Master Thesis
HALMSTAD
UNIVERSITY
Master's Programme In Electronics Design, 60 credits
Designing a Wireless Step Counter
Master Thesis, 15 credits
Halmstad 2018-10-09
Bavithra Vijayaganapathi
1
2
i
Preface
I have taken efforts in this project. However, it would not have been possible without the support and help of many individuals. I would like to extend my sincere thanks to all of them.
I am highly indebted to Pererik Andreasson, for his dedication and keen interest above all his overwhelming attitude to help his students had been solely and mainly responsible for completing my work, his timely advice, and scientific approach have helped me to a very great extent to accomplish this task.
I would like to express my special thanks of gratitude to Hans-Erik Eldemark, who gave me the golden opportunity to do this wonderful project, which also helped me in doing a lot of research and I came to know about so many new things.
I am extremely thankful to Per-olof Karlsson, for providing me necessary technical suggestions during my research pursuit.
I thank profusely, Christoffer Lindhe and Michael Svedberg for giving me a such great attention and time. Their kind help, co-operation and providing necessary information throughout my research has helped me in successful completion of this project.
My thanks and appreciation also go to all the professors at Halmstad University (Håkan Pettersson, Struan Gray and Emil Nilsson) who taught us the valuable information and the individuals, who have helped me out with their abilities.
I hope you will enjoy reading this thesis and wish you to benefit from it.
Bavithra Vijayaganapathi October 2018
3
4
5
Abstract
This work focuses on designing a wireless step counter to count the total number of steps taken by the Xtend prosthetic foot until it needs to be replaced. This foot is especially designed for amputees to get rid of wheel chairs.
An algorithm is designed which translates the acceleration movement of the amputee’s foot during walking. The motion of the foot is detected by a 3- axis digital MEMS accelerometer (ADXL345), interfaced to a low-cost precision analog microcontroller ADuC7024. The devised algorithm then calculates the number of steps.
6
7
Table of Contents
Preface ... 2
Abstract ... 5
1 Introduction ... 10
2 Objectives ... 12
3 Pedometer Principles ... 13
3.1 Introduction ...13
3.2 Mechanical Pedometers ...13
3.3 Electronic Pedometers ...13
4 Accelerometer ... 14
4.1 ADXL345 ...15
5 Microcontroller ADuC7024 ... 16
6 Development Board ... 18
7 Principle behind the analysis... 19
7.1 Related works...20
7.2 System Architecture ...21
7.2.1 Data collection phase ... 22
7.2.2 Filtering Phase ... 22
7.2.3 Dynamic threshold and precision phase ... 23
7.2.4 Peak Detection phase ... 24
7.2.5 Time window and count regulation phase ... 25
8 Acceleration data ... 25
9 Programming setup ... 27
9.1 Flowchart ...28
10 Experimental results and analysis ... 30
11 Conclusions ... 32
11.1 Summary ...32
11.2 Consequence and reliability ...33
11.3 Future works ...33
12 References ... 34
13 Appendix ... 36
Hardware ... 36
8
9
ACRONYMS
ADC Analog-to-Digital Converter ARM Advanced RISC Machine CW Continuous Wave
DAC Digital-to-Analog Converter DB Data Base
DMA Direct Memory Access DSP Digital Signal Processor FIFO First-In-First-Out
I²C Inter Integrated Circuit MCU Microcontroller unit PCB Printed Circuit Board SoC System on Chip
SPI Serial Peripheral Interface
10
1 Introduction
Figure 1 shows Christoffer Lindhe, Founder and owner of the Lindhe Xtend company. He had a railway accident when he was 17 years old and lost both of his legs and one arm in the accident. As he was a swimmer, he competed in the Paralympics in Peking 2008 and in London 2012.
Christoffer tried many prosthetic foot designs in partnership with the orthopedic technicians and the universities but was never satisfied. The prosthetic foot on the market were too unstable so that also relatively small obstacles, e.g., pebbles, would too often lead to a fall or other accident.
Christoffer started to develop his own prosthetic foot which ended up with the Xtend foot, which nearly works like a normal human foot. The designed foot is also a water-proof and his discovery is the world’s first and foremost innovation to the amputees to get rid of the wheel chair.
Figure 1 : Christoffer Lindhe and Xtend foot
11
Figure 1 shows the Xtend foot [1], a foot prosthesis which functions to all intents and purposes like a human foot. It is intended for both above and below knee leg amputees, who have a moderate to high-level of activity, particularly outdoors and on uneven terrain. The prosthetic foot accommodates both double and single amputees up to 125 kg in weight.
The average moderately active person takes around 7500 steps per day.
The average life time of the Xtend foot is approximately one to two years based on the usage of the Amputees.
12
2 Objectives
Pedometers have in recent years become very popular as an everyday exercise progress monitor and to encourage individuals to lead a more active life and become healthier. The main objective of this work is to accommodate a warranty on the Xtend foot on how many steps the foot will last. Most specifically, I have addressed the following main questions in this thesis:
• How does the pedometer work?
• How does the pedometer get power?
• How does the pedometer integrate with the Xtend foot?
• Is the pedometer reliable to the users?
The following section provides the answers to them.
13
3 Pedometer Principles
3.1 Introduction
A pedometer, or step counter, is a device that counts each step a person takes by detecting the motion of hip, arms and foot.
The pedometer can be generally divided into two main categories:
mechanical pedometers and the electronic Pedometers. The principle of a pedometer is to utilize upward and downward oscillations while the user is walking.
3.2 Mechanical Pedometers
The mechanical pedometer utilizes a spring coil to suspend a lever arm and to provide stress for its upward and downward movements. The lever arm will move upward and downward for every walking steps, but the disadvantage is that the detection can only be measured in a single direction and this direction must be vertical ground surface. Moreover, when these devices are shaken one can hear a metal ball sliding back and forth or a pendulum striking stops as it swings.
3.3 Electronic Pedometers
This Pedometer type utilizes an accelerometer to detect the changes in acceleration in different directions. These pedometers rely on Micro Electro Mechanical System (MEMS) inertial sensors and software which are developed to a high degree of complexity to detect true steps with high probability. MEMS inertial sensors permit more accurate detection of steps and fewer false positives compared to mechanical pedometers.
14
4 Accelerometer
An accelerometer is an electromechanical device that measures acceleration forces. The basic principle of operation behind the MEMS accelerometer is the displacement of a small proof mass etched into the silicon surface of the integrated circuit and suspended by small beams.
Consistent with the Newton’s second law of motion (F=ma), a force applied to a mass results in an acceleration in the direction of the force which displaces the mass. The support beam act as a spring, and the fluid trapped inside the Integrated circuit act as a damper. This is the source of the limited operational bandwidth and non-uniform frequency response of accelerometers.
There are many different ways to make an accelerometer. Some uses the piezoelectric effect in which the crystal structures that get distorted by the accelerative forces generate a voltage. Another way to make an accelerometer is by sensing changes in capacitance [2]. This thesis is based on the MEMS accelerometer which uses the capacitive sensing.
The capacitive sensing is independent of the base material and relies on the variation of capacitance when the geometry of a capacitor is changing. The parallel-plate capacitance is given by [3]:
C = 𝜖𝐴
𝑑
where, ϵA = ϵoϵA and A is the area of the electrodes.
d is the distance between them.
ϵ is the permittivity of the material separating the plate electrodes.
A change in any of these parameters is measured as the change of capacitance.
15
Typical MEMS accelerometers are composed of a movable proof mass with plates that is attached through a mechanical suspension system to a refence frame. The capacitors are represented by the movable plates and the fixed outer plates. The deflection of proof mass is measured using the capacitance difference [3].
4.1 ADXL345
The ADXL345 is a small, thin low power capacitive 3-axis sensing accelerometer with high resolution acceleration measurement at up to
±16g. Digital output data is formatted as 16-bit two’s complement and is accessible through either SPI or I²C digital interface.
The ADXL345 is well suited for mobile device applications. It measures the static acceleration of gravity in tilt-sensing applications, as well as dynamic acceleration resulting from motion or shock. Its high resolution enables measurement of inclination changes less than 1.0 degree.
Several special sensing functions are provided. Activity and inactivity sensing detect the presence or lack of motion and if the acceleration on any axis exceeds a user-set level. Tap sensing detects single and double taps. Free-fall sensing detects if the device is falling. These functions can be mapped to one of two interrupt output pins.
16 3-axis
sensor
Sense electronics
Analog digital convert -er
Digital filter
Control and interrupt
logic
32 level
FIFO Serial I/O
Power management
INT1 1 INT2
Vs VDD
SDA/SDI /SDIO SDO /ALT ADR SCL/SCLK ESS
GND CS
Figure 2 : Internal description of ADXL345
5 Microcontroller ADuC7024
An ADuC7024 is a fully integrated microcontrol unit featuring a 1 megasamples per second (MSPS)/12-bit data acquisition system, a high performance multichannel analog-to-digital converter (ADC), a 16-bit/32- bit micro-controller unit (MCUs), and Flash memory on a single chip.
17
The ADC consists of up to 12 single-ended inputs. An additional four inputs are available but are multiplexed with the four DAC output pins.
The ADC can operate in single-ended or differential input mode. The ADC input voltage is 0 V to reference voltage. A low drift band gap reference which produce constant voltage regardless of power supply variations, temperature sensor, and the voltage comparator complete the ADC peripheral set.
The devices operate from an on-chip oscillator and a Phase-locked loop (PLL) generating an internal high frequency clock of 41.78 MHz. This clock is routed through a programmable clock divider from which the MCU clock operating frequency is generated. The microcontroller is based on ARM7 processor, which offers up to 41 million instruction per second (MIPS) peak performance. 8 kilobytes of Static Random-access memory (SRAM) and 62 Kilobytes of non-volatile/EE memory provided on-chip.
The parts operate from 2.7 V to 3.6 V and are specified over an industrial temperature range of -40ºC to +125ºC. When operating at 41.78 MHz, the power dissipation is typically 120×10ˆ3 µA.
Typical industrial applications include:
• Industrial control and automation
• Smart sensors, precision instrumentation
• Base station systems, optical networking
18
6 Development Board
The ADXL345 Development board (Figure 3) is an easy-to-use tool designed to shorten application development time by providing a ready-to- use platform for data collection and firmware development.
Figure 3 : ADXL345 Development board
The board is pre-configured as a datalogger that can be used to gather data for refining algorithms, tuning thresholds, and generally familiarizing oneself with accelerometer data. Powered by two AAA batteries, the board is completely untethered and integrates seamlessly into portable applications. Logged data is stored on a microSD card, providing essentially unlimited memory capacity and operating system versatility.
Data is stored in a text file, no software installation is required to operate the board or read data. A 2 GB microSD card and a USB microSD card reader are provided with the board. A schematic with the internal connections of development board is given in Figure 4.
19
ADuC7024 ADXL345
IRQ0
IRQ1
SCL
SDA
GND
INT1
INT2
SCL
SDA
SDO
PC 3V batteries
Figure 4 : simplified schematic of hardware system
Additionally, the board is fully programmable providing a hardware platform on which firmware can be tested in parallel with development of application hardware. Communications and processing are performed by an ARM-7 based ADuC7024 microcontroller, and the firmware is written in C.
7 Principle behind the analysis
To analyse the step detection, we choose acceleration as the relevant parameter. The three components of motion for an individual are forward,
20
vertical and side components. The accelerometer sensor ADXL345 senses the acceleration along its three axes x, y, and z
The accelerometer sensor can be in any orientation and therefore the measurement accuracy should not depend on relationship between the motion axes and the accelerometer measurement axes.
Normally, a single step results in changes in the vertical and forward directions. In the typical pattern of x, y, z measurements corresponding to vertical, forward and side acceleration of a walking person, at least one axes will have relatively large periodic acceleration changes.
In this paper, peak detection algorithm and dynamic threshold algorithm are used for detecting the steps with high accuracy when compared to the paper [20] based on a pedometer algorithm with fixed threshold in indoor positioning system. The dynamic threshold algorithm uses the new technique to count the peaks in which miscounting peaks are reduced by implementing the time window.
7.1 Related works
Various methodologies and techniques have been proposed and evaluated for step counting. Researchers have presented different approaches based on fixed step length.
On an accelerometer-based joint step detection and adaptive step length estimation using hand-held devices paper, Liu et al presented an algorithm for adaptive step length estimation based on an empirical formula and back-propagation neural network for a hand-held device. [5]
In paper [6], the author developed a waist-mounted based pedestrian dead reckoning (PDR) system to estimate the horizontal walking distance using Pythagoras theorem for the height changes of the waist. They used Pythagoras theorem to estimate each step length and vertical acceleration to detect the step.
21
In paper [7] explains the work on acceleration and gravity sensors to count the steps, where two different thresholds are used for detection of peaks and valleys. If there are two adjacent points one of them corresponds to a peak and the other to a valley, then there is a possibility to record a new step. But, if both the points are either peaks or valleys then the point will be deleted. If the time period between two points is very short, then the point will also be deleted.
In paper [8] another mobile phone-based technique, Finite state machine (FSM) approach is used with the threshold values to estimate user steps.
They also relied on support vector machine classifier for estimating the varying stride length of the user.
Other related works uses the accelerometer sensor MMA9553XL for detecting the step and used as a pedometer. But, the main problem for all those devices is that the sensor is not in production.
A keen observation from the literature review related to step counting and travelled distance estimation shows that some of the studies assume a fixed step length while the others are based on adopting a dynamic step length.
This work is based on the following algorithms:
• Dynamic threshold precision algorithm
• Peak detection algorithm.
The usage of Dynamic threshold algorithm mitigates the error in the distance estimation using fixed step length-based approaches.
7.2 System Architecture
Our Step counting approach includes the following sections:
• Data collection phase
22
• Filtering phase
• Dynamic threshold and precision phase.
• Peak detection phase
• Time window and count regulation phase
7.2.1 Data collection phase
In this phase, the raw data from the 3-axes accelerometer sensor is derived while the person is walking. The derived data includes the x, y, z and t values.
7.2.2 Filtering Phase
The sensors suffer from various problems as it is too sensitive to the acceleration variation. Hence, the raw acceleration values (x, y, and z) from the accelerometer sensor must be digitally filtered to avoid noise and outlier values such as acceleration changes due to false steps or miscounting.
The digital filter used here is the moving average filter. This filter is optimal for reducing the random noise. It operates by averaging number of points from the input signal to produce each point in the output signal. The equation form for the moving average filter is given as
𝑦[𝑖] = 1
𝑀 ∑ 𝑥[𝑖 + 𝑗]
𝑀−1
𝑗=0
Where x [i+j] is the input signal, y [i] is the output signal,
and M is the number of points in the average.
23
By passing the data points through this digital filter, the filtered data of the most active axis of the pedometer worn by a walking person can be extracted. In this approach four registers and a summing unit can be used to reduce noise.
The figure 5 shows the four points moving average filter. The data 0,1,2 and 3 are the acceleration data from the most active axes of the accelerometer sensor. The amount of noise reduction is equal to the square root of the number of points in the average. The 4-point moving average filter thus reduces the noise by a factor of 2. We can use more register but the problem of adding the register will decrease the response time.
Figure 5 : Moving average filter.
Normally, the operation of this digital filter [12] is determined by a program stored in the processor’s memory. The smoothening of the data by this digital filter helps in improving the accuracy of the system because the accelerometer sensor is very sensitive.
7.2.3 Dynamic threshold and precision phase
For every 50 samples, the maximum and minimum values of three axes accelerations are continuously updated by the system. The dynamic threshold value is defined as the average value which is given by
(max + min)/2. This value decides the steps for the following 50 samples.
24
For further filtering, the dynamic precision is also used to increase the accuracy of the system. In addition to the dynamic threshold, Linear shift registers are used to detect the steps.
The Linear shift register contains two registers namely sample new register and the sample old register. The data in these registers are called sample new and sample old respectively. When a new data sample comes, sample new is shifted to the sample old register unconditionally. But, the sample result will be shifted into the sample new register based on the condition:
If the changes in acceleration are greater than a predefined precision which is the value we set while programming the microcontroller, the sample result is shifted to the sample new register, otherwise the sample new register will remain unchanged.
A detection of step will happen if there is a negative slope of the acceleration plot (sample new value < sample old value).
7.2.4 Peak Detection phase
Changes in acceleration takes place upon taking a step. For instance, a vertical movement, such as raising a leg, leads to large acceleration changes along the Z-direction.
Some papers [6 and 10] uses the fixed step length algorithm to detect the peaks. This algorithm utilizes the linear shift register to detect the real peak values as well as the time when they are taken place. Since the shift registers contains the value that fall above the step threshold, then the maximum value should represent the real peak.
25
7.2.5 Time window and count regulation phase
Sometimes, the step counter is too sensitive in registering a step because when the pedometer vary very rapidly or very slowly from a cause other than walking or running, a step is registered. To avoid this miscounting and to find the true real steps, the time window and count regulations are used.
Time window
Normally the person can run fast as five steps per second or walk as slowly as one step every two second, thus the interval between two valid steps is define within the time window [0.2s – 2.0s], the steps which have the time interval within this time interval is considered as a step otherwise the step count will be discarded.
The user-selectable output data rate of ADXL345 is used to implement the time window.
Count regulation
The count regulation determines whether the steps are in a rhythmic pattern. The step counter has two working states: searching regulation and found out regulation. When the pedometer starts working, it works in searching regulation mode. If the regulation exists after four continuous valid steps, then the result is refreshed and displayed, and the pedometer will work in found out regulation mode. On this mode, the step count would be refreshed after every valid step. But if even one step is invalid, the step counter will return to searching regulation mode and search for four continuous valid steps.
8 Acceleration data
The Figure 6 shows the 3-axes acceleration data from the development board. The z-axis acceleration value gets affected by the user. The number of steps was calculated according to the vertical vibration of the body during each step. The acceleration along the y-axis suddenly drops to
26
negative values as the user stops walking. This is because the acceleration is defined as the rate of change of velocity with the time. It is given by the equation
𝑎 = 𝑑𝑣𝑑𝑡 Where, a is the acceleration,
dv is the rate of change of velocity, and dt is the rate of change of time.
So, the acceleration value becomes negative as the user halts.
figure 6 : x, y, and z acceleration values. (The left-hand scale is for x and y axis while the right- hand scale is for z axis).
8.8 9 9.2 9.4 9.6 9.8 10 10.2
-0.5 0 0.5 1 1.5 2 2.5 3
0 10 20 30 40 50 60 70 80 90 100
Acceleration in m/s²
Acceleration in m/s²
Number of steps
X-axis Y-axis Z-axis
27
9 Programming setup
The following algorithm is implemented on the development board to count the steps.
The ADXL345 accelerometer sensor detects the acceleration value of the three axes and the Inter integrated circuits transmits the acceleration information to the microcontroller ADuC7024. The digital filter which is programmed in the microcontroller is used to filter the acceleration values.
The program implemented in the microcontroller initiates the maximum and minimum values for comparing. “Sam” in the flowchart refers to the sampling rate which is fixed to 50 Hz. The values are checked for the above condition. If the result is yes, then the system resets the sampling rate to 0 and again initiates the maximum and minimum values for comparing.
The dynamic threshold level, which is the average value of maximum and minimum. This threshold value decides whether a step has been taken or not. The values in the sample new register and sample old register are shifted continuously only when the changes in the acceleration are greater than the predefined decision. If the old value is greater than the new fixed value, then the program implements the time window which reduces the miscounting peaks. If the acceleration curve crosses the dynamic threshold the step is happened and saved as a result.
28 9.1 Flowchart
Start
Get the values from the accelerometer sensor
Filter the acceleration values
Initiate the values for maximum and minimum
A
The connector A in the flowchart represents the continuation on next page.
29
The continuation from the previous page is indicated by the connector A.
N N
Y Y
Y
N Is sam
=50?
Reset sampling to 0
Initiate max and min values
If fixed>pr
ecision
Result= New fixed
> old fixed
Find the largest acceleration change
New value=new fixed>old fixed
Set the dynamic
precision value If old
val>ne wfixed
Implement the time window
Step is saved as a result A
stop
30
10 Experimental results and analysis
The number of steps was derived from the number of detected events where the foot, or prosthetic device, was lifted off the ground and put back down again in the process of ambulation.
Three experiments were conducted to assess the accuracy of the pedometer. The pedometer was worn on the foot of the male user for all three experiments. The user walked and stopped walking at will for 100 continual steps to validate the proposed pedometer.
Experiment 1: The user worn the pedometer on the foot and walked on a horizontal flat surface for 100 steps.
Experiment 2: The user had the pedometer in same place but, climbed upstairs for 100 steps.
Experiment 3: The user climbed downstairs with the pedometer on the foot and 100 steps was taken in to the account.
The walking accuracy rate was calculated based on these three experiments. The formula for calculating accuracy rate is given as
𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 𝑟𝑎𝑡𝑒 = 100% −|𝑑𝑒𝑡𝑒𝑐𝑡𝑒𝑑 𝑠𝑡𝑒𝑝𝑠 − 𝑎𝑐𝑡𝑢𝑎𝑙 𝑠𝑡𝑒𝑝𝑠|
𝑎𝑐𝑡𝑢𝑎𝑙 𝑠𝑡𝑒𝑝𝑠 ∗ 100%
Where, detected steps are the number of steps detected by the pedometer and Actual steps are the number of real steps taken by the user.
The results of the experiments are shown in Table 1.
31
Table 1: Result from the development board
ACTIVITY
ACTUAL NUMBER OF STEPS
ALGORITHM STEP COUNT
ACCURACY
WALKING IN FLAT
SURFACE
100 96 96%
WALKING UP-
STAIRS 100 89 89%
WALKING
DOWN-STAIRS 100 87 87%
It can be observed that the accuracies span a range between 87% to 96%
and walking along a flat surface often results in good performance compared to climbing upstairs and downstairs.
The accuracies can be further improved by optimizing the circuit design for this application specifically, instead of using the development board which is a universal ready-to-use development platform.
Jerome and Albright [19] have compared the performance of five commercially available pedometers with the experiments conducted on 13 vision impaired adults and 10 senior adults and observed the step detection accuracy rate. The accuracy rate was poor for all of them. While walking on the flat surface, an accuracy of (41%-67%) was derived which is significantly lower than 96%. Moreover, the accuracy reported for walking upstairs was as low as (9%-28%) which is much worse than our accuracy of 89%. Walking down stairs, the reported accuracy was (11%-41%) which is much worse than the 87%\ deduced using our proposed method.
Waqar et al [20] have used an accelerometer-based pedometer algorithm with fixed threshold in indoor positioning system. The author has reported a mean accuracy rate of 86.67% in their 6 trails of 40 steps each, with the
32
minimum accuracy of 82.5% and maximum of 95%. Our approach results in slightly improved performance of 87% and 96% respectively.
11 Conclusions
11.1 Summary
Three main topics have been discussed in the thesis: Accelerometer sensor, microcontroller and algorithm. A pedometer/step counter design has been done in this thesis according to several requirements by the company. A dynamic threshold algorithm has been chosen in order to have a better performance than the fixed step length-based approaches. This leads to a more accurate step counting rather than the old design. The proposed approach outperforms the problem of fake peaks and the difference on the length of each step. The accuracy of the proposed approach is better than the one with conventional method.
Our approach has been proposed in order to detect accurate and precise step. The proposed approach enhanced the step estimation by taking into account the step length and the error in counting the peaks, which are not solved perfectly in most of the existing approaches.
This paper implemented a practical pedometer utilizing a three-axis acceleration sensor. A threshold algorithm is used to filter the number of steps taken during walking. The presented electronic pedometer can be placed with any orientation to remove the disadvantage of general mechanical pedometers, which can only be oriented in single direction.
Moreover, the proposed pedometer can get rid of the need of using a continuous movement mechanism as error estimation criteria in general pedometers. Therefore, this system is good at detection accuracy and achieves a compact pedometer for usage in Xtend foot.
33 11.2 Consequence and reliability
Even though the batteries in the circuits have long life, leakage is still possible. When the prosthetic foot is not used for too long the battery can leak leading to corrosion in the circuits and prosthetic foot.
The reliability of the pedometer was tested by conducting the same experiments three times. Each time the pedometer gives the same accuracy rates. The consistency in the accuracy of the proposed pedometer makes it suitable for a prosthetic foot.
11.3 Future works
Enhancements on the development board has to be made because the size of the PCB is of critical importance. In our approach, the problem is mitigated but not perfectly solved.
In future work, we will also consider the shaking problem as the pedometer vibrates very rapidly, or very slowly from a cause other than walking or running because this might lead to false counting steps. The other important work is to find an alternative powering of the board to avoid the usage of the battery. One possible energy source stem from harvesting electrical energy from the mechanical work done during the compression of the prosthetic foot.
34
12 References
[1] Xtend foot. “The foot prosthesis Xtend foot” [ONLINE] Available at:
https://www.lindhextend.com/en/products/prosthetic-foot-amputee
[2] S. Beeby, G. Ensell, M. Kraft, N.White, MEMS mechanical sensors (Artech house inc., USA, 2004)
[3] S. E. Lyshevski, Mems and Nems: systems, devices and structures (CRC Press LLC, USA, 2002)
[4] the structure of accelerometer is available online at http://mafija.fmf.uni- lj.si/seminar/files/2007_2008/MEMS_accelerometers-koncna.pdf
[5] Y Liu et al., "Accelerometer Based Joint Step Detection and Adaptive Step Length Estimation Algorithm Using Handheld Devices," Journal of Communications, vol. 10, no. 7, 2015.
[6] W Y Shih, L Y Chen, and K C Lan, "Estimating walking distance with a smart phone," in 2012 Fifth International Symposium on Parallel Architectures, Algorithms and Programming, 2012, pp. 166—171.
[7] Q Zeng, B Zhou, C Jing, N Kim, and Y Kim, "A Novel Step Counting Algorithm Based on Acceleration and Gravity Sensors of a Smart-Phone," parameters, vol. 9, no.
4, 2015.
[8] M Alzantot and M Youssef, "UPTIME: Ubiquitous Pedestrian Tracking using Mobile Phones," in 2012 IEEE Wireless Communications and Networking
Conference (WCNC, 2012, pp. 3204-3209.
[9] C. C. Shih, The Implementation of A G-Sensor-Based Pedometer Department of Computer Science and information Engineering, National Central University, Master Thesis, 2010.
[10] M. J. Mathie, N. H. Lovell, A. C. F. Coster, and B. G. Celle “Determining activity using a triaxial accelerometer,” Annual Conference and the Annual Fall Meeting of the Biomedical Engineering Society EMBS/BMES Conference, vol. 3, pp. 2481–2482, 2002.
35
[11] X. H. Wu, The Study of G-Sensor-Based Systems and Their Applications, Department of Computer Science and information Engineering, National Central University, Master Thesis, 2011.
[12] Z. Shiqi, Y. Chun and Z. Yan, “Handwritten character recognition using orientation quantization based on 3D accelerometer,” Annual International Conference on Mobile and Ubiquitous Systems, 2008.
[13] Analog devices “Microcontroller ADuC7024 datasheet”
[ONLINE] Available at: http://www.analog.com/media/en/technical- documentation/data-sheets/aduc7019_20_21_22_24_25_26_27_28_29.pdf
[14] Analog Devices. “ADXL345 data sheet” [ONLINE] Available at:
http://www.analog.com/media/en/technical-documentation/data- sheets/ADXL345.pdf.
[15] Analog Devices. “Development board user guide” [ONLINE] Available at:
http://www.analog.com/media/en/technical-documentation/user-guides/UG- 065.pdf
[16] Analog Devices. “Application note AN-900” [ONLINE] Available at:
http://www.analog.com/media/en/technical-documentation/application- notes/47076299220991AN_900.pdf
[17] Analog Devices. “Application note AN-602” [ONLINE] Available at:
http://www.analog.com/media/en/technical-documentation/application- notes/513772624AN602.pdf
[18] Xtend foot. “The foot prosthesis Xtend foot” [ONLINE] Available at:
https://www.lindhextend.com/en/products/prosthetic-foot-amputee.
[19] G. J. Jerome and C. Albright. (2011, June). “Accuracy of five talking pedometers under controlled conditions,” The Journal of Blindness Innovation and Research [On-line] vol.1(2),
Available: www.nfbjbir.org/index.php/JBIR/article/view/17/38 [Oct. 27, 2011].
[20] W.Waqar, A.Vardy and Y.Chen. “Motion modelling using smartphones for indoor mobile phone positioning,” in 20th Newfoundland Electrical and Computer Engineering Conference [Online], Newfoundland, Canada, 2011, Available:
http://necec.engr.mun.ca/ocs2011/viewpaper.php?id=55&print=1.
36
13 Appendix
Hardware
Figure 7 : ADXL345 pin description
37
Figure 8: ADuC7024 pin description
38
Figure 19 : schematic file Figure 9 : schematic file
39
Figure 10 : development board layout
Figure 11 : to program the development board
40
Table 2: Sample acceleration value from the development board
41
42
PO Box 823, SE-301 18 Halmstad Phone: +35 46 16 71 00
E-mail: registrator@hh.se www.hh.se
Bavithra Vijayaganapathi,
Masters In Electronic Design,Sweden.
Bachelor In Electronics And Communication Engineering, India
