Makondo, NRosman, Benjamin SHasegawa, O2015-11-302015-11-302015-11Makondo, N, Rosman, B.S. and Hasegawa, O. 2015. Knowledge transfer for learning robot models via local procrustes analysis. In: IEEE-RAS International Conference on Humanoid Robots, Seoul, Korea, November 3-5, 2015http://www.benjaminrosman.com/papers/humanoids15.pdfhttp://hdl.handle.net/10204/8314IEEE-RAS International Conference on Humanoid Robots, Seoul, Korea, November 3-5, 2015Learning of robot kinematic and dynamic models from data has attracted much interest recently as an alternative to manually defined models. However, the amount of data required to learn these models becomes large when the number of degrees of freedom increases and collecting it can be a timeintensive process. We employ transfer learning techniques in order to speed up learning of robot models, by using additional data obtained from other robots. We propose a method for approximating non-linear mappings between manifolds, which we call Local Procrustes Analysis (LPA), by adopting and extending the linear Procrustes Analysis method. Experimental results indicate that the proposed method offers an accurate transfer of data and significantly improves learning of the forward kinematics model. Furthermore, it allows learning a global mapping between two robots that can be used to successfully transfer trajectories.enTransfer learningRobot kinematicsRobot dynamicsModel transferProscutes analysisConference PresentationMakondo, N., Rosman, B. S., & Hasegawa, O. (2015). Knowledge transfer for learning robot models via local procrustes analysis. http://hdl.handle.net/10204/8314Makondo, N, Benjamin S Rosman, and O Hasegawa. "Knowledge transfer for learning robot models via local procrustes analysis." (2015): http://hdl.handle.net/10204/8314Makondo N, Rosman BS, Hasegawa O, Knowledge transfer for learning robot models via local procrustes analysis; 2015. http://hdl.handle.net/10204/8314 .TY - Conference Presentation AU - Makondo, N AU - Rosman, Benjamin S AU - Hasegawa, O AB - Learning of robot kinematic and dynamic models from data has attracted much interest recently as an alternative to manually defined models. However, the amount of data required to learn these models becomes large when the number of degrees of freedom increases and collecting it can be a timeintensive process. We employ transfer learning techniques in order to speed up learning of robot models, by using additional data obtained from other robots. We propose a method for approximating non-linear mappings between manifolds, which we call Local Procrustes Analysis (LPA), by adopting and extending the linear Procrustes Analysis method. Experimental results indicate that the proposed method offers an accurate transfer of data and significantly improves learning of the forward kinematics model. Furthermore, it allows learning a global mapping between two robots that can be used to successfully transfer trajectories. DA - 2015-11 DB - ResearchSpace DP - CSIR KW - Transfer learning KW - Robot kinematics KW - Robot dynamics KW - Model transfer KW - Proscutes analysis LK - https://researchspace.csir.co.za PY - 2015 T1 - Knowledge transfer for learning robot models via local procrustes analysis TI - Knowledge transfer for learning robot models via local procrustes analysis UR - http://hdl.handle.net/10204/8314 ER -