IEEE Robotics & Automation Magazine - December 2018 - 25

The work we present here is intended to reduce the num-
ber of robot operators required offshore-hence, reducing
costs and staffing requirements-by facilitating operations
from an onshore control center and narrowing the gap between
low-level teleoperation and full autonomy (Figure 1). The
basic idea is that the user interacts with a real-time simulation
environment, and a cognitive engine (CE) analyzes the user's
control requests and turns them into movement primitives the
ROV needs to autonomously execute in the real environment,
independently of communication latencies.
This article focuses on the results of intensive field trials
held 26 June-7 July 2017 in the Mediterranean Sea near Mar-
seille, France. Seven extended experimental dives were per-
formed with the ROV while connected via satellite to the
command center in Brussels, Belgium. Four different sites
were used with different water depths (8, 30, 48, and 100 m).
System Components
Overview
Our work here is targeted at a high technology readiness
level of six, i.e., it is developed and validated beyond only lab-
oratory experiments. The research vessel Janus II from Comex
with a 2,500-msw rated SubAtlantic Apache ROV is used for
this purpose (Figure 2). For our research, the ship is equipped
with satellite communications, explained in the "Satellite
Communication" section, to allow control of the ROV by
pilots located in a command center in Brussels, as discussed
in the section "Control Center and the Exoskeleton." Fur-
thermore, a skid is added to the ROV to carry additional
components used for our research, i.e., an electric manipu-
lator and two manipulators in a bimanual setup (as
described in the section "Underwater Manipulators") and a
vision system.
Underwater Manipulators
Our manipulator was designed beginning with the underwa-
ter modular arm. Two kinds of electrically driven joints hav-
ing either one or two motion axes are complemented by a set
of links to connect the joints. Different kinematic configura-
tions can be obtained by varying the number of basic modules,
i.e., joints and links, and/or the way they are interconnect-
ed. The arm is characterized by six degrees of freedom
(DoF), obtained by connecting three modules, each with
two DoF forming a pitch-roll configuration. The overall
length when totally stretched is slightly more than 1 m.
However, the arm is also fully foldable to minimize its size
when parked in the ROV skid during the navigation phas-
es. Both a single-arm and a dual-arm setup can be used.
During the 2017 trials, a mockup of grippers currently
under development was used.
Vision System
An intelligent underwater vision system with computing
power on board the ROV is used to minimize the traffic over
the umbilical cable from the ROV to the vessel. The system is

based on high-resolution firewire (IEEE 1934b) cameras in
pressure housings connected to an embedded computer,
which can be used for vision processing and adaptive video
compression on board the ROV.
The firewire bus supports, among others, the synchroniza-
tion of the cameras. They can therefore be used for multicam-
era stereo setups to generate depth information from different
views with a known relative geometry. Because of the Apache
ROV's payload constraints, a stereo setup with two cameras
was used in the 2017 trials. The compute bottle of the vision
system on board the robot also services the core navigation
sensors in the form of a LinkQuest NavQuest 600 P Micro
Doppler velocity log (DVL) and a Xsens MTi-300 inertial
measurement unit (IMU).
Satellite Communication
Satellite communications services for mobile offshore mari-
time operations are associated with bandwidth limitations
(uplink and downlink), inherent delays, and disruptions; in
addition, they require a complex stabilized satellite tracking
antenna. In the context of this research, we employ a mari-
time very-small-aperture terminal from a service provider
(Omniaccess) that includes a Ku-band Cobham Sailor 800
tracking antenna, its controller, and the related modems. The
nominal data bandwidth for the uplink from the vessel is
768 kB/s, and the downlink to the vessel is 256 kB/s, with an
inherent nominal round-trip delay of 620 ms.
Control Center and the Exoskeleton
The onshore control center in Brussels consists of a monitor-
ing and control room that features a double 7-DoF arm and
6-DoF hand-force feedback exoskeleton. It is based, in part,
on a design for the European Space Agency [1] that was fur-
ther improved on in the EU-FP7 project ICARUS [2]. It is
designed as a modular solution, allowing each arm and hand
exoskeleton subsystem to be easily and conveniently con-
nected and removed from the rest of the setup. Furthermore,
it features a passive gravity compensation system that con-
nects to the arm exoskeletons and can be calibrated to com-
pensate for the full mass of the exoskeleton's physical setups
as well as the mass of the user arms. The user is, hence, given
the impression of operating in neutral buoyancy, i.e., as a
diver typically would. This reduces user fatigue during the
ROV operation.
A Test Panel for System Validation
For validation, a test panel was developed; it also served as a
target for the trials to emulate different scenarios, e.g., offshore
oil and gas facilities or the handling of archeological artifacts
(Figures 2 and 3). The panel consisted of three sides, each
equipped with mockup elements. One side was used to test
functionalities in offshore oil and gas interfaces based on the
International Standards Organization 13628 standard includ-
ing, e.g., valves or wet-mate connectors. In addition, a biologic
panel (including mockup corals) and an archeological box
(including mockup ceramics) were included.
december 2018

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

25
IEEE Robotics & Automation Magazine - December 2018

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2018
