IEEE Robotics & Automation Magazine - June 2018 - 29

Most industrial robots are characterized by very narrow
impedance bandwidths, large payload capabilities, precise
actuators, and extensive workspaces. Humans, on the other
hand, have large impedance bandwidths, flexibility and dexterity, but low payload capabilities. Combining the advantages
of both may yield highly effective industrial systems. This
approach led to the development of human-friendly robotic manipulators [1], notably, in the field of pHRI [2], [3]. To
allow fine manipulation, pHRI manipulators should closely
match the varying human mechanical impedance, i.e., if the
apparent impedance of a manipulator can be minimized, the
human operator can then deploy his/her own impedance,
which yields a very intuitive interaction. This is especially
important in assembly operations where the matching of
parts requires a very low-impedance interaction to be effective and intuitive to the human operator.
In most industrial pHRI applications, force/torque sensors are used to sense and regulate the interaction between
the human operator and the robot. An admittance controller
is then used to emulate different impedances [4], [5]. In
some cases, a proportional-integral controller [6], or even
lead and lag compensators [7] are used. However, the hardware dynamics limit the apparent impedance reduction [8],
and any attempt to go below a certain fraction of the intrinsic
inertia leads to unstable behaviors [9]. According to [4], [5],
and [7], based on such techniques, the apparent inertia can
be reduced by a factor of five to seven with respect to the real
inertia. Other approaches making use of force sensors
include the appending of compliant material to mechanically
filter the high-frequency interactions [10]. Nevertheless,
large inertia reduction ratios are achievable only by overstepping the concept of passivity [11], [12], which means
that physical contacts are limited to specific ranges of environmental dynamics.
When physically interacting with robots for long periods
of time, e.g., during a work shift in an assembly plant,
intuitiveness is of utmost importance. If a system lacks
responsiveness, the human user may have the experience of
constantly dragging or fighting with the robot, which
becomes tiring and frustrating. The use of force/torque sensors typically requires filters to reduce the sensor noise, which
yields a lack of responsiveness caused by delays and induces
the dragging impression. This issue may be addressed by
replacing the high-impedance force/torque sensor with a
low-impedance mechanical interface that can be considered
a position sensor. In fact, force/torque sensors based on strain
gauges can also be considered position sensors; however, they
have very high impedance since they are based on the measurement of the extension of the strain gauges.
This approach was used in [13] and [14], where the authors
proposed the concept of underactuation redundancy to provide very low mechanical impedance to a human operator.
In this concept, a mini passive LIP manipulator, which provides the low-impedance interaction, is mounted at the end
effector of a macro HIA active manipulator that provides
the workspace and force capabilities. Such an arrangement

yields apparent impedance that is lower than that of any
actuated mechanism.
This concept applies to task spaces with limited DoF, e.g.,
to 4-DoF selective compliance assembly robot arm-type
tasks. In such tasks, the space is divided into the space of
operational DoF and the space of constrained DoF. In the
space of operational DoF, all of the work on the payload,
except for the gravity compensation forces, is performed by
the human, while, in the constrained space, the constraint
forces are provided by the robot. Therefore, only the operational DoF need to render the low impedance. Examples of
similar mechanisms are cable-suspended intelligent assist
devices, as shown in [15] and [16]. Unfortunately, cable-suspended devices cannot constrain rotational motion and cannot handle off-centered payloads.
The robot proposed in [14] consists of a macro 3-DoF
active gantry system with a mounted passive 3-DoF mini
mechanism. The effectiveness of the low-impedance rendering is demonstrated in assembly operations, e.g., peg-in-hole
tasks. The robot demonstrates responsiveness, intuitiveness,
and provides very high bandwidth for small-range precision
tasks; when the robot is moving autonomously, possible collisions can be detected, and contact forces in the occurrence
of a collision are very low compared to those encountered in
active systems.
Moreover, the HIA mechanism constituting the macro
component of the robot is located outside the human operator's workspace, which ensures safety, i.e., the human operator
interacts strictly with the LIP mechanism. However, the mini
manipulator proposed in
[14] is based on the serial
Human-robot collaboration
arrangement of three,
complex 1-DoF mechais a rapidly growing field
nisms that aim to decouple the three translational
wherein actuated systems
directions of motion
while providing a central
either interact with
equilibrium configuration
so the mini manipulator
humans or are actively
smoothly returns when
there no external force
controlled by humans.
applied. One of the drawbacks of this mechanism
is its complexity; a large
number of links and joints are required to ensure proper kinematics. Another drawback of the passive mechanism proposed in [14] is that the forces that return the mechanism to
its neutral position when displaced horizontally are proportional to the mass of the payload attached to the end-effector. When large payloads are handled, this effect can be
detrimental because the human operator is required to use
significant force.
In this article, we propose a novel mini LIP mechanism.
This mechanism is based on a decoupled parallel mechanism
known as the Tripteron [17], which makes it much more
compact than the mini passive mechanism proposed in [14].
JUNE 2018

IEEE ROBOTICS & AUTOMATION MAGAZINE

IEEE Robotics & Automation Magazine - June 2018

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - June 2018