IEEE Robotics & Automation Magazine - December 2022 - 54

high-pass and low-pass filters. Second, considering that the
signal amplitude may still not be enough to meet the needs of
engineering applications, a dynamic gain with an amplification
range of 1-64 times is further added based on the sampling
feedback. In this way, the total amplification of the
original signal is 100-6,400 times. Finally, the fast Fourier
transform (FFT) technique is applied to the amplified signal
to extract its amplitude and phase, which are then sent to the
CPU (i.e., a Raspberry Pi) of the circuit to be further calculated
to obtain the perception information.
Task
η
PF-Based
Electric Sense
η
Fusion
Planning
(x, y)
Control
u
Dynamic Model
ψ
IMU
V
(a)
Task
Planning
η
PF-Based
Electric Sense
(b)
ψ
Control
IMU
V
Task
Planning
(x, y)
η
Fusion
Control
u
Dynamic Model
ψ
IMU
(c)
Figure 4. Three localization methods. (a) Electric sense + IMU + dynamic model,
(b) electric sense + IMU, (c) IMU + dynamic model.
20
40
60
-60
-40
-20
-80
10-2
10-1
100
101
Frequency (kHz)
Figure 5. Amplitude-frequency response characteristics of
the bandpass filter. The resistance's tolerance is 1%, and the
capacitance's tolerance is 5%. The blue line is the result of the
standard parameters, the red line is the upper limit, and the
yellow line is the lower limit.
54 * IEEE ROBOTICS & AUTOMATION MAGAZINE * DECEMBER 2022
102
103
AC-Based Electric Sense
Now we introduce how to realize the
underwater electric sense using our
designed hardware solution. As we
have mentioned, the electric emitter
including two electrodes is placed in
the underwater environment, while
the electric receiver, which consists of
four electrodes, is equipped on the
robot. Relative to one emitter, we build
a coordinate system in which we are
able to describe the pose [including
the position (x, y) and the orientation
ψ] of the robot. When the robot
moves, the four electrodes of the
receiver fixed on the robot continuously
measure the electric field generated
by the emitter(s). Then, according
to the theoretical model of the electric
field (see the section " Model of the
Electric Field " ) and the known physical
parameters of the hardware, one
can get the position information of the
electrode from its measurement. Since
the receiver has four electrodes, we get
three independent measurements,
,
Vi ,,, that further determine
i 123=
the three variables (, ,),xy } i.e., the
pose of the robot.
We also show the advantages of
our ac-based solution by calculating
its signal-to-noise ratio (SNR) and
comparing it to that of the dc-based
solution [13]. From the theoretical
model of the electric field, we know
that, under the same conditions, the strength of the signal
measured by the receiver is identical under both ac- and
Table 2. The main parameters of the filters.
Filter
High-pass second-order
Chebyshev
Low-pass fourth-order
Butterworth
Low-pass second-order
Chebyshev
High-pass second-order
Chebyshev
30 dB 10 kHz
0 dB 7 kHz
0 dB 2 kHz
Gain Passband Stopband
10 dB 2 kHz
-40 dB@50 Hz
-30 dB@40 kHz
-15 dB@40 kHz
-40 dB@50 Hz
Amplitude (dB)
IEEE Robotics & Automation Magazine - December 2022

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2022
