IEEE Robotics & Automation Magazine - December 2019 - 62

capabilities. On the left, a flexible, 4-DoF Hardware Embedded Reduced Intricacy (HERI) II hand [11] allows for a
higher payload and exhibits higher compliance and robustness. The HERI II hand was developed specifically to overcome the limitation of typical underactuated hands in
executing basic dexterous motions, such as pinching and
triggering tools, while maintaining the main advantages of
underactuated designs.
With respect to the sensor setup, KHG reported that the
operation of its robots, based solely on camera images, puts a
high cognitive load on the operators and requires extensive
training to derive correct
situation assessments
from those images. The
A semiautonomous
lack of assessment of the
overall kinematic system
stepping controller is used
state and of the robot's
environment causes
for walking across irregular uncertainty and stress.
Therefore, Centauro
terrain.
incorporates a multimodel sensor setup that provides comprehensive
visualizations to the operators and input to (semi)autonomous control functionalities. It includes a 3D laser scanner
with spherical field of view, an RGB depth (RGB-D) sensor,
and three RGB cameras in a panoramic configuration at the
robot head. A pan-tilt mechanism allows for adaptation of the
RGB-D sensor pose, facilitating manipulation at different
locations. Furthermore, an inertial measurement unit (IMU)
is mounted in the torso. Two RGB cameras under the robot
base provide views on the feet. An RGB-D sensor at the right
wrist creates an additional perspective during manipulation.
Finally, two 6-DoF force/torque sensors are mounted between
the robot arms and hands.
To the best of our knowledge, no system on the market
provides hybrid locomotion combined with bimanual
manipulation and the required sensor setup. The robot has
been mainly fabricated of an aluminum alloy. To drive the
robot, five sizes of torque-controlled actuators have been
developed. The compactly integrated series elasticities provide mechanical robustness to impacts and torque sensing
via deformation monitoring. Based on an effort analysis, a
suitable actuator was chosen for each joint [12]. In addition, the robot body houses a Wi-Fi router, a battery, and
three PCs for real-time control, perception, and higher-level behavior.
One consideration in designing a disaster-response
robot is to include hardening to withstand radiation. This
can be realized through either massive lead cases for the
computational hardware, which would add significant
weight, or the use of radiation-immune computer hardware, which is severely limited, compared to commercial
off-the-shelf (COTS) hardware. We developed the CENTAURO system as a scientific demonstrator that aims at
pushing the state of the art. Hardening the system against
62

IEEE ROBOTICS & AUTOMATION MAGAZINE

DECEMBER 2019

radiation would have affected our research by considerably
limiting the modular hardware design and the use of
cheaper COTS components. We thus see system hardening
as a task for further development according to the demands
of the end user.
Software Description
For low-level hardware communication, we developed
XBotCore, an open-source, real-time, cross-robot software
platform [13]. This software makes it possible to seamlessly
program and control any robotic platform while providing
a canonical abstraction that hides the specifics of the
underlying hardware. Higher-level motion commands are
processed in the developed CartesI/O motion controller
[14], which can consider multiple tasks and constraints
with given, situation-specific priorities. The controller
resolves a cascade of quadratic programming (QP) problems, each associated with a priority level, and considers
the optimality reached in all of the previous priority levels.
The high computational cost of the successive QP problems
is alleviated by assessing our inverse kinematics (IK) loop
in a range between 1,000 and 250 Hz.
The Operator Station
KHG reported that during its real-world missions, human
concentration suffers from a high cognitive load. Thus, operators are usually exchanged approximately every 15 min. The
kinematics of Centauro are considerably more complex,
which means the cognitive load would likely be higher. In discussions with KHG, we identified several reasons for these
high cognitive loads.
1) Camera images are not comprehensive enough to enable
operators to fully understand the robot's situation, especially if contact with the environment has to be assessed.
2) Operators have problems understanding the robot kinematics if they cannot see them and cannot assess joint limits and self-collisions.
3) Operation of the robot requires continuous direct control
of all robot capabilities, which demands an exhausting,
continuous, and high level of concentration.
We derived several requirements for the CENTAURO
operator station from these experiences. Regarding visualization, we aimed at providing a more intuitive understanding of the situation through a 3D virtual model of the robot
and its environment. This model is generated from joint
states, the IMU, the laser scanner, and the RGB-D sensors. It
is enriched with camera images from several perspectives.
We refer to these digital counterparts of the robot and its
environment as digital twins. Moreover, we added force
feedback to improve situational assessment, especially if
environment contacts are involved. Regarding control, several interfaces with different strengths were implemented.
The main operator intuitively controls the robot through a
full-body telepresence suit, while support operators are able
to command additional control functionalities on different
levels of autonomy.

IEEE Robotics & Automation Magazine - December 2019

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2019