An Interactive Machine Learning (IML) approach for training a dribbling engine for humanoid biped robots in RoboCup competitions (Standard Platform League) is presented. The proposed dribbling approach solves two decision problems: the determination of the dribbling direction and the calculation of the walking velocities required for pushing the ball toward the desired direction. Moreover, the prediction of the position of moving balls is used for improving the dribbling performance, when it is needed to intercept a moving ball. A combination of batch and incremental learning is used for shaping the policies of the dribbling controller. Results obtained from previous RoboCup competitions, and also from specific experiments, validate the proposed methods.
CITATION STYLE
Celemin, C., Perez, R., Ruiz-Del-Solar, J., & Veloso, M. (2018). Interactive machine learning applied to dribble a ball in soccer with biped robots. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11175 LNAI, pp. 363–375). Springer Verlag. https://doi.org/10.1007/978-3-030-00308-1_30
Mendeley helps you to discover research relevant for your work.