning space by framing the problem as a contextual multi-armed bandit problem, in which the actions are a set of pre-computed plans, and the state is the position of the free kick on the field. During execution, we model the reward function for different free kicks using Gaussian Processes, and perform online learning using the Upper Confidence Bound algorithm. Results from a physics-based simulation reveal that the robots are capable of adapting to various different realistic opponents to maximize their expected reward during free kicks.},\r\n keywords = {adversarial action planning in competitive robotic domains,acquisition of planning models for robotics,learning action and task models}\r\n}\r\n\r\n","author_short":["Mendoza, J. P.","Simmons, R.","Veloso, M."],"key":"icaps16-195","id":"icaps16-195","bibbaseid":"mendoza-simmons-veloso-onlinelearningofrobotsoccerfreekickplansusingabanditapproach","role":"author","urls":{"Paper":"http://www.aaai.org/ocs/index.php/ICAPS/ICAPS16/paper/view/13180"},"keyword":["adversarial action planning in competitive robotic domains","acquisition of planning models for robotics","learning action and task models"],"metadata":{"authorlinks":{}}},"search_terms":["online","learning","robot","soccer","free","kick","plans","using","bandit","approach","mendoza","simmons","veloso"],"keywords":["adversarial action planning in competitive robotic domains","acquisition of planning models for robotics","learning action and task models"],"authorIDs":[],"dataSources":["iMkx859KiXcegwsin","EZtZjCTnxcdTTyeij"]}