Tech Briefs Magazine - March 2021 - ST-20

Tech Briefs
devices, for example, to measure blood
flow in superficial arteries for health
monitoring applications, " said Assistant
Professor Benjamin Tee.
The next step for the NUS team is to further improve the conformability of their
material for different wearable applications and to develop artificial intelligence

(AI) applications based on the sensors.
" Our long-term goal is to predict cardiovascular health in the form of a tiny smart
patch that is placed on human skin. This
TRACE sensor is a step forward towards
that reality because the data it can capture
for pulse velocities is more accurate and
can also be equipped with machine learn-

ing algorithms to predict surface textures
more accurately, " explained Tee.
Other applications the NUS team aims
to develop include uses in prosthetics,
where having a reliable skin interface
allows for a more intelligent response.
For more information, contact Carolyn
Fong at carolyn@nus.edu.sg

New Mathematical Tool Can Select the Best Sensors for
The Job
A comprehensive mathematical framework can help engineers make informed decisions about
which sensors to use and where they must be positioned in aircraft and other machines.
Texas A&M University, College Station, TX
n the 2019 Boeing 737 Max crash,
where to put the sensor is critical
the recovered black box from the
because one might place an expenaftermath hinted that a failed pressive sensor at a location where it is
sure sensor may have caused the illnot needed. Thus, he said the ideal
fated aircraft to nosedive. This incisolution balances cost and precident and others have fueled a largsion by optimizing the number of
er debate on sensor selection,
sensors and their positions.
number, and placement to prevent
To test this rationale, Bhattacharya
the reoccurrence of such tragedies.
and his team designed a mathematTexas A&M University reical model using a set of equations
searchers have now developed a
that described the model of an Fcomprehensive
mathematical
16 aircraft. In their study, the
framework that can help engineers
researchers' objective was to estimake informed decisions about
mate the forward velocity, the
which sensors to use and where
direction of wind angle with
they must be positioned in aircraft New mathematical framework automates the decision-mak- respect to the airplane (the angle
ing process for selecting which sensors to use and where to
and other machines.
of attack), the angle between
place them in aircraft and other complex machines. (Image:
" During the early design stage Getty Images)
where the airplane is pointed and
for any control system, critical decithe horizon (the pitch angle), and
sions have to be made about which senaccuracies may be defined differently
pitch rate for this aircraft. Available to
sors to use and where to place them so
depending on the system and the applithem, were sensors that are normally in
that the system is optimized for measurcation. For instance, some systems may
aircraft for measuring acceleration,
ing certain physical quantities of interrequire that noise in the predictions
angular velocity, pitch rate, pressure,
est, " said Professor Raktim Bhattacharya.
does not exceed a certain amount, while
and the angle of attack. In addition, the
" With our mathematical formulation,
others may need the square of the noise
model was also provided with expected
engineers can feed the model with inforto be as small as possible. In all cases,
accuracies for each sensor.
mation on what needs to be sensed and
prediction accuracy has a direct impact
Their model revealed that not all of
with what precision, and the model's
on the cost of the sensor.
the sensors were needed to accurately
output will be the fewest sensors needed
" If you want to get sensor accuracy
estimate forward velocity; readings from
and their accuracies. "
that is two times more accurate, the cost
angular velocity sensors and pressure
Whether a car or an airplane, comis likely to be more than double, " said
sensors were enough. Also, these sensors
plex systems have internal properties
Bhattacharya. " Furthermore, in some
were enough to estimate the other physthat need to be measured. For instance,
cases, very high accuracy is not even
ical states, like the angle of attack, prein an airplane, sensors for angular velocrequired. For example, an expensive 4K
cluding the need of an additional angle
ity and acceleration are placed at specifHD vehicle camera for object detection
of attack sensor. In fact, these sensors,
ic locations to estimate the velocity.
is unnecessary because first, fine feaalthough a surrogate for measuring the
Sensors can also have different accuratures are not needed to distinguish
angle of attack, had the effect of introcies. In technical terms, accuracy is
humans from other cars and second,
ducing redundancy in the system, resultmeasured by the noise, or the wiggles, in
data processing from high-definition
ing in higher system reliability.
the sensor measurements. This noise
cameras becomes an issue. "
Bhattacharya said the mathematical
impacts how accurately the internal
Bhattacharya added that even if the
framework has been designed so that it
properties can be predicted. However,
sensors are extremely precise, knowing
always indicates the fewest sensors that

www.techbriefs.com

Cov

ToC

Sensor Technology, March 2021

http://www.techbriefs.com

Tech Briefs Magazine - March 2021

Table of Contents for the Digital Edition of Tech Briefs Magazine - March 2021