A Customizable Wearable Robust 3D Printed Bionic

Arm: Muscle Controlled

S. Said1,2 , M. Sheikh1 ,F. Al-Rashidi1, Y. Lakys1,T. A. Nait-ali2
Beyrouthy1, S. alkork1 2
Laboratoire Images, Signaux et Systèmes Intelligents
1 (LISSI), Université Paris-EST 122 rue Paul Armangot
College of Engineering and Technology
American University of the Middle East 94400 Vitry sur Seine, France
Eqaila, Kuwait
[email protected]
to have a new arm. The surgical method is very costly. Some
problems may happen due to the surgical arm. For example,
Abstract— In this research paper, customizable wearable sometimes the nerves may cause problem when they are
robust 3D printed bionic arm been designed and implemented for damaged totally, which could make it hard to perform surgery.
a right arm. An experimental test has been conducted for a
person who has been born without a right arm. A control of a 9
In addition, the surgical method causes heart disease and back
DOF hand is accomplished successfully using EMG signals pain in some patients. The amputees face nociceptive and
acquired by MYO armband. Different gestures are recognized neuropathic pain, due to bone and soft tissue injury [3]. On the
and mapped to control different movements of the hand. Basic other hand, the prosthetic arm has less problems when
daily activities have been accomplished after training of that compared to the surgical arm. Prosthetic arm avoids many
person. The arm is a lightweight for daily use for long time. The medical issues that may results from the surgery procedure.
cost of the arm is cheap compared to available solutions available There are many techniques to control a robotic arm. One
on the market. technique is to use an EEG device. The EEG is a headset,
which records the brain waves when the person is thinking of
Keywords—Myo armband; Bionic arm; Prothetics; Gesture
Recognition; Robotics an action or implementing a facial expression. The EEG will
read signals and then convert them to commands, to send them
to the arm [4]. The second technique is to use Muscle Activity
I. STATE OF ART
Sensors, which is called electromyography (EMG) sensors.
Many people are suffering from disability, which is preventing EMG signal is the electric potential generated by muscle cells
them from doing their daily live activities. Recent statistics [7] when these cells are electrically or neurologically
study shows 15% of the world’s population suffers from some activated. The signals can be analyzed to detect medical
forms of disability [1]. Amputee’s number is around 10 abnormalities, activation level, or recruitment order, or to
million out of the world’s population, in which 30% of them analyze the biomechanics of human. EMG signal are to be
are arm amputees [2]. Moreover, due to many reasons, the processed for multiple hand gesture and movement
overall rate of amputees and limb dysfunction patients is recognition. EMG monitors the electrical signals under human
increasing. Arm amputees are being classified as either being skin that are produced by the muscles. Myo is a wearable
born without an arm or part of it or completely lost of arm due bracelet that can be used as a commander for any connected
to disease or accident. Both scenarios are suffering while device [8]. The Myo has no cameras, no keyboard and mouse,
doing their daily life routine by themselves and they need it just involves the hands. The Myo can detect any tiny
assist from others [3]. There are different proposed solutions electrical activation (contraction) produced by the muscle.
to help the amputees, but they are either being incredibly Myo Armband detects the electrical activity in forearm
expensive not everyone can afford, hard to install and muscles just below the elbow. The human forearm has
maintain, or it may require surgical procedures in this case it different types of muscles, each of which has a different
relies on the nerves which in some cases might be damaged arrangement, and these muscles control the movement of the
[5]. The mind-controlled 3D printed arm has the potential to wrist, such as moving fingers, making a fist, turning left or
help amputees to normally preform many of their daily right [9]. Myo Armband has been used in a very efficient way
activities [6], provide a better life, and improve the quality of by researchers at Johns Hopkins University [10] to control a
life. In addition, it helps amputees to regain functionality with prosthetic limb using electric impulses transmitted from an
natural control via brain waves. There are several existing amputee’s mind to his limb. The armband works by reading
solutions such as surgical arm, myoelectric-controlled arm and the electromyographically (EMG) impulses triggered by a
cosmetic restoration. Each type has advantages and thought from a person’s brain sending a signal to a limb, which
disadvantages. One of the solutions for amputees is having a causes a movement. A transradial myoelectric prosthesis
prosthetic arm. Prosthetic arm needs to be measured and based on an innovative mechanism called Adam’s hand has
specified to the patient needs. The second solution is the been developed. It can actuate five three-phalanx fingers (15
surgical limbs, where the patient will have a surgical operation degrees of freedom). Adam's Hand fingertips are provided
with temperature and pressure sensors, while the user Figure 1 shows the process starting from acquiring EMG
myoelectric signals are acquired wirelessly by means of the signal using Myo Armband; the signals are transferred
Myo armband [11]. Different control approaches have been wirelessly to the PC. In the PC, the gesture done is recognized
implemented in controlling prosthetic arms. A regression- and then a control signal is sent to the bionic arm controller to
based control approach that allows for simultaneous and perform the desired action.
proportional control of two degrees of freedom implemented
A. Wearable Armband
and evaluated on five prosthetic end users [12]. Tasks
mimicking daily life activities have been tested, the regression Many wearable systems are available on the market for
approach was robust over multiple days and only slightly acquiring bio-signals [20]. Myo Armband is the basic concept
affected by changing in the arm position. Proportional and of Human Computer Interaction (HCI) in which humans
simultaneous control of multiple degrees of freedom of the interact with computers and systems in an interactive way. The
prosthesis can allow achieving more natural and dexterous structure of Myo consists of eight EMG (electromyography)
control using unsupervised or supervised methods [13], [14]. sensors and an IMU which consists of a gyroscope,
Unsupervised methods are usually based on signal accelerometer and a magnetometer [21]. Myo can be worn on
factorization (e.g., through Non-Negative Matrix the forearm below the elbow which this person has. Myo is
Factorization, NMF), they require a short calibration phase detecting the human gestures and movements. Myo detected
and they are relatively independent on the number and exact basic gestures wave left, wave right, fist and fingers open. The
location of the electrodes [15], [16]. Supervised methods [17], training of the person is done for the four gestures. He kept
[18] are usually based on regression techniques (e.g., Linear practicing controlling his forearm muscles and check on the
Regression, LR, Artificial Neural Networks, ANN, Support PC monitor the recognised gesture until he has succeffuly got
Vector Machines, and SVM) that require a reliable ground able to control his muscles to do certain action easily.
truth for hand kinematics.
II. METHODOLOGY
The bionic arm presented in this paper inspired due to a need
of a person to a bionic arm that can solve the problems he
faced in his life. This person is 23 years old born without a
right arm. When he was 12 years old, he experienced his first
prosthetic arm manufactured specially for him. This arm can
open and close only by EMG signal without any other
movements. He used his arm for 3 years, but he complained
about the comfort level. He used to take it out every 2 hours
and it was heavy. When he reached 16 years old, He travelled Figure 2 Gestures Recognition training using MYO
to Italy for producing a cosmetic arm, which does not have any EMG signals from the eight electrodes are used to define the
ability to move. After 2 years, he travelled to Germany to muscle activity. The combination of the signals is used to
produce a bionic arm that can move. The problems he faced detect a specific gesture of the user. The user signals are
was the heave weight, he must take it out every now and then recorded to analyze its power level for all of the defined
due to the uncomfortable feeling. All these points have been gestures. The IMU signals available in the Myo are not used
taken into consideration while designing an arm for him that since it is not equipped with a motor at the wrist. Figure 3
can solve all the problems he experienced. He was involved in shows the EMG signals acquired by Myo while the user is
the design phase getting his feedback in every step of the performing Fist action.
design.

Figure 1 Schematic drawing of controlling the bionic arm

 The 3D model of the bionic arm is shown in Figure 3; The
Bionic Arm design is based on multiple parameters listed
below:
 Comfort: Comfort is one of the most important factors to
prolong the usage of the arm; many thoughts are given in
this process. The main factor that decides the comfort level
is the attachment where the arm goes in. Many different
materials were considered to achieve a comfortable grip.
Inspired by high-end mattresses, the grip is made of
multilayered foam that has hard foam at the bottom for good
grip and a thin layer of soft memory foam.
 Appearance: To make it look more realistic the arm was
made as a replica of the non-affected arm.
 Brunel Hand: A 3D assembly of the Brunel hand is made
to ensure the fitting.
Figure 3 Eight EMG signals  Bicep Support: The extension of the arm adds weight and
the support holding it upright is the insertion point. This
B. Bionic Arm Design adds pressure to this point on the bones and makes the
overall experience uncomfortable. Therefore, additional
The loss of a limb comes with both emotional and functional
support is required to ease the pressure. A Bicep support is
damage thus making the design process very important not
added using readily available straps.
just in terms of comfort but appearance as well. In this
 Myo Integration: The Myo band was integrated in the arm
research, the case studied is for person who has been born
at the positon based on the correct signals.
without most of his right arm, which made it harder since he
 Light Weight & Breathable: The arm is made to be
did not control his forearm muscles before. The focus is to
breathable and lightweight by strategically removing
provide an arm that can add features for him in his daily life
unnecessary parts without compromising the strength of the
activities. The current devices available in the market range
arm.
from 4000 up to 20000 US Dollars [22]. The main target is to
manufacture a cheap bionic arm affordable for amputees.  Center of gravity: Having the center of gravity behind the
Nowadays, the advancement and easy access of 3-D printing center point helps in lifting the arm while providing more
technology, has reduced the cost of manufacturing the bionic support.
arms and provides easier solutions for the design of prosthetic  Strength and Durability: To be functional for daily usage
arms customized for users. the arm should be strong and durable. This property was
tested using a Finite Element Analysis. With the fixture at
The Bionic arm is designed customizable for a user; the the insertion point (Figure 5: green arrows) and the load at
dimension of his non-affected arm (Left Arm) was first the far end of the arm (Figure 4: purple arrows). With the
measured. A mirrored geometry was then made using a test load of 2.4 kg (0.4kg for the Brunel hand), the results
computer aided design software. The initial prototype has show that the arm should not under go any major stress issue
been drawn then modified after measuring the small part he when lifting of 2kg load.
has in his right arm after the elbow. The dimensions of the
affected arm were taken and embedded in the design to
develop a wearable arm with an empty space for Myo
Armband to fit and be invisible in his arm.

Fig 5. Stress analysis of the designed arm

 Electronics and Battery: To ensure safe and organized
assemble, passages for the cables and the electronics were
designed. For minimizing the weight of the arm, the battery
was used as an external unit with a wire attached to the arm.
After completion and evaluation of the design, a large-scale 3D
printer is used to print the whole arm in one print. Table 1 shows
the properties of the print and the price of the electronics and
Fig 4. Bionic Arm Design
actuators used in the hand. The total cost of the whole arm, parts The system flowchart is shown in Figure 8. All the Myo
and electronics is around $255. This price is very cheap armband data is acquired wirelessly through Bluetooth to the
compared to other available solutions in the market. PC serially. This serial data is processed to recognize the user
Table 1 Cost analysis gesture and the arm movements. GUI screen for interfacing
with the user is developed to present for the user the gestures
Index Property Value
detected, and the position of the arm as shown in Figure 9. The
1 Time to Build 10 hours screen is showing the connection status with Myo. The
Total Weight without 362.45g detected gesture by Myo EMG sensors is mapped to perform
2
Support material hand movement for example, closing the hand, close one
3 Support material weight 78.78g finger or two fingers. These movements are achieved by
4 Material Cost $12.41 controlling the motion of the actuators. The control signals of
the actuators are transferred by C program serially to Chestnut
5 Hand print $20
board, which is sending all the commands serially to chestnut
6 Electronics $20 board.
7 Actuators $200
Total Cost $255

Figure 6 User wearing the bionic arm

C. Electronics and control

The arm starting from the elbow until the hand attachment
point is a 3D printed part with a slot for a screen. The hand has
the Chestnut board embedded inside it [23] featuring the ARM
Cortex M0+ Processor.

Fig 8 System Flowchart

Fig. 7 Chestnut board pins configurations

The Chestnut board is designed to be embedded within

robotics projects, from balancing robots to robotic hands. It
can control up to four motors simultaneously, and can be
controlled via USB, UART, I2C and analogue input. It has an
IMU embedded with it. The mass of the board is 15 g and it is Fig. 9 Interface Screen for the user
dimensions are small 57 x 45 x 9 mm.
 III. RESULTS success rate. In two actions combination, the user performed
Testing scenarios have been proposed to the user for testing Fist then wave out with success rate of 100 % but the user
the arm design and the EMG control. The user has been asked performed Fist then wave in and wave in then wave out with
to wear the Myo armband in his forearm. The user performed 95 % success rate as shown in Figure 10.
each trained gesture using his muscles for 20 times
consequently to validate them. Then the user is asked to
perform two different gestures consequently for 20 times to
validate the use of daily activities for the user.

The testing scenarios showed the ability of the user to control

the bionic hand accurately after training phase. The bionic
hand movements are optimized to allow the user to perform
different activities, holding tiny objects, holding a bottle, and
grasping of a card. In single action testing, the user has been
asked to perform one action at a time. The actions include
(Fist, spread fingers, one finger close, two fingers close) as
shown in Figure 9.

Figure 10 Results

IV. CONCLUSIONS AND FUTURE WORKS

This research work details the design and control of a
myoelectric prosthesis bionic arm, which has been realized,
with the aim of improving the quality of life of upper-limb
amputees. The prosthesis is designed to be affordable and
(a) lightweight in order to provide the users with comfort and easy
use during his daily activities.
ACKNOWLEDGMENT
Special thanks to AUM for funding this research project and
for using AUM robotics research and innovation center
facilities to develop the bionic arm.
REFERENCES
[1] Hawking, S. W., "World Report on disability," Geneva: World
Health Organization, 2011..
(b)
[2] M. LeBlance, "Give a hope," 2011. [Online]. Available:
https://web.stanford.edu/class/engr110/Newsletter/lecture03a-
2011.html.
[3] S. Hasan, K. Al-Kandari, E. Al-Awadhi, A. Jaafar, B. Al-
Farhan, M. Hassan, S. Said and S. AlKork, "Wearable Mind
Thoughts Controlled Open Source 3D Printed Arm with
Embedded Sensor Feedback System," in CHIRA, Seville,
Spain, 2018.
[4] Taha Beyrouthy, Samer Al. Kork, J. Korbane, M. Abouelela,
"EEG Mind Controlled Smart Prosthetic Arm – A
Comprehensive Study", Advances in Science, Technology
and Engineering Systems Journal, vol. 2, no. 3, pp. 891-899
(2017). google scholar
[5] Junhua Li ; Gong Chen ; Pavithra Thangavel ; Haoyong Yu ;
(c) Nitish Thakor ; Anastasios Bezerianos ; Yu Sun, "A Robotic
Figure 9 (a) Two fingers close (b) Fist (C) One finger close Knee Exoskeleton for Walking Assistance and Connectivity
Topology Exploration in EEG Signal," in IEEE RAS/EMBS, U
The user performed each action repetitively for 20 consecutive town, 2016.
times. The results of each single action testing are 100 %
[6] Ahmed Roshdy, Samer Al Kork, Sherif Said, Taha Beyrouthy, the surface electromyographic signal," in IEEE Trans. Biomed.
"A Wearable Exoskeleton Rehabilitation Device for Paralysis– Eng., 2009.
A Comprehensive Study," Advances in Science, Technology
and Engineering Systems , Vols. Journal Vol. 4, , no. No. 1, , [16] Muceli, S., Jiang, N., and Farina, D., "Extracting signals robust
pp. 17-26 , 2019. to electrode number and shift for online simultaneous and
proportional Myoelectric control by factorization algorithms,"
[7] G. Robertson, G. Caldwell, J. Hamill, G. Kamen and S.
in IEEE Trans. Neural Syst. Rehabil. Eng., 2014.
Whittlese,, " Research Methods in Biomechanics," Human
Kinetics, vol. 2nd Edition, 2014. [17] Nielsen, J. L. G., Holmgaard, S., Jiang, N., Englehart, K. B.,
Farina, D., and Parker, P. A. , " Simultaneous and proportional
[8] Sherif Said, Samer AlKork, Amine Nait-Ali, "Wearable
force estimation for multifunction Myoelectric prostheses
Technologies in Biomedical and Biometrics Applications," in
using mirrored bilateral training," in IEEE Trans. Biomed.
Biometrics under Biomedical Applications, Springer, 2019, p.
Eng., 2011.
Chapter 10.
[18] Hahne, J. M., BieBmann, F., Jiang, N., Rehbaum, H., Farina,
[9] A. Ganiev, H.-S. Shin and K.-H. Lee,, " "Study on Virtual
Control of a Robotic Arm via a Myo Armband for the Self D., Meinecke, F. C.,, "Linear and nonlinear regression
Manipulation of a Hand Amputee,"," International Journal of techniques for simultaneous and proportional Myoelectric
Applied Engineering Research, vol. vol. 11, pp. pp. 775-782, control," in IEEE Trans. Neural Syst. Rehabil. Eng., 2014.
2016.. Ameri, A., Kamavuako, E. N., Scheme, E. J., Englehart, K. B.,
[10] "https://www.cantechletter.com/2016/01/thalmic-labs-myo [19] and Parker, P., "Support vector regression for improved real-
armband-allows-amputee-to-control-prosthetic-limb/," time, simultaneous Myoelectric control," in IEEE Trans.
[Online]. Neural Syst. Rehabil. Eng., 2014.
[11] Federico Gaetani, Patrizio Primiceri, Giovanni Antonio S. Said, S. Alkork, T. Beyrouthy and M. Fayek, "Wearable
Zappatore, Paolo Visconti, "Hardware design and software [20] Bio-Sensors Bracelet for Driver’s Health Emergency
development of a motion control and driving system for
Detection," in Biosmart, Paris, France, 2017.
transradial prothesis based on a wireless myoelectric
armband," IET Science, Measurement & Technology, 2018. S. Rawat, S. Vats and P. Kumar, "Evaluating and Exploring
[12] Janne M. Hahne, Meike A. Schweisfurth, Mario Koppe, and [21] the MYO Armband," in International Conference on System
Dario Farina, "Simultaneous control of multiple functions of Modeling & Advancement in Research Trends, Moradabad,
bionic hand prostheses: Performance and robustness in end India, 2016.
users," Science Robotics, vol. 3, no. 19, 2018. J Zuniga, D Katsavelis, J Peck, J Stollberg, M Petrykowski, A
[13] Fougner, A., Stavdahl, Ø., Kyberd, P. J., Losier, Y. G., and [22] Carson, C. Fernandez, "Cyborg beast: a low-cost 3d-printed
Parker, P. A. , "Control of upper limb prostheses: terminology prosthetic hand for children with upper-limb
and proportional Myoelectric control-a review," in IEEE differences", BMC Research Notes, vol. 8, no. 10, pp. 1-8,
Trans. Neural Syst. Rehabil. Eng, 2012. 2015.
[14] Farina, D., Jiang, N., Rehbaum, H., Holobar, A., Graimann, B., [23] https://openbionicslabs.com/shop/chestnut-board
Dietl, H.,, "The extraction of neural information from the
surface EMG for the control of upper-limb prostheses:
emerging avenues and challenges.," in IEEE Trans. Neural
Syst. Rehabil., 2014.
[15] Jiang, N., Englehart, K. B., Parker, P. A., and Englehart, K. B.
, "Extracting simultaneous and proportional neural control
information for multiple degree of freedom prostheses from