An invited session on SICURA has been organized at the national meeting SIDRA'10 held in L'Aquila, on September 13, 2010.

There were presentations by the five project partners, illustrating selected research results obtained during the second year of the project and on-going activities. The session was well attended also by other members of the italian robotics and automatic control communities.

Pdf or pps presentations are available for download in the right column (sorry, none of the shown videos). 
Actual presenters are highlighted.

presentations	pdf/pps
UNIBO Università di Bologna  "Advances in design of a dexterous anthropomorphic hand based on a compliant structure" Giovanni Berselli, Gabriele Vassura The presentation wil cover the main issues of the development of a robotic hand explicitely oriented to anthropomorphic design.Two main topics will be described in detail:  the optimization of articulated finger structures based on compliant integrated joints and the characterization of compliant pads for cover of robotic limbs. A complete prototype of a hand integrated into a purposely designed robotic forearm will be presented and first experimental results will be commented.
UNINA Università di Napoli Federico II "Fast visual grasp of unknown objects with robotic multi-fingered hands" Vincenzo Lippiello, Fabio Ruggiero, Bruno Siciliano, Luigi Villani In this work a control algorithm for fast visual grasping of unknown objects with a multi-fingered hand is investigated. The algorithm is composed of an object surface reconstruction algorithm and a kinematic motion control, evolving in parallel, providing the input for the low-level interaction controller. The reconstruction algorithm makes use of images taken by a camera carried by the robot, and virtually places an elastic elliptical reconstruction surface, whose axes and dimensions are assigned by a preshaping process, around the object. The surface is let to evolve dynamically under the action of reconstruction forces so as to shrink toward the object until some parts of the surface intercept the object visual hull. This process shapes the surface around the unknown object. At the same time, the motion controller moves the fingertips on the current available reconstruction surface, achieving a planar equilateral grasp according to suitable hand kinematic indices. The fingers keep moving towards local minima depending on the evolution of the reconstruction surface deformation. Experiments are presented, showing the effectiveness of the proposed algorithm.
UNIPI Università di Pisa "On the problem of measuring variable impedance" Giorgio Grioli, Alessandro Serio, Antonio Bicchi The need for adaptability to the environment, energy conservation, and safety during physical interaction with humans of many advanced robotic applications has prompted the development of a number of Variable Stiffness Actuators (VSA). These have been implemented in a variety of ways, using different transduction technologies (electromechanical, pneumatic, hydraulic, but also piezoelectric, active polymeric, etc.) and arrangements with elastic elements. All designs share a fundamentally unavoidable nonlinear behavior. The control schemes proposed for these actuators typically aim at independently controlling the position (or force) of the link, and its stiffness with respect to external disturbances. Although effective feedback control schemes using position and force sensors are commonplace in robotics, control of stiffness is at present completely in open loop. In practice, instead of measuring stiffness, it is inferred from the mathematical model of the actuator. Being this in most cases only roughly known, model mismatches affect severely stiffness control, undermining its utility. It should be noticed that, while for constant stiffness elements an accurate calibration of the model is possible, the same approach is hardly viable for variable stiffness systems. We propose a method for estimating the time-varying impedance of a VSA system. The presented method is an extension of a method proposed in a previous work of the same authors: it integrates a stiffness observer with an Extended Kalman Filter for estimation of mass and damping coefficient. Using instantaneous measurements of force and position at one of the ends of the system, it derives a measure that converges to the current value of mechanical impedance. Some preliminary results are provided, illustrating the performance of the proposed measurement method.
UNIROMA1 Università di Roma "La Sapienza" "Stiffness estimation for flexible transmissions" Fabrizio Flacco, Alessandro De Luca We present a new approach to estimate the stiffness of flexible motion transmission elements in robots intended for safe HRI. The unknown stiffness may be constant (such as in elastic joint robots), may have a nonlinear characteristic, or be variable over time (as in VSA devices). The stiffness estimator is based on the knowledge of the dynamic parameters on the motor side (inertia and viscous friction coefficient) of the joint and uses the motor and link position and velocity measurements. Therefore, link dynamics and external torques need not to be known, nor an extra joint torque sensor or acceleration sensing/estimation are necessary. Two different strategies are considered, an off-line stiffness estimator based on a parametric model and an on-line black-box stiffness estimator, both consisting of a two-stage algorithm. For the off-line model-based estimator, the first stage uses a suitable residual (as in fault detection methods), which is a first-order filtered version of the flexibility torque of the transmission; in the second stage, a weighted least-square fitting method is used to estimate the parameters of the stiffness model. Instead, the on-line estimation method uses a second-order residual generator, which is a filtered version of the joint  stiffness multiplied by the transmission deformation rate, followed by a weighted regressor to identify the current stiffness value. In this second strategy, the stability of the regressor has been analyzed and enforced by resorting to a simple dynamic saturation scheme. Both strategies can be conveniently used to identify the total stiffness of VSA antagonistic devices, by estimating the stiffness of each transmission independently and then merging the results. Simulations performed with constant, nonlinear or variable stiffness transmissions demonstrate the effectiveness of the approach. The estimated stiffness can be used within recently developed nonlinear decoupling feedback laws for the simultaneous control of motion and stiffness.
UNIROMA2 Università di Roma "Tor Vergata" "Linearization of robotic manipulators subject to impacts" Antonio Tornambè Fully actuated robotic manipulators are probably the first physically motivated example of a dynamic system that can be linearized by state/input transformations. Nevertheless, the presence of impacts may destroy the linear behavior thus obtained. In this presentation, motivated by the case of robotic manipulators, it is presented a class of nonlinear impulsive systems for which the joint linearization of both the continuous and discrete dynamics is addressed geometrically, using the concept of orbital symmetry. Some examples in the robotic area are presented.