IEEE Robotics & Automation Magazine - December 2020 - 13

and effectively approach objects and perform dexterous
grasping tasks, including retrieving objects from deep aper-
tures in overhang environments. This robotic system has the
potential to scale up to make shallow water collection tasks
safer and more efficient.
Deep Sea Challenges
Oceanic exploration is considered a frontier for under-
standing our planet and its changes, searching for new
resources, sustaining populations, and even discovering
novel medical therapies. At present, however, less than
five percent of our oceans has been thoroughly ex--
plored. Oceanic exploration can be costly, dangerous,
impractical, and logistically challenging [1]. During
recent years, demand for robots capable of underwater
exploration and manipulation has grown immensely
and is expected to continue to increase during the com-
ing decades. Researchers and marine enterprises require
reliable and low-cost underwater robots to unveil the
mysteries of the oceans and boost the so-called blue
economy [2].
Underwater Mobile Manipulation
The two main categories of underwater robots are remotely
operated vehicles (ROVs) and autonomous underwater vehi-
cles (AUVs). ROVs are usually employed for interaction tasks,
with manipulators mounted to the main frame to facilitate
remote operation. This subcategory of ROVs, known as the
underwater vehicle-manipulator system (UVMS), is essentially
a mobile robot that performs so-called floating manipula-
tion [3]. This complex task requires a shared controller
between the operator and the vehicle: high-level requests
from the operator to hold a position or interact with an
object are transformed into an actuator command for a
vehicle's thrust and manipulation systems. But optimal
UVMS performance is challenged by the mathematical
complexity of the robotic system, imprecise modeling of
thruster dynamics, limited sensing capabilities, and harsh
and rapidly changing environmental conditions [4].
Researchers have devoted considerable attention to improv-
ing the floating manipulation and control of UVMSs during
recent years. In simulations [5], undisturbed swimming
pools [6], and pools with a current-like disturbance,
researchers have demonstrated control capabilities that have
errors on the order of a few centimeters. Despite these
promising results, most ROVs still need to rest on the sea-
floor for stability when manipulation is required [7].
Another class of underwater vehicles, called benthic
crawlers, facilitates interaction with underwater structures,
without increasing control complexity, by maneuvering on
the seabed via tracks and wheels. They are routinely used for
heavy work duties, shallow water investigations, and longterm monitoring despite the fact that they are limited to
substrates where tracks and wheels can be used [8]. A few
recent examples of legged benthic robots, developed to
withstand high currents [9], [10], move across uneven

terrain [11], operate without disturbing the environment
[12], and work in shallow water, promise to integrate the
advantages of benthic crawlers with the dexterity of legged
robots. With legged systems, researchers envision the possi-
bility of precise and swift seabed interaction without the
added complexity of controlling underwater stability. In
addition, improved visibility can be achieved by the ben-
thic crawler's legs (compared to propellers) while perform-
ing locomotion on sandy substrates. The actions of
propellers often raise sand particles, which can reduce visi-
bility; in our case, precise
and slow positioning of
the legs could prevent
Demand for robots capable
this from happening.

of underwater exploration

Underwater Soft
Manipulation
and manipulation has
A robot's manipulator sys-
tem is another essential
grown immensely and
component to facilitate
optimal underwater per-
is expected to continue
formance. Traditional
underwater hydraulic ro--
to increase.
botic arms and grippers
are designed for missions
with heavy payloads and
high levels of force [13]. These metal components are poorly
suited for grasping fragile and squishable objects. Addi-
tionally, a massive metal arm has a large inertia, making
underwater mobility challenging for a robot in unsteady
conditions. Soft materials and bioinspired structures have
immense potential to integrate flexible, lightweight ele-
ments into manipulators to reduce the mass and decrease
the chance of damaging fragile objects. For example, pre-
vious work on an octopus-inspired soft arm achieved
tethered motion under water [14], and a dexterous subsea
hand [15] has been tested as a gripper. More recently,
studies on soft robotic manipulation have begun to focus
on underwater applications [16], [17], a silicone-rubber
gripper [18], a modular soft robotic wrist [19], a dexter-
ous glove-based soft arm [20], and a jamming gripper
[21] have all been tested to grasp delicate underwater
organisms at shallow-to-deep-sea depths. Thus, a soft
manipulator-a combination of a soft arm and a grip-
per-may be a practical tool for real-world mobile under-
water manipulation.
In this article, we investigate, for the first time, an integrat-
ed mobile benthic platform and a soft manipulator (Figure 1).
For the benthic platform, we use SILVER2 [22], a six-legged,
crab-inspired robot developed for exploration and environ-
mental monitoring. For the manipulator, we use a soft device
with a four-fingered soft gripper that can move in a 3D
domain and perform delicate grasping. We experimentally
employ spatial manipulation with inverse kinematics specifi-
cally for collecting tasks in a natural underwater environ-
ment. Both systems have been separately designed and
DECEMBER 2020

IEEE ROBOTICS & AUTOMATION MAGAZINE

IEEE Robotics & Automation Magazine - December 2020

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