# Dual-arm Manipulation of Elastic Rods

### Wednesday, November 20th, 2019, 16:10

### Checkpoint 480

### Dual-arm Manipulation of Elastic Rods

### Avishai Sintov, TAU

### Abstract:

Industrial manipulation of rigid objects has been automated for quite a long time, while the handling of deformable objects is usually done manually due to the lack of feasible motion planning algorithms. Motion planning algorithms have mainly focused on the manipulation of rigid bodies by one or more robots. Consequently, much less attention has been given to motion planning for the manipulation of deformable objects in general, and elastic rods in particular.

In the talk I will use an analytical description of the configuration space of elastic rods to present a motion planning algorithm for robotic manipulation that is easy to implement and works well in practice. Previous work has shown that the configuration space of an elastic rod, i.e., the set of all equilibrium configurations, is a six-dimensional smooth manifold. This remarkable discovery enables the use of standard sampling-based motion planners to easily plan the manipulation of such rods with two robotic arms. However, this usually results in high computational costs due to a large amount of time spent in solving ordinary differential equations. Hence, I will present an advanced approach to plan motions over two dimensional slices in the free configuration space of the rod. These slices are based on the scale invariance property of the rod, they show that the configuration space of a rod is path-connected and they enable low-cost computations.

In the talk I will use an analytical description of the configuration space of elastic rods to present a motion planning algorithm for robotic manipulation that is easy to implement and works well in practice. Previous work has shown that the configuration space of an elastic rod, i.e., the set of all equilibrium configurations, is a six-dimensional smooth manifold. This remarkable discovery enables the use of standard sampling-based motion planners to easily plan the manipulation of such rods with two robotic arms. However, this usually results in high computational costs due to a large amount of time spent in solving ordinary differential equations. Hence, I will present an advanced approach to plan motions over two dimensional slices in the free configuration space of the rod. These slices are based on the scale invariance property of the rod, they show that the configuration space of a rod is path-connected and they enable low-cost computations.