IEEE Robotics & Automation Magazine - June 2020 - 21

"Machine Learning for Active Gravi-
ty Compensation in Robotics" by Axier
Ugartemendia et al. improves the per-
formance of mechatronic systems used
in rehabilitation, bringing their capabili-
ties closer to those of human physio-
therapists. The article compares various
methods for learning to compensate
for gravity. "Decoding Motor Skills of
Artificial Intelligence and Human Poli-
cies" by Kai Yuan et al. offers the oppo-
site perspective. It shows how using
machine learning to discover how a
humanoid robot's push recovery strate-
gies can be used to better understand
and study human balance control.
Those discovered strategies can then be
employed for control design.
"Assured Runtime Monitoring and
Planning" by Esen Yel et al. addresses
one of the biggest open questions in
robot learning: safety and verification.
This article proposes to speed up online
reachability analysis with a pretrained
neural network where the neural net-
work itself has been verified. The
approach is demonstrated experimen-
tally with quadrotors.
"Multifidelity Reinforcement Learn-
ing With Gaussian Processes" by Varun
Suryan et al. shows how data collected
on less realistic simulations can be used
to minimize the amount of data that
need to be collected on a real robot. The

approach is validated with a navigation
task on a Pioneer robot. "Unsupervised
Pedestrian Pose Prediction" by Xiaoxiao
Du et al. addresses a problem that is
highly relevant for autonomous driving,
i.e., not only detecting pedestrians but
also predicting their future behavior. In
particular, the authors present a method
that does not require labeled pedestrian
data for training, which has been a
major bottleneck in the practical appli-
cation of such methods. "Deep LearningBased Localization and Perception
Systems" by Zhe Liu et al. presents a sys-
tem, based on deep learning, that can
very accurately localize itself. This is an
essential building block toward enabling
autonomous navigation, in this case in
the Hong Kong International Airport,
without requiring modifications to
the environment.
As a whole, the 11 articles in this spe-
cial issue cover a broad range of chal-
lenges, ranging from crucial theoretical
problems (manifold structure of data,
temporal features, unsupervised learn-
ing, and multifidelity reinforcement
learning) to key transversal needs
(assurance and verification of autono-
mous operations, gravity compensation,
and human skill decoding) while cover-
ing a wide and diverse set of applications
and robotic morphological structures. In
addition, the article on the CUDA

parallel computing platform and pro-
gramming model by Enric Cervera
suggests some interesting technical in--
sights that can be useful for the practical
implementation of machine/deep learn-
ing systems. Further, the article by David
St-Onge et al. on planetary exploration
provides an example of an application
where those approaches could have
huge benefits in the future.
We hope that, while providing a
rich picture of the state of the art in
the application of machine and deep
learning in intelligent robotics, this spe-
cial issue will inspire further research
from both the theoretical and applica-
tion standpoints.
Fabio Bonsignorio is with Heron Robots,
Genoa, Italy. Email: fabio.bonsignorio
@heronrobots.com.
David Hsu is with the National Univer-
sity of Singapore. Email: dyhsu@comp
.nus.edu.sg.
Matthew Johnson-Roberson is with the
University of Michigan, Ann Arbor,
United States. Email: mattjr@umich.edu.
Jens Kober is with the Delft University of
Technology, The Netherlands. Email:
j.kober@tudelft.nl.


IEEE connects you to a universe of information!
As the world's largest professional association dedicated to advancing
technological innovation and excellence for the benefit of humanity,
the IEEE and its Members inspire a global community through its
highly cited publications, conferences, technology standards,
and professional and educational activities.
Visit www.ieee.org.
Publications / IEEE Xplore ® / Standards / Membership / Conferences / Education
IMAGE LICENSED BY INGRAM PUBLISHING

JUNE 2020

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

21
http://www.ieee.org
IEEE Robotics & Automation Magazine - June 2020

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