selection of the inertial measurement unit sensors for

SELECTION OF THE INERTIAL MEASUREMENT
UNIT SENSORS FOR UNMANNED AERIAL
VEHICLES
Abstract
The Inertial Measurement Unit is the heart of every robotic vehicle, because it gives essential
information for the attitude stabilization system and for the navigation system. All type of unmanned
vehicle need to have such a sensor system but they are also play important role in the case of manned
aircraft because they are the basis of the instrumented flight.
The inertial measurement unit has long history. The demand for reliable and precise inertial
measurement unit comes from the dawn of rocket technology. Since than most of the rockets, missiles,
submarines, aircraft has such a system and they are still essential part of the human or even the
unmanned spaceflight.
With the evolution of Unmanned Aerial Vehicle (UAV) and Micro Aerial Vehicle (MAV) the
requirements were changed, because beside the reliability and precision the small size, low weight and
low energy consumption became the highest priority.
There are several types of the inertial sensors even they are providing similar information they are
based on different principles and have different advantage or drawback.
This article is an overview and comparison of the inertial measurement systems considering unmanned
aerial vehicle as they primary usage.
Keywords: unmanned aerial vehicle, inertial measurement unit, robotic vehicle, UAV, IMU
Introduction
Application of unmanned aerial vehicles (UAV) becomes more and more widespread not just in
military applications but also in civilian - inspecting, rescue and observation - role. These highly
autonomous vehicles have advanced onboard digital computer and sensor system. Usually the human
operator plays a decision making and mission commanding role. All flight control and stability
operations are done onboard making the whole system more reliable and independent from the
communication connection.
There is a tendency of reducing the size of the civilian UAVs, so they can be easily delivered
(for example in a backpack or in a trunk of a car) to the operation area where they can be easily and
quickly deployed for a rescue or observation mission.
The UAV’s small size and low cost is achievable when the size, cost and power consumption
of the inertial measurement units is lowered. Therefore several new inertial measurement principles,
devices were born in the recent years.
Inertial Measurement Unit
The inertial measurement unit provides the aircraft attitude (orientation) information in an earth fixed
coordinate system and usually it also gives acceleration, velocity and earth magnetic vector (compass)
data.
This information is primary used in the flight control system to stabilize the aircraft in the air
by changing the deflection of the control surfaces or engine thrust. This means the IMU must provide
enough precise and enough frequent data for the flight control system to achieve stabile flight even in
severe weather condition.
Nowadays the primary navigational data source is the Global Positioning System (GPS),
however the information from the IMU also can be used as a navigational data source in the case of
GPS signal jamming or losing the signal completely. This is the only form of navigation that does not
rely on external references. The Inertial Navigation System (INS) method demands high precision and
very high stability over the time for the IMU sensors.
Because of the wide spectrum of the requirements the IMU sensors needs to be selected and
the complete IMU system need to be designed very carefully.
The expected basic IMU output information is shown on the Figure 1. A full (6 degree of
freedom) IMU produces three angular information around the main axes of the aircraft as well as the
acceleration information along those axes. This information can be further processed, with their
integration or derivation acceleration, speed and position values can be obtained.
Vertical axis
Longitudinal axis
Lateral axis
X
Y
(Pitch)
Z
(Yaw)
Figure 1. Aircraft orientation axes (pitch, roll, and yaw)
(Roll )
History of the inertial measurement unit
The history of the first applications of the inertial measurement unit goes back to the end of the 19th
century (R. Christensen, N. Fogh, 2008). They were simple gyro compasses and were able to
determine the direction of true north. Under WW2 the development of the INS was refined, and the V2
rocket utilizes two free gyroscopes (a horizon and a vertical) for lateral stabilization, and an
accelerometer for engine control.
Further development of the gyros lead to even more precise INS during the 1950’s. Until the
1970’s only the gimbaled (mechanical) systems had been investigated but in the late 1970’s the
development of the strapdown INS (SINS) began. In a SINS, the sensors are rigidly mounted to the
body of the vehicle, hence the name “strapdown”. The development of the SINS is primarily due to the
introduction of the Ring Laser Gyro (RLG) in the 1960’s and the Fiber Optic Gyro (FOG) in the
1970’s. These gyros eventually enabled strapdown INS to obtain a degree of accuracy comparable to
low-end gimbaled systems but with a lower price tag.
This made INS solutions applicable to military aircraft and the first commercial aircraft
Boeing 757. The advantages of a non-mechanic system with low price and low weight were the source
of this development. The lack of computer processing power postponed the introduction of SINS
system until the 1980’s. The gimbaled system still achieved better precision but the SINS had a
precision which made it applicable in lower-cost applications.
The sensor evolution continued and in the recent decade the semiconductor manufacturing
technology made possible of producing small mechanical components on silicon wafer. This
technology is called Micro-Electro-Mechanical Systems or MEMS.
Figure 2 Size comparison of an early mechanical gimbaled system1 and a recent low cost MEMS
system2
1
SPIRE (G. T. Schmid, 2009)
http://www.robotshop.com/world/content/images/sfe-ultimate-imu-triple-axis-accelerometer-gyromagnetometer-large.jpg
2
Features of IMU sensor technologies
The main features of IMU sensors are the accuracy, stability, size, cost and power consumption.
Nearly all inertial navigation systems, the largest errors are due to the inertial sensors (G. T. Schmid
2009). Whether the inertial sensor error is caused by internal mechanical imperfections, electronics
errors, or other sources, the effect is to cause errors in the indicated outputs of these devices.
For the gyros, the major errors are in measuring angular rates. For the accelerometers, the
major errors are in measuring acceleration. For both instruments, the largest errors are usually a bias
instability (measured in deg/h for gyro bias drift, or micro g (µg) for the accelerometer bias), and
scale-factor stability (which is usually measured in parts per million (ppm) of the sensed inertial
quantity). The smaller inertial sensor error provides better quality for the instruments, it improves the
accuracy of the resulting navigation solution but it produces higher the cost of the system. The next
two figures (G. T. Schmid 2009) give a general accuracy, stability overview of different sensor
technologies.
Figure 3 Comparison of gyro sensor technologies
Figure 4 Comparison of acceleration sensor technologies
Beside the mentioned features - especially in the case of UAV application - the power
consumption and the mechanical size would be another important factor. The size simply limited by
the available space in the vehicle, for example the MAVs dimensions are in the cm range. The weight
usually comes together with the size and it is again an issue in the case of MAVs where the overall
weight of the complete vehicle is in the gram range.
The next table shows the comparison of the important sensor technologies used in UAV
IMUs. As the table shows the MEMS sensors has significant advantage over other technologies almost
in all features except accuracy/stability. Which leads to the conclusion that even using MEMS sensors
has several important advantage the signal processing algorithm needs to be more sophisticated to be
able to compensate the accuracy drawback.
Sensor Type
Mechanical
Optical
MEMS
Price
Accuracy/Stability
Power
Size
++
-+
+
++
++
++
Table 1 Comparison of sensor technologies
Cost
-++
Signal processing algorithm (Sensor fusion)
One sensor is not able to produce all information needed by IMU therefore several sensors need to be
used. For the complete 3D operation all sensors must be able to measure their value along three axes.
The principal sensors are the angular velocity sensor (gyro) and acceleration sensor. The gyro
measures the angular velocity which needs to be integrated once to get the aircraft orientation. The
accelerometer measures the acceleration force which needs to be integrated twice to produce position
information.
Both sensors have many sources of the inaccuracy. The gyro has high offset and drift error
which makes the sensor to very instable for longer time.
The acceleration sensor also measures the gravity force of the earth which needs to be
considered while processing its value.
These errors can be somewhat compensated by combining their value using a sensor fusion
algorithm. The algorithm tries to filter sensor data and tries to compensate of their error using a value
from the other sensor. The error can be further reduced introducing a magnetic field sensor, which acts
as a compass i.e. measures the magnetic field of the earth. Fusion of data from all sensors might
reduce the overall inaccuracy. The fusion algorithm is usually based on Kalman-filtering or artificial
intelligence methods.
The basic results of the sensor fusion algorithm are the attitude information and speed and
acceleration information as it can be seen on the next figure.
Figure 4 Data flow of the IMU
Typical MEMS sensors for IMU application
There are several MEMS sensors suitable for IMU usage available on the market. This section shows
the basic parameters of some widely available typical IMU MEMS sensor. These sensors (gyro,
acceleration, magnetic field) have very small size, power consumption and low price. As it was
mentioned earlier their accuracy needs to be improved in the signal processing chain.
Three axis acceleration sensor
•
•
•
•
Three axis gyroscope
1.8V to 3.6V supply
Low Power: 25 to 130uA @ 2.5V
SPI and I2C interfaces
Up to 13bit resolution at +/-16g
• Digital-output X-, Y-, and Z-Axis angular rate
sensors (gyros) on one integrated circuit
• Digitally-programmable low-pass filter
• Low 6.5mA operating current consumption for
long battery life
• Wide VDD supply voltage range of 2.1V to
3.6V
• Standby current: 5µA
• Digital-output temperature sensor
• Fast Mode I2C (400kHz) serial interface
• Optional external clock inputs of 32.768kHz or
19.2MHz to synchronize with system clock
Three axis magnetometer
•
•
•
•
•
Simple I2C interface
2.5 to 3.3VDC supply range
Low current draw
7 milli-gauss resolution
Low-cost
Table 2 Typical IMU sensors3
Conclusions
The inertial measurement unit is an essential and critical part of any unmanned aerial vehicle design.
Therefore its sensors need to be selected carefully because it will affect the overall performance and
3
Source: http://www.sparkfun.com
stability of the vehicle. The mechanical (gimbal) sensors are usually can’t be used because of their big
size and high power consumption. The optical sensors are too expensive and their size and power
consumption is still not in the range as required for UAV operation and especially not suitable for
micro air vehicle application.
The optimal choice is the solid-state (MEMS) sensor system, which has very low power
consumption, low price and low size. However their accuracy and especially stability versus time is
worse than their mechanical counterpart. This inaccuracy can be almost fully compensated by an
appropriate and sophisticated signal processing algorithm. Considering nowadays processor power,
price and power consumption parameters this drawback can be compensated relatively easily in the
signal processing chain.
References
A. Turóczi, 2006. Pilóta nélküli légi járművek navigációs berendezései, Bolyai Szemle 2006(1): 179193.
A. D. KING, 1998. Inertial Navigation – Forty Years of Evolution, Gec Review 13: 140-149
G. T. Schmid, 2009. INS/GPS Technology Trends, Massachusetts Institute of Technology
R. Christensen, N. Fogh, 2008. Inertial Navigation System, Master project, Aalborg University