IEEE Robotics & Automation Magazine - December 2013 - 131

Achieving
Humanlike
Motion
Resolving Redundancy
for Anthropomorphic
Industrial Manipulators

By Andrea Maria Zanchettin, Luca Bascetta,
and Paolo Rocco
© ISTOCKPHOTO.COM/SURICOMA

esearch interest in human-robot interaction (HRI)
has been increasing in recent decades and has now
reached an appealing degree of maturity, which
encourages first steps toward a technology transfer
from research centers to robot manufacturers.
Making robots able to cooperate with humans in a safe and
intuitive manner is no longer a futuristic vision but a concrete
opportunity that is just around the corner. On the other hand,
measuring the quality of this interaction is a novel field in
robotics research and will be extremely useful for future
assessments of all efforts in the direction of more effective HRI.
First results exist, ranging from a simple approach, like the
so-called "uncanny valley" proposed in [1], to a more scientific
interpretation supported by physiological measurements [2].
Pioneer studies on human-machine interaction were
based on the assumption of a physical exchange of forces/
wrenches [3]. This led researchers to investigate compliant
actuators, humanlike control, and sensing strategies [4] to
achieve intrinsically safe robots.
This cooperation should be not just physically safe but also
psychologically comfortable to the humans: they must, at all
times, feel assured that they understand the robot's intentions,
and the robot must execute its tasks with motion profiles that
humans perceive as natural. For this reason, research in the

Digital Object Identifier 10.1109/MRA.2013.2283650
Date of publication: 30 October 2013

field of human kinematic analysis has gained interest in
recent years and has fostered new ideas and expectations.
Next-generation manipulators are thus expected to resemble
human behavior at the kinematic level to avoid causing
unease or discomfort to the nearby humans.
The main challenge in synthesizing the human motion is
redundancy resolution [5]. In fact, most of the tasks in human
common activities entail only a reduced number of constraints with respect to the total number of available degrees
of freedom (DoF). Therefore, a consistent and reliable way to
resolve the redundancy should be achieved.
Several studies on the motion of the human arm are
already available, ranging from pure kinematic approaches
to more complicated neuromuscular analyses. Pioneers of
neurophysiological studies proposed the existence of neural strategies to simplify the dimensionality of motor control [6]. These concepts, which are well known nowadays
and referred to as synergies, were later refined and
exploited in different works, e.g. to explain natural grasping
postures [7]. In [8], the existence of an uncontrolled manifold is conjectured. By observing several repetitions of the
same task, the authors noticed that the configuration space
can be divided into two orthogonal spaces: one of them,
depending on the task, is accurately controlled and stabilized, while the other may contain a higher variability and
seems to be uncontrolled. As reported in [9], this variability might be very small compared to the variability allowed
by joint redundancy.
DECEMBER 2013

IEEE ROBOTICS & AUTOMATION MAGAZINE

131

http://www.ISTOCKPHOTO.COM/SURICOMA

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2013