IEEE Robotics & Automation Magazine - September 2023 - 60

force and the moment applied to the end effector. When
a relationship between any of the components of W E
and one or more sound features
CC ,,C12 f, is
sound = "
established, a mapping function gS can be derived to
estimate one or more components of WE (defined by
index i) as
Wg .Csound
Ei, = ^h
S
(3)
A practical example of (3) is given by the a near-linear
relationship between spindle torque x and spindle rotation
velocity ~ in air-driven spindles for machining tools
[23]. The machining sound generated from the interaction
between tool and workpiece has a soundwave frequency f
directly proportional to .~ Thus, the relationship between
spindle torque x and sound frequency f can be obtained as
in Figure 3(d).
Based on the preceding theory, the touching-the-sound
mode builds the bridge between haptic feedback and process
sound. When the system turns into this mode, the process
sound is acquired by a remote microphone and processed
to extract relevant features. Then, according to the mapping
function between force and sound features, previously calibrated,
the interaction force is calculated in real time (block B
of Figure 2). After the wrench acting on the robot is evaluated,
the haptic wrench Ws is obtained by scaling the measuring
force as
WA TWSS IE
= $$1E
(4)
where
AS is a user-defined scaling factor of the measured
force (similar to AI ), as shown in Figure 3(c).
AN EXAMPLE: HAPTIC CONTROL OF MACHINING TASKS
To validate the proposed method, we report an example in which
the robot executes a machining operation for a teleoperated repair
of airplane engine components. In this scenario, no onboard and
environmental sensors can be used, as the space allowed by narrow
access ports (<15 mm in diameter) is taken by robot and tool
actuation. The following section details a solution to realize haptic
control with the proposed method.
SYSTEM HARDWARE
The system hardware, presented in Figure 4, is composed by
three main parts corresponding to the control blocks of Figure
2. A first subsystem with a microphone array and signal
processing hardware acquires and processes sound. The haptic
device acts as an interface for human-robot interaction,
receiving motion commands and providing force feedback.
The last subsystem is a teleoperated robot with a machining
tool as the end effector.
SOUND ACQUISITION AND PROCESSING
This first subsystem records sound and extracts relevant features
to estimate the machining force. In the reported setup, a
XE
ZI
(PI,RI)
T1E
XI
YI
l1
{PE,RE}
(a)
YE
ZI
WS
AS *TIE3×3
-1
XI
YI
Workpiece
(c)
(d)
FIGURE 3. The mapping between the haptic device and follower robot. (a) The ideal behavior of the system, with the robot always
following the haptic device's motion according to the desired kinematic mapping. (b) The robot motion control mode, with a
springlike force generated to make the user aware of a difference between the pose commanded through the haptic device and
the actual pose of the robot. (c) Touching the sound, with the user experiencing a force feedback based on process sound features.
(d) An example of haptic feedback acquisition from sound, with an air-driven machining process in which the spindle torque x can be
directly derived from sound frequency f as
x~
==
60 IEEE ROBOTICS & AUTOMATION MAGAZINE SEPTEMBER 2023
gg gf22 1
() (( )).
WE,i
Workpiece
ZE
g2(ω)
τ
XE
Spindle
ω
g1(f)
f
XI
YE
ZE
l2
WI
YI
{PI
S,RI
S}
T 1
IE
IE
(b)
-1
{PE
S,RE
S}
{PM,RE
E
M}
l3
ZI
{PM,RI
I
M}
TIE
IE
YE
ZE
XE
IEEE Robotics & Automation Magazine - September 2023

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2023
