EMG controlled Prosthetic Arm

Implementation of a 3D printed prosthetic arm controlled by EMG signals

1. Paper ID 39 Design and Implementation of an EMG Controlled 3D Printed Prosthetic Arm Authored by Nazmus Sakib and Md. Kafiul Islam Presented by Md. Kafiul Islam, PhD , SMIEEE Asst. Prof., Dept. of EEE, IUB [email protected] 1

2. Presentation Outline Introduction Motivation and Objectives Electromyography (EMG) Basics State-of-the-Art Prosthetic Systems Proposed System Design & Implementation Analog Front-End (EMG Recorder) Design & PCB Implementation Signal Analysis in MATLAB Prosthetic Arm Design & 3D Printing Control Circuitry Experiment and Testing Performance Evaluation & Comparison Conclusion & Future work 2

3. Motivation and Objectives 3 Motivation: > 4000 people get injured in road accident every year and live without one or more body parts. Rana plaza massacre and many other left many people cripple for life time. Dependency on imported expensive prosthesis. Goal: Provide an affordable solution of prosthetic limb for our country Objectives: Design an Analog Frontend circuit to record the electrical activity of muscle (EMG). Design and 3D print a prosthetic arm Detect muscle contraction to generate a command signal to control prosthetic arm.

4. Electromyography (EMG) 4 Electromyography (EMG) detects electrical potential generated by muscle cells upon voluntary contraction of muscle. Brain commands in spinal cord transmitted to muscle fiber by Motor neuron. When motor unit is activated, muscle fibers contract . Motor Unit Action Potential ( MUAP )

5. EMG Signal Characteristics 5 Typical MUAP Waveform Biopotential Frequency Range Signal Amplitude Electrode Electromyogram (EMG) 20 Hz – 1000 Hz Maximum Usable Energy : 50 Hz - 150 Hz 10 µV – 2mV Surface

6. 6 Currently Available Prosthetic Systems* Touch Bioinics Open Hand Project Bebionics Price range: USD 10K Products: • i -limb revolution • i -limb ultra • i -limb digits • livingskin Price range: USD 1K They are using 3D printer technology Price range: Between USD 11-60K . Most advanced prosthetic technology Human like hand movement system. Price range: USD 6K Products: Brunel Hand 2.0 Hero Arm Open Bionics *Ref: [2]-[5]

7. 7 Proposed System

8. Electrodes & Connector Using surface EMG (SEMG), made of Ag/ AgCl . Permits electron conduction from the skin to the wire. The connectors of these electrodes have three conductor sensor cable with electrode pad leads. Using electrolytic gel between skin and electrode can reduce electrode impedance. Multi-useable electrodes, positive, negative and ground, to the skin/surface covering the muscle, in order to detect muscle movement. 8 Connectors & Electrodes used in this Project Electrodes

9. Dual DC Voltage Supply 9 To provide constant voltage and protect circuit IC components. 5V Boost converters are used to provide +5V and -5V output. 3.7V Lipo or Li-ion both type of batteries can be used.

10. Instrumentation Amplifier 10 AD620: Low cost, high accuracy with high input impedance, low DC offset, low noise and very high open-loop gain. High CMRR: >100dB Amplitude of input: 0.01mV to 1 mV . Gain = 1+ 5000 (for Full swing of the ADC) Takes the difference between two electrodes and amplify it.  

11. Active High-Pass Filter 11 To remove low frequency noise interference and DC offset. Cutoff Frequency: = Order of filter: 2 nd order Gain: Unity Output: Non inverted  

12. Active Low-Pass Filter 12 Remove high frequency noise interference. Cutoff Frequency: = Order of filter: 2 nd order Gain: Unity Output: Non inverted  

13. 13 Analog Filters Simulation LM358 dual Op-Amp IC was used in this Project. Both filters ware designed and simulated in Porteous using LM358. The PCB (printed circuit board) was also designed in Porteous Software LM358

14. 14 PCB Design of Analog FE 3D view of PCB with components PCB Layout

15. 15 EMG Recorder Circuit (Analog FE) AD620 LM358 R Gain Pins to Connect Power Supply Electrode Connector Pins Output

16. 16 Complete EMG recorder Circuit Complete EMG Recorder FE

17. Recorded EMG Signals EMG signal Recorded shown on Oscilloscope 20/40 18

18. Signal Analysis in MATLAB 21/40 19

19. Signal Analysis in MATLAB (Cont…) Filtering in Digital domain Infinite impulse response (IIR) Filter Notch Filter To remove 50Hz power line noise Order of filter: 2 nd order Stability: Stable HPF Cutoff frequency: 10Hz Order of filter: 5 th order Stability: Stable LPF Cutoff frequency: 150Hz Order of filter: 4 th order Stability: Stable Power line noise & other noise removal 20

20. Trial No. SNR with Background Noise (dB) SNR after Filtering (dB) Improvement in SNR after Filtering (dB) 1 38.08 65.89 27.81 2 36.52 65.11 28.58 3 39.12 67.74 28.61 4 38.28 66.64 28.35 5 42.05 70.74 28.68 Average 38.81 67.22 28.41 Signal Analysis in MATLAB (Cont…) Signal Improvement Analysis by calculating SNR EMG signal before and after filtering   21

21. Prosthetic Arm 3D Design Open Source Design Designed and modified in TinkerCAD online free software 22

22. No. Name of the part No. of joints/parts Length /Weight 1 Thumb finger 2 5.5 cm 2 Index finger 3 6 cm 3 Middle finger 3 8.5 cm 4 Ring finger 3 7.5 cm 5 Pinky finger 3 5.5 cm 6 Palm 1 10 cm 7 Wrist 4 23 cm 8 Diameter at end of wrist ̶̶̶ 10 cm 9 Total length of the Arm ̶ 41.5 cm 10 In-fill ̶ 20% 11 Weight ̶ 300gm (approx.) 3D Printing (Technical Specs) Printed in PRUSA mk3 PLA filament Speed 6.09gm/min 23

23. 3D Printed Arm 5 Actuators (MG90s) for 5 Fingers 5 DOF Elastic foe flexibility Strong thread as Tension wire Extra supports for more stable finger movement 24

24. Control Circuitry Circuit includes 2 Arduino pro-mini EMG recorder Circuit Voice recognition module Prosthetic Arms actuators (MG90s servo motors) 1 st Arduino It gets signal from EMG recorder Process the signal in digital domain using the coefficients calculated in MATLAB. detects muscle construction by setting up a threshold and send a pulse to the 2 nd Arduino 2 nd Arduino It detects the pulse from 1 st Arduino. For the 1 st pulse the Arduino sends signal to the arm to perform a gesture and when it gets another pulse it sends a signal to relax the arm. This Arduino parallelly checks the voice command. Whenever a different voice command is given the prosthetic arm performs a different gesture when it gets a pulse from 1 st Arduino . 25

25. Gesture Mode Thumb Index Middle Ring Pinky Grab 130° 150° 160° 155° 130° Pick 130° 150° 0° 0° 0° Open 0° 0° 0° 0° 0° Control Circuitry (Cont…) Prosthetic Control Each finger is connected to a servo motor by a tension wire. Whenever the servo rotates the fingers of the arm move to perform a gesture Servos are controlled by PWM signal with fixed frequency of 50 Hz. All servos receive individual PWM signal from the 2 nd Arduino. By varying the pulse width the angular rotation varies. For different gestures the PWM signal changes differently for each finger as a result the each individual motor has individual angular rotation Initial Angle Maximum Angle 26

26. 2 Different gestures performed for different task Grabbing of Objects by the Arm 27

27. Prosthetic Arm 1st Arduino EMG Recorder Circuit Servo motors Voice Recognition Module 2nd Arduino Complete System 28

28. 28 Performance Evaluation ( Accuracy ) Average Accuracy per subject is around 84 %!   6 subjects Each subject had 5 trials And 6 commands in each trial In total 180 trial samples for accuracy calculation

29. 29 Performance Evaluation ( Response Time & Power Consumption ) Supply Voltage Current Consumption (No load) Current Consumption (Full load) Total Power Consumption (Full load) Overall System Response Time 5V 0.5A 2.6A 13W 1.5 sec ( approx.) 7.4V 6000mAh Li-ion battery Regulated using 7805 voltage regulator

30. (Component) Cost Analysis 30 SI Component Unit Cost (BDT) 1 Electrodes 3 30 2 Connectors 1 20 3 PCB 1 50 4 AD620 1 220 5 LM350 Op-Amp 1 10 6 Electrical components (resistor capacitor, connector, wires, LED, Vero-board, regulators etc.) ̶ 200 7 Lipo Batteries 2 400 8 Li-ion Batteries 6 300 9 Boost Converters 2 100 10 Arduino Pro-Mini 2 300 11 Prosthetic Arm (printing cost) 1 1000 12 Servo motor MG90s 5 750 13 Voice recognition module V3 1 2000 14 Others (glue, tension wire, elastic etc.) ̶ 100   Total cost   5480 Commercial EMG recorder available in Bangladesh ( Price: 5150 BDT)

31. 31 Comparison with Other Works Ref No . Biomedical recorder Robotic/Prosthetic hand Extra feature 1 Commercial EMG recorder In house 3D printed Prosthetic Hand No Extra Feature 2 In house-built EMG recorder Not enough info available 3 Commercial EMG recorder No prosthesis / simulated controlling 4 Commercial biomedical signal recorder In house built Robotic arm 5 In house-built EMG recorder Commercial robotic arm 6 Not enough info No prosthesis / simulated controlling 7 Commercial EMG recorder Commercial 3D printed Prosthetic Hand 8 Commercial EMG recorder No prosthesis / simulated controlling 9 Not enough info In house built Prosthetic Hand 10 Commercial EMG recorder No prosthesis / simulated controlling 11 Commercial EMG recorder In house-built hand rehabilitation robotic system 12 Commercial EMG recorder In house 3D printed Prosthetic Hand 13 Not used Simulation test Voice Recognition 14 In house 3D printed Prosthetic Hand 15 Commercial robotic arm This One In House built EMG recorder In house 3D printed prosthetic arm Voice Recognition

32. Conclusion The EMG signal and Voice command controlled Prosthetic system developed in this project is less expensive and can be affordable for people in developing country like Bangladesh. The system was successfully designed and tested on an amputee individual. Since the whole system is designed and developed here, so if any modification required, can be possible. 32

33. Future Works Making the prosthetic arm more user friendly Using Dry electrodes Making an in-house built voice recognition module with local recources Improving the design of the prosthetic arm Increasing the degree of freedom of the prosthetic arm Making the project available for the people so that they can have the benefits of this project by improving their quality of life. EMG signal based diagnosis of neuromuscular diseases. 33

34. 34 Sample Video of Experiment

35. References 35 Yazicioglu , Refet Firat , Chris Van Hoof, and Robert Puers . Biopotential readout circuits for portable acquisition systems. Springer Science & Business Media, 2008. https://www.ottobockus.com/prosthetics/upper-limb-prosthetics/solution-overview/bebionic-hand/ (assessed on 02/09/2019) https://www.touchbionics.com/(assessed on 02/09/2019) http://www.openhandproject.org/(assessed on 02/09/2019) https://openbionics.com/(assessed on 02/09/2019) Mounika, M. P., B. S. S. Phanisankar , and M. Manoj . "Design & analysis of prosthetic hand with EMG technology in 3-D printing machine." Int. J. Curr . Eng. Technol 7, no. 1 (2017): 115-119. Trzmiel , Grzegorz, Dariusz Kurz , and Wiktor Smoczyński . "The use of the EMG signal for the arm model control." In ITM Web of Conferences , vol. 28, p. 01024. EDP Sciences, 2019. Ganiev , Asilbek , Ho-Sun Shin, and Kang- Hee Lee. "Study on virtual control of a robotic arm via a myo armband for the selfmanipulation of a hand amputee." Int. J. Appl. Eng. Res 11, no. 2 (2016): 775-782. Minati , Ludovico, Natsue Yoshimura, and Yasuharu Koike. "Hybrid control of a vision-guided robot arm by EOG, EMG, EEG biosignals and head movement acquired via a consumer-grade wearable device." Ieee Access 4 (2016): 9528-9541. Bitzer, Sebastian, and Patrick Van Der Smagt . "Learning EMG control of a robotic hand: towards active prostheses." In Proceedings 2006 IEEE International Conference on Robotics and Automation, 2006. ICRA 2006. , pp. 2819-2823. IEEE, 2006. Blana , Dimitra, Theocharis Kyriacou , Joris M. Lambrecht, and Edward K. Chadwick. "Feasibility of using combined EMG and kinematic signals for prosthesis control: A simulation study using a virtual reality environment." Journal of Electromyography and Kinesiology 29 (2016): 21-27. Said, S., M. Sheikh, F. Al-Rashidi, Y. Lakys , T. Beyrouthy , and A. Nait-ali . "A Customizable Wearable Robust 3D Printed Bionic Arm: Muscle Controlled." In 2019 3rd International Conference on Bio-engineering for Smart Technologies ( BioSMART ) , pp. 1-6. IEEE, 2019. Wong, Tat Hang, Davide Asnaghi , and Suk Wai Winnie Leung. "Mechatronics Enabling Kit for 3D Printed Hand Prosthesis." In Proceedings of the Design Society: International Conference on Engineering Design , vol. 1, no. 1, pp. 769-778. Cambridge University Press, 2019. Borisov, Ivan I., Olga V. Borisova , Sergei V. Krivosheev , Roman V. Oleynik , and Stanislav S. Reznikov. "Prototyping of emg -controlled prosthetic hand with sensory system." IFAC- PapersOnLine 50, no. 1 (2017): 16027-16031. Visconti, P., F. Gaetani, G. A. Zappatore , and P. Primiceri . "Technical features and functionalities of Myo armband: an overview on related literature and advanced applications of myoelectric armbands mainly focused on arm prostheses." Int. J. Smart Sens. Intell . Syst 11, no. 1 (2018): 1-25. Abdallah, Ismail Ben, Yassine Bouteraa , and Chokri Rekik. "DESIGN AND DEVELOPMENT OF 3D PRINTED MYOELECTRIC ROBOTIC EXOSKELETON FOR HAND REHABILITATION." International Journal on Smart Sensing & Intelligent Systems 10, no. 2 (2017). Hetherington, Austin T. "Integration of a Sensory Driven Model for Hand Grasp Function in 3D Printed Prostheses." (2018). Gruppioni , E., B. G. Saldutto , A. G. Cutti , Elena Mainardi , and A. Davalli . "A voice-controlled prosthesis: test of a vocabulary and development of the prototype." In Proceeding of MEC (Myoelectric Control Conference) . 2008. Asyali , Musa Hakan , Mustafa Yilmaz, Mahmut Tokmakci , Kanber Sedef, Bekir Hakan Aksebzeci , and Rohin Mittal. "Design and implementation of a voice-controlled prosthetic hand." Turkish Journal of Electrical Engineering & Computer Sciences 19, no. 1 (2011): 33-46. House, Brandi, Jonathan Malkin , and Jeff Bilmes . "The VoiceBot : a voice-controlled robot arm." In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems , pp. 183-192. ACM, 2009 .

36. Thank You! 36