Advances in Science, Technology and Engineering Systems Journal Vol. 2, No.

3, 891-899 (2017)
ASTESJ
www.astesj.com
ISSN: 2415-6698
Special Issue on Recent Advances in Engineering Systems

EEG Mind Controlled Smart Prosthetic Arm – A Comprehensive Study

Taha Beyrouthy*, Samer Al Kork, Joe Akl Korbane, Mohamed Abouelela

American University of the Middle East, Egaila, Kuwait

ARTICLE INFO ABSTRACT

Article history: Recently, the field of prosthetics has seen many accomplishments especially with the
Received: 05 April, 2017 integration of technological advancements. In this paper, different arm types (robotic,
Accepted: 11 June, 2017 surgical, bionic, prosthetic and static) are analyzed in terms of resistance, usage, flexibility,
Online: 26 June, 2017 cost and potential. Most of these techniques have some problems; they are extremely
expensive, hard to install and maintain and may require surgery. Therefore, our work
Keywords:
introduces the initial design of an EEG mind controlled smart prosthetic arm. The arm is
EEG
controlled by the brain commands, obtained from an electroencephalography (EEG)
Prosthetic arm
headset, and equipped with a network of smart sensors and actuators that give the patient
Sensors
intelligent feedback about the surrounding environment and the object in contact. This
Mind-controlled
network provides the arm with normal hand functionality, smart reflexes and smooth
Robotics
movements. Various types of sensors are used including temperature, pressure, ultrasonic
Emotiv
proximity sensors, accelerometers, potentiometers, strain gauges and gyroscopes. The arm
Signal processing
is completely 3D printed built from various lightweight and high strength materials that
can handle high impacts and fragile elements as well. Our project requires the use of nine
servomotors installed at different places in the arm. Therefore, the static and dynamic
modes of servomotors are analyzed. The total cost of the project is estimated to be relatively
cheap compared to other previously built arms. Many scenarios are analyzed
corresponding to the actions that the prosthetic arm can perform, and an algorithm is
created to match these scenarios. Experimental results show that the proposed EEG Mind-
controlled Arm is a promising alternative for current solutions that require invasive and
expensive surgical procedures.

1. Introduction amputees to control their prosthetic devices by merely thinking

about the action they want to perform as stated by the John
In recent studies, the World Health Organization (WHO)
Hopkins Applied Physics Laboratory.
reported that about 15% of the world's population suffers from a
form of disability, half of which cannot afford health care [1]. Due A team at the John Hopkins University developed a robotic arm
to various political, economic, scientific, and demographical in their physics lab that is controlled by brain signals [5]. The arm
reasons, the overall rates of amputees and limb dysfunction has 26 joints, and can lift up to 45 pounds. It is based on an idea,
patients are increasing [2]. There are over 10 million amputees which can be altered to fit any need, from someone missing just a
worldwide, out of which 30% are arm amputees [3]. Although hand to an entire limb. Engineers from Ossur, one of the biggest
prosthetic limbs exist since decades, they are not very natural in bionic arms markets in the world, designed an artificial bionic
terms of operation and interaction with the environment. They advanced leg and it was tested on a farmer who lost a leg [6]. A
require undergoing an invasive surgical procedure [4]. The main comparison was made between the bionic and the mechanical
goal of such complex procedures is to reassign nerves and allow artificial legs, in which the engineers noticed some flaws in the
mechanical leg, and thus causing the user to fall while walking if
*
Taha Beyrouthy, American University of the Middle East, +965 2225 1400 Ext. wrong amount of pressure was applied to the toe where the leg is
1134, Email: taha.beyrouthy@aum.edu.kw

www.astesj.com 891
https://dx.doi.org/10.25046/aj0203111
 T. Beyrouthy et al. / Advances in Science, Technology and Engineering Systems Journal Vol. 2, No. 3, 891-899 (2017)
connected. The bionic leg, on the other hand, was constructed out Similar to the Very high Overshadowed by the
surgical limbs, flexibility surgical and the

Prosthetic
of more than one piece. The knee, the foot and the leg were but less (custom made for robotic limbs
assembled to create a fully functional bionic limb [7]. This limb medical each person
attention using exact
has a brain of its own, and can sense what surrounds it by the measurements)
processors that analyze the inputs. According to the head of this No Not flexible No potential
group of engineers, the leg costs more than a few sedan cars.

Static
mechanical
use
Prosthetic limbs need to be measured and fitted to the patient
for his needs [8]. To apply prosthetics on a patient, intense medical
observation and a training course for the patient are needed so that
he can use the limb comfortably. There are several techniques used The control techniques used are summarized and compared with
as a means of controlling robotics arms, and the top three methods each other in Table 2 hereafter in terms of cost, installation,
are highlighted hereafter. The first method is to use an degree of control and accuracy.
electroencephalogram (EEG) device [9], which will record the
Table 2. Control techniques comparison
person’s brain waves when he is thinking of a certain action or
implementing a facial expression. These readings are then Type Approximate Cost Installation Degree of Accuracy
(U.S. Dollars) Control
converted to commands for the arm. The author in [10] states that
the mind regulates its activities by electric waves registered in the EEG 100 – 400 Detachable Complete Accurate
brain that emits electrochemical impulses having different control
frequencies, which can be registered by an electroencephalogram. Surgical 10,000 – 120,000 Permanent Complete Very
For instance, beta waves are emitted when a person feels nervous control accurate
or afraid with frequencies ranging from 13 to 60 Hertz. Alpha Sensors Below 100 Detachable Limited Accurate
waves are emitted when a person feels relaxed mentally and
physically with frequencies from 7 to 13 Hertz. On the other hand,
The EEG method is not only cost effective, but it is also
delta waves are emitted when a person is in a state of
unconsciousness. The advancement in technology made it possible accurate and gives the patient complete control of the arm. It also
to process these EEG frequencies and data directly in real time by gives the user the luxury of taking it off when feeling discomfort.
the use of a brain-computer interface which is a combination of EEG is a noninvasive method of monitoring brain activity.
hardware and software. Typically, it uses electrodes placed on the outside of the head, and
The second method is the surgical implantation. The arm is measures voltage oscillations in the neurons of the brain caused by
surgically connected to the person’s torso. Connections are also ionic current. It has been used in medical applications for a very
made to the nerves to allow the reading of electrical signals so that long time. The Emotiv EPOC is an example of an EEG headset
these signals can then be filtered and converted to commands. The with 14 sensors and having an internal sampling rate of 2048 Hz.
last control method consists in using sensors, which will be After filtering the signals, it sends the data to the computer at
connected to the robotic arm in order to take specific readings.
approximately 128 Hz. The signals are transferred from the
Some of the most common sensors used in this case are EMG,
gyroscope, and accelerometer sensors. This will allow the user to headset to the computer through wireless technology. This offers
be aware of the position his arm can be in as well as expand and much greater mobility, and instead of requiring a special gel, the
enclose it. All the arm types used are summarized and compared electrodes of the EPOC simply need to be dampened using a saline
with each other in Table 1 hereafter in terms of usage, flexibility, solution that in disinfectant and common.
cost potential.
The project presented in this paper aims to develop a low-cost
Table 1. Arm Types comparison
and versatile human-like prosthetic arm controllable via brain
Arm Usage Flexibility Potential activity using EEG neuro-feedback technology. The arm is
Type
equipped with a network of smart sensors and actuators that give
Using mounts Depends on the Big potential with the
motors, material rise of 3D printers the patient intelligent feedback about the surrounding
Robotic

and design environment and the object in contact. It also allows the arm to
react and execute pre-programmed series of actions in critical
Surgically Depends on the Very high especially
cases (extremely hot or fragile objects, etc.) A first prototype has
Implanted training and the for the war victims been developed to test the prosthetic arm with the embedded
Surgical

physiotherapy electro-mechanical system. This prototype is controlled using flex

sensors integrated within a wearable glove. A microcontroller is
Very effective Depends on how Good potential, added to the system, thus allowing to perform programmed
for everyday well can the especially with the
usage processor microprocessor actions and tasks. The arm is created using a 3-D printer in order
Bionic

analyze the revolution for it to be cost efficient. All parts are printed separately then
surroundings
assembled together. This prototype focuses on the arm-
environment interaction. A second prototype based on the EEG
www.astesj.com 892
 T. Beyrouthy et al. / Advances in Science, Technology and Engineering Systems Journal Vol. 2, No. 3, 891-899 (2017)
control has also been developed and still under test. Preliminary
experimental results show that the EEG technique is a promising
and good alternative to other existing techniques.

2. System Architecture

The proposed system is divided into 4 major units [11] as shown

in Figure 1:

1- The Input Unit – EEG sensors

2- The Processing Unit – Pattern recognition
3- The Electro-Mechanical Unit - The arm
4- The Interface Unit – Smart sensor network

Figure 2. EEG 14-sensor arrangement

2.2. Processing Unit

The EEG signals provided by the input unit were sampled and
processed on a lightweight wearable device – the Processing Unit.
The processing activity consists of two main parts: a pattern
recognition part that identifies different brain behavior captured by
the input unit, and a command part that generates a series of
commands to be sent to the mechatronics system of the arm.
This unit was programmed to distinguish between several
states of the mind representing different levels of “meditation” and
“focus”. Every mind state was captured and encoded to represent
a set of desired tasks to be performed by the arm. Due to the
diversity and the complexity of brain wave activities among
different humans, machine-learning techniques were required to
train patients to specific arm movements according to a set of mind
states.
2.3. Electro-mechanical Unit

This unit was designed and built from various lightweight high-
strength materials that can handle high impacts and fragile
elements as well. It integrated servos capable of handling 800 oz.-
in. of stall torque. These servos were strategically placed to
minimize hardware and facilitate complex moves. A
Figure 1. Mind-controlled smart prosthetic arm architecture microcontroller was also integrated to this setup to provide the
interface between the Mechanical Unit and the Processing Unit. It
2.1. Input Unit can also be programmed to perform a series of predefined
movements, allowing the arm to have a sophisticated and realistic
In this unit, brain signals were captured by an array of advanced
real hand behavior.
EEG sensors communicating with a Signal Processing Unit via
low-power and secure connectivity using Bluetooth technology. 2.4. Interface Unit – Smart Sensor Network
This device has an internal sampling rate of 2048Hz and 14 This unit is composed of a network of smart sensors, including
sensors arranged according to the international 10-20 System as temperature, skin pressure and ultrasonic proximity sensors,
shown in Figure 2 in order to cover the most relevant area over accelerometers, potentiometers, strain gauges and gyroscopes. The
main features of this unit allow the arm to interact with and adapt
the brain. This allows a maximum and efficient coverage of the to the surrounding environment. Moreover, a bi-directional
brain activity. EEG signals are acquired using the Emotiv EPOC communication was required to give commands to the arm and
wireless recording headset bearing 14 channels (AF3, F7, F3, FC5, provides feedback to the patient. This integrated network of
T7, P7, O1, O2, P8, T8, FC6, F4, F8), referenced to the common sensors and actuators required custom communication protocols
and networking techniques that allow seamless interaction and
mode sense [12]. control between the arm and patient. By default, controlling the
www.astesj.com 893
 T. Beyrouthy et al. / Advances in Science, Technology and Engineering Systems Journal Vol. 2, No. 3, 891-899 (2017)
arm was handled by the brain (patient); however, it can be component number 4 is the control unit, which is Arduino based;
transferred to the arm to proactively protect itself against damage. component number 5 is the XBee driver (wireless
Due to its unique features, the proposed Mind-controlled Smart communication with the processing unit).
Prosthetic Arm should be able to improve the quality of life for The Data Sampling Unit includes 14 EEG sensors
millions of patients and their families around the world. Its low implemented in the EEG-based headset. This unit converts EEG
cost design makes it accessible for a wide range of beneficiaries, signals to digital signals and send them to the processing unit
especially those with limited or no access to advanced health care. through Bluetooth communication channel.
3. Technical Specifications The Wireless Communication Unit includes low power
3.1. Hardware Overview connection Bluetooth module interfaced with the Raspberry III
microcomputer. This part implements a communication protocol
The Smart Prosthetic Arm architecture was based on an between the user (EEG sensors installed on the head) and the
advanced electro-mechanical system controlled by EEG neuro- embedded microcontroller in the prosthetic arm.
feedback technology. This architecture is divided into 4 main The Computing and Processing Unit: Sampled EEG signals
units in terms of hardware (see Figure 3 and Figure 4) [13]: are channeled through the wireless communication unit to reach
the Computing and Processing Unit, which is embedded in the
arm and includes a Raspberry Pi III microcomputer, interfaced
On board the arm
with an Arduino Mega microcontroller that handle the mechanical
servo units installed in the arm. The main function of the
processing unit is to treat the digitized EEG signals. It is
Control programmed to compare between the headset reading and a set of
Communication
Communication

Unit premeasured patterns related to different states of the mind.

Different mind states are captured and encoded to represent a set
Feedback About Surroundings

of desired tasks to be performed by the arm. This Unit is also

Xbee
WiFi

EEG Signal programmed with multiple hand reflexes, triggered by the smart
Processing
Headset Unit Mechatronic sensor network embedded in the arm. It gives the arm a human-
like behavior via smooth movements and smart reflexes.
System
Finally, the Electro-Mechanical Unit includes nine servo
(The arm) motors installed on the 3D designed model. The 3D hand model is
built from various lightweight high-strength materials that can
handle high impacts and fragile elements as well. This unit
integrates all servos, which are capable of handling 800 oz.-in. of
Sensor stall torque. It is embedded in the arm and links servos to joints to
Network
perform different motions. The servos are strategically placed to
minimize hardware and facilitate complex moves. A
microcontroller is also integrated to this setup to provide the
Figure 3. System overview – block diagram
interface between the Mechanical Unit and the Processing Unit.
One of the main features of this arm is its affordability, which
is accomplished mainly due to the fact that the arm is entirely 3D
printed, which is one of the most affordable, yet durable ways of
manufacturing such a product in this day.
For the process of building, different materials were used to
create a full arm. The usage of different materials provided a more
practical build.
The hand was created using a material called EcoPLA, which
is PLA mixed with certain chemicals to provide a sense for heat
recognition. This material can change color when exposed to heat.
The added sensors provide visual aid to the user of the arm so that
the user can estimate the heat of the held item. EcoPLA is
Figure 4. System overview – photo of the actual components considered to be a clean material, mainly extruded from starch,
corn, and sugar cane, which makes it more environmental friendly.
As shown in Figure 4, component number 1 is the processing unit
that receives the EEG waves from the headset and communicates For the mechanical moving parts that need better endurance and
with the control unit; component number 2 is the Emotiv EEG will be exposed to more pressure, PLA is also used as the material
headset with 14 sensors; component number 3 corresponds to the for the build because of its strength, flexibility, and heat resistance,
embedded sensors (temperature, proximity and touch); which makes it more reliable; therefore, it is the best material to
www.astesj.com 894
 T. Beyrouthy et al. / Advances in Science, Technology and Engineering Systems Journal Vol. 2, No. 3, 891-899 (2017)
execute the job with no flaws. It is also easily sanded and Table 3. Load calculation for a servo motor
machined.
Mass Force Wrist S. Hand S. Biceps S. Shoulder S.
(Kg) (N) Torque Torque Torque Torque
For the forearm, shoulder, and elbow, PLA is used to complete
1 Kg 9.8 1.96 3.43 6.615 7.889
the build. PLA is a printing material that is easily machined under 2 Kg 19.6 3.92 6.86 13.23 15.778
normal conditions, on the other hand ABS needs special heating 3 Kg 29.4 5.88 10.29 19.845 23.667
bed during the printing. PLA is also an environment friendly 5 Kg 49 9.8 17.15 33.075 39.445
plastic which can be created using a mixture of different plant 7 Kg 68.6 13.72 24.01 46.305 55.223
substances, such as potatoes and corn, yet, it does not bio-degrade 10 Kg 98 19.6 34.3 66.15 78.89
easily. PLA is also considered strong and rigid material, yet it has
a lower melting temperature which makes it unreliable under 3.2. Dynamic Modeling of Prosthetic Arm
pressure. In the time domain, the mechanical model of the Prosthetic
Arm can be represented as a generalized second order differential
The project required the use of nine servomotors in different
equation form for the translation motions as shown in (3):
places in the arm. To make sure the hand is able to carry a load
𝑑𝑑 2 𝑥𝑥 𝑑𝑑𝑑𝑑
varying from 3-5 kg, first we had to know how much torque each 𝑓𝑓(𝑡𝑡) = 𝑚𝑚 + 𝐶𝐶1 + 𝐾𝐾1 . 𝑥𝑥(𝑡𝑡). (3)
𝑑𝑑𝑡𝑡 2 𝑑𝑑𝑑𝑑
servo motor needed. To get that, the following formula shown in
where: f(t) represents the applied tension forces in the two sides of
(1) for the static mode is used:
the tendons to create displacement between fingers at different
+M = (L1 × cos(θ1 ) + L2 × cos(θ2 ))x9.8 (1) configurations, m is the mass of the movable part, x(t) is the
+M is the momentum; L1 is the length between the force and the displacement between fingers at different time, C1 is the friction
first angle θ1 ; L2 is the length between the force and the second damping coefficient for the translation motion , K1 is the spring
angle θ2 as shown in Figure 5. constant . the same model can be restated for the torque 𝑇𝑇(𝑡𝑡) at
the elbow and shoulder joints as shown in (4):
𝑑𝑑 2 𝜃𝜃 𝑑𝑑𝑑𝑑
𝑇𝑇(𝑡𝑡) = 𝐽𝐽 + 𝐶𝐶2 + 𝐾𝐾2 . 𝜃𝜃(𝑡𝑡). (4)
𝑑𝑑𝑡𝑡 2 𝑑𝑑𝑑𝑑

where: J is the mass polar moment of inertia (kg.m2) of the rotated

part, θ(t) is the angular displacement (rad.), C2 is the friction
damping coefficient for the rotational motion , K2 is the rotational
spring constant .
From the multibody dynamic model of the different parts of the
Prosthetic Arm, we can get the required toque for each of the 8
motors at all possible movement configurations, the selection
design value for the motors specifications was based on the
maximum calculated required torque & Power.
3.3. Power Consumption

The power study of the proposed Smart Prosthetic Arm is

Figure 5. Prosthetic Arm Technical Specifications divided into 2 major parts.
For the dynamic mode of the servo motors, friction of the arm The EEG headset uses a lithium (Li) battery that provides up
parts along with some other parameters were neglected. The to 12 hours of continuous use when fully charged. If fully
following formula shown in (2) is used: discharged (after 12 hours of usage), it can take up to 4 hours to
be recharged. However, the charging process takes between 30
𝑀𝑀 = 𝐹𝐹 × 𝑙𝑙 + 𝐽𝐽 × 𝛼𝛼 (2) minutes to 2 hours when not fully discharged. Therefore, it is
𝐹𝐹 is the force applied on the hand; 𝑙𝑙 is the length of the arm; 𝐽𝐽 is recommended to recharge it before 12 hours of continuous use.
For safety reasons, the headset does not work while charging it.
the polar movement of inertia; 𝛼𝛼 is the angular acceleration or
polar acceleration. The electro-mechanical system is composed of eight servo
motors of rated power 0.5 Watts and 3.5 Watts, which are placed
The equation for the dynamic mode shown in (2) changes from on different parts of the prosthetic arm (wrist, elbow, shoulder and
one part to another (shoulder, wrist, hand or elbow). fingers). Three servomotors with a rated power of 3.5 Watts are
placed on the wrist, elbow and shoulder area to provide three
The next calculations are obtained while the hand is in the static degrees of freedom. In addition, five servomotors with a rated
mode, not the dynamic mode. While the angles are equal to zero, power 0.5 Watts each are used to control the five fingers. Thus,
the servos will be outputting their highest torque in the static mode all the servomotors consume a total power of 13 Watts. A 5V
as shown in Table 3. battery with an output current of 2.6 Ampere is required to provide

www.astesj.com 895
 T. Beyrouthy et al. / Advances in Science, Technology and Engineering Systems Journal Vol. 2, No. 3, 891-899 (2017)
power to all servomotors embedded in the arm. The wireless human hand, etc…). This integrated network of sensors and
communication unit operates when a 50 mA is supplied at 3.3 V, actuators requires custom communication protocols and control
which requires a power of 0.165 Watts to operate. mechanisms techniques that allow seamless interaction and
In addition, a network of smart sensors is used, including control hand over between the arm and the user.
temperature, skin pressure and ultrasonic proximity sensors, By default, the brain signals control the arm movements semi-
accelerometers, potentiometers, strain gauges and gyroscopes. autonomously via a wireless connection. The EEG headset has a
These sensors consume a total current of 100 mA. The power proprietary wireless USB dongle that can be connected to the
consumption of the low-power single board computer processing unit via a Bluetooth module. It reads the neuro-
(Computing and Processing Unit) is 0.1 Watts corresponding to a electrical signals and interprets them as a set of predefined outputs
current of 30 mA. To power all the units listed above, the system that reflect facial expressions, mood and conscious intentions.
requires a power source with an output current not less than 2.8 A These predefined outputs are received by the processing unit,
and 5V as output voltage. compared with user-dependent library of pattern and then
Two 10,000 mAh lithium ion batteries are chosen with an converted into functions. These functions are then labeled using
output current of 2A each. They include a charging circuit (via variables and sent to the Arduino microcontroller through via
USB Cable), and a boost converter that provides 5V DC. These UART channel. Based on these variables, a certain movement of
batteries have an 80% efficiency loss on both ends, meaning that the arm occurs according to the mapping that is done between the
it is not recommended to operate the arm while the battery is being variables and the readings.
charged. With these two batteries, the arm will be operational for 4. Project Cost
7 continuous hours knowing that the average hand movements per
person during a day are equivalent to 1 to 3 hours of continuous The cost of the project was divided into components and
movements depending on the daily activity performed. In equipment cost (servo motors, sensors, Bluetooth, Raspberry PI II
conclusion, the whole system can be operational for 2 full days. and EEG-based headset, two USB battery packs), labor costs and
the 3D printing cost of the prosthetic arm. All the required
3.4. Smart Sensors components are relatively cheap off-shelf products that can be
The sensors included in the proposed solution can be classified purchased from multiple providers. The total cost of the system is
in two categories: user-end sensors and environment-end sensors. estimated to be around 1140 USD as shown in Table 4.
The first category consists of the 14 EEG sensors presented Table 4. Cost of project components
previously which are installed on the user headset.
Components needed Approximate Cost
(U.S. Dollars)
As for sensors of the second category, their main feature is to
allow the arm to interact and adapt to the surrounding EEG-based headset (including battery and 400
software)
environment, by providing intelligent feedback about critical
Servo motors 135
condition, such as high temperature or pressure, etc. When
interfaced with the embedded microprocessor installed on the arm, Bluetooth module 95
this network will give the prosthetic arm a human-like behavior Smart network sensors 150
with smart reflexes and smooth movements. Note that the Raspberry PI II & Arduino Mega 60
feedback coming from some of these sensors will not only be used
Batteries 100
to operate some servos of the arm, but will also be displayed on
3D model arm 200
small LCD screen mounted on the forearm.
Total cost 1140
3.5. Control

The proposed system was based on both fully autonomous and The total cost of the proposed solution is estimated to be
semi-autonomous control. relatively affordable and more economical when compared to the
existing solutions (surgical, bionic and static prosthetic arms)
A bi-directional communication channel was implemented
manufactured by different companies around the world.
between the smart sensor network and the embedded
microprocessor in such a way to autonomously control the 5. Testing and Results
electro-mechanical unit and provide feedback to the user by
displaying it on an LCD mounted on the arm. This setup offered During the training session performed, the signals
the arm the ability to have smart reflexes when it counters delicate, corresponding to valence, engagement, frustration, meditation,
dangerous and critical situations such as protecting the arm from short and long-term excitement detected by the Emotiv EPOC
very hot surface contact or over squeezing fragile objects (glass, headset for 30 and 300 seconds are shown in Figure 6.

www.astesj.com 896
 T. Beyrouthy et al. / Advances in Science, Technology and Engineering Systems Journal Vol. 2, No. 3, 891-899 (2017)
The scenarios simply define the smart reflexes of the arm which
can protect the user from the surrounding environment or the
people who are communicating with the user. Initially, the arm
automatically checks the three sensors (pressure, IR temperature,
and proximity). If all of the sensors are in a safe state, the
following scenarios can be initiated:

Figure 6. Different signals detected by the Emotiv headset

A 3D cube was displayed, and a training session was available

in order to guide the user to move the object either to the left or to
the right with his thoughts as shown in Figure 7. The process Figure 9. Prosthetic arm – opening hand
started by recording the user’s mind when he was not thinking of
Scenario 1: User will be able to initiate brain command and
anything. Then, the user could select an action he prefered to
control the arm if all the sensors are in a safe state.
move the object. His thoughts were recorded for the time interval
when he was thinking of the object. Scenario 2: If the user initiates “Action 1” command, this will
enable the handshake. The servomotors in the hand will increment
the angle and check if it has the optimum pressure for the
handshake. If it reaches it, it will stop automatically.

Scenario 3: If the user imitates “Action 2” command, this will

enable the arm to grab a soda can and pour it in a cup. The servo
motor in the bicep moves up and stops at a certain degree, then
the servo motor in the wrist is rotated so that the soda will be
poured in the cup.

On the other hand, if one of the three sensors is not in a safe state,
the following scenarios will be taken:
Figure 7. Training session with the 3D cube
Scenario 4: If the hand holds an object, the temperature sensor,
Thinking of moving the cube to the left stimulated the hand of
which is embedded within the hand, will measure the temperature.
the prosthetic arm to close (see Figure 8). However, thinking of
If the temperature is too hot for human skin and reaches the pain
moving the cube to the right stimulated the hand to open (see
threshold, “Action 3” will be automatically initiated. Once Action
Figure 9).
3 is initiated a red LED lights up.

Scenario 5: It is the same as the previous scenario, but in this case

if the temperature is too hot for the material of the hand itself,
“Action 4” is automatically initiated. The servomotors in the hand
open the fingers and release the object.

Scenario 6: Here if the proximity sensor which is placed in the

back side of the arm reads a distance which is not in the safe range
(too close to an object), the servo motors within the shoulder will
move the arm forward to keep the arm within a safe range. This is
defined as “Action 5”.
Figure 8. Prosthetic arm – closing hand
Finally the code on the Arduino was updated so that all the
Coming up for the scenarios and the algorithms of the arm are individual codes were combined together. The codes for each
one of the things that makes this arm considered as a smart arm. sensor, the codes for testing the servomotors and the codes for the
www.astesj.com 897
 T. Beyrouthy et al. / Advances in Science, Technology and Engineering Systems Journal Vol. 2, No. 3, 891-899 (2017)
different gestures based on the input were all put together in a way 6. Conclusion
for the arm to function effortlessly. Within the code, the first thing
The proposed arm hosts state-of-the art technological
that was taken into consideration and checked is the sensor
advancement, communication protocols, control systems, and
network. If the sensors are not in a safe state, they will follow the
human interfacing. This gives it great potential in many
scenarios as shown previously. Otherwise, the serial port will be applications related to the health care field as well as other fields.
checked if it is receiving any data. The serial port is checked As long as health care is considered, the idea could be expanded
because the Arduino communicates with the XBee through a to other body parts as well as to patients having other dysfunctions
serial connection. If data is being received, it will go through as nerve damage. On the other hand, many industrial and
different if statements, each one checking if the input is commercial applications can utilize many features of the proposed
representing a certain letter. Each letter corresponds either to a arm. Within the health care field, there exists a class of patients
raw action (move arm up) or a certain gesture. who need extra help with their daily lives. This includes elderly
people, people under rehabilitation, and people with limited
The Emotiv EPOC headset has 16 electrodes, which it uses to mobility, etc. The proposed arm may be interfaced to a robotic-
measure the electromagnetic pulses from the brain. These structure and function as a helper or caregiver to this group of
electromagnetic pulses are created by intentional thoughts, facial people. It can be programmed to do various functions according
expressions and even the mental state of the user. What we will to specific patient needs. This may vary from cooking to
be using for the inputs are the intentional thoughts and the facial assistance with bathing or dressing. Another example in the
expressions. medical field is remote high precision surgical procedures, where
surgeons can undergo operations remotely with the aid of the
Before wearing the headset, Saline solution must first be robotic arm. Many industries employ robots in the manufacturing
applied to the electrodes. Once the electrodes are damp, they can process, many of which can make use of a modified version of the
be connected to the headset. Saline solution is applied in order to proposed arm. Based on a specific application, this smart arm can
increase contact and the quality of the signal read; however, if too be programmed to execute a series of predefined actions, and
much is applied, it could lead to affecting the electrical equipment. customized with dedicated sensors, actuators and customized
Once the user slides the headset on to his head, he/she must insure algorithms (such as image and signal processing, gesture and
that the electrodes are placed in the correct position and are in voice recognition etc…). In addition, connecting the arm to the
contact with the scalp; this ensures a good signal (green). To get Internet, and making it part of an Internet of Things network (IOT)
the most accurate output, it is best that all electrodes are green. will increase the performance and productivity of many industry
applications. A first prototype is designed, built and was under
The flowchart of the algorithm is shown in Figure 10. test. The testing required long training sessions in order first to
build a user-dependent library of brain activity patterns, and
second to make the user more familiar and comfortable using this
hand.

Conflict of Interest

The authors declare no conflict of interest.

References
[1] S. W. Hawking, “World report on disability,” World Health Organization,
Geneva, Switzerland, 2011.
[2] NBC News. (2010, March 20). Limb loss a grim, growing global crisis
[Online].Available:
http://haitiamputees.nbcnews.com/_news/2010/03/19/4040341-limb-loss-a-
grim-growing-global-crisis
[3] M. LeBlanc. (2011, January 14). Give Hope – Give a Hand [Online].
Available: https://web.stanford.edu/class/engr110/Newsletter/lecture03a-
2011.html
[4] C. Moreton. (2012, August 4). London 2012 Olympics: Oscar Pistorius
finally runs in Games after five year battle [Online]. Available:
http://www.telegraph.co.uk/sport/olympics/athletics/9452280/London-
2012-Olympics-Oscar-Pistorius-finally-runs-in-Games-after-five-year-
battle.html
[5] Y. Jeong, D. Lee, K. Kim and J. Park, “A wearable robotic arm with high
force-reflection capability,” in 9th IEEE International Workshop on Robot
and Human Interactive Communication, Osaka, 2000, pp. 411-416.
[6] A. Bennett Wilson Jr., B. (n.d.). Retrieved Octobor 17, 2015, from
oandplibrary: http://www.oandplibrary.org/alp/chap01-01.asp
Figure 10. Flowchart of the algorithm

www.astesj.com 898
 T. Beyrouthy et al. / Advances in Science, Technology and Engineering Systems Journal Vol. 2, No. 3, 891-899 (2017)
[7] E. Sofge (2012, May 28). Smart Bionic Limbs are Reengineering the
Human. [Online]. Available:
http://www.popularmechanics.com/science/health/a7764/smart-bionic-
limbs-are-reengineering-the-human-9160299
[8] R. M. Coupland, War Wounds of Limbs: surgical management. Geneve,
Switzerland, ICRC, 1993.
[9] Jerkey. Brain-Controlled Wheelchair [Online]. Available:
http://www.instructables.com/id/Brain-Controlled-Wheelchair
[10] H. Heyrman. Brainwaves [Online]. Available:
http://www.doctorhugo.org/brainwaves/brainwaves.html
[11] S. Sequeira, C. Diogo and F.J.T.E. Ferreira, “EEG-signals based control
strategy for prosthetic drive systems,” in IEEE 3rd Portuguese Meeting in
Bioengineering, Braga, 2013, pp. 1-4.
[12] V. Charisis, S. Hadjidimitriou, L. Hadjileontiadis, D. Ugurca and E. Yilmaz,
“EmoActivity – An EEG-based gamified emotion HCI for augmented
artistic expression: The i-Treasures paradigm,” in Springler-Verlag Berlin
Heidelberg, Berlin, 2011.
[13] T. Beyrouthy, S. K. AlKork and J. A. Korbane, “EEG mind controlled smart
prosthetic arm,” in IEEE International Conference on Emerging
Technologies and Innovative Business Practices for the Transformation of
Societies, August 2016.
[14] S. K. Al Kork, "Development of 3D finite element model of human elbow to
study elbow dislocation and instability," in ASME 2009 Summer
Bioengineering Conference, American Society of Mechanical Engineers,
2009.
[15] T. Beyrouthy and L. Fesquet, “An event-driven FIR filter: design and
implementation,” in 22nd IEEE International Symposium on Rapid System
Prototyping (RSP), May 2011.
[16] C. Heckathorne, “Upper-limb prosthetics: Components for adult externally
powered systems,” in Atlas of Limb Prosthetics, H. K. Bowker and J. W.
Michael, Eds. Rosemont, IL: Am. Acad. Orthoped. Surg., 2002

www.astesj.com 899