Coupled dynamical system based arm-hand grasping model for learning fast adaptation strategies

摘要: Performing manipulation tasks interactively in real environments requires a high degree of accuracy and stability. At the same time, when one cannot assume fully deterministic static environment, must endow robot with ability to react rapidly sudden changes environment. These considerations make task reach grasp difficult deal with. We follow Programming by Demonstration (PbD) approach problem take inspiration from way humans adapt their motions perturbed. This is sharp contrast previous work PbD that uses unperturbed for training system then applies perturbation solely during testing phase. In this work, we record kinematics arm fingers human subjects perturbed motions. demonstrations, target's location changed suddenly after onset motion. Data show strong coupling between hand transport finger hypothesize enables subject seamlessly motion coordination posture. To our competence, develop coupled dynamical based controller, whereby two systems driving are coupled. offers compact encoding reach-to-grasp ensures fast adaptation zero latency re-planning. simulation on iCub smooth ''human-like'' demonstrate performance model under spatial, temporal type perturbations which reaching target coordinated hand-arm necessary success task.