Robotic Manipulation of Elongated and Elastic Objects

While we are observing the increasing number of areas where robots are used to manipulate objects, their perception of shape and physical properties remains highly limited. Today's industrial robots are designed to perform their tasks with very high speed and precision without taking into account the physical parameters of objects. While this approach works effectively in a mass production, any non-standard manipulation-based task, performed on objects that are not precisely of the same shape, still has to be supervised by a human. In our work, we propose a robotic workstation for manipulation of elongated and elastic objects, where the prior knowledge of an object's shape is used to perform an efficient grasp. Grasping is done by a human-like robotic hand with force sensors mounted in fingertips, which enables us to perform exploratory moves to estimate the elasticity modulus of a manipulated object.