A method based on support vector regression (SVR) is proposed to learn the batting policy to return the ball to a desired location for a 7-DoF (degree of freedom) anthropomorphic table tennis robot. Firstly, table tennis playing process is formalized as the batting evaluation function, which maps the state of the incoming ball and the parameters of the batting trajectory to the reward. Then, an exploration method based on the confidence region of the physical model is proposed to collect training data efficiently, and the batting evaluation function is obtained by generalizing the training data using ε-support vector regression (ε-SVR). Finally, the optimal batting trajectory is computed during decision process by maximizing the batting evaluation function using multi-start Quasi-Newton method. The proposed method is applied to a 7-DoF table tennis robot, and the results verifies its effectiveness.
 Andersson R L. A robot ping-pong player: Experiments in real-time intelligent control[M]. Cambridge, USA: MIT Press, 1998. Miyazaki F, Takeuchi M, Matsushima M, et al. Realization of the table tennis task based on virtual targets[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA: IEEE, 2002: 3844-3849. Matsushima M, Hashimoto T, Takeuchi M, et al. A learning approach to robotic table tennis[J]. IEEE Transactions on Robot-ics, 2005, 21(4): 767-771.  Mülling K, Peters J. A computational model of human table tennis for robot application[C]//2009 Autonome Mobile System. Dordrecht, Netherlands: Springer, 2009: 57-64. Kober J, Mülling K, Kromer O, et al. Movement templates for learning of hitting and batting[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA: IEEE, 2010: 853-858. Mülling K, Kober J, Kromer O, et al. Learning to select and generalize striking movements in robot table tennis[J]. International Journal of Robotics Research, 2013, 32(3): 263-279.  Xiong R, Liu Y, Zheng H B. A humanoid robot for table tennis playing[C]//IEEE Workshop on Advanced Robotics and Its Social Impacts. Piscataway, USA: IEEE, 2011: 66-67. Xiong R, Sun Y C, Zhu Q G, et al. Impedance control and its effects on a humanoid robot playing table tennis[J]. International Journal of Advanced Robotic Systems, 2012, 9(9): 1-11. 苏虎, 徐德, 黄艳龙, 等.基于模糊自调节算法的乒乓球机器人回球速度计算[J].自动化学报, 2012, 38(6):923-931. Su H, Xu D, Huang Y L, et al. The computation of desired ball velocity after striking based on self-tuning fuzzy algorithm for robotic table tennis[J]. Acta Automatica Sinica, 2012, 38(6): 923-931. Yang P, Zhang Z T, Wang H W, et al. Design and motion control of a ping pong robot[C]//8th World Congress on Intelligent Control and Automation. Piscataway, USA: IEEE, 2010: 102-107. 张奇, 谢宗武, 刘宏, 等.基于无源性理论的柔性关节控制器设计[J].机器人, 2013, 35(1):23-31. Zhang Q, Xie Z W, Liu H, et al. Flexible joint controller design based on passivity theory[J]. Robot, 2013, 35(1): 23-31. Ramanantsoa M, Duray A. Toward a stroke construction model[J]. International Journal of Table Tennis Sciences, 1994, 2: 97-114. Liégeois A. Automatic supervisory control of the configuration and behavior of multibody mechanisms[J]. IEEE Transactions on Systems, Man and Cybernetics, 1977, 7(12): 868-871.  Smola A J, Schölkopf B. A tutorial on support vector regression[J]. Statistics and Computing, 2004, 14(3): 199-222.  Schölkopf B, Smola A J. Learning with kernels: Support vector machines, regularization, optimization, and beyond[M]. Cambridge, USA: MIT Press, 2001. Sutton R S, Barto A G. Reinforcement learning: An introduction[M]. Cambridge, USA: MIT Press, 1998. Carreira-Perpinan M A. Mode-finding for mixtures of Gaussian distributions[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(11): 1318-1323.  Chang C C, Lin C J. LIBSVM: A library for support vector machines[J]. ACM Transactions on Intelligent Systems and Technology, 2011, 2(3): 27. Free Software Foundation, Inc.. GNU scientific library reference manual[M/OL]. 3rd ed. (2011-05-06)[2013-04-16]. http://www.gnu.org/software/gsl.