IEEE Robotics & Automation Magazine - September 2021 - 26

injector onto the patient's arm, a task in which haptic feedback
plays an important role. With his or her left arm, the
teleoperator can then press a button on the injector device
to administer the injection [Figure 13(e)-(g)]. In this demonstration,
no real glucose injector was used: imitation was
employed that paints a dot on the patient's arm when the
button is pressed successfully. In Figure 13(h), the teleoperator
places the injector onto the kitchen counter.
One day, robotic systems
like Justin and EDAN will
support patients and
caregivers and relieve
them from repetitive and
exhausting tasks that
include no or few social
interactions.
Conclusions and Future Work
The use of assistive robots to support caregivers and those
in need of care is an aspirational goal in robotics research.
Care is based on the interpersonal relation and communication
of a human caregiver and the person in need. Work
in care robotics involves a large number of stakeholders:
primary (users/patients, clinicians, and caregivers), secondary
(engineers, environmental service workers, and health
administrators), and tertiary
(policy makers,
insurers, and advocacy
groups) [2]. It is our aim
to respect this large variety
of stakeholders and
involve them in the eventual
development as far
as possible.
People in different life
situations demand individual
levels of support
and, thus, tailored solutions.
Furthermore, providing
support for this
vulnerable part of society
demands high requirements
for the safety, reliability,
and functionality
of assistive technologies.
The human individual must be kept at the center throughout,
and robotic solutions need to be designed accordingly.
One day, robotic systems like Justin and EDAN will support
patients and caregivers and relieve them from repetitive
and exhausting tasks that include no or few social
interactions. Using the time, availability, and energy gained
by this support, better-quality care can be delivered. We
certainly do not aim to replace human caregivers. A more
complex and in-depth discussion of the ethical compliance
of robotic systems in caregiving is vital for bringing robotic
assistants into application, but it is beyond the scope of
this article.
To contribute toward the technical solution, we presented
a holistic ecosystem that includes various robots
and concepts to meet the individual requirements of people
in need of care. In the current state of implementation,
the ecosystem employs robots such as the humanoid Justin,
the wheelchair-based assistant EDAN, and the haptic
teleoperation device HUG. While Justin can serve as a
26 * IEEE ROBOTICS & AUTOMATION MAGAZINE * SEPTEMBER 2021
helping hand in households or support the elderly in
retirement homes, EDAN is designed to support people
with severe disabilities.
Within the ecosystem, different control modes and user
interfaces are available, which allows for a tailor-made
application for the user. These range from an intuitive and
simple voice command that triggers fully autonomous
operation to overcoming one's own physical limitations by
using a direct-control interface to command a robotic
manipulator. The ecosystem presents a concept to combine
all of these approaches in a holistic way, thereby providing a
solution for the individual and including all stakeholders:
caretakers, caregivers, and relatives.
The ecosystem is complemented with a trained teleoperator,
who can take control of the robots whenever necessary.
This additional interface is a key aspect of the
ecosystem, as it mitigates the risk of system failure, even
without permanent, fully autonomous behavior. Additionally,
a teleoperator with a medical or nursing background
can help the user in emergency situations. In exemplary
scenarios derived from the needs of the envisioned users,
we have shown how the various parts of our proposed ecosystem
could provide support in practice. It is now important
to evaluate these findings in a real-life environment
together with those in need of support.
As shown in the previous sections, the current status of
our development already allows for a wide range of applications
in the context of health care. However, further developments
and improvements of some technologies are
foreseen that will enhance the ecosystem and its interaction
and, thus, promote operation in a real environment. In particular,
we are currently working on making the different
levels of autonomy available across all systems and user
interfaces. We envision that users of the EDAN system can
benefit from fully autonomous robot behavior, especially
when they are temporarily not sitting in EDAN's wheelchair.
Here, they may make use of the EMG- or joystickbased
interface to activate tasks in the supervised-autonomy
mode. Having all autonomy levels available across all systems
allows users to freely decide which mode to use in a
particular situation.
As a result, a smooth transition between the interaction
modalities or autonomy levels among the different robot
systems has to be further developed. On the one hand, this
requires the creation of a unified action representation
from which task execution at various levels of autonomy
can be derived. On the other hand, when switching among
the autonomy levels or from and to telepresence, a state
estimation is needed to continuously evaluate the actions
being performed and seamlessly switch back to autonomous
operation afterward [34]. Additionally, we will investigate
how our shared-control approaches can be applied to
assist the teleoperator and increase efficiency of the haptic
teleoperation in the scenario of medical assistance and better
support the teleoperator by automatically coordinating
redundant DoF in terms of multitasking [35].
IEEE Robotics & Automation Magazine - September 2021

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2021
