Dimensionality Reduction for Dynamic Movement Primitives and Application to Bimanual Manipulation of Clothes

• Published in: IEEE transactions on robotics 2018-06, Vol.34 (3), p.602-615
• Main Authors: Colome, Adria, Torras, Carme
• Format: Article
• Language: English
• Subjects: Artificial intelligence; Classificació INSPEC; Couplings; Covariance matrices; Cybernetics; Dimensionality reduction; Indexes; Informàtica; Intel·ligència artificial; learning (artificial intelligence); Learning curves; Linearity; Machine learning; machine learning algorithms; manipulators; Parameter robustness; Parameterization; Reduction; Rescaling; Robot dynamics; Robot kinematics; robot learning; robot motion; Task analysis; Trajectory; Àrees temàtiques de la UPC
• ISSN: 1552-3098; 1941-0468
• DOI: 10.1109/TRO.2018.2808924

Dynamic movement primitives (DMPs) are widely used as movement parametrization for learning robot trajectories, because of their linearity in the parameters, rescaling robustness, and continuity. However, when learning a movement with DMPs, a very large number of Gaussian approximations needs to be performed. Adding them up for all joints yields too many parameters to be explored when using reinforcement learning (RL), thus requiring a prohibitive number of experiments/simulations to converge to a solution with a (locally or globally) optimal reward. In this paper, we address the process of simultaneously learning a DMP-characterized robot motion and its underlying joint couplings through linear dimensionality reduction (DR), which will provide valuable qualitative information leading to a reduced and intuitive algebraic description of such motion. The results in the experimental section not only show that we can effectively perform DR on DMPs while learning, but we can also obtain better learning curves, as well as additional information about each motion: linear mappings relating joint values and some latent variables.