IEEE Robotics & Automation Magazine - June 2021 - 114

Intelligent Systems (MPI-IS) in 2020. It was based on the
remote execution of submitted software on a robotic hand.
There was a fixed set of tasks, such as grasping and pushing,
which did not require mobility. The IEEE International Conference
on Soft Robotics also holds a competition (http://
www.robosoft2019.org/robosoft_competition.html) with a
manipulation challenge that emphasizes soft manipulators.
Similarly, the tests do not require mobility.
There have also been recent competitions that targeted
mobile manipulation. The FetchIt! Mobile Manipulation
Challenge was held at the 2019 IEEE International Conference
on Robotics and Automation [10]. The task was to
assemble a kit formed by six objects obtained from stations
around a designated arena, combining navigation and
manipulation skills. Similarly, the RoboCup@Home competition
(https://athome.robocup.org/), using the Toyota
Human Support Robot (HSR) [11] as the official platform,
includes a set of tidying-up and service tasks in living room
and kitchen setups, requiring mobile manipulation. RoboCup@Home
also encourages teams to make " open challenge "
demonstrations (i.e., free demonstrations determined
by the teams, instead of a fixed set of tasks), although these
are not the main focus, as they are performed during off
hours and do not necessarily include awards [12]. The
Smart Cities Robotics Challenge [13], which is organized as
part of the European Robotics League and builds on the
success of the European Robotics Challenge (EuRoC) [14],
also includes a fixed set of mobile manipulation tasks, such
as delivering coffee shop orders and shopping pick-andpack
procedures.
The unique feature of our hackathon, compared to the
preceding competitions, is that it positions mobile manipulation
together with open demonstrations at center stage. As
explained, multiple mobile manipulation competitions have
focused on fixed sets of tasks. This has the advantage of creating
benchmarks that enable progress to be objectively measured.
Therefore, they are crucial to the community. However,
we believe that an open format also has a place. It enables 1)
teams to demonstrate their core research innovations more
directly and 2) the community/audience to be informed
about the state of the art for a rich variety of tasks. With the
Mobile Manipulation Hackathon, our goal has been to push
teams to perform their own research demonstrations and to
identify tasks that the community is working on.
The Field of Mobile Manipulation
Merging mobility and manipulation, mobile manipulation
systems need to overcome some of the most difficult challenges
in robotics, including the following:
●
●
Uncertainty: The ability to locomote, the required generality
in task execution, and the use of multiple sensors and
actuators make it impractical to engineer an entire environment
for a task. As a result, mobile manipulation systems
have to explicitly address problems that arise due to
the uncertainty of sensing and actuation.
●
System complexity: Mobile manipulation systems require
the integration of a large number of hardware components
for sensing, manipulation, and locomotion as well as the
orchestration of algorithmic capabilities in perception,
manipulation, control, planning, and so on.
The mobility of these systems can take multiple forms
depending on the environment: air/space (drones, planes, helicopters,
and satellites), water (ships and submarines), and land
(wheeled and legged robots). In air and space, mobile manipulation
systems often take the shape of aerial vehicles carrying
some sort of manipulator [15], [16], e.g., grippers [17] and
multilink arms [18], [19] attached to a rotorcraft; they may
also be built as manipulators endowed with some flying mechanism,
e.g., rotors [20]. A significant challenge for these systems
is to maintain flight stability during object manipulation,
which limits the range of operations that can be performed.
This coupling between the control of mobility and manipulation
also exists in water, where robots need to maintain a stable
pose while experiencing additional forces due to object manipulation
[21], [22]. Land is the most common environment for
mobile manipulation. Humans live on land, and therefore a
larger variety of mobile manipulation tasks can be found there.
Furthermore, the control of mobility and manipulation can be
decoupled more easily on land, compared to in-air and underwater
manipulation. A land robot can attain a statically stable
configuration and, for small enough forces, avoid the need to
balance during manipulation.
Two common forms of mobility on land are legs and
wheels. Legged locomotion and bimanual manipulation are
typically combined in humanoid robots, e.g., [23]. Even
though planning and control for legged locomotion can be
more complex than for wheeled locomotion, legs can be
advantageous depending on ground characteristics. Particularly
for search-and-rescue operations, where debris, obstacles,
and steps are expected, legged mobile manipulation is
preferred. Such systems dominated, for example, the DARPA
Robotics Challenge [24].
The most common and versatile mobile manipulation
Generality: Mobile manipulation systems must perform a
variety of tasks, acquire new skills, and apply those abilities
in novel situations. They must be able to continuously
adapt and improve their performance.
●
High-dimensional state space: Versatile robotic systems must
be equipped with many actuators and sensors, resulting in
high-dimensional state spaces for planning and control.
114 * IEEE ROBOTICS & AUTOMATION MAGAZINE * JUNE 2021
systems, however, are wheeled. They strike the right balance
between ease of mobility and manipulation and access
to most human environments. The development of
wheeled mobile manipulators has unfolded during the past
35 years. The first prototype of a mobile manipulator was
the Mobile Robot (MORO), in 1984 [25]. Initial attempts to
mount robotic arms on mobile platforms happened during
the 1990s, with robots such as the Hostile Environment
Robotic Machine Intelligence Experiment Series (HERMIES)
(Hostile Environment Robotic Machine Intelligence
Experiment Series) [26] and KAMRO (Karlsruhe Autonomous
Mobile Robot) [26] and the Karlsruhe Autonomous
http://www.robosoft2019.org/robosoft_competition.html http://www.robosoft2019.org/robosoft_competition.html https://athome.robocup.org/
IEEE Robotics & Automation Magazine - June 2021

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