IEEE Robotics & Automation Magazine - December 2015 - 70

An overview of techniques for three-dimensional (3-D)
object grasp synthesis with multifingered robotic hands is proposed in [3], and a focus on analytical and empirical
approaches is provided in [4]. A more general overview of
grasping is presented in [5].
The offline computation of a context-independent system
(i.e., where only the robot gripper and the target object are
considered) and a dense set of grasp configurations instead of
a small set of configurations, which is regarded as optimal
using a given criterion, are proposed in [9]. The achieved set is
then used online, allowing the robot to quickly choose a suitable grasp for a given situation.
The combination of several grasp quality indices for the
synthesis of optimal grasp configurations has been employed
by a number of methods proposed in the literature. Several
nondimensional performance indices are proposed in [11]
and [12], where the problem of merging quality indices with
different physical meanings is specifically addressed.
The computational complexity of the optimal grasp search
algorithms is an important aspect to take into account. In the
case of polyhedral objects,
which are the most investigated ones, the evaluaThe ability of a robotic
tion of the force-closure
regions can be cast into a
hand to guarantee a firm
linear programming
problem [15], [16], which
grasp is an essential
is computationally efficient, whereas the optimal
requirement for the
grasp configuration can
be chosen by solving a
manipulation of an object.
nonlinear programming
problem [17]. In [18], a
solution for grasp synthesis and fixture layout design in the
discrete domain is presented.
In [19], the effect that the number and the type of
engaged postural synergies has on the choice of grasping
forces and on the ultimate quality of the grasp are investigated. Numerical results have been presented showing the
role played by different synergies in making a number of different grasps possible. Anthropomorphism is exploited in
[20] and [21] for the development of a humanlike grasping
approach based on the synergic motions that can be
observed in the human hand. In [22], through the analysis of
a quasi-static model, grasp structural properties related to
contact force and object motion controllability are defined.
Based on these recent results, we can see that, on one hand,
new compact grasp quality indices can be derived and
employed for the grasp synthesis, and, on the other hand, the
grasping of unknown objects can be achieved by employing
complaint models, e.g., soft synergies, or suitable mechanical
designs to control the interaction forces and ensure both
force closure and manipulability of the grasp.
In [24], the finger-positioning error during grasping and
its consequence to the force-closure property are considered. In particular, the concept of independent contact
70

IEEE ROBOTICS & AUTOMATION MAGAZINE

DECEMBER 2015

regions (ICRs) has been proposed to provide robustness to
the grasp, i.e., the force closure is guaranteed when finger
contact occurs anywhere inside each of these regions,
despite the exact contact position. A realistic modeling of
the contact between the robotic fingers and an object is
used in [25] for the synthesis of multifingered grasps. A
patch contact model is adopted to locally approximate the
contact between a rigid object and a deformable finger as a
set of ICRs.
In this article, the work presented in [26] and [27] is
specialized to the case of anthropomorphic robotic hands
to quickly synthesize optimal n-fingered grasps for 3-D
objects with any shape, ensuring the minimization of gravitational and inertial effects. The computational complexity
is reduced by performing a suitable discretization of the
object surface and a simultaneous selection of the pieces of
surfaces suitable to generate optimal grasps. Unlike existing
approaches, this step is made under the lead of constraints
derived from important grasp quality indices that allow a
strong reduction of the computational complexity. In
detail, the presented method requires an initial evaluation
of regions that can generate grasps with minimal gravitational and inertial effects. The achieved set of regions is
further discretized on the basis of the local surface curvature, resulting in a set of groups of regions with uniform
curvatures (e.g., planar regions, concave regions, convex
regions, and angular regions). The fingertip size and the
object model uncertainty are considered to further decompose the achieved regions. The search space of the optimization algorithm is, hence, reduced by applying linear
constraints derived from the kinematics of the anthropomorphic hand and by considering the peculiarity of the
thumb. Finally, a set of grasp quality indices is applied,
ranking the grasps according to a serial approach that takes
into account the computational complexity of the chosen
quality measures. The proposed method is particularly useful for the manipulation of heavy objects, compared with
the hand capabilities, because of the choice of the inertial
quality metrics. A number of objects with several shapes
are presented to show the effectiveness of the proposed
approach in terms of both the computational time and
quality of the grasp configuration with respect to a given
anthropomorphic hand.
A MATLAB demo toolbox endowed with wizards to
generate both suitable 3-D MATLAB models from standard
stereolithography (STL) files and optimal grasps to test the
proposed approach is available at: http://wpage.unina.it/
lippiell/docs/pubs/OGS_RAM2013_ Matlab_Demo.zip.
Wizards have been provided both for the construction of
suitable MATLAB 3-D models from standard STL files and
for the optimal grasp synthesis. All of the STL files of the
employed 3-D models are included to recreate the figures in
this article. Note that this code is designed for demonstration purposes only and is not optimized like the one written
in C++ that is employed for the performance tests of the
article. However, although some simplifications have been

http://wpage.unina.it/

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2015