Coupled Dynamical System Based Hand-Arm Grasp Planning under Real-Time Perturbations

Abstract

Abstract-Robustness to perturbation has been advocated as a key element to robot control and efforts in that direction are numerous. While in essence these approaches aim at "endowing robots with a flexibility similar to that displayed by humans", few have actually looked at how humans react in the face of fast perturbations. We recorded the kinematic data from human subjects during grasping motions under very fast perturbations. Results show a strong coupling between the reach and grasp components of the task that enables rapid adaptation of the fingers in coordination with the hand posture when the target object is perturbed. We develop a robot controller based on Coupled Dynamical Systems that exploits coupling between two dynamical systems driving the hand and finger motions. This offers a compact encoding for a variety of reach and grasp motions that adapts on-the-fly to perturbations without the need for any re-planning. To validate the model we control the motion of the iCub robot when reaching for different objects.

Learning by Demonstration and Robust Control of Dexterous In-Hand Robotic Manipulation Skills

“…Both approaches modify the DMP formulation by adding a coupling term that affects the behavior of the second manipulator. Another related approach [13] based on SEDS representation, uses the state of one manipulator as the phase of another to achieve coupling. In their application, the evolution of hand system is driven by the arm system.…”

Modeling of movement control architectures based on motion primitives using domain-specific languages

“…b) Catching objects in flight [13]: Catching objects with a catching point not located at the center of mass and highly non-linear dynamics, e.g. a tennis racket or a halffilled water bottle.…”

“…The analysis covered the range of point-to-point [12], [13], [17] and periodic movements [12], [17] modeled by autonomous dynamical systems and assessed functional and non-functional properties of their diverse Adaptive Modules and their implementations. Functional properties in this survey covered parametrization, data representation, learning algorithms (e.g.…”

“…To allow adaption to new situations, new environments and additional learning input, Adaptive Modules optionally contain one or more Learners that adapt the Dynamical Systems. A common case for adaption is to shape the dynamics as done with kinesthetic teaching in the catching [13] and upper body control [17] examples. The Adaptive Module concept has ingoing connections from Spaces to receive goals during execution, sensor values or learning data (optional).…”

“…The state reported by Adaptive Components represents the status of the performed movement: i) converged / tracking: The reference of the movement is reached (in case of a goal-directed movement) or tracked (in case of a periodic movement) and ii) ongoing: Movement is still ongoing. This can be used for goal directed movements as done in the catching example [13].…”