Coupled Multiple Dynamic Movement Primitives Generalization for Deformable Object Manipulation

Dynamic Movement Primitives (DMP) are widely applied in movement representation due to their ability encode tasks using generalization properties. However, the coupled multiple DMP cannot be directly solved based on original formula. Prior works provide satisfactory performance for rigid object manipulation, but extension deformable objects may degrade intrinsic uncertainty of model structure and parameters. This letter introduces an adaptive term replace fixed couple generalizations classic mass-spring-damper model. Based modeling, manipulation a can treated as second-order system, which provides additional implementation flexibility robustness transportation. To validate proposed method, we perform extensive simulations cooperatively transporting rope thin film, imitating with semi-ellipse trajectory M-shape trajectory. We further implement our method dual-arm robot platform depth visual feedback. Both simulation experiment results demonstrate generalization, collision avoidance, configuration preservation.