El‑Sheimy and Youssef Satell Navig (2020) 1:2

https://doi.org/10.1186/s43020-019-0001-5
Satellite Navigation
https://satellite-navigation.springeropen.com/

REVIEW Open Access

Inertial sensors technologies for navigation

applications: state of the art and future trends
Naser El‑Sheimy and Ahmed Youssef*

Abstract
Inertial navigation represents a unique method of navigation, in which there is no dependency on external sources of
information. As opposed to other position fixing navigation techniques, inertial navigation performs the navigation in a
relative sense with respect to the initial navigation state of the moving platform. Hence, inertial navigation systems are
not prone to jamming, or spoofing. Inertial navigation systems have developed vastly, from their occurrence in the 1940s
up to date. The accuracy of the inertial sensors has improved over time, making inertial sensors sufficient in terms of
size, weight, cost, and accuracy for navigation and guidance applications. Within the past few years, inertial sensors have
developed from being purely mechanical into incorporating various technologies and taking advantage of numerous
physical phenomena, from which the dynamic forces exerted on a moving body could be computed accurately. Besides,
the evolution of inertial navigation scheme involved the evolution from stable-platform inertial navigation system, which
were mechanically complicated, to computationally demanding strap-down inertial navigation systems. Optical sensory
technologies have provided highly accurate inertial sensors, at smaller sizes. Besides, the vibratory inertial navigation
technologies enabled the production of Micro-electro-machined inertial sensors that are extremely low-cost, and offer
extremely low size, weight and power consumption, making them suitable for a wide range of day-to-day navigation
applications. Recently, advanced inertial sensor technologies have been introduced to the industry such as nuclear mag‑
netic resonance technology, cold-atom technology, and the re-introduction of fluid-based inertial sensors. On another
note, inertial sensor errors constitute a huge research aspect in which it is intended for inertial sensors to reach level in
which they could operate for substantially long operation times in the absence of updates from aiding sensors, which
would be a huge leap. Inertial sensors error modeling techniques have been developing rapidly trying to ensure higher
levels of navigation accuracy using lower-cost inertial sensors. In this review, the inertial sensor technologies are covered
extensively, along the future trends in the inertial sensors’ technologies. Besides, this review covers a brief overview on
the inertial error modeling techniques used to enhance the performance of low-cost sensors.
Keywords: Gyroscopes, Accelerometers, Optical inertial sensors, Micro-electro-machined, Fluid-based inertial
sensors, Stochastic modeling

Introduction the platform in either two-dimensional (2-D) or three-

The state of motion of any moving platform could be dimensional (3-D) space [1].
determined through a process known as Navigation. Navigation techniques are classified into two major
Whereas, navigation is done by determining the navi- categories. Namely, position fixing and dead reckoning.
gation states of the moving platform. The navigation Position fixing is performed by determining the naviga-
states represent the position, velocity, and orientation of tion states with respect to a set of well-known positions.
An example of position fixing technique is the global nav-
*Correspondence: [email protected]
igation satellite systems (GNSS). On the other hand, dead
Mobile Multi‑sensor Systems Research Team, Department of Geomatics reckoning determines the navigation states of a moving
Engineering, University of Calgary, 2500 University Drive N.W, Calgary, AB platform by measuring recursively the progression of
T2N 1N4, Canada

© The Author(s) 2020. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and
the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material
in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material
is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the
permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat​iveco​
mmons​.org/licen​ses/by/4.0/.
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 2 of 21

such navigation states with respect to their initial values. sensors, include biases scale factor, scale-factor non-
Inertial navigation is an example of the dead reckoning linearity, and cross-coupling of sensitive axes measure-
navigation technique [1]. ments. A bias in an inertial sensor is a constant shift in
The need for dead-reckoning navigation arises from the the measured quantity from the actual input to the sen-
limitations of typical position fixing techniques which sor. Whereas, a scale factor is an error that represents
require a direct line of sight between the platform, to be the mismatch between the input quantity to an inertial
navigated, and the well-known fixed positions. To clar- sensor, and the reported output quantity of the sensor.
ify, navigation using GNSS requires a direct line of sight Typically, one should expect an inertial sensor to report
between the GNSS receiver and at least four satellites to an output value equivalent to whichever input value
acquire the navigation states of the navigated platform. imposed upon the sensor. Hence, the expected input–
Such condition is not usually met practically, especially output ratio should be equal to one. However, a scale
when navigation takes place in urban or indoor envi- factor would manifest as deviation of the input–output
ronments. Consequently, GNSS-denied environments relation of an inertial sensor from being equal to one.
require the use of dead-reckoning, especially inertial Another form of the systematic errors is the scale-factor
navigation, to provide a navigation solution for periods in non-linearity. Typically, the input–output relation of a
which position fixing solution is not possible [2]. sensor is expected be a linear relation. However, due to
If one could measure the resultant force exerted on a environmental impacts and some sensors designs, the
moving body, and deploy Newton’s second law of motion, input–output relation ship of the inertial sensor might
the linear and/or rotational rates of motion could be not be a linear relation, which is a systematic error that
deduced. Hence, inertial sensors emerge. Inertial sensors should be accounted for. Due to improper mounting of
are classified into two main categories: accelerometers inertial sensors within a geometric assembly of an IMU,
and gyroscopes. Broadly, accelerometers measure spe- cross-coupling error occurs. Cross-coupling is caused
cific forces or accelerations, while, gyroscopes measure by the non-orthogonality of the sensitive axes of iner-
angular velocities. tial sensors. Consequently, the inertia sensors (either
When fitted into specific geometric forms that guar- accelerometers or gyroscopes) measure residual inertial
antees capturing the motion of any given platform, the measurements from another axis that is supposed to be
inertial sensors assembly is referred to as an inertial meas- orthogonal to its sensitive axis [3].
urement unit (IMU). Whereas, IMUs are usually coupled Evidently, inertial sensors endure random errors that
with some form of basic on-board data processing to would manifest as noises within the inertial measure-
convert the raw measurements to sensible specific forces ment signals acquired from inertial sensors. The ran-
or angular velocities. A typical IMU comprises a triad of dom errors can be attributed to electrical or mechanical
accelerometers and a triad of gyroscopes mounted along sources, depending on the design and manufacturing of
three mutually orthogonal axis to capture the 3-D motion the inertial sensors. Nonetheless, the order of magni-
of any given platform to which it is mounted. tude and impact of such random errors on the inertial
Nevertheless, inertial navigation is done by process- navigation solution is dependent upon the technology,
ing the inertial measurements that are acquired from design, and manufacturing techniques of the inertial sen-
IMUs. The inertial measurements are mathematically sors. Consequently, the performance of any given IMU
reduced into variations in position, velocity, and orienta- in terms of providing an accurate navigation solution is
tion for the moving platform. Consequently, the naviga- defined by the order of magnitude of the systematic and
tion states could be accumulated over time to identify the random errors included in its measurements. Hence,
position, velocity, and orientation of the platform at any IMUs are classified into grades as per their performance
given instant. Therefore, a system that utilizes the meas- and accuracy.
urements of IMUs to acquire the navigation states of any However, it is understandable that there exists a high
moving platform to which it is mounted is known as an correlation between the performance of any given IMU,
inertial navigation system (INS). An INS is a system that its underlying technology, and its cost. Whereas, IMUs
would include an IMU along with some means to process are classified into: strategic, navigation, tactical, and con-
the inertial measurements into a full navigation solution. sumer grades. Table 1 provides a summary of IMUs clas-
Inertial sensors suffer from errors, which are either sification as per performance, accuracy, and cost, which
systematic errors or random errors. Systematic errors is modified after [1]. It is noted that performance param-
can be modeled mathematically and can be mitigated eters upon which the IMUs are classified are discussed
through calibration. Systematic errors, in inertial afterwards.
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 3 of 21

Table 1 A summary of IMUs classification as per performance, from [1]

Grade Strategic grade Navigation grade Tactical grade Commercial ­gradea

Positional error 30–100 m/h 1 nmib/h or 0.5 m/s 10–20 nmi/h Large variation
Gyroscope drift 0.0001–0.001°/h < 0.01°/h 1–10°/h 0.1°/s
√ √ √
Gyroscope random walk – < 0.002◦ / h 0.05– < 0.02 ◦ / h Several ◦ / h
Accelerometer bias 0.1–1 μg < 100 μg 1–5 mg 100–1000 μg
Applications Submarines General navigation Integrated with GPS for Research
Intercontinental ballistic High precision georeferencing mapping Low cost navigation
missiles Mapping Weapons (Short time) Pedometers
Anti-locking breaking
Active Suspension
Airbags
Cost ~ $1 million ~ $100,000 ~ $2000–$50,000 $1 for accelerometers
$10 for gyroscopes
a
Also called automotive grade
b
1 nautical mile (nmi) ≈ 6076 ft ≈ 1851 m

This paper introduces an overview over the classifi- General classification of inertial navigation
cation of the INSs based on their operation mode, into systems
either stable-platform or strap-down INSs, as shown in Another form of classifying the INSs is depending on
“General classification of inertial navigation systems” their mechanical operation scheme. INSs can be classi-
section. The differentiation between the two categories fied into stable platform INSs, and strap-down INSs. An
of INSs resides in the mechanical system mounts, and argument can be made that this classification is a chrono-
mathematical process implemented to acquire a full navi- logical one. Whereas, the stable platform INSs have been
gation solution and their differences. “Inertial navigation gradually replaced over the years by the strap-down INSs.
technologies” section covers an overview on the state- However, stable platform INSs are not being entirely
of-the-art inertial sensor technologies, and a historical replaced, and are still used for some navigation applica-
overview over the development of inertial sensor tech- tions. Stable platform INSs are commonly referred to as
nologies. “Future trends in inertial sensors technologies” mechanical INS, as discussed hereafter. Figure 1 shows a
section provides a summary of future trends in inertial schematic of the difference between stable-platform and
navigation sensor technologies. “Stochastic error mode- a strapdown INSs, from [1].
ling for inertial navigation” section covers an overview on
the stochastic error modeling methods, with which iner- Stable‑platform inertial navigation systems
tial sensor errors are modeled into navigation algorithms, This type of INSs require mounting the inertial sensors
which is a major aspect of acquiring an error-bounded on a stable platform that is mechanically isolated from
navigation solution from a dead-reckoning INS. the rotational motion of the vehicle. Such requirement

Fig. 1 Shows a schematic of the difference between stable-platform and a strapdown INSs, from [1]
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 4 of 21

could be achieved by utilizing mechanical inertial sen- navigation systems” section. However, for strap-down
sors, specifically gyroscopes. Within its internal struc- INSs, the mechanical stabilization that was provided
ture and mechanism, a typical mechanical gyroscope within stable-platform systems are replaced by a com-
comprises a rotating rotor means, with high moment putational model to achieve the same output navigation
of inertia about a given spin axis, which is rotated by states. Since, the computational model of strap-down
mechanical means, and are presumed to maintain high INSs is a mathematical realization of the mechanical sta-
rotational speeds. These conditions lead such rotor to bilization in mechanical systems, it is referred to as INS
maintain spatial rigidity in space, as per the law of con- mechanization [4].
servation of momentum. Such spatial rigidity of the rotor INS Mechanization can be defined as the process by
allows it to maintain a stable direction in space. A gimbal which Newton’s law of motion are applied, along with
connection is connected to said rotor means, which typi- the geometric transformations of reference frames, to
cally constitute three free rings connected through pure acquire the required navigation states from IMU meas-
hinged connections and are free to rotate in 3D. Conse- urements. The navigation states include the position,
quently, the rotation rate of the moving platform can be velocity, and attitude of the moving platform.
detected by utilizing a pick-off means to determine the INS Mechanization in general can be realized using
rotation of the gimbal tings with respect to the spatially any set of sensors that would be able to provide the raw
rigid rotor [4]. measurements, that when processed can give the naviga-
Nonetheless, mechanical gyroscopes were the means to tion states within the chosen reference frame dimensions
provide a mechanically stable platform that helped real- (i.e. 1-D dimensional, 2-D, or 3-D navigation frames).
ize inertial navigation in the first place. It is noted that Consequently, the mechanization process transforms
such theoretical assumption of having a stable spin axis the measurements to the navigation frame as a basic
direction for the mechanical gyroscope rotor was not component of the process. Then, the mechanization
entirely satisfied practically. Whereas, mechanical INSs process includes an integration over time to acquire the
suffered from various sources of errors. Chief among navigation states from the raw measurements of the IMU.
those errors, is the precession error caused by exter- Such measurements include the rotation rates from the
nally applied torques to the spinning rotor which would gyroscopes that are integrated to acquire the attitude
in turn affect its spatial rigidity in space and would lead angles, and the specific forces from the accelerometers
the rotor to deviate from its assumed direction [4]. Such that are integrated to acquire the velocities and the posi-
external torques could be caused by improper balance of tions [1].
masses within the gyroscope design, or under the impact Strap-down INSs comprise most of the state-of-the-art
of external shocks. Despite that, stable platform INSs are INSs. Strap-down INSs provide optimal alternative for
considered very accurate and reliable. Whereas, stable the stable-platform INSs, because strap-down systems
platform INSs are still used for applications that require provide lower cost, smaller size INSs, that have compa-
very accurate estimates of navigation data such as ships rable reliability to the stable-platform systems. Besides,
and submarines. However, the downsides of such systems strap-down INSs remove most of the mechanical com-
are being large in size, being of high cost, and having high plexity associated with the stable-platform systems. Such
mechanical complexity. advantages enable the strap-down INSs to be utilized for
a wider range of applications, that demand high perfor-
Strap‑down inertial navigation systems mance and light weight. On the other hand, strap-down
A modern alternative for the stable-platform INSs are systems endure some drawbacks that include the sub-
the strap-down INSs. From the nomenclature, strap- stantial increase in computational complexity and high
down INSs imply that the inertial sensors are strapped demand for on-board processing power. However, due to
down rigidly to the vehicle to which they are mounted. recent advances in computer technology with the devel-
Generally, it is noted that the concept of inertial naviga- opment of suitable sensors, such strap-down systems
tion depends of acquiring inertial measurements of the have been successfully realized, and dominate the major
moving vehicle with respect to an inertial non-rotating aspects in the state-of -the-art inertial navigation.
non accelerating reference coordinate system, or frame.
However, the navigation states should be represented Inertial navigation technologies
with respect to a navigation frame. Hence, for stable- There are numerous technologies that comprise the
platform INSs, such measurements are acquired by uti- state-of-the-art commercialized inertial sensors that are
lizing the mechanical stabilization of the inertial sensors utilized to build strapdown INSs. However, there are
as explained in the previous “Stable-platform inertial basic technologies which dominate the market of inertial
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 5 of 21

navigation. In this section, the dominant state-of-the-art Angular rate sensors technologies
technologies in inertial navigation are discussed in terms As stated earlier, an IMU comprises an angular rate sen-
of sensor basic operation principles and expected perfor- sor that is used to measure the angular rates of any mov-
mances. This section highlights the main technologies for ing platform, form which the attitude of the platform is
angular rate sensors, and accelerometers, as well. determined. Angular rate sensors vary over a large range
An inertial sensor, regardless of its underlying tech- of accuracy, which can be described holistically by bias of
nology, comprises three main components that consti- the angular rate sensors. The range of biases encountered
tute a fully-functional inertial sensor. An inertial sensor with different angular rate sensors vary from less than
includes a motion transduction mechanism, a signal con- 0.0001°/h up to 1°/s class or worse. Generally, most angu-
ditional mechanism, and a sensor read-out component lar rate sensors are sensitive to their operation environ-
[5]. ment, which leads to some undesirable effects.
A motion transduction mechanism is responsible for
detecting the physical phenomenon upon which the iner- Dynamically tuned gyroscopes (DTG)
tial sensor operates and transforms such detection into a State-of-the-art angular rate sensors vary in terms of the
quantifiable signal. Hence, a motion transduction mecha- utilized technologies and operation principles, which
nism includes a motion indicator, or a sensing element, a in turn is representative of their accuracy. Angular rate
pick-off mechanism, and a damping mechanism [5]. The sensors include mechanical-based gyroscopes, which
motion indicator of an inertial sensor represents the seis- are modernized to be used in strapdown mode, such as
mic element that reacts to any externally applied motion. dynamically tuned gyroscopes (DTG), and flex gyro-
A pick-off mechanism represents the means by which the scopes [4]. Such type of gyroscopes basically operates
reaction of the motion indicator is detected and trans- using a highly spinning mass. However, in these strap-
formed into a signal. Whereas, a damping mechanism is down-type sensors, the operation concept relies on the
responsible for restoring the sensing element to its null gyroscopic precession effect that occur because of exter-
position whenever the external stimulus is unapplied nal torques acting upon a spinning mass. The torques are
to the sensor. It is noted that the damping mechanism applied intentionally to the spinning mass, using restrain-
depends mainly on the design configuration of the iner- ing means as hinges or the like. Consequently, the spin-
tial sensor. ning mass precession is considered as a measure of the
To clarify, an inertial sensor could be design with angular rate, which is measured using appropriate pick-
either an open-loop or closed-loop design configuration. off means. Figure 2 shows a schematic 3D view of a DTG,
Whereas, an open-loop design configuration relies on modified after [4].
the physical phenomenon upon which the sensor oper-
ates to restore its sensing element to its null state with- Vibratory gyroscopes
out any interference from an external source, depending Another technology for angular rate sensors fabrication
solely on the governing physical equilibrium. On the is the vibratory gyroscopes. The operation principle of
other hand, a closed-loop design configuration provides vibratory gyroscopes was discovered by Foucault [4]. The
an external source of balance that provides a feedback operation principle of the vibratory gyroscopes implies
to the sensor’s sensing element to restore it to its null that a vibrating element, such as a rod, maintains its
state, and is referred to in literature as the feedback, or plane of vibration regardless of the motion of the sensor
rebalance, loop. Closed-loop inertial sensors generally
provide higher sensitivity sensors with low order of mag-
nitude errors; however, it can be characterized of being
relatively high cost in comparison to open-loop inertial
sensors.
Nevertheless, a signal conditioning mechanism is
a means by which the detected signal from the pick-
off mechanism is converted into the desired measur-
able quantity from an inertial sensor (i.e., specific forces
for accelerometers, or angular rates for gyroscopes).
Whereas, a sensor read-out is the interface by which the
measured quantities are conveyed to the user. However, a
sensor read-out is not often considered as a core compo-
nent of an inertial sensor, as it varies as per the intended
application for the inertial sensor. Fig. 2 A schematic 3D view of a DTG, modified after [4]
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 6 of 21

in space. Consequently, the angular rates and linear accel- shocks. Figure 3 shows a schematic 3D view of a vibra-
erations can be measured relying on such concept. The tory fork gyroscope, as an example of vibratory gyro-
vibrating element can take various forms, such as: a scopes, modified after [4].
string, hollow cylinder, rod, tuning fork, beam, and hemi-
spherical dome. To clarify, the vibratory motion of an Optical gyroscopes
element creates an oscillatory linear velocity. If the sen- One of the most effective and accurate technologies in
sor is rotated about an axis orthogonal to the oscillatory gyroscopes fabrication is the optical gyroscopes. The
linear velocity, a Coriolis acceleration is induced, which operation concept of the optical gyroscopes depends on
modifies the motion of the vibrating element. If such the Sagnac Effect. The Sagnac effect implies that, for a
motion variations, due to Coriolis force, can be detected closed loop interferometer, the phase of light traveling
and measured, the angular rate can then be deduced. The within the interferometer is proportional to the exter-
vibratory gyroscopes can be characterized by its suscep- nal angular rate, to which it is subjected. Optical gyro-
tibility to be miniaturized and being mass produced at scopes can be realized in many forms, most effective of
relatively cost. Hence, vibratory gyroscopes can be con- which are ring laser gyroscopes (RLG), which represents
sidered as the base technology upon which micro-elec- an active optical gyroscope architecture, and fiber optics
tro-mechanical-system (MEMS) based gyroscopes are gyroscopes (FOG), which represents a passive optical
produced. However, vibratory gyroscopes can be charac- gyroscope architecture [6].
terized by having high drift rates, limits on input range To clarify, the FOG gyroscope consists of a Fiber
of the gyroscopes due to limits of natural frequency of Optics cable coiled in a closed form. For simplicity, the
the resonator (i.e., vibrating element), and sensitivity to Fiber Optics cable is coiled into one turn; however, typi-
environmental effects such as temperature variations and cally it is turned into more than one turn. It is noted that
the number of turns affects the sensitivity of the Gyro-
scope. Nevertheless, the FOG gyroscope also consists of
a light source, light detector, and a set of coupling lenses.
The light source emits light which passes through a beam
splitter. The light waves, afterwards, flow into two oppos-
ing paths, and are then rejoined through the beam split-
ter towards the light detector. The interference pattern of
the two sets of light beams remains constant, as per the
detector, until the gyroscope is subjected to an external
rotation about its sensitive axis [6]. Afterwards, the inter-
ference pattern varies with a magnitude proportional to
the angular rate. Table 2 provides a summary of typical
performance characteristics for a range of angular rate
sensors suitable for strapdown application, from [4].
Whereas, Fig. 4 shows a very simple schematic of a sin-
gle turn FOG, to clarify the operation concepts and main
Fig. 3 A schematic 3D view of a vibratory fork gyroscope, as an components of FOG.
example of vibratory gyroscope, modified after [4]

Table 2 A summary of typical performance characteristics for a range of angular rate sensors suitable for strapdown
application, from [4]
Characteristic DTG Flex gyro Vibratory gyro RLG FOG

g-Independent bias (°/h) 0.05–10 1–50 360–1800 0.001–10 0.5–50a

g-Dependent bias (°/h/g) 0.01–10 1–10 36–180 0 <1
An-isoelastic bias (°/h/g2) 0.1–0.5 0.05–0.25 18 0 < 0.1
Scale-factor non-linearity (%) 0.01–0.1 0.01–0.1 0.2–0.3 5–100 0.05–0.5
Bandwidth (Hz) 100 200 500 > 200 > 100
Maximum input rate (°/s) 1000 > 500 > 1000 > 1000 > 1000
Shock resistance Moderate Moderate > 25,000 g Good Good
a
√
For high-end interferometric FOG, the bias instability < 0.0003°/h, random walk < 0.00008 ◦ / h , and scale factor of < 0.5 ppm
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 7 of 21

Fig. 4 A very simple schematic of the main components of a FOG

It is noted that gyroscopes performance and costs are representative of the gyroscope technology and the asso-
directly related to the applications requirements, and ciated noises therewith [6].
the utilized technologies therewith. Figure 5 shows the
relation between the scale factor instability in parts per Fluid‑based angular rate sensors
million (ppm) which reflects the quality of the gyro- Whereas, for fluid-based angular rate sensors, the sub-
scopes in depicting angular rates without being sensitive categorization includes: rate integrating gyroscopes
to undesirable impacts, as a function of the bias stabil- (RIGs), dual-axis rate transducers (DART), magnetohy-
ity expressed in (◦ /h), which is intrinsically a quantity drodynamic (MHD) angular rate sensors, jet flow angular

Fig. 5 The relation between the scale factor instability in parts per million (ppm) as a function of the bias stability expressed in (◦ /h), from [6]
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 8 of 21

Fig. 6 A schematic view of the internal structure of the rate integrating gyroscope, along with a 3D perspective view of the section from an
embodiment of the sensor, from [4]

rate sensors, porous fluid flow gyroscopes, and vortex

rate gyroscopes, integrating angular accelerometers,
micromachined fluid-based gyroscopes (thermal gas
gyroscopes), and micromachined biomimetic gyroscopic
sensors.

Rate integrating gyroscopes RIGs were initially intro-

duced in the 1960s. Many US patents were introduced
in such period, developing various versions of rate inte-
grating gyroscopes, such as [7–12]. Nonetheless, RIGs do
not implement fluid as its inertial mass. However, RIGs
make use of the fluid properties, and the conservation of
momentum to measure the angular velocity. RIGs use a Fig. 7 A view of the internal structure of a DART sensor, from [4]
high-rate spinning rotor with high moment of inertia to
keep its spin axis fixed in 3-D, unless affected by exter- Dual‑axis rate transducers DART uses fluid as a part of
nal torque. In such case, the rotor would precess, and the its motion transduction mechanism. As most fluid-based
spin axis would follow the direction of the applied torque inertial sensors, DART utilizes the law of conservation of
axis [4]. Figure 6 shows a schematic view of the internal momentum for fluid bodies. To clarify, DART sensors use
structure of the rate integrating gyroscope, along with a a sphere of heavy liquid which rotated via a driver motor
3D perspective view of the section from an embodiment at high speed such that it would acquire high angular
of the sensor, from [4]. momentum. Across the contained fluid body, deflectable
paddles are fixed along the circumference of the spherical
Magnetohydrodynamic gyroscopes MHD angular rate cavity encapsulating the fluid. The paddles are connected
sensors make use of the magnetohydrodynamic effect in to piezoelectric crystals, which act as the motion pick-off
conductive fluids, such as mercury, to detect the angular mechanism. However, a DART sensor is sensitive to angu-
rates. The magnetohydrodynamic effect is the induction lar rates along two mutually orthogonal axes, which are
of electric current into a fluid, because of being subjected normal to the spin axis of the spherical cavity [4]. Figure 7
to varying magnetic fields. The induced current can depicts a view of the internal structure of a DART sensor,
polarize the flow, which in turn affect the applied mag- from [4].
netic field itself. The magnetohydrodynamic effect can be
mathematically modeled using the Navier–Stokes equa- Jet flow and vortex rate gyroscopes Jet flow gyroscopes
tions for momentum conservation for incompressible flu- measures the temperatures or pressure variations of a
ids, and Maxwell’s equations for electromagnetism. The fluid body entrapped in container. Whereby, external
variations in the magnetic field are detected as indication angular rates cause variations in either temperature or
of the external angular motion imparted to these types of pressure, which are measured, and are proportional to the
sensors [4]. applied angular rates [4].
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 9 of 21

Vortex rate gyroscopes use a container, where a 2-D than respective MEMS-based inertial sensors. How-
sink flow is generated, which causes a vortex with a well- ever, fluid-based sensors are limited in terms of sensi-
defined pattern of stream-lines. Whenever the sensor is t tivity and bandwidth, when compared to MEMS-based
at rest, the vortex pattern remains unchanged. However, sensors [24]. The concept behind those microma-
when the sensor is subjected to angular motion, the vor- chined angular rate sensors are quite straightforward.
tex pattern is changed in the form of an additional vortex Micromachined fluid-based angular rate sensors use
being superimposed to the initial vortex pattern, which either jet flow or thermal flow to measure angular
manifests as a combined vortex flow. The measurement velocities. For jet flow sensors, a micro pump is used to
of the fluid field velocities of the combined vortex flow generate a laminar gas flow that is affected by external
would lead to the determination of the imposed angular rotation of the sensor, and its disturbance is an indica-
rate to the sensor. Various designs have been reported for tion of such rotation. Figure 8 shows schematic view
the concept of vortex rate gyroscopes [13–21]. of a uniaxial micromachined jet flow gyroscope, along
with a diagram showing the jet flow with and without
Micromachined fluid‑based angular rate sen‑ applied rotation, from [16].
sors Micromachining of the fluid-based angular rate Whereas, thermal flow angular rate sensors measure
sensors have been attempted numerously. A microma- fluid flow because of thermal convection. Thermal gas
chined biomimetic fluid rotor angular rate sensor was gyroscopes utilize the same concepts used for thermal
reported by Andreou et al. in 2014 to be used for vestibu- accelerometers, explained earlier. However, in a thermal
lar prostheses [22]. It is noted that the vestibular system is gas gyroscope, the temperature sensors are mounted in
the part of the human anatomy that is responsible for the a 2-D configuration to measure rotation rates instead of
detection of the head orientation in 3-D space. The ves- linear accelerations [24]. Figure 9 a schematic view of a
tibular system is a crucial for self-motion and body bal- thermal gas gyroscope, and the shape of the temperature
ance, adjusting body posture and helps stabilize the vision profiles formed within the sensor before and after appli-
during movement [22, 23]. cation of rotation rate, from [16].
If one compares micromachined fluid-based iner-
tial sensors to MEMS-based IMUs, it is found out that Accelerometer technologies
fluid-based inertial sensors utilize fluid instead of solid The second constituent of an IMU is the accelerometer.
mass in their motion transduction mechanisms. Hence, Accelerometers are inertial sensors that measure the
they can have simpler structures and be at lower costs magnitude of an accelerating force, referred to as the

Resonance driver

Y sensors 1

pump
X ω
nozzle
Z Jet flow
Detecting Driving Detecting
chamber chamber chamber sensors 2

ωy X
Y

Z temperature profile at jet flow

approaching with rotation ω
Jet flow
Sensors Sensors
channels temperature profile at jet flow
departure with rotation ω
Jet flow without rotation

Jet flow with rotation ωy

Fig. 8 A Schematic view of a uniaxial micromachined jet flow gyroscope, along with a diagram showing the jet flow with and without applied
rotation, from [24]
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 10 of 21

Sensors 1 Sensors 3

heater
Upper silicon cavity ω
thermal flow 1 thermal flow 2
Sensors 2 Sensors 4
Chamber filled with gas

Lower silicon cavity

Heater
temperature profile
Sensor without rotation ω

temperature profile
with rotation ω
Fig. 9 A schematic view of a thermal gas gyroscope, and the shape of the temperature profiles formed within the sensor before and after
application of rotation rate, from [24]

specific force, as mentioned earlier in Sect. 0. It is noted the case with vibrating quartz accelerometers, and sili-
that the inertial sensors industry is more mature in terms con accelerometers [4]. It is noted that vibratory devices
of accelerometers in comparison to the angular rate in accelerometers design provide the basis upon which
sensors. Nevertheless, state-of-the-art accelerometers MEMS-based accelerometers are fabricated.
include a huge variety of sensors which adopt different
operation principles and technologies. State-of-the-art Optical and surface acoustic wave accelerometers
accelerometers include mechanical sensors which use Moreover, solid-state technologies include the realiza-
the classical pendulum principle, up to the modern solid- tion of optical accelerometers, such as the vibrating
state sensors. Consequently, accelerometers are found fiber optics accelerometers [4]. Such optical accelerom-
over a vast range of accuracies. eters imply the use of fiber optics in the fabrication of
the accelerometers and utilizes different approaches and
Mechanical accelerometers designs to acquire the applied acceleration to the sensor
Mechanical accelerometers can be considered of top tier from the measuring the light properties passing through
in terms of accuracy as inertial sensors. Nonetheless, various designs of fiber optics cables. Figure 11 shows
mechanical accelerometers can be realized in different a schematic of a vibratory fiber optics accelerometer.
forms with various designs. Mechanical accelerometers Another form of the sold-state technologies in manufac-
comprise a pendulum hinge mechanism, which is respon- turing linear accelerometers if the surface acoustic wave
sive to the applied linear acceleration. Such mechanical (SAW) devices, which contains a pendulum hinge mech-
sensors are, in some cases, filled with fluid to enhance the anism in the form of a cantilever beam to which a proof
damping effect within the sensor. Besides, the pendulum mass is attached; however, the pick-off mechanism in
can be constrained to very small displacements, through SAW accelerometers depends on the existence of a reso-
implementing the sensor in closed-loop configuration, nator on top of the cantilever beam, whose frequency is
with a rebalance loop existent within the sensor design. altered systematically as a function of any applied linear
Such closed-loop configuration aids in enhancing sen- acceleration [4]. Figure 12 shows a schematic 3D view of
sor accuracy and increases its input range. An example of a SAW accelerometer, modified after [4].
such mechanical accelerometers are the force-feedback It is noted that solid-state technologies enable small,
pendulous accelerometers [4]. Figure 10 shows a sche- reliable, and relatively less expensive inertial sensors to be
matic 3D section of a force-feedback pendulous acceler- realized. Besides, it is states that sold-state technologies
ometer, modified after [4]. produce sensors that are mainly operated in an open-
Another form of linear accelerometers are the accel- loop configuration. However, in some cases, closed loop
erometers that utilize the solid-state technologies. Such design configurations can be attained using solid state
accelerometers can be achieved by using various tech- technologies [4]. Table 3 shows a summary of typical
niques including the use of vibratory devices, which is
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 11 of 21

Fig. 10 A schematic 3D section of a force-feedback pendulous accelerometer modified after [4]

performance characteristics for a range of accelerometers

as per their underlying technologies, from [4].

Fluid‑based accelerometers
For fluid-based linear accelerometers, the sub-categoriza-
tion includes viscous float type accelerometers, pressure-
driven accelerometers, and micro-machined thermal
accelerometers.

Viscous float type accelerometers Viscous float type

accelerometers rely on the concept of viscous drag and
buoyant forces in a fluid medium and its interaction with
Fig. 11 A schematic of a vibratory fiber optics accelerometer, solid bodies. Viscous float type accelerometers use a float
modified after [4] placed within a chamber filled with a viscous fluid, as its
sensing element. In a viscous float type accelerometer, the
fluid filled chamber is usually rotated about the sensitive
axis to provide a vorticity capable of lifting the solid float
and keeping it buoyant within the swirling fluid. When-
ever the sensor is subjected to an acceleration along its
sensitive axis, the solid float is shifted within the cham-
ber. The shift of the solid float is proportional to the input
acceleration. Nonetheless, the shift of the float is detected
using the pick-off mechanism of the sensor, which depends
on the sensor design. Viscous float type accelerometers,
themselves, are categorized into four types, depending on
the type of pick-off mechanism utilized within the sensor
Fig. 12 A schematic 3D view of a SAW accelerometer modified after design. Viscous float type accelerometers can be induc-
[4] tive, capacitive, photo-electric, and variable resistance
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 12 of 21

Table 3 A summary of typical performance characteristics for a range of accelerometers as per their underlying
technologies, from [4]
Characteristic Accelerometer type
Force-feedback Vibratory fiber optic Vibrating quartz SAW Silicon
pendulous

Input range (g) ± 100 ± 20 ± 200 ± 100 ± 100

Scale-factor stability (%) 0.1 0.001 0.01 0.1–0.5 0.5–2
Scale-factor non-linearity (% full 0.05 0.05 0.05 < 0.1 0.1–0.4
scale)
Fixed bias (mg) 0.1–10 1 0.1–1 < 0.5 < 25
Threshold (μg) 10 1 < 10 1–10 1–10
Bandwidth (Hz) 400 100 400 400 400

direct measure of the linear acceleration. To clarify, it is

known that from the Navier–Stokes equations that the
pressure gradient along with the viscous forces should be
in equilibrium with the external inertial force imparted
to the fluid. Consequently, if one manages to measure the
pressure gradient, for a constant viscous force, the linear
acceleration can be computed thereafter from the Navier–
Stokes equations. However, the reported invention just
uses an analogy derived from Navier–stokes equation,
where it is postulated that linear acceleration is directly
proportional to the generated pressure gradient.
Nonetheless, the sensor basic internal structure consti-
tutes a porous cylindrical container, which can be made
of porous ceramics or sintered powdered metal com-
pacts. The porous container in filled with a charge of
non-wetting heavy liquid, such as mercury, metal liquid,
Fig. 13 A schematic 3D view of an example of a viscous float type or the like. The non-wetting liquid is designed, such that
accelerometer, as described in [13] it does not fill the entire porous cylindrical container.
The porous container is sealed on both its bases by two
endcaps, which share an intricate design. Each end cap is
accelerometers [25–32]. It is noted that viscous float type designed to have a means of gas inflow that is connected
accelerometers are found in both open-loop and closed- to a plenum fitted therewith. The plenum is then con-
loop design configurations. The common limitation of nected to the non-wetting liquid through a series of pas-
viscous float type accelerometers is being affected with sages, and orifices. It is noted that gas inflow means has a
the thermal variations which have huge impact on the series of filters, valves, and inlet metering orifices to keep
fluid viscosity, which affected the sensor output. Figure 13 track of the mass flow rate being diffused through the sys-
shows a schematic 3D view of an example of a viscous tem. Besides, the gas plenums are also provided with gas
float type accelerometer, as described in [31]. outflow orifices, which are connected externally to a set
of pressure manometers to measure the pressure differ-
Pressure‑driven accelerometers Another form of fluid- ences. The pressure manometers are arranged, such that
based linear accelerometers was introduced by Schmidlin there is a pressure manometer that is connected to both
[33], and is referred to hereafter as pressure-driven lin- gas outflow orifices from both end caps. Whereas, there
ear accelerometer. The sensor was designed as an open- are other two pressure manometers, where each is con-
loop system, where the sensor was designed to not have nected to one of the end caps, with their other branches
a feedback mechanism, as the sensor does not contain a subjected to atmospheric pressure to measure the abso-
proof mass to detect the linear acceleration to which it lute pressure of the gas within each end cap. The series of
is subjected. The scientific concept behind the sensor pressure manometers represent the pick-off mechanism
design is to use the differential hydrostatic pressure cre- of the sensor.
ated on a fluid charge entrapped within a container as a
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 13 of 21

When the pressure-driven accelerometer is station- double-checked by the two absolute pressure manom-
ary, as constant gas flow rate is pumped through the gas eters. Consequently, the linear acceleration can be com-
inflow means through the endcaps, and to the non-wet- puted depending on the proportionality with pressure
ting fluid, where the excess volume of gas is being dif- difference. Figures 14 and 15 show a schematic 3D view
fused out of the sensor through the porous walls of the of the pressure-driven accelerometer with its reported
cylindrical container. In such case, the pressure at both pick-off mechanism in [33]. However, there other designs
end caps is equal; hence the pressure gradient is equal for pressure driven accelerometers as reported in [30].
to zero, and the accelerometer theoretically reads zero-
acceleration output. Micro‑machined fluid‑based accelerometers Micro
On the other hand, when the pressure driven acceler- machined accelerometers represent a different form of
ometer is subjected lo linear acceleration along its sen- fluid-based linear accelerometers. The micro-machined
sitive axis, which is an axis parallel to the longitudinal thermal accelerometers operation principle is based on
axis of the porous cylindrical container, the liquid charge the free-convection heat transfer of a tiny hot air bubble
tends to follow the law of conservation of momentum. in an enclosed chamber. The sensor comprises a heater
Consequently, pressure builds up on one endcap, and that is placed at the center of chamber. The heater is oper-
reduces at the other endcap. This difference in pressure ated at high joule power. Adjacent to the heater, two tem-
would affect the pressure by which the gas is pumped perature sensors are placed, one at each side of the heater.
into the system, and which is monitored by the pressure The temperature sensors are designed to operate at low
manometers. To clarify, the end cap that experiences joule power. The temperature sensors are either built as
elevated pressure for the gas outflow due to motion is thermistors or thermopiles. A gas fills the chamber that
going to register high absolute pressure value through contains the heater and the temperature sensors.
the connected manometer. Whereas, on the other hand, When the sensor is stationary, the heater heats the
the other end cap would register a lower absolute pres- entrapped gas, and lowers its density. In such case, free
sure value for the gas outflow. Consequently, the gener- convection is induced, and builds a temperature profile
ated pressure difference between both endcaps can be that is symmetrical, such that the two temperature sen-
measured using the differential pressure manometer and sors read the same value. However, when the sensor is

Fig. 14 A schematic 3D view of pressure-driven accelerometer as reported in [15]

El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 14 of 21

No. Element Identification

10 Sensor Main Body – See Figure (5)
28,30 Gas Inlet Orifices
32,34 Gas Outlet Orifices
50 Conduit System
52 Gas Filter
54 Valve
56 Rotometer
58 Pressure Manometer – for Inflow Gas
60 Absolute Manometer – Outflow Gas @ 32
62 Absolute Manometer – Outflow Gas @ 34
64 Differential Manometer – Outflow Gas 32,34

Fig. 15 The reported pick-off mechanism for a pressure-driven accelerometer, from [33]

subjected to lateral acceleration, the temperature sensors thermal accelerometer was introduced by Leung et al. in
read different values. Hence, the temperature profile is 2011, which constituted a buckled cantilever assembly
deflected towards a direction opposing to the direction upon which the heaters and sensors are mounted [36].
of the applied acceleration. Nonetheless, the temperature Figure 16 shows a schematic view of a uniaxial thermal
difference is fed into a conditioning circuit which gener- gas accelerometer, along with the form of the tempera-
ates and electric signal that is proportional to the applied ture profiles as measured by the deployed thermal sen-
acceleration. sors before and after motion, as depicted in [16].
The first thermal convective single axis accelerometer
was introduced by Leung et al. in 1997 [24, 34]. However, Future trends in inertial sensors technologies
a dual axis version of the sensor was introduced by Leung This section provides a summary of the future trends
et al. in 1998 [24, 35]. The sensor followed the same in inertial sensor technologies, and the recent advance-
operation concept, and motion transduction mechanism. ments occurring in designing and fabrication of inertial
However, instead of using a pair of thermal sensors, the sensors of lower cost, higher performance, and smaller
dual axis thermal accelerometer utilized four thermal in size, in comparison to current state-of-the-art inertial
sensors that are placed on two orthogonal axes in a com- sensors. The advancements in inertial sensor technolo-
mon plane. Hence, the sensor can sense accelerations gies would enable the use of inertial navigation in a wide
along two orthogonal axes. Advancements were made area of applications.
to this form thermal accelerometers to try and enhance
their performance. Attempts have been made to reach
a monolithic triaxial thermal accelerometer. A triaxial

Fig. 16 Schematic view of a uniaxial thermal gas accelerometer, along with the form of the temperature profiles as measured by the deployed
thermal sensors before and after motion, as depicted in [24]
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 15 of 21

Nuclear magnetic resonance gyroscopes The cold-atom inertial sensors rely on laser cooling of
The physical phenomenon governing the operation of atoms, without cryogenic cooling. To clarify, subjecting
nuclear magnetic resonance (NMR) gyroscopes is the the atoms of some materials to laser beams at frequen-
quantum spin of atoms nuclei [2]. In quantum mechan- cies close to the atomic resonance of said materials would
ics, it is well-known that many atoms exhibit a property trap atoms. Thus, the atoms would lose their kinetic
referred to as nuclear spin. A nuclear spin entails that a energy and drive their temperatures to micro-Kelvin
nucleus, of a given atom, acts as if it were spinning. This ranges [2].
is the case in most atoms with odd mass, odd atomic Consequently, the trapped atoms would show some
number or both [37]. However, the property of nuclear quantum mechanical behaviours, such that their states
spin do not occur for atoms with even mass number or could be altered with various techniques. The trapped
atomic number or both. These atoms are characterized atoms could be manipulated by the application of tuned
by having quantized spin angular momentum and mag- laser pulse and letting the atomic wavefunctions interfere
netic moment, which is calculable. The relevant aspect of after moving along a designated path in space [40], in a
the nuclear spin is that the nuclear spin is affected under similar way to that way in which FOG operates; however,
the impact of external magnetic field. Hence, one can instead of using light waves, the atomic wavefunctions
drive a transition between different spin states by apply- are used.
ing radio-frequency electromagnetic radiation [38]. Other than implementing the cold-atom technique in
To clarify, the nuclear spin is an indication of the a similar way of interfering wavefunctions (as mentioned
total angular momentum of the nucleus. Although, the above), there are various ways to implement the cold-
nucleus is composed of neutrons and protons, it acts as a atom technique for inertial measurements. One way is
single entity with an intrinsic angular momentum, which that if the trapped atoms are released from the laser cool-
is the nuclear spin. The nuclear spin is dependent on the ing, they would act as free-falling masses. Afterwards, the
mass number of the atom. acceleration of the inertial sensor case can be measured
Consequently, if the nuclear spin could be affected relative to these atoms. However, other approaches have
by the externally applied electromagnetic radiation, been implemented to manipulate the trapped atoms and
the nuclear spin could be polarized, and some how the use them inertial measurements. It has been reported
nuclear spin states could be manipulated. Hence, it has that the atoms can be set to a fixed velocity or can be
been concluded that a spin polarized nucleus will precess guided for acceleration ad rotation measurements [41].
at a certain frequency when subjected to a magnetic field. Nevertheless, cold-atom inertial sensors are charac-
The precession frequency is known as Larmor frequency terized of having high performance in comparison to
and depends on the atom whose spin is polarized, and on typical optical gyroscopes, because the effective atom
the magnetic field applied to the atom [38]. Nonetheless, wavelength is less than that of RLG of FOG by ten orders
the rotation of a cell containing spin-polarized-nuclei of magnitude. Additionally, due to low temperature of
will modify the precession frequency proportional to the the trapped cold atoms, they provide low noise meas-
applied angular rate. urements and high signal to noise ratio (SNR) relative to
NMR gyroscopes have reached considerable develop- optical inertial sensors [2].
ment in major aspects of its design and implementation.
Specifically, research advancements have led to enhanced Micro‑opto‑electro‑mechanical‑systems (MOEMS) inertial
method of controlling the atomic spin, applied magnetic sensors
field, and the development of better pick-off mechanisms In contrast to MEMS-based inertial sensors, which uti-
to record the gyroscope signal. lize capacitive pick-off mechanisms, a new approach
It is stated that NMR gyroscopes approaches naviga- emerged known as micro-opto-electromechanical sys-
tion requirements in terms of error budget [39]. Besides, tems (MOEMS). MEMS-based IMUs usually include an
NMR gyroscopes are unaffected by vibrations because electronic capacitive means by which the motion of the
of the motion transduction mechanism of such sensors, micro inertial seismic mass is detected and transformed
which do not require a solid proof mass. To sum up, into signals. However, MOEMS-based inertial sensors
NMR gyroscope technology is expected to reach high utilize optical pick-off mechanisms which eliminate a
levels of accuracy in high dynamic applications within a vast range of errors that occur in MEMS-based sensors.
decade [2]. Various pick-off mechanisms have been introduced and
researched. According to [4], the optical pick-off mecha-
Cold atom inertial sensors nisms are classified into interferometric approaches or
Another recent technology that has been implemented attenuation approaches. The interferometric pick-off
in inertial sensors fabrication is cold-atom technology. mechanisms utilize similar concepts applied in FOG,
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 16 of 21

which offer low noise and high-resolution inertial sen- PIVINS consists of two triads of fluid flow channels
sors. On the other hand, attenuation pick-off mechanisms place along 3 mutually orthogonal planes, with an imag-
use means to interrupt the light beam from a diode. ing sensor that is facing each plane to track a particle that
Additionally, the optical pick-off mechanism is also is placed in each flow channel. One triad represents a set
researched for optimizing the installation and harmo- of orthogonal gyroscopes, and the second triad represent
nization of the optical source (i.e., light source) and its the accelerometers. The particles are chosen to be of the
detector. Nonetheless, MOEMS sensors are also opti- same density of the fluid filling up the flow channels to be
mized to achieve low-cost manufacturing while achieving neutrally buoyant.
high accuracies [42–44]. Two patents have been filed and are currently pending,
and research is currently ongoing to reach a fully func-
Particle imaging velocimetry inertial sensors tional inertial navigation system [45, 46]. The PIVINS is
Recently, a fluid-based inertial sensors technology nearly a drift-free inertial navigation system with minor
is being developed by the Mobile Multi-sensor Sys- bias instability. Whereby, The PIVINS endures only sys-
tems research group at the University of Calgary. The tematic errors that can be modeled either analytically or
research proposes a fluid-based inertial navigation sys- experimentally. Hence, the PIVINS is comparable at each
tem, referred to as particle imaging velocimetry inertial level to a high-end tactical grade IMU, and yet extremely
navigation system (PIVINS), depends on fluid dynam- efficient in terms of bias instability. It is expected that the
ics to provide the inertial measurements. The concept PIVINS reaches such level of performance with a simple
is that when an inertial force is imposed upon a control low-cost design.
volume of fluid, a flow is generated in response to such
force in order to satisfy Newton’s second law of motion. Hemispherical resonator gyroscope technology
Hence, once the flow can be detected, and through the Hemispherical resonator gyroscope (HRG) technology
implementation of fluid dynamics theories, the actual follow the same underlying scientific principles imple-
inertial measurement can be acquired. The fluid dynam- mented in vibratory gyroscopes, as discussed earlier.
ics theories, which are of interest to design this system, The HRG depends on Coriolis force in detection of input
are the law of conservation of momentum implemented angular rate. Hence, HRG technology lies within the
for incompressible fluids, or alternatively referred to as more specific taxonomy of gyroscopes, which is referred
Navier–Stokes equations, and the continuity equation to as Coriolis Vibratory Gyroscopes [47]. The underly-
for an incompressible fluid. The concept is applicable for ing scientific principle states that a supported resonat-
either linear or rotational types of motion. ing object at a given frequency would remain resonating
From the proposed nomenclature, PIVINS employs a along the same plane even if its supports are rotated.
version of particle imaging velocimetry to perform the Typically, an HRG includes a hemispherical dome res-
inertial measurements. Particle Imaging Velocimetry is a onator of a piezo-electric material as an inertial sensing
branch of fluid dynamics science in which the properties element that is resonated at a predetermined frequency
of fluids and fluid flows can be determined through track- generating a standing wave across the hemispherical
ing particles that are neutrally buoyant, which mimic the dome [4]. HRG utilizes a piezo-electric (for example:
actual dynamics of the flow. Particle imaging velocime- quartz) which senses the locations of the nodes and anti-
try implies the use of an imaging sensor that are used to nodes of the standing wave pattern of the hemispheri-
track the particles, whether those imaging sensors are set cal resonator. To clarify, the pick-off mechanism utilizes
up in fixed or moving positions along the examined flow. a series of pick-off electrodes that are placed across the
Research has been extensive in particle imaging veloci- outer circumference of the resonator. The electrodes and
metry with vast technical advancements in digital image the resonator act as a series of capacitors that produce
processing to acquire more precise, computationally effi- capacitive electrostatic charges between the metal coated
cient, and high rates of data. surfaces on the quartz components. Another use of the
However, in this adoption of the concept within PIV- pick-off electrodes is to sustain the standing wave pat-
INS, only one particle is monitored within a predefined tern. Besides, the resonator driving mechanism, which
control volume flow channel, and the particle monitoring takes the form of a forcer electrode ensures that hemi-
is done using a fixed imaging sensor with respect to the spherical dome resonates at a specific frequency. Fig-
pre-located and fixed flow channel. The motion of such ure 17 shows a schematic of the HRG main components
particle is determined from the acquired sequence of and main operation concept, as found in [4].
images via a series of digital image processing techniques. When the HRG is stationary, the sensor is designed
such that the resonator generates a standing wave whose
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 17 of 21

nodes are located midway between the successive pick- fused quartz for optimal isotropy and minimizing energy
off electrodes, as shown in Fig. 17. However, when a rota- dissipation. Finally, the metallic coating of the electrodes
tion is applied to the HRG a tangential force develops due ensures minimal energy dissipation within the electrodes.
to the generated Coriolis acceleration, which could cause The HRG technologies was first developed in the 1980s
the standing wave pattern to shift along the circumfer- by Delco (now the Northrop–Grumman Corporation,
ence of the resonating hemispherical dome, as shown in Litton Systems) [4, 48]. Their HRG sensors has 58 mm
Fig. 17. In such case, the shifted standing wave pattern is resonator diameter and had a bias instability 0.0001 ◦ /h,
rotated at a precession angle with respect to the station- scale factor accuracy less
√ than 1 ppm, and an angular ran-
ary standing wave pattern. The precession shift angle is dom walk of 0.0008 ◦ / h.
proportional to the input rotation rate. Afterwards, attempts have been made to scale down
The main advantages of the HRG design over other the size of the HRG to be suited for bore hole applica-
CVGs lies within some key aspects. The HRG ha a reso- tions. Besides, the HRG needed to be accommodate
nator that is axisymmetric about the sensor’s rotation mechanical shock and vibration [49]; however, the results
axis. Such setup ensures optimal decoupling between did not meet the cost requirements for mass production
the resonator and the outer world. This can be achieved of HRG. Whereas, attempts have been made in Russia to
when the resonator’s support is placed at a vibration achieve HRG, which have reported an HRG of 30 mm
node of the generated standing wave pattern. Besides, the resonator diameter [50].
resonating flexural waves are controlled by electrostatic Recently, SAFRAN, a leading international corporation
forces, which minimizes the requested energy and mini- specializing in aerospace, aviation, guidance and navi-
mizes the errors induced by the electronics imperfec- gation applications, have developed a new HRG design
tions. Additionally, the topology of a hemispherical dome that ensures achieving a gyroscope of higher perfor-
enables each gram of the resonator to flexural energy mance, smaller size and a lower cost. The SAFRAN HRG
storage, in comparison to other CVG resonators. Moreo- design is advantageous over its predecessors due to some
ver, the hemispherical resonator is made of amorphous optimizations and enhancements that SAFRAN have

Fig. 17 Shows a schematic of the HRG main components and main operation concept, as found in [4]
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 18 of 21

implemented to the scientific operation principle of the Moreover, inertial sensor errors vary as per their sources,
HRG. which is mainly attributed to their operation concept and
SAFRAN exploited the fact that the performance of an their underlying technologies. Additionally, such errors
HRG does not depend on the size of its resonator. Rather, affect the classification of the IMUs as per magnitude
the performance of the HRGs depend on the Q-factor within the inertial sensors’ measurements.
and the quality of how the flexural standing waves are Consequently, inertial sensors errors can be, in some
driven. Besides, they have found a solution to the major sort, a measure for the sensor performance. Hence, such
drawback of the HRG, having high manufacturing costs. errors, or performance measures, provide a basis of com-
The high manufacturing costs of the HRG result from parison between inertial sensors technologies. Never-
the requirement of the high precise manufacturing and theless, all inertial sensors share a common ground of
assembly of the sensor [47]. discrepancies that are inevitable.
Therein SAFRAN introduced their patented electrodes The institute of electrical and electronics engineers
support design that ensures that the resonator driving (IEEE) have published a standard for inertial sensor ter-
forces are generated at the equatorial plane of the hemi- minology that include the scientific definitions for ter-
spherical dome such that the assembly of the resonator minologies that are encountered, when studying inertial
is only crucial along the 2D equatorial plane [47, 51]. sensors operation [52]. Nonetheless, the standards pro-
Hence, the resonator becomes isotropically gapped from vide a set of definitions to the most common inertial
the standing flexural wave driving electrodes. sensor performance measures. These performance meas-
SAFRAN’s HRG reached a size of 20 mm resonator at ures are grouped hereafter to provide a basis upon which
low cost that can satisfy the demand of the mass mar- inertial navigation technologies, that are discussed after-
ket. SAFRAN claims that their HRG can replace optical wards, are evaluated. Interested reader is referred to the
technologies such as FOG and RLG and even replace IEEE standards document [52] for complete definitions
Cold Atom technology, as well. SAFRAN’s HRG has been of said performance measures.
implemented and tested for various applications includ-
ing but not limited to: strategic, navigation and tactical Stochastic error modeling for inertial navigation
grade IMUs, cost -effective marine compasses, tripod Inertial navigation systems provide high accuracy naviga-
mounted north finders, space launchers, INSs for com- tion states over short time intervals. The high-accuracy
mercial aircrafts, weapons guidance systems. navigation solution is function of the grade of the uti-
Such applications could only be fulfilled by sensors lized IMU. Nonetheless, whenever the operation period
of extremely high performance. As per SAFRAN, their of IMUs increase, the error budget within the acquired
HRG√ is reported to have angular random walk 0.0002 navigation solution increases as well. Hence, to acquire
◦ / h . Besides, SAFRAN have not reported a specific an accurate navigation solution from an INS, the stochas-
value for the bias stability of their HRG on account of the tic errors within the included inertial sensors must be
requirement of a longer testing time for Allan Variance modelled.
to provide a proper estimate of the bias stability value. Generally, random errors, within inertial sensors, are
However, SAFRAN projected the bias instability value all modeled as random processes within the navigation
to reach 0.0001°/h as root mean square error (rms) over solution scheme whenever the inertial measurements
2000 h test period. are integrated with an update source to end up with an
There are various aspects that are open for intensive error bounded navigation solution [1]. However, the
research, as reported by SAFRAN [47]. Such research model determination of a certain random process to
aspects include: bias stability optimization, enhancement be included into a navigation algorithm encompasses a
of control electronics of the HRG and the implementa- major research aspect in inertial navigation. There are
tion of more adequate calibration and compensation various methods by which the stochastic errors of inertial
models. sensors can be characterized and mathematically model.
This section gives a brief discussion of the most effective
Inertial sensors performance methods of modeling inertial sensor errors.
As stated earlier, inertial sensors are categorized, as
per their performance, into different grades which are Autocorrelation function method
separated by their respective costs, and suitable appli- As mentioned above, the inertial sensors random errors
cations. The basis upon which IMUs classifications are are incorporated into navigation filters (e.g., the Kalman
built are the errors that occur within their output meas- filter) as random processes. The random processes are
urements. Such errors vary between inertial sensors in expected to describe how the inertial sensor errors evolve
terms of their nature being systematic or random errors. over time. Therefore, it is crucial to accurately determine
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 19 of 21

the parameters of the chosen random processes that are Nonetheless, the ACF endures some drawback when it
used to describe the inertial sensors errors. Otherwise, comes to implementation. The common practice in iner-
the estimation algorithm would yield inaccurate naviga- tial navigation is that the random process parameters,
tion solution, which would grow even less accurate over which are deduced based on ACF are usually non-opti-
time. mal and require some tuning when being used in navi-
Consequently, it is a major aspect of performing navi- gation filters. Besides, ACF method would render itself
gation applications to better describe the inertial sensors inadequate when used for modeling inertial sensor errors
error terms and include them in the navigation filter to in high dynamics applications, or for higher order ran-
be implemented. One of the most popular methods in dom processes. On the other hand, when using the PSD
estimating the stochastic errors in inertial sensor meas- method, it had been found to be sufficient for high fre-
urements is deducing the autocorrelation function (ACF) quency noise processes, such as White Noise. However,
for the inertial sensor error signal. PSD lacks accuracy when it comes to determining the
The stochastic errors of inertial sensors are typically low frequency noise parameters [54].
characterized by being correlated over time [1]. This is
the reason behind modeling the stochastic inertial errors Allan variance method
as random processes. Besides, as a general assumption, There have been several variance techniques that could
the inertial sensor errors are assumed to be stationary be used for stochastic modeling of errors in inertial
processes, which implies that the statistical properties of sensors. The simplest of those techniques is the Allan
the error signal of any given inertial sensor would be the variance, reported initially in [55]. The Allan variance
same over time [53]. The most common random process method utilizes the root mean square (RMS) random-
which are considered when using the ACF method is the drift error as a function of averaging time. For a detailed
1st order Gauss Markov. Whereas, most inertial sensor description of the derivation of the Allan variance, the
stochastic errors are assumed to follow a 1st order Gauss reader is referred to [55, 56]. However, by computing the
Markov process. Allan deviations across the entire length of data, a char-
When one computes the ACF of any given inertial sen- acteristic curve is acquired, which is usually represented
sor error signal, the underlying time-correlated low fre- in a log–log scale, showing the averaging time versus the
quency random process could be deduced. Moreover, Allan deviation values. By further examining the charac-
the parameters that could describe such random process teristic curve, one could discern the major error compo-
(usually, 1st order Gauss-Markov process) could be easily nents within the acquired inertial sensor data.
acquired from the auto-correlation sequence. Allan variance can be used to identify the underlying
To compute the autocorrelation function of experi- stochastic errors that could occur in the read-out signals
mental inertial data, a substantially long period, which of inertial sensors. Besides, Allan variance provide the
could reach 8 h, of inertial sensor data is acquired in a capability of classifying the stochastic error components.
well-known controlled environment. The acquired iner- To be specific, the Allan variance method can character-
tial measurements should have extremely accurate refer- ize five stochastic error terms that occur in inertial sen-
ence values. Hence, it is usually recommended to acquire sors measurements, which are: the quantization noise,
long periods of inertial measurements while the sensor the random walk, the white noise, bias in-stability, and
is stationary in a laboratory environment. Afterwards, rate ramp [57].
to get rid of the high frequency noise, which inversely Nevertheless, once the Allan variance is computed for
occur over short period of time, denoising is imple- the different error terms, the acquired Allan variance
mented using wavelet denoising. It is noted that the level can be utilized through different models to compute the
of wavelet denoising should be considered not to affect parameters that could better represent the random pro-
the required sampling rate for the respective inertial cess that could be incorporated into a navigation filter.
navigation application for which the inertial sensors are Besides, the Allan variance provide information on the
intended. The detailed method of deriving the ACF for type and magnitudes of various error terms. The Allan
inertial measurements is provided in [53]. variance method has been used to identify and charac-
In some instances, the power spectral density (PSD) terize the stochastic error components in various inertial
function is drawn instead of the autocorrelation of the sensors of different grades, as reported in [49, 50].
inertial error time series. However, the PSD only repre- However, the main drawback of the Allan variance
sents the Fourier transform (FT) of the autocorrelation method is being statistically inconsistent in modeling
function. Consequently, the drawn information about the stochastic errors in small and low-cost MEMS-based
stochastic error parameters should coincide from using IMUs [54].
wither approaches to acquire them [53].
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 20 of 21

Generalized model of wavelet moments (GMWM) method transformed from being bulky in size and high in cost
The generalized model of wavelet moments (GMWM) into reliable, small, and low-cost systems, that could
method is a statistical method that has been utilized be utilized for high accuracy navigation applications.
recently for inertial sensors stochastic error modeling. Besides, future trends in inertial navigation technologies,
It represents a futuristic trend in inertial sensor error which are under continuous development, provide con-
modeling techniques. However, the GMWM method was siderable promise in terms of the achievable size, weight,
introduced initially in [58]. power consumption, and accuracy. Ergo, INSs could be
The GMWM is an estimation method which uti- extended to further applications with high reliability and
lizes the Generalized Method of Moments by using the accuracy even without the availability of aiding or fusion
wavelet variance (WV) as an auxiliary parameter [54, with other navigation systems such as GNSS.
59]. Whereby, the GMWM utilizes the relation between Furthermore, a brief overview has been introduced on
the WV and the parameters of random processes, to state-of-the-art techniques that are being used to model
be included in stochastic model, to estimate the latter the stochastic errors of inertial sensors, that are intended
by minimizing the distance between the empirical and to be used in inertial navigation applications. It has been
model-based WV. Interested readers are referred to the shown that accurately modeling the inertial sensors sto-
detailed description of the GMWM method for iner- chastic error components could significantly improve the
tial sensors error modeling in [58]. However, it is noted navigation solution acquired from said inertial sensors
that the random processes that are commonly encoun- for longer periods of unaided operation. Furthermore,
tered within inertial sensor measurements, and which are the method of GMWM has been introduced, which rep-
found extensively in literature are: the quantization noise, resent a futuristic trend stochastic error modeling tech-
white noise, auto-regressive process of 1st order, and drift nique. With further enhancements being introduced to
ramp, as shown in [54]. Additionally, in most cases the the GMWM method, it is expected to have better sto-
1st order auto-regressive process could be transformed chastic models that could better characterize the iner-
mathematically into 1st order Gauss-Markov process. tial sensors errors and provide more accurate navigation
Additionally, enhancements have been made for the solutions in stand-alone operation mode of the INSs.
GMWM model to better model the stochastic error To sum up, vast research achievements have been
terms of low-cost inertial sensors. Such enhancements made into using extremely low-cost inertial sensors in
include the used of Multi-signal GMWM method. To navigation applications and the error modeling of said
clarify, the need for the Multi-Signal GMWM arises from sensors. This research endeavors aim at having higher
the fact that the GMWM method assumes a stationary accuracy navigation solution with the use of extremely
process when it comes to the nature of the error signal low-cost sensors in accuracy-critical navigation applica-
acquired from a given inertial sensor. However, this is tions including oil ang gas industry, aerospace naviga-
assumption becomes unrealistic when trying to handle tion, pedestrian dead-reckoning (PDR), wearable devices,
stochastic errors of low-cost MEMS-based IMUs. Conse- indoor navigation applications, unmanned aerial vehicles
quently, the Multi-signal GMWM provides an approach (UAVs), and self-driving cars.
to overcome the stationarity assumption and enables
Acknowledgements
handling non-stationary inertial error signals. Further Not applicable.
explanation and discussion of the method can be found
in [54]. Authors’ contributions
NS identified the relevant resources and references; additionally, NS devised
Nevertheless, The GMWM method has been proven the manuscript structure and general contents and structure and writing parts
effective and advantageous over typical methods includ- of the manuscript. AY assisted in summarizing and writing the manuscript.
ing the Allan variance method in terms of its accuracy in Both authors have read and approved the final manuscript.
characterizing the inertial sensors error components and Funding
the computed parameters are more suitable for direct This work was supported by Dr. Naser El-Sheimy research funds from NSERC
implementation in navigation filters. and Canada Research Chairs programs (Grant No. RT691875).

Availability of data and materials

Summary Data sharing is not applicable to this article as no datasets were generated or
In this review paper, inertial navigation technologies analysed in this review article.
have been broadly covered. Nevertheless, it has been Competing interests
shown that the inertial navigation technologies have sig- The authors declare that they have no competing interests.
nificantly matured. Whereas, inertial sensors have been
Received: 2 July 2019 Accepted: 29 November 2019
developed enough to meet the high-accuracy demand
of broad navigation applications. Hence, the INSs have
El‑Sheimy and Youssef Satell Navig (2020) 1:2 Page 21 of 21

References designed for automated wirebonder assembly. In IEEE 21st international

1. Noureldin, A., Karamat, T. B., & Georgy, J. (2013). Fundamentals of inertial conference on micro electro mechanical systems, 2008. MEMS 2008, pp.
navigation, satellite-based positioning and their integration (1st ed.). Berlin: 880–883.
Springer. 37. Povh, B., Rith, K., Scholz, C., Zetsche, F., & Rodejohann, W. (2015). Particles
2. Tazartes, D. (2014). An historical perspective on inertial navigation and nuclei: An introduction to the physical concepts (7th ed.). Berlin:
systems. In 2014 international symposium on inertial sensors and systems Springer.
(ISISS), pp. 1–5. 38. Duer, M. J. (2007). The basics of solid-state NMR. In M. J. Duer (Ed.), Solid-state
3. Groves, P. D. (2015). Navigation using inertial sensors [Tutorial]. IEEE Aero- NMR spectroscopy principles and applications (pp. 1–72). Hoboken, NJ: Wiley.
space and Electronic Systems Magazine, 30(2), 42–69. 39. Larsen, M., & Bulatowicz, M. (2012). Nuclear magnetic resonance gyro‑
4. Titterton, D., & Weston, J. L. (2011). Strapdown inertial navigation technol- scope: For DARPA’s micro-technology for positioning, navigation and
ogy (Vol. 17). London: IET. timing program. In 2012 IEEE international frequency control symposium
5. Beckwith, T. G., Buck, N. L., & Marangoni, R. D. (1982). Mechanical measure- proceedings, pp. 1–5.
ments. Bostoon: Addison-Wesley Pub. Co. 40. Kitching, J., Knappe, S., & Donley, E. A. (2011). Atomic sensors—A review.
6. Passaro, V. M. N., Cuccovillo, A., Vaiani, L., De Carlo, M., & Campanella, C. E. IEEE Sensors Journal, 11(9), 1749–1758.
(2017). Gyroscope technology and applications: A review in the industrial 41. Gauguet, A., Canuel, B., Lévèque, T., Chaibi, W., & Landragin, A. (2009).
perspective. Sensors, 17(10), 2284. Characterization and limits of a cold atom Sagnac interferometer. Physical
7. Pittman, R. (1967). Rate integrating gyroscope. 3,359,806. December 26, Review A, 80(6), 063604.
1967. 42. Lu, S., Chen, S., & Zhao, Y. (2010). MOEMS gyroscope based on acous‑
8. Zeldman, M. I., & Feinberg, R. (1966). Fluid rate-integrating gyro. US3261213 tooptic mode coupling. In Optical sensors and biophotonics II (2010), paper
A. July 19, 1966. 79900I, p. 79900I.
9. Lahde, R. N. (1960). Integrating rate gyroscope. US2951377 A. September 43. Trigona, C., Andò, B., & Baglio, S. (2016). Fabrication and characterization
06, 1960. of an MOEMS gyroscope based on photonic bandgap materials. IEEE
10. Diamantides, N. D. (1968). Rate gyroscope. US3367194 A. February 06, Transactions on Instrumentation and Measurement, 65(12), 2840–2852.
1968. 44. Zhang, B., & Li, W. F. (2012). Development of a micro accelerometer based
11. Summers, T. O. Jr. (1960). Rate integrating gyro. US2951375 A. September MOEMS. In 2012 international conference on manipulation, manufacturing
06, 1960. and measurement on the nanoscale (3 M-NANO), pp. 250–253.
12. Huvers, M. E. (1978). Rate integrating gyroscopic aiming method and device 45. Youssef, A., & El-Sheimy, N. Gyroscope using torus shaped channels and
therefor. US4087919 A. May 09, 1978. image processing. U.S. Patent Application 62/796,231. January 2019.
13. Ogren, H. D. (1965). Vortex rate sensor. US3203237 A. August 31, 1965. 46. Youssef, A., & El-Sheimy, N. Particle based accelerometer. U.S. Patent Appli‑
14. Senstad, P. D. (1966). Magnetically and electrically rebalanced vortex rate cation 62/796,266. January 2019.
sensor. US3230765 A. January 25, 1966. 47. Delhaye, F. (2018). HRG by SAFRAN: The game-changing technology. In
15. Barrett, D. (1966). Vortex rate sensor. US3240060 A. March 15, 1966. 2018 IEEE international symposium on inertial sensors and systems (INER-
16. Camarata, F. J. (1969). Twin vortex angular rate sensor. US3447383 A. June TIAL), pp. 1–4.
03, 1969. 48. Remillieux, G., & Delhaye, F. (2014). Sagem coriolis vibrating gyros: A
17. Sieracki, L. M. (1969). Vortical comparator. US3452768 A. July 01, 1969. vision realized. In 2014 DGON inertial sensors and systems (ISS), pp. 1–13.
18. Burke, J. F., Dunn, J. L., & Scudder, K. R. (1969). Pick-off for fluid angular rate 49. Loper, E. J., Lynch, D. D., & Stevenson, K. M. (1986). Projected performance
sensor. US3454023 A. July 08, 1969. of smaller hemispherical resonator gyros. Presented at the PLANS’86—posi‑
19. Neradka, V. F. (1970). Ac vortex rate sensor. US3529613 A. September 22, tion location and navigation symposium, pp. 61–64.
1970. 50. Bodunov, B. P., Lopatin, V. M., Bodunov, S. B., & Kovshov, G. N. (1999).
20. Heilmann, T. G., Kwok, C. C. K., & Lapinas, Z. J. (1971). Angular rate sensor. Gyroinclinometer for surveying during the drilling process. In DGON
US3604273 A. September 14, 1971. proceedings, gyro technology symposium, Stuttgart.
21. Frederick, G. L. (1990). Constant gain laminar jet angular rate sensing device. 51. Jeanroy, A., & Leger, P. (2002). Gyroscopic sensor and rotation measurement
US4945764 A. August 07, 1990. apparatus constituting an application thereof. US6474161B1, November 05,
22. Andreou, C., Pahitas, Y., & Georgiou, J. (2014). Bio-inspired micro-fluidic 2002.
angular-rate sensor for vestibular prostheses. Sensors, 14(7), 13173–13185. 52. (2009). IEEE standard for inertial systems terminology. IEEE Std 1559-2009,
23. Andreou, C. M., Pahitas, Y., Pilavaki, E., & Georgiou, J. (2013). Bio-mimetic pp. c1–c30.
gyroscopic sensor for vestibular prostheses. In 2013 IEEE biomedical 53. Nassar, S. (2003). Improving the inertial navigation system (INS) error model
circuits and systems conference (BioCAS), pp. 17–20. for INS and INS/DGPS applications. PhD. Thesis, University of Calgary,
24. Liu, S., & Zhu, R. (2017). Micromachined fluid inertial sensors. Sensors, Department of Geomatics Engineering, Calgary, Alberta, Canada.
17(2), 367. 54. Radi, A., Bakalli, G., Guerrier, S., El-Sheimy, N., Sesay, A. B., & Molinari, R.
25. Bosch, M. T., & Kishel, J. F. (1960). Accelerometer. 2,943,493. July 05, 1960. (2019). A multisignal wavelet variance-based framework for inertial sen‑
26. Lees, S. (1961). Null type integrating accelerometer. US2988920 A. June 20, sor stochastic error modeling. IEEE Transactions on Instrumentation and
1961. Measurement, 68(12), 4924–4936.
27. Orrange, R. J. (1961). Integrating accelerometer with digital readout. 55. Allan, D. W. (1966). Statistics of atomic frequency standards. Proceedings of
2,993,382. July 25, 1961. the IEEE, 54(2), 221–230.
28. Parker, B. (1962). Gas bearing accelerometer. 3,068,704. December 18, 1962. 56. El-Sheimy, N., Hou, H., & Niu, X. (2008). Analysis and modeling of inertial
29. Aske, V. H. (1964). Accelerometer. 3,142,990. August 04, 1964. sensors using Allan variance. IEEE Transactions on Instrumentation and
30. Entin, L. P. (1965). Accelerometer. 3,175,404. March 30, 1965. Measurement, 57(1), 140–149.
31. Bentley, E. P., & Speas, C. A. (1965). Acceleration sensitive device. 3,195,357. 57. Hou, H. (2004). Modeling inertial sensors errors using allan variance. M.Sc.
July 20, 1965. Thesis, University of Calgary, Department of Geomatics Engineering,
32. Raymond, H. A. (1970). Fluid accelerometer. 3,550,457. December 29, 1970. Calgary, Alberta, Canada.
33. Schmidlin, A. E. (1971). Flueric accelerometer. 3,577,786. May 04, 1971. 58. Guerrier, S., Skaloud, J., Stebler, Y., & Victoria-Feser, M.-P. (2013). Wavelet-
34. Leung, A. M., Jones, J., Czyzewska, E., Chen, J., & Pascal, M. (1997). variance-based estimation for composite stochastic processes. Journal of
Micromachined accelerometer with no proof mass. In Technical digest, the American Statistical Association, 108(503), 1021–1030.
international electron devices meeting, 1997. IEDM’97, pp. 899–902. 59. Hansen, L. P. (1982). Large sample properties of generalized method of
35. Leung, A. M., Jones, J., Czyzewska, E., Chen, J., & Woods, B. (1998). moments estimators. Econometrica, 50(4), 1029–1054.
Micromachined accelerometer based on convection heat transfer. In
Proceedings of the eleventh annual international workshop on micro electro
mechanical systems, 1998. MEMS 98, pp. 627–630. Publisher’s Note
36. Tsang, S.-H., Ma, A. H., Karim, K. S., Parameswaran, A., & Leung, A. M. (2008). Springer Nature remains neutral with regard to jurisdictional claims in pub‑
Monolithically fabricated polymermems 3-axis thermal accelerometers lished maps and institutional affiliations.