IEEE Robotics & Automation Magazine - December 2015 - 53

A different design approach is inspired by biological systems. In particular, several studies demonstrated that, notwithstanding the human hand complexity, only a limited set
of variables are able to describe most of the variance of hand
postures during grasping tasks. Neuroscientists often refer to
this low-dimensional representation of hand postures as postural synergies [5].
Reducing the number of control inputs from a fully actuated
to an underactuated or joint-coupled solution obviously limits
the dimension of the force and motion controllability subspaces,
reducing the dexterity of the given robotic grasp. If we define the
actual joint variables as a linear combination of as few inputs as
those defined by postural synergies, as in [6] and [7], hand postures would rigidly lie on a linear manifold and the hand would
rigidly move along given directions in the joint space. In this
case, it would be impossible for the hand to adapt its posture to
grasp objects with different shapes and dimensions. An
improvement in the adaptability and robustness level of robotic
hands can be obtained using a compliant definition of synergies.
Soft synergies, therefore, represent a joint displacement aggregation corresponding to a reduced dimension representation of
the hand reference movements. The actual hand posture will
differ from the reference one because of its compliance. We
defined this type of hand joint coupling as compliant postural
synergies or soft synergies in [8] and [9].
The toolbox, presented in its preliminary version in [10],
is particularly suited to the modeling and analysis of underactuated hands, hands with joint coupled movements, and
soft synergy-actuated hands, but it can be easily adapted to
fully actuated hands. Several compliance sources can be
modeled with SynGrasp: 1) contact compliance due to the
fingertip and object deformation, 2) joint compliance due to
mechanical structure deformations and actuator static control gains, and 3) synergy compliance induced by soft synergy organization of hand motions.
The availability of numerical tools for the definition of the
main grasp properties is useful to analyze and compare the
performance of the existing solutions and also to design and
optimize new devices. Such tools are also useful to analyze
and better understand the behavior of the human hand when
performing a grasp operation. For instance, it is possible to
evaluate the contact force distribution in grasping, how the
synergies influence grasp quality, and so on.
Simulation tools have a long tradition in robot grasping,
especially for the evaluation of grasp planning algorithms.
Among those, it is worth mentioning GraspIt! [11] and
OpenGrasp [12] for the specificity of their functions. The former has been frequently used in the literature and includes
several models of robot hands, allowing for the study of the
dynamics of contacting bodies and a grasp planner for a
Barrett hand model. OpenGrasp is built upon the OpenRAVE
simulator [13] and includes a robot hand editor together with
a physics abstraction layer, enabling an interface to different
physics engines. Both of these simulators allow for the definition of complex environments, everyday life objects, and different manipulation platforms, but they do not embed a large

set of functions for grasp analysis, especially if underactuated
hands are considered.
SynGrasp is particularly suitable for sensitivity analysis
and optimization procedures due to the limited set of instructions necessary to define a hand and a grasp using the available functions. The toolbox includes several functions for
grasp analysis and optimization that are usually not included
in other software packages, e.g., controllable internal force
subspace, rigid body motion subspace, grasp stiffness, and
manipulability ellipsoids.
We used the MATlAB [14] platform since this language
is widely diffused in the scientific and technical world. The
MATlAB programming environment allows us to easily
exploit other specific
tools and built-in math
functions, enabling the
The toolbox, presented in
exploration of multiple
approaches and integraits preliminary version in
tion with other analysis
tools, e.g., statistical pro[10], is particularly suited
cessing of experimental
data, optimization, and
to model and analyze
dynamic models and
simulations.
underactuated hands.
Moreover, MATlAB is
well known outside the
robotics community,
which makes our toolbox useful in other fields, such as for
experiment design and validation in neuroscience, physiology,
and haptics.
Overview of the Toolbox
How to Use SynGrasp
There are two possible ways to use SynGrasp: 1) in scripting mode or 2) with a GUI. The former allows the user to
write MATlAB in scripts including SynGrasp functions.
This solution is preferred if customization is needed. The
user can add his or her own functions and/or modify those
already existing. A list of the main functions provided with
the toolbox is shown in Table 1 and will be detailed in the
rest of this section. A complete description of all of the
toolbox functions and their usage is provided in [15].
A set of functions can be used in scripting mode to let the
user have a simple graphical representation of the manipulator and the object. The function SGplotHand() draws the
hand in the configuration specified by the user defining the
joint variable values. The function SGhandFrames()plots a
scheme of the hand, highlighting its kinematical structure,
joints, links, and orientation of the local frames for each link.
It is also possible to draw some simple objects to be grasped.
We included a sphere, a cylinder, and a cube that can be
drawn using SGplotCylinder(), SGplotSphere(), and
SGplotCube(), respectively.
The GUI allows the user to load a hand structure and
interactively perform hand and grasp analysis. As shown in
DECEMBER 2015

IEEE ROBOTICS & AUTOMATION MAGAZINE

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