The Application of “Off-the-shelf” Components for

2014 International Conference on Indoor Positioning and Indoor Navigation, 27th -30th October 2014
The Application of “Off-the-shelf” Components for
Building IMUs for Navigation Research
Nimsiri Abhayasinghe
Department of Electrical and Computer Engineering
Curtin University
Perth, Western Australia
[email protected]
Abstract—Inertial measurement units (IMU) are commonly
used in pedestrian and robotic navigation applications and research. Although many IMUs are commercially available, almost
all of them are non-customizable and they process the collected
raw data before presenting them to the user. However, this creates
a limitation for researchers due to the fact that they have to rely
on a set of per-processed data. Further, available resources and
features such as SD card slots, wireless connectivity, available
in the IMU may not suit one’s research. This paper provides a
survey on availability and usage of different off-the-shelf devices
to build a custom made IMU. The authors considered opensource microcontroller platforms, low cost MEMS sensors and
low cost accessories in this survey so that the IMUs will be
affordable to many people. A range of sensors, their features,
available processor options and different types of wired and
wireless communication options available are discussed. Particular
emphasis is made on the ability to modify or add functionality
to commonly available hardware. Possible technical issues in
assembling the IMU and calibrating sensors are also discussed
in this paper. Technologies available for constructing a housing
and mounting systems for the IMU best suited to the application
are also discussed in this paper. As an example, IMUs developed
and implemented by the authors with different housing designs
specifically created for particular applications are presented. This
survey indicated that off-the-shelf components can effectively be
used to build custom-made IMUs to suit the particular research
interest or application best.
Keywords—Human gait analysis; inertial measurement units;
indoor navigation
Iain Murray
Department of Electrical and Computer Engineering
Curtin University
Perth, Western Australia
[email protected]
a different package or options than those currently owned,
then purchase of a new module is required. In addition to
that, performing unknown processing inside the IMU before
output is produced restricts the users’ access to raw data and
the flexibility to perform operations and computations desired.
This paper presents the outcomes of a survey conducted to
identify the off-the-shelf resources available to build an IMU
that exactly suits one’s requirements. It also presents techniques
available for building custom made housings for the IMUs
to suit the application. This paper also discusses two custom
made IMUs that are designed and build by the authors to suit
with two applications as examples. The work discussed in this
paper is a part of a project conducted at Curtin University,
Perth, Western Australia that develops a navigation aid for
vision impaired people. Two commercially available IMUs
are discussed in the “Commercially Available IMUs” section
of this paper while available electronics circuit options are
discussed in the “Options Available for Electronic Circuitry”
section. Possibilities for implementing custom made housings
are discussed in the “Enclosure Design and Implementation”
section and the IMUs built by the authors, issues faced in
building those and how they could be overcome are discussed
in the “IMUs Implemented by the Authors”.
II. C OMMERCIALLY AVAILABLE IMU S
I. I NTRODUCTION
Many researchers use Inertial Measurement Units (IMU) to
track movement of sections of body in human navigation and
tracking systems. IMUs are commonly used in other areas such
as robotics too. Usage of bare sensor boards and microcontroller
boards is not uncommon in most cases in areas like robotics.
However, securely enclosed devices will have to be used in
applications with human interaction and involvement, such as
navigation and tracking systems.
Although IMUs are commercially available in a variety of
packages, they are usually expensive and noncustomizable.
Some limitations of using such ready made IMUs are that
they often do some processing of data before presenting to the
user, availability of limited packaging options and availability
of limited options (Bluetooth, SD card, etc.). If one requires
Although many IMUs are available in the market, most of
them are either sensor ICs or development boards. Details of
these are discussed in “Sensors” sub section of this paper. Two
ready to use IMUs available in the market will be discussed in
this section. These IMUs come in a casing and easy to use.
One of them is the x-IMU by x-IO Technologies [1]. This
IMU consists of a 3–axis accelerometer, a 3–axis gyroscope
and a 3–axis magnetometer. It has an SD card for data logging
and the IMU can be connected to a computer using a USB
cable or Bluetooth. x-IO provides software needed to access
the IMU and configure it and one can develop software for this
as the software is open source. However, their firmware is not
open source, hence one has to rely on its per-processed output.
The x-IMU comes with an enclosure and a battery as shown
in Fig. 1 and priced at £309.00.
2014 International Conference on Indoor Positioning and Indoor Navigation, 27th -30th October 2014
Table I: Sensors and Their Usage
Sensor
Accelerometer
Gyroscope
Magnetometer
Pressure Sensor
Temperature Sensor
Ambient Light Sensor
Fig. 1: x-IMU [1]
The second is 3-Space IMU series by YEI Technologies.
They have several versions of the IMU with different features
such as SD card slot, Bluetooth and wireless. Each version has
its own set of features and there is no version which supports
both wireless communication and SD card. These IMUs also
have a 3–axis accelerometer, a 3–axis gyroscope and a 3–
axis magnetometer. YEI Technologies also provide software to
work with the IMUs and an API for software developing, but
the firmware of the IMUs are not open source. The Bluetooth
version of 3-Space IMUs is shown in Fig. 2 which is priced at
US$309.00.
Further technical details of these two sensors are discussed
in [3] and can be found in their websites.
III. O PTIONS AVAILABLE FOR E LECTRONIC C IRCUITRY
There are many different options available for sensors,
processing units, data storage, communication, batteries and
chargers. Some of the low cost, easy to use options are
discussed in this section.
A. Sensors
Although there are many different types of sensors and
sensor technologies available in the market, the authors have
selected Microelectromechanical Systems (MEMS) sensors in
this survey as they have become more popular and affordable
nowadays. Sensors that may be used in robotics and indoor
navigation applications are listed with their usage in Table I.
The authors have considered sensors that comes on a board
(breakout board) in this study, so that they can be easily used
to assemble the IMU.
Fig. 2: 3-Space Bluetooth [2]
Parameter Measured by the Sensor
Acceleration
Angular velocity (rotation)
Magnetic field strength
Atmospheric pressure
Atmospheric temperature
Light level
There are many MEMS accelerometers and magnetometers
produced by different manufacturers that come on breakout
boards in Sparkfun store [4]. However, there is a limited number
of MEMS gyroscopes available. The most common one is ITG3200 by Invensense [5] for which Sparkfun has a breakout
board. Although these individual sensors may be used in IMUs,
the complexity and the size of the product increases when they
are used.
One solution for this issue is to use a breakout board that has
two or three of these sensors as necessary. There are some such
boards in Sparkfun. Although this option is better than using
one breakout board per sensor, larger size of such breakout
boards limits the miniaturization of the IMU. Further, using
several inertial sensors will introduce errors due to offset of
sensors on the Printed Circuit Board (PCB). A solution for this
is to select a sensor that includes two or three inertial sensors
in a single electronic chip. One example for such a sensor is
MPU6050 by Invensense [5], which has a 3–axis accelerometer
and a 3–axis gyroscope in a single chip. A breakout board for
this is available in Sparkfun. When such a 6–axis sensor is
used, one still has to use another sensor for the magnetometer.
Depending on the application a better solution may be a 9–
axis sensor that contains 3–axis accelerometer, gyroscope and a
magnetometer. The authors of this paper have found four 9–axis
sensors during the survey. They are MPU-9150 and MPU-9250
from Invensense [5], LSM9DS0 from STMicroelectronics [6]
and BMX055 from Bosch Sensortec [7]. Breakout boards for
both MPU-9150 and LSM9DS0 are available in Sparkfun while
breakout board for BMX055 is available in ThanksBuyer [8].
All these sensor breakout boards are also available in on-line
stores [9]. Prices of these sensor boards are shown in Table II.
Some of the specification that has to be considered when
selecting a digital inertial sensor are resolution, measurement
ranges, sensitivity, zero-point offset and noise density. A comparison of these parameters for the above sensors is shown in
Table III. This indicates the the performance of each sensor of
these IMUs are slightly different. All these have comparable
measurement ranges. The sensitivity of the accelerometer of
MPU-9150 and LSM9DS0 is better than that of BMX055 where
as the sensitivities of the gyroscope of all three IMUs are in
the same range. The zero-point offset of the accelerometer
is worst in MPU-9150 and best in LSM9DS0, but BMX055
has the lowest zero-point error in the gyroscope. Although
the noise density of the accelerometer of MPU-9150 is worse
than BMX055, the noise density of the gyroscope of MPU9150 is better than that of BMX055. This implies that none
of these sensors is a best: all have some parameters better
2014 International Conference on Indoor Positioning and Indoor Navigation, 27th -30th October 2014
than the others. Because of this reason, there would not be
any major advantage or disadvantage of selecting any of these
three sensors.
There are many digital temperature, pressure and ambient
light sensors available in on-line stores. Examples are TMP102
digital temperature sensor, ISL29125 RGB Light Sensor and
MPL115A1 barometric pressure sensor [4].
B. Processors
There are many 8-bit microcontrollers available in the market. Most these microcontrollers suit the needs of an IMU,
where the main task is to read sensors and either store them onto
a storage device or send them to a computer or both. For connecting sensors, storage devices and communication devices,
these microcontrollers are required to include communication
interfaces such as I2 C (Inter Integrated Circuit) and SPI (Serial
Peripheral Interface). For one to implement the IMU without
designing and producing a PCB, a development board has to be
used. Although most microcontrollers have their development
boards, authors have selected Arduino open source platform
[10], as the best option because it is easy to use and there is a
number of different boards with different sizes and capabilities.
The Arduino platform has several available boards with
a reduced footprint, such as Arduino Micro, Arduino Nano,
Arduino Pro Mini and Arduino Fio. All these boards are about
the same with Fio being slightly larger. Micro and Nano boards
work at 5 V while Pro Mini works either at 5 V or 3.3 V
and Fio works at 3.3 V. Although 5 V devices may be used
in wired applications, 3.3 V devices are better for wireless
devices due to their suitability to work with battery voltages.
Although Micro and Nano are programmable directly using a
micro USB cable, an external programming cable or an FTDI
breakout board. The processors used in all these boards are 8bit Atmel microcontrollers with comparable performance and
all these have I2 C, SPI and serial interfaces.
C. Data Storage and Communication Options
The most common storage option for embedded devices
is the Secure Digital (SD) memory card. The version of SD
cards used in most small size devices is microSD. All SD
card versions support SPI bus mode and SD bus mode [11].
Therefore, they can be connected with a microcontroller using
the SPI bus and there is an Arduino library to work with SD
cards [10]. microSD card breakout boards are available in both
Sparkfun and in on-line stores.
There are many communication options that one can use with
an IMU. The wired options are serial (RS-232) and USB. USB
is a better option out of these two due to the fact that most
computers and laptops are equipped with USB ports and the
serial port is rarely found in laptops nowadays. Arduino boards
can communicate with a computer through the programming
interface (FTDI). If RS-232 is opted, then a level shifter has to
be used. RS-232 level shifters are available in Sparkfun and in
on-line stores.
There are few commonly used wireless communication
options that can be used in an IMU. Two commonly used
alternatives are Bluetooth and NRF24 wireless transceivers.
Both these modules are available both in Sparkfun and in online stores. Bluetooth transceivers communicate with the host
microcontroller using RS-232 interface while NRF24 uses SPI
interface. Using a Bluetooth transceiver will be advantageous
when sensor data is to be sent to a smartphone or a PDA as
these devices are equipped with Bluetooth interfaces. However,
for long rage (up to 100 m) communications, NRF24 will be
better, but it needs a separate receiving device connected with
the computer. Details of this are discussed in “IMUs Implemented by the Authors” section. The Current consumption of
the HC-06 Bluetooth modules after pairing is less that 10 mA
[12] where as for NRF24, that is about 12 mA [13].
D. Batteries and Battery Chargers
Lithium-ion polymer batteries (LiPo) are the commonly used
batteries in embedded and robotics applications. LiPo is well
suited due to the availability of capacity and dimensions - suits
housing design. One can select the shape, size and the capacity
depending on their design and time of run required. Small size
USB powered LiPo chargers are also available in the market
so that one can use a USB port or a USB charger to charge
the battery. Some battery and charger options are discussed in
“IMUs Implemented by the Authors” section of this paper.
IV. E NCLOSURE D ESIGN AND I MPLEMENTATION
Prototyping with filament type 3D printers has now become
affordable. One main advantage when using a 3D printer to
print the IMU enclosure is that a custom made enclosure
can be made that suits the application. Although high end
3D printers that give much better qualities are also available,
they are not affordable as home and hobby use 3D printers
do. Some low-end 3D printer makes are Leapfrog, Makerbot,
PrintrBot, Afinia, Rostock, Ultimaker, Reprap, Bits from Bytes,
Makergear, Airwolf3D and Bukobot. Although 3D printers are
available for prices under USD 1000, a decent quality printer
will be in the range of USD 3000 and the filament price is in
the range of USD 30-60 per 1 kg roll. The authors of this paper
have designed three different enclosure designs and printed with
a Leapfrog Creatr Dual Extruder printer, which are discussed
in “IMUs Implemented by the Authors” section of this paper.
Table II: Prices of 9–Axis Sensor Breakout Boards [4], [8], [9]
Sensor
MPU-9150
MPU-9250
LSM9DS0
BMX055
Sparkfun/ThanksBuyer Pricea.
USD 34.95
–
USD 29.95
USD 16.36
Ebay Price
USD 11.50
USD 10.00
USD 33.50
USD 13.00
a. Prices at Sparkfun and ThanksBuyer are without shipping and prices at Ebay are with shipping to Perth, Australia.
V. IMU S I MPLEMENTED BY THE AUTHORS
Two IMUs implemented by the authors for different requirements are discussed in this section. Devices used for each IMU,
the enclosure designs, some concerns when implementing IMU
and remedies for those are discussed in relevant subsections.
2014 International Conference on Indoor Positioning and Indoor Navigation, 27th -30th October 2014
Table III: Comparison of Key Specifications of the Inertial Sensors [5], [6], [7]
Accelerometer
Gyroscope
Magnetometer
Specification
MPU-9150
LSM9DS0
BMX055
MPU-9150
LSM9DS0
BMX055
MPU-9150
LSM9DS0
Resolution
N/A
N/A
0.98 mg
N/A
N/A
0.004 °/s
N/A
N/A
0.3 µT
Measurement
±2 g, ±4 g,
±2 g, ±4 g,
±2 g, ±4 g,
±250 °/s,
±245 °/s,
±125 °/s,
±1200 µT
±200 µT,
±1300 µT (x,y),
Ranges
±8 g, ±16 g
±6 g, ±8 g,
±8 g, ±16 g
±500 °/s,
±500 °/s,
±250 °/s,
±400 µT,
±2500 µT (z)
±1000 °/s,
±2000 °/s
±500 °/s,
±800 µT,
±1000 °/s,
±1200 µT
±16 g
±2000 °/s
BMX055
±2000 °/s
Sensitivity
1024 LSB/g ,
131 LSB/°/s,
8.75 m°/digit,
262.4 LSB/°/s,
8197 LSB/g,
512 LSB/g,
65.5 LSB/°/s,
17.5 m°/digit,
131.2 LSB/°/s,
0.016 µT/LSB,
4096 LSB/g,
5464 LSB/g,
256 LSB/g,
32.8 LSB/°/s,
70 m°/digit
65.6 LSB/°/s,
0.032 µT/LSB,
2048 LSB/g
4098 LSB/g,
128 LSB/g
16.4 LSB/°/s
32.8 LSB/°/s,
0.048 µT/LSB
Zero-point
± 80 mg (x, y),
Offset
± 150 mg (z)
√
400 µg/ Hz
1 µT/µT
16.4 LSB/°/s
± 60 mg
± 70 mg
± 20 °/s
N/A
√
150 µg/ Hz
√
0.005 °/s/ Hz
± 10 °/s, ±
± 1 °/s
±1000 LSB
N/A
± 40 µT
√
0.014 °/s/ Hz
N/A
N/A
0.6 µT
15 °/s, ± 25 °/s
A. Strap-Mount IMU
The requirement for this IMU was to record inertial data of
the lower human body (legs). Logging of data from multiple
sensors was also a requirement.
The sensor selected for this application was MPU-9150
mainly due to the fact that it contains all 3 sensors (accelerometer, gyroscope and magnetometer) in a single chip and hence
the size of the boards is smaller. A picture of the MPU-9150
board is shown in Fig. 3a.
A clone of Arduino Pro-mini 3.3 V was selected as the
microcontroller board because of its small size and as it is
working at 3.3 V. Although the 3.3 V version runs at 8 MHz,
it is sufficient to cater the demands of the IMU. A picture of the
board is shown in Fig. 3b. This clone of Pro-mini was selected
because of its pin placement is helpful in assembling the IMU.
nRF24L01+ RF transceiver was selected to communicate
with the host device because it can communicate simultaneously with 6 slaves. A “dongle”, which is connected with
the computer, was designed using Arduino Uno board and
nRF24L01+ to communicate with 6 IMUs and transfer data
to the computer. All IMUs are synchronized to the time of the
dongle, so that the relation between the movement of different
sections of the leg can be studied.
A 3.7 V 900 mAh flat LiPo battery of size of 48 mm
× 30 mm × 5.5 mm was used so that the circuit can be
mounted directly on the battery and the finished IMU to have
(a) MPU-9150
0.008 µT/LSB,
16384 LSB/g ,
8192 LSB/g,
1366 LSB/g
Noise Density
0.3 µT/LSB
16384 LSB/g ,
(b) Pro-mini
(c) nRF24L01+
Fig. 3: Devices used for strap-mount IMU
N/A
proper aspect ratio. The battery selected has a built-in protecting
circuit. An external USB charger was used to charge the battery,
so that the size of the IMU and internal heating is a minimum.
The charger contains a charging controller to avoid over charge
of the battery. The battery and charger are shown in Fig. 4.
The assembled IMU is shown in Fig. 5a. The enclosure was
designed to have loops to strap down. The 3D printed enclosure
is shown in Fig. 5b. The program of the IMUs was designed to
sample all sensors at 100 samples per second and send data to
the dongle and the program of the dongle was designed to read
data from 6 IMUs and send them to the USB port. A serial
reader software was used to read data from the USB port and
log them. Although the dongle theoretically support 6 IMUs,
it was observed that 4 IMUs is the practical limit for reliable
operation due to collisions and subsequent packet loss.
The dimensions of the strap-mount IMU are 55 mm × 41 mm
× 23 mm without the loops and the total cost was about
AUD 35.
B. IMU for White Cane
The second version of the IMU was designed to be attached
to the white cane to study the synchronization of the leg
movement with the white cane movement. Therefore, it was
designed to be longer, but narrower. Sample data was to be
sent to a smartphone for further analysis. Therefore, HC06 Bluetooth was selected as the communication interface as
smartphones are equipped with Bluetooth. A longer (58 mm
× 19 mm × 7 mm) 600 mAh battery was selected so that it
matches the design of the IMU. The battery and HC-06 are
(a) Battery
(b) Charger
Fig. 4: Battery and the charger
2014 International Conference on Indoor Positioning and Indoor Navigation, 27th -30th October 2014
(a) Assembled IMU
(b) IMU with enclosure
Fig. 5: Completed strap-mount IMU
(a) Assembled
IMU
shown in Fig. 6. All other units are as same as the units used
for the strap-mount IMU.
Two enclosure designs were used, one for the white cane and
the other for strapping to the thigh. The completed IMU and
the two enclosures are shown in Fig. 7. The dimensions of the
IMU for white cane are 91 mm × 25 mm × 25 mm without
the mounting mechanism and the total cost was about AUD 35.
(b) White cane mounted IMU
(c) Thigh mounted IMU
Fig. 7: Completed IMU for white cane
C. Sample Outputs
In one of the experiments, two strap-mount IMUs were
used to check the synchronization of the two legs. In this
experiments, two strap-mount IMUs were attached to the two
thighs of the subject and the subject was instructed to walk
on flat surface. The output of the two IMUs were logged in
the computer and the thigh angles were computed by fusing
the accelerometer and the gyroscope. The outcome is shown
in Figure 8. This is similar to the thigh angle waveform
discussed in [14], [15] and [16] that has been captured using
a smartphone. However, it was observed that there are packet
losses sometimes, which makes the waveform non-smooth at
some points.
Further results obtained using both these IMUs are discussed
in [17] with more details on development steps of the IMU for
White Cane.
D. Concerns and Remedies
The main concern when building both IMUs was the alignment of the sensor with the enclosure of the IMU. Three
remedial measures were taken to reduce and compensate the
alignment error. The first is to align the sensor board with the
microcontroller board as much as possible while assembling
the IMU. Solid connector pins were used to mount the sensor
on the microcontroller board so that there is no free movement
between them. Suitable holding mechanisms were designed in
the enclosures to avoid any movements of the circuit inside the
(a) Battery
(b) HC-06
Fig. 6: Devices used for IMU for white cane
Fig. 8: Thigh angle of the two legs computed from data of two
strap-mount IMUs
enclosure. As these two solutions cannot avoid misalignment
fully, an axis calibration was done after building the IMU, so
that the output of the IMU is aligned with the enclosure.
VI. C ONCLUSIONS
This paper presented a survey of available off-the-shelf
devices for building IMUs for navigation research and possible
custom-made enclosure making technologies. This also presented two IMU built for two different applications with different requirements. It is concluded that off-the-shelf components
and devices can effectively be used, without great difficulty, to
build IMUs well suited to each application with an enclosure
designed specifically for the application .
R EFERENCES
[1] x-io Technologies. “x-IMU” [Online]. Available: http://www.xio.co.uk/products/x-imu/, [July 17, 2014].
[2] YEI Technology. “YEI 3-Space Sensor” [Online]. Available:
http://www.yeitechnology.com/yei-3-space-sensor, [July 17, 2014].
[3] R. Chandrasiri, N. Abhayasinghe and I. Murray. (2013, Oct.). “Bluetooth
Embedded Inertial Measurement Unit for Real-Time Data Collection,” in
Forth International Conference on Indoor Positioning and Indoor Navigation (IPIN2013)[Online], Montbéliard, France, 2013, pp. 510–513. Available: http://ipin2013.sciencesconf.org/conference/ipin2013/eda_en.pdf,
[Nov. 19, 2013].
[4] Sparkfun. [Online]. Available: https://www.sparkfun.com [July 17, 2013].
[5] Invensense. [Online]. Available: http://www.invensense.com, [July 17,
2014].
2014 International Conference on Indoor Positioning and Indoor Navigation, 27th -30th October 2014
[6] STMicroelectronics.
“LSM9DS0
iNEMO
inertial
module:3D
accelerometer, 3D gyroscope, 3D magnetometer” [Online]. Available:
http://www.st.com/web/catalog/sense_power/FM89/SC1448/PF258556,
[July 17, 2014].
[7] Bosch Sensortec. “BMX055” [Online]. Available: http://www.boschsensortec.com/en/homepage/products_3/9_axis_sensors_5/ecompass_2/
bmx055_1/bmx055, [July 17, 2014].
[8] ThanksBuyer. “BMX055 IMU 9DoF Sensor Board Integrated
9-Axis Attitude Sensor Module SPI/I2C” [Online]. Available:
http://www.thanksbuyer.com/bmx055-imu-9dof-sensor-board-integrated9-axis-attitude-sensor-module-spi-i2c-28310, [July 17, 2014].
[9] Ebay. [Online]. Available: http://www.ebay.com, [July 17, 2014].
[10] Arduino. [Online]. Available: http://arduino.cc/
[11] Wikipedia. (2014, Jul. 23). “Secure Digital” [Online]. Available:
http://en.wikipedia.org/wiki/Secure_Digital [Jul. 25, 2014].
[12] Exp Tech. “HC Serial Bluetooth Products User Instructional Manual”
[Online]. Available: http://www.exp-tech.de/service/datasheet/HC-SerialBluetooth-Products.pdf [Jul. 25, 2014].
[13] Nordic
Semiconductor.
“nRF24L01
Single
Chip
2.4GHz
Transceiver
Product
Specification”
[Online].
Available:
http://www.nordicsemi.com/eng/nordic/download_resource/8041/1/
62622944 [Jul. 25, 2014].
[14] K. Abhayasinghe and I. Murray. (2012, Nov.). “A novel approach for
indoor localization using human gait analysis with gyroscopic data,”
in Third International Conference on Indoor Positioning and Indoor
Navigation (IPIN2012) [Online], Sydney, Australia, 2012. Available:
http://www.surveying.unsw.edu.au/ipin2012/proceedings/submissions/
22_Paper.pdf [Mar. 5, 2013].
[15] S. Jayalath, N. Abhayasinghe and I. Murray. (2013, Oct.). “A
Gyroscope Based Accurate Pedometer Algorithm,” in Forth
International Conference on Indoor Positioning and Indoor Navigation
(IPIN2013)[Online], Montbéliard, France, 2013, pp. 510–513. Available:
http://ipin2013.sciencesconf.org/conference/ipin2013/eda_en.pdf [Nov.
19, 2013].
[16] N. Abhayasinghe and I. Murray, “Human Gait Phase Recognition Based
on Thigh Movement Computed using IMUs”, in 2014 IEEE Ninth
International Conference on Intelligent Sensors, Sensor Networks and
Information Processing (ISSNIP)[Online], Singapore, April, 2014. Available: IEEE Xplore [Jun. 19, 2014]
[17] R. Dumont, “IMU based method to study the synchronization between the
white cane usage and the human gait cycle by vision impaired people,” M.
S. Thesis, Faculty of Engineering, University of Mons, Mons, Belgium,
2014.

Download Report