IEEE Robotics & Automation Magazine - September 2020 - 57

and other fixtures may hinder robot motion or cause
undesired collisions.
●● Grocery handling for packaging: A very common example is the recent practice of dark storing, whereby
goods are well organized but come in a wide variety of
shapes, textures, weights, and sizes. They are often delicate and can be easily damaged by improper manipulation [Figure 2(b)].
●● Raw-food handling: As in grocery handling, the shape,
weight, size, orientation, and stiffness of the objects differ
extensively after harvesting [Figure 2(c)]. Furthermore,
raw food is usually placed into crates without any particular organization.
●● Waste sorting: In this task, the irregularity and randomness
of the objects are extreme, as are the physical properties
(density, weight, shape, stiffness, and so on) of items, such
as cardboard, packaging materials, glass fragments, and
hard pieces of scrap metal or wood [Figure 2(d)].
All of these cases share the possible issue of inaccuracy
in the detection of the object pose. This may cause a grasp
to fail or, in some cases, generate large interaction forces
that can damage the robot or object. These different use
contexts can present several possibilities in terms of object
placement. Sometimes, even in the same context, objects
can be placed randomly or in a fixed and predefined configuration. This is the case, for example, with food handling in dark storage areas. In this scenario, objects can be
placed in bins randomly (e.g., zucchini or bananas) or in
specific grids (e.g., apples or peaches). Such variability is
one of the leading factors for the need for versatile handling systems.
In recent years, these tasks have received more and
more attention from the robotic community. Soft Robotics
Inc. [25] develops soft grippers for bin picking and rawfood handling, while Ocado [26] proposes solutions for
raw-food and grocery handling. Amazon Robotics [27]
sets challenges for improving technologies for grocery
handling. ZenRobotics Ltd. [28] designs waste-sorting systems, and Pick-it N.V. [29] develops vision systems for
bin-picking tasks.
In this article, although we present results in the fields of
grocery handling and bin picking, we especially focus on

(a)

(b)

raw-food handling, applying the proposed system to this scenario in particular. In the state of the art, there are several
investigations related to this topic [18]-[20]. In [20], a robotic
system for the manipulation of onions and artichokes is presented; the robot consists of a rigid delta robot with a vacuum
suction cup. The differences with our solution are that our
manipulator has flexible joints and, thus, is able to exert lower
interaction forces. Furthermore, the end effector used in [20]
is specifically designed for onions and artichokes, whereas our
solution can be applied to several different objects and does
not rely on suction systems; therefore, it does not require
pressured air to function.
Historically, suction grippers have been broadly used in
the contexts discussed in this article, but such systems present
limits when grasping porous objects, such as as fresh fruit and
vegetables, or objects with slippery surfaces, like those in in
waste-sorting scenarios (i.e., in the presence of humidity or
dust) or bin picking (i.e., with oils or moisture). Moreover
suction grippers often need a different design or control strategy depending on the item to be grasped. To grasp heterogeneous objects and develop versatile systems, it is our opinion
that an adaptive fingered gripper is preferable to end effectors
based on suction technologies.
SoftHandler Design
The SoftHandler is obtained by the combination of a novel
soft articulated end effector, the Pisa/IIT SoftGripper, and
parallel manipulator equipped with soft actuators, the SoftDelta. The entire system is then mounted on a 1 × 1 × 1-m
rigid structure. In the following sections, we describe the
design of each component as well as present and discuss the
control architecture and the different control algorithms that
can be employed on the system. We also consider different
circumstances and scenarios of use.
Pisa/IIT SoftGripper
Simplicity (i.e., underactuation), robustness, and adaptivity
(i.e., softness and compliance) are the main features that led to
the development of the Pisa/IIT SoftHand [2]. The same
driving ideas underlie the design of the Pisa/IIT SoftGripper
[Figure 3(a)]: the 12 degrees of freedom that make up this
device are driven by only one degree of actuation, while the

(c)

(d)

Figure 2. Some examples of industrial pick-and-place tasks that can benefit from the use of soft robots: (a) bin picking, (b) grocery
handling, (c) raw-food handling, and (d) waste sorting. Handling heterogeneous, unordered, or fragile objects is a challenge for
traditional rigid robots.

SEPTEMBER 2020

IEEE ROBOTICS & AUTOMATION MAGAZINE

IEEE Robotics & Automation Magazine - September 2020

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2020