Article citation information:
Mammadov, A.
Analysis of micro-electromechanical inertial measurement units for
unmanned aerial vehicle applications. Scientific
Journal of Silesian University of Technology. Series Transport. 2022, 117, 129-138. ISSN: 0209-3324. DOI: https://doi.org/10.20858/sjsutst.2022.117.9.
Aftandil MAMMADOV[1]
ANALYSIS OF MICRO-ELECTROMECHANICAL INERTIAL MEASUREMENT UNITS FOR
UNMANNED AERIAL VEHICLE APPLICATIONS
Summary. Typically, an
inertial navigation system (INS) is used to determine the position, speed, and
orientation of an object moving relative to the earth's surface. The navigation
information (position, speed and orientation) of an unmanned aerial vehicle
(UAV) is needed to control its flight. Since the resistance of INS to
interferences is very high, it is possible to ensure reliable flights in
conditions of high-intensity noise. This article explores the principles of
constructing inertial measurement units (IMU) that
are part of the INS and indicates perspective directions for their development.
Micro-electromechanical inertial measurement units were studied in this work,
and functional and principal electrical circuits for connecting units of
inertial measurements to the microcontroller were developed. The results of
practical measurements of units without calibration and after calibration were
obtained using the created laboratory device. Based on the obtained results,
the necessity of sensor calibration was revealed, and accuracy was improved by
performing calibration with the Kalman filter
algorithm. The Kalman filter is the heart of the
navigation system. In a low-cost system, IMU errors
like bias, scale factor error and random walk noise dominate the INS error
growth.
Keywords: unmanned
aerial vehicles, inertial measurement unit, gyroscope, accelerometer,
micro-electromechanical system, roll, pitch, yaw, acceleration
1. INTRODUCTION
Imagining the modern world without
modern navigation systems is almost inconceivable, that is, without inertial
sensors covering all areas of human life. The creation and
study of this type of sensor led to their use in modern navigation complexes as
part of inertial navigation systems (INS) [1]. The
inertial navigation system consists of the inertial measuring module (IMU) or the inertial reference unit (IRU) and the navigation determinants for the calculation of
the freelance acceleration [2]. The structural scheme of INS with two
sensor types is shown in Figure 1.
Fig. 1. Structural scheme of INS with two
sensor types
When the accelerator measures the motion of the
acceleration of the air vehicle, Earth’s gravitation area also affects
its work. The navigation system, upon obtaining this
value, that is, the freelance acceleration – g, from the measured value
of the acceleration, gives the real value.
Here, the navigation system on
integrating the signal from the gyroscope and measuring the acceleration of the
object’s movement in the moment of integration, 2 time destinies the
coordinate. Based on the found coordinates there is determined the g value
and deducted from the measured acceleration. The
main function of the INS is to calculate the acceleration and information on an
aircraft’s roll pitch and yaw angular and linear acceleration and
transmit it to the appropriate systems. This
information is used for navigation calculation and roll-pitch displays [2].
Due to their complex structures and micro-size,
the development of micro-mechanical sensors and inertial measurement units (IMU) requires high technologies from countries with
developed industries. Micro-electromechanical systems (MEMS) - inertial sensors include the
gyroscope and accelerometer, as well as a combination of gyroscope, accelerometer,
and magnetometer [3, 4]. The accelerometer is used to measure inertial
acceleration. While the gyroscope, on the other hand,
measures angular rotation. Both sensors typically have three degrees of
freedom to measure from three axes. The magnetometer
measures the bearing magnetic direction, thus it can improve the reading of the
gyroscope. MEMS components are small, light,
inexpensive, and have low power consumption and short start-up times. Also, their accuracy has
significantly increased over the years.
The main function of the INS is to
calculate and transmit data to the necessary systems about the coordinates,
speed, as well as angles of roll, pitch and yaw, based on the measurement and
integration of the acceleration, and angular velocities of the UA in the
spatial coordinate system. The processing and application of such data in autonomous transport
management systems in recent years allow the achievement of successful results.
Examples of this can be the determination of the direction of autonomous
vehicles carrying cargo in ports, the successful use of unmanned vehicles with
the ability to explore submarines for various purposes, the creation of
autonomous vehicles that can move without a pilot in urban and intercity
transport or the development of bomb disposal robots. Autonomous vehicles
are the types of vehicles equipped with camera systems, sensors, controllers,
and wireless communication modules (Global Positioning System (GPS)/INS),
produced as a result of the rapid development of digital technology added to
the conventional vehicles used today.
Inertial sensors are also frequently used for
pose estimation of cars, boats, trains and aerial vehicles [4, 6].
Over the years, INS has been
improved through the electromechanical devices that control missiles to the
semiconductor devices that are now used in many modern vehicles. The INS, developed through
inertial sensors, which has passed rapid development in recent times, operating
without external influence, has now become a crucial part of aircraft, ships,
missiles, and spacecraft, as a standard part of civil and military navigation
systems. Furthermore, it seems that the systems
developed in this form, especially the autonomous systems applied to various
vehicles in recent years, have been developed, with high results achieved.
Presently, the mass production and
application of light and ultra-light unmanned aerial vehicles (UAV) tightens
the weight and size requirements for inertial navigation systems, making it
impossible to use traditional INS. Recently, a sharp increase in demand for drones has led to the expansion
and deepening of research in this direction. The development
and application of satellite-inertial navigation systems that provide the
maximum possible light, small-sized and precise controllability for this type
of vehicle are especially critical.
The rapidly developing
micro-electromechanical systems (MEMS) - sensor inertial measurement units (IMU) are already
been widely used in the navigation and control systems of missiles, and land
and air vehicles, in recent years. In particular, MEMS-based IMUs
are gaining significant application in unmanned aerial vehicles (UAV) and
robotics due to their low cost, low power consumption and small size offered by
advanced MEMS manufacturing technology. Although the MEMS-based IMU provides a reduction in size and cost compared to
traditional IMU based on fiber optic or laser
gyroscopes, it suffers from more non-linear or random errors, which causes
navigation solutions in MEMS INS to vary greatly over time [6-8].
Due to failures in the IMU
components (gyroscopes and accelerometers), it cannot show the position
perfectly. Those failures cause errors in the determined position that increase
over time. These failures can be accepted in vehicles
making short-term flights. For performing long-term
missions, the navigation system needs periodic corrective actions to bring the
failures caused by INS as close to zero as possible. Thus,
complex filtering algorithms are applied to reduce failures for data processing
in navigation systems, as well as algorithms for sensor calibration [9].
2. SUBJECT RELEVANCE
Autonomous
vehicles, which are seen as the future of many technological developments, have
become a critical element for countries that possess this technology with the
advanced features and capabilities provided by technological opportunities. The improvement and production of
these vehicles of strategic importance should be determined as a target. We have decided to work toward the development of
autonomous systems to participate in the change of this potentially strategic
situation. Also, this issue is significant for us, given its applicability
to various fields of industry.
The
purpose of this study is to analyze and evaluate the results obtained
during practical measurements with the installation of MEMS-based IMU that determine the orientation of small-sized aircraft
in space on the Arduino platform and its application. Thus, our research work is very
relevant for defining and developing the technical parameters of new-generation
navigation systems for light and ultra-light aircraft.
MEMS-based IMU
containing gyroscopes, accelerometers, and magnetometers, are widely used in
position and navigation measurement due to their increasing accuracy, small
size and low cost. However, due to various non-linear errors in MEMS-based IMU, errors of MEMS-based autonomous INS have significantly
increased over time. Therefore, IMU is calibrated and integrated with GPS to provide reliable
navigation solutions through MEMS INS.
Today,
there is an extensive choice of inertial sensors and units with different
features at different prices. Currently, many companies [10-12] produce
inertial measurement units that include gyroscopes, accelerometers, temperature
sensors, barometers, and even magnetometers. IMU
manufacturers are leading companies such as STMicroelectronics, InvenSense, Analog Devices, Honeywell, XSens,
Teknol and others. To improve the accuracy of these
types of sensors and IMUs in them, it is necessary to
perform an appropriate calibration procedure.
3. ISSUE SOLVING AND DISCUSSION
Based on our previous study, we found it more appropriate to use 3x3x1 mm MPU-9255 with a
three-axis gyroscope, accelerometer and magnetometer and 4x4x0.9
mm MPU-6050 MEMS sensors with a three-axis gyroscope
and accelerometer manufactured by InvenSense with
suitable parameters for creating a sufficiently accurate inertial measurement
system that can be applied in UAVs. Currently, the
leading company in the development of MEMS inertial sensors, InvenSense [12], manufactures MPUs
(Motion Processing Units) with up to 9 axes, installing a magnetic compass (MPU-91xx and MPU-92xx).
Inertial measurement units such as MPU-6050
with a three-axis gyroscope and three-axis accelerometer and MPU-9250 with a three-axis gyroscope, three-axis
accelerometer and three-axis magnetometer manufactured by InvenSense
were used in the study. The main advantage of these
units is providing relatively high characteristics at a low cost.
MPU-6050 type unit with a three-axis gyroscope, three-axis accelerometer and
temperature sensor is designed to detect tilt angles in the X, Y and Z axes. The supply voltage of the unit is 3.3-5 V, the measuring range of the
accelerometer is ±2, ±4, ±8 and ±16 g, and the
measuring range of the gyroscope is ±250, ±500, ±1000 and
±2000 0/sec. When the unit is placed on a flat surface, it will measure
0 g on the X and Y axis, and +1 g on the Z-axis.
Another unit used in the study is InvenSense’s
second-generation MPU-9250, with the smallest
nine-degree-of-freedom. Two crystals are connected in
the body of the microchip. One crystal houses a three-axis gyroscope and a three-axis
accelerometer, and the second crystal houses a three-axis magnetometer.
Accordingly, the MPU-9250 incorporates a nine-axis
motion tracking unit.
MPU-9250 unit supply voltage is 3-5 V, the interface is I2C
(400 kHs)/SPI (1 MHs),
accelerometer measurement ranges are ±2, ±4, ±8 and
±16 g, gyroscope measurement ranges are ±250, ±500,
±1000 and ± 2000 0/sec, magnetometer full-scale range is
±4800 μT, the size is 15x25mm.
In addition, the I2C auxiliary
port of the MPU-9250 unit is designed to connect
several non-inertial digital sensors, such as pressure sensors. MPU-9250 unit uses a total of nine 16-bit analog-to-digital converters to digitize the output data of
the gyroscope, accelerometer, and magnetometer. The data received from
the sensors is digitized in a 16-bit analog-to-digital
converter, processed by the DMP (Digital Motion
Processor) signal processor using Motion Fusion algorithms and transmitted to
an external microcontroller via the I_2C/SPI bus
interface. Motion Fusion algorithms work with all
internal sensors to collect complete data [13]. In
addition, a programming digital filter and temperature sensor are also placed
in the unit. Communication with all device
registers is done using either 400 kHs I2C or 1 MHz SPI.
To conduct this study, the
assembly diagram of connecting the IMU to the
microcontroller and the display was developed. According to the initial connection structure
on the breadboard, the microcontroller that performs the function of the
"brain" and each of the other intelligent sensors are added to the
first breadboard and connected to the digital pins by appropriate protocols
through buses in the abovementioned diagram. Pins
are assigned in the corresponding structure to the columns and rows with holes
on the breadboard, which is considered the first experimental diagram.
In addition, the main
electrical diagram for connecting inertial measurement units to the
microcontroller was built using the EasyEda online program (Figure
2).
Fig. 2. Main electrical
diagram of the IMU connection to a microcontroller and display
Arduino Nano platform was
used as a microcontroller to process data from MPU-6050 and MPU-9250 units, and LCD and Oled
displays were used to display the processed values in this diagram. Arduino Nano board with an Atmega328 microcontroller from Atmel was used in the
functional diagram. It has 14 digital I/O pins (6 of
which can be used as PWM outputs), 8 analog inputs, 16 Mhs crystal, a USB port, an ICSP
connector and a reset button. It can be
connected to a computer via a USB cable equipped with an adapter or battery.
FTDI FT232 USB-serial
converter was also included for downloading programs and connecting with a
computer.
Since the Oled display, MPU-9250
and MPU-6050 units used in the diagram use the i2c protocol, each is connected to the analog inputs A4 and A5 of the microcontroller. LCD
16x2, which is capable of displaying 16 lines 2
columns data, is connected with its 6 digital d7, d6, d5, d4,
e, rs outputs into the controller inputs d2, d3, d4,
d5, d11, d12, respectively.
Fig. 3. Laboratory
stand
After loading the created code to the microcontroller, the time diagrams
of the output values of the accelerometer and gyroscope without calibration in
the MPU-6050 unit are obtained, as depicted in Figure
4.
a)
Fig. 4. Uncalibrated
characteristic of the MPU-6050 unit on the X, Y, Z axes:
a) accelerometer;
b) gyroscope
As seen from the
characteristics, the output values of the gyroscope and accelerometer gradually
decrease along the time axis X and Z axes, while along the Y-axis, the value of
the output signal first decreases slowly, then increases at a high rate and
finally decreases again at the same rate. Unit calibration is
required to obtain more stable and accurate output signals. Thus, a Basic AHRS calibration code was used in the unit. The calibrated characteristics of the gyroscope and
accelerometer on the X, Y and Z axes in the MPU 6050 unit are presented in Figure 5.
a)
Fig. 5. Calibrated
characteristic of the MPU-6050 unit on the X, Y, Z axes:
a) accelerometer;
b) gyroscope
According to the
results of practical measurements, the average shift of the accelerometer
values of the unit without calibration was determined to be 0.0872 m/sec2 on the X-axis, 0.06 m/sec2
on the Y-axis, 0.277 m/sec2 on the Z-axis, and after
calibration, the average shift value 0.04 m/sec2 (4.7
mg) on the X-axis, 0.026 m/sec2 (2.65 mg) on the
Y-axis and 0.038 m/sec2 (3.87 mg) on the Z-axis,
which means 46% improvement as a result of the calibration. In the output signals of the gyroscope, the
average shift of the gyroscope without calibration was determined to be 0.66 deg/min on the X-axis, 1.88 deg/min
on the Y-axis, 4 deg/min on the Z-axis, and after
calibration, the average shift 0.0040 deg/min on the
X-axis, 0.0038 deg/min on the Y-axis and 0.0065 deg/min on the Z-axis, then it is indicated that the zero
shift of the gyroscope is ±20 rpm (±1200 rpm) and the shift of
the accelerometer is ±50 mg on the X and Y axes and ±80 mg on the
Z-axis in the technical characteristics of the MPU-6050
unit.
Thereafter,
measurement results were obtained without calibration by connecting the MPU-9250 module to the stand (Figure 6).
a)
Fig. 6. Uncalibrated
characteristic of the MPU-9250 unit on the X, Y, Z axes:
a) accelerometer;
b) gyroscope
Based on the results
obtained from the MPU-9250 unit, the output signals
were found to be unstable and inaccurate. Therefore,
using the unit without any calibration will make its application in a UAV
inconvenient. If we run the unit without any
calibration and read the output values, the accuracy and stability of the
result will be low. Unit calibration is required to obtain more stable and
accurate output signals. After using the calibration
code, the three-axis output characteristics of the unit were obtained (Figure
7).
a)
Fig. 7. Calibrated
characteristic of the MPU-9250 unit on the X, Y, Z
axes:
a) accelerometer;
b) gyroscope
Based on the results of practical measurements, the
average shift of the accelerometer values on the X-axis without calibration was
determined to be 0.0568 m/sec2, on the Y-axis 0.0355 m/sec2, on the Z-axis
0.6506 m/sec2, and after calibration, the average shift value is 0.0207 m/sec2
(2.11 mg) on the X-axis, 0.0238 m/sec2 (2.43 mg) on the Y-axis and 0.4206
m/sec2 (42.8 mg) on the Z-axis, which is 28% improvement as a result of the
calibration. In
the output signals of the gyroscope, the average shift of the gyroscope without
calibration was determined to be 1.68 deg/min on the X-axis,
1.52 deg/min on the Y-axis, 0.02 deg/min on the Z-axis, and after calibration,
the average shift 0.0321 deg/min on the X-axis, 0.0286 deg/min on the Y-axis
and 0.0064 deg/min on the Z-axis, then it is indicated that the zero shift of
the gyroscope is ±5 rpm (±300 rpm) and the shift of the
accelerometer is ±60 mg on the X and Y axes and ±80 mg on the
Z-axis in the technical characteristics of the MPU-9250
unit.
Fig. 8. Output values
of MPU-9250 and MPU-6050 units after calibration
As
observed from the simulation, the change of angular states on all three axes
after calibration is more stable and smoother, which shows us the possibility
of applying these modules in small-sized UAVs. From the study, IMU could be used in many applications. For future projects, this study of IMU would be applied to small-sized UAVs, which will be used as
flight stabilizer controllers.
4. CONCLUSION
The parameters of
MEMS-based IMU for light and ultra-light aircraft are
defined, and the algorithm and software of inertial navigation system
management are created. The function of IMU, which can measure the pitch, roll and yaw
of a UAV, is the main type of sensor that must be used for that application.
Besides, it can be implemented with the autopilot system.
The main advantages
of the proposed MEMS-based IMU are autonomy,
universality and durability to obstruction. IMU sensors can also be combined with other
sensors, such as GPS, for accurate navigation, guidance and controlling system.
Based on the results of this study, the calibration
herein can be considered appropriate for the use of the reviewed unit in the
integrated navigation system of light and ultra-light aircraft.
References
1.
Manon Kok,
Jeroen D. Hol, Thomas B. Schon. 2017. ”Using Inertial
Sensors for Position and Orientation Estimation”. Foundations and Trends in Signal Processing 11(1-2): 1-153. DOI: 10.1561/2000000094.
2.
Набиев Р.Н.,
А.З. Маммадов.
2021. “Модели
построения
инерциальной
навигационной
системы для беспилотных
летательных
аппаратов”
Авиакосмическое приборостроение
1: 12-22. DOI: 10.25791/aviakosmos.1.2021.1197. [In Russian: Nabiyev R.N., A.Z. Mammadov. 2021. “Models of construction of inertial
navigation system for unmanned aerial vehicle”. Aerospace Instrument-Making].
3.
Ishak Ertugrul,
Ulkir Osman. 2020. “Analysis of MEMS-IMU Navigation System Used in Autonomous Vehicles”. In:
Autonomous Vehicle and Smart Traffic.
Edited by Sezgin Ersoy, Tayyab Waqar.London, UK: IntechOpen. DOI: 10.5772/intechopen.92985.
4.
Ahmad N., R.A.R. Ghazilla, N.M. Khairi, V. Kasi. 2013.
“Reviews on various inertial measurement unit (IMU)
sensor applications”. International
Journal of Signal Processing Systems 1(2): 256-262.
5.
H. Chao, Cao Y., Chen Y. 2010. “Autopilots for
small unmanned aerial vehicles: A survey”. The International Journal of Control, Automation and Systems 8(1): 36-44.
6.
Skog I., P. Handel. 2009. “In-car positioning
and navigation technologies – A survey”. IEEE Transactions on Intelligent Transportation Systems 10(1): 4-21.
7.
Hardy J., et al. 2016. “Unmanned aerial vehicle
relative navigation in GPS denied environments”. In: 2016 IEEE/eION Position, Location and
Navigation Symposium (PLANS): 344-352. Savannah, GA. DOI:
10.1109/PLANS.2016.7479719.
8.
Perlmutter M., L. Robin. 2012. “High-performance,
low cost inertial MEMS: A market in motion!” In: Proceedings of the IEEE/ION Position Location and Navigation Symposium
(PLANS): 225-229. Myrtle Beach, South Carolina, USA.
9.
Choe N., H. Zhao, S. Qiu, Y. So. 2019. “A sensor-to-segment calibration
method for motion capture system based on low cost MIMU”.
Measurement 131: 490-500.
10. Analog Devices. 2021.
Available at: http://https://www.analog.com/en/parametricsearch/11172#/.
11. Xsens Technologies. 2021.
Available at: http://www.xsens.com.
12. İnvensense. 2021. Available at:
https://invensense.tdk.com/products/motion-tracking.
Received 20.06.2022; accepted in
revised form 05.10.2022
Scientific Journal of Silesian University of Technology. Series
Transport is licensed under a Creative Commons Attribution 4.0
International License
[1] Department of Aerospace devices,
Faculty of Physics and Technology, National Aviation Academy, Baku, Azerbaijan,
Mardakan ava.30. Email:
aftandil855@mail.ru. ORCID:
https://orcid.org/0000-0003-0842-4230