Template LAAR 2021

Abstract-- Replacement of missing arms by an active prosthetic arm can improve the quality of life of an amputee. We demonstrate the development of a prosthetic arm by analyzing the electromyography (EMG) signals received from the EMG sensor connected to the muscles of a cleft arm to support disabled people. The prosthetic arm was designed using SolidWorks simulator. After convincing simulation results, we developed the prosthetic arm using 3D printing technology. The robust prosthetic arm was made of polymer for light weight and long durability. We analyzed the EMG signals of different muscles such as forearm and bicep received from several people of different gender and age group. Based on the EMG data, we designed the control unit of the prosthetic arm. Furthermore, we investigated the performance of the EMG sensor actuated prosthetic arm. We believe that, the proposed EMG sensor based prosthetic arm can perform almost like a normal arm.

Keywords-- Prosthetic arm, electromyography sensor, cleft arm, SolidWorks simulator, 3D print.

I. INTRODUCTION

Prosthesis can be considered as an artificial device that replaces a human organ and performs like it (Sudarsan and Sekaran, 2012). The World Health Organization (WHO) estimated a total of 30 million people in need of prosthetic and orthotic devices by the year of 2010 where 75% of them were living in developing countries (World Health Organization, 2005). A prosthetic arm can be installed in place of an amputated limb of an amputee to support his regular activities. Depending on the types and degrees of disabilities of a human, the design and required functions of the prosthetic arm are different. However, the prosthetic arms or hands can accomplish very few functional motions of human arms and hands (Verma et al., 2013). Currently three types of prosthesis are available as a replacement of upper limb of an amputee: body-power controlled and cable-driven device with a split hook, electroencephalogram (EEG) and electromyography (EMG) sensors actuated electronically powered prosthesis, and a cosmetic replacement with a passive prosthetic hand (Heger et al., 1985).

Because of the significant importance, development of prosthetic limbs comes to the forefront of the research field. Large numbers of research works had been conducted by many research groups to develop electronically power driven prosthetic arms (Selvan et al., 2021; Sudarsan and Sekaran, 2012; Verma et al., 2013; Heger et al., 1985; Bright et al., 2016; Beyrouthy et al., 2016; Quraishi et al., 2018; Neogi et al., 2011; Mavani et al., 2014; Xu et al., 2016; Li et al., 2011) and legs (Parajuli et al., 2019; Sawake et al., 2014; Islam et al., 2011) where the researchers mostly utilized EEG (Bright et al., 2016; Beyrouthy et al., 2016; Quraishi et al., 2018) and EMG (Sudarsan and Sekaran, 2012; Verma et al., 2013; Heger et al., 1985; Neogi et al., 2011; Mavani et al., 2014; Xu et al., 2016; McDonald et al., 2020; Li et al., 2011) sensors. Arm movement signals coming out from the EEG head sensors are difficult to distinguish which in turn affects the decision making during the movement of the prosthetic limb. As a result, EEG sensors actuated prosthetic arms suffer from inaccuracy and poor precision. Compared to the EEG signals, EMG signals of any limb provide specific signals depending on the desire of the human to make accurate and precise decision for appropriate movement of any prosthetic limb. Human brain sends movement signals to the cleft arm from the motor cortex by means of the spinal cord through a spinal nerve up to a network near the shoulder of the disabled people. These signals for each finger are directed are sent along either the Radial nerve, Median nerve, or the ulnar nerve. EMG muscle sensors can collect these signals from the human body and can be utilized to develop prosthetic organs.

Researchers analyzed the muscle activities by means of EMG signals and processed the signals towards the development of a prosthetic arm (Sudarsan and Sekaran, 2012). Their research work was primarily focused on processing EMG signals to generate pulses. Application of such pulses for prosthetic arm movement was missing in the research work. Verma and his research group proposed the development of a myoelectric arm using a single actuator (Verma et al., 2013). However, the developed arm was bulky and able to perform very limited tasks. Another research group developed an electrically powered muscle sensor actuated prosthetic arm for upper limb amputation (Heger et al., 1985). Although the developed prosthetic arm was a remarkable achievement in mid 1990s, the weight and structure of the prosthetic arm made it unsuitable for commercial use in modern era. Another research group developed a very simple prosthetic arm with very few degrees of freedoms by utilizing EMG signals coming out from bicep and thus less effective for amputees (Neogi et al., 2011). Researchers also proposed the design of a myoelectric prosthetic arm without developing it (Mavani et al., 2014). Another research group proposed pattern recognition based EMG sensor actuated prosthetic arm with limited degrees of freedoms (Xu et al., 2016). Some other researchers also introduced pattern recognition assisted EMG signal actuated upper limb prosthesis without developing the prosthetic arm (Li et al., 2011). This research work was primarily focused on pattern recognition from the EMG signals received from arm muscles for basic six kinds of wrist and hand movements. Pattern recognition assisted EMG sensor based prosthetic arm was also proposed by another research group (Bright et al., 2016). Another research group proposed the development of a prosthetic leg based on EMG signals for above knee amputees (Sawake et al., 2014). One research group developed an EMG sensor assisted microcontroller driven prosthetic leg (Islam et al., 2011). Although the authors of both of the research works claimed the development of prosthetic legs, real-life performance analysis showing the walking capability of any disabled human with the help of the developed prosthetic legs was absent in the research works. Although a large variety of prosthetic arms and legs have been proposed by different research groups, still the development of EMG sensor actuated prosthetic arm with high degree of freedoms are challenging and deserves more investigation.

In this work, we reported the development of an EMG sensor actuated prosthetic arm to offer wide range of activities to the disable people with high degree of freedoms. The prosthetic arm was designed and simulated using SolidWorks simulator and fabricated by means of 3D printing technology. The EMG signals, coming out from the biceps or forearms, have been converted to associated movement command to open or close the fingers of the prosthetic arm. In order to design the control unit of the prosthetic arm, we analyzed the EMG signals of different muscles from the people of different age group and gender. To control the prosthetic arm, a disabled person simply needs to integrate the light-weight prosthetic arm with the cleft arm where the control unit is placed inside the prosthetic arm. The muscle sensor must be attached with the appropriate location of the cleft arm to receive EMG signals. The polymer made prosthetic arm was robust enough to carry and hold a large dynamic of materials of different size and shape having maximum weight of ~ 2 kg. Most importantly, the proposed model has the flexibility of designing and fabricating person-specific prosthetic arm depending on the level of disability. Integration of a long-life rechargeable Lithium-Polymer (Li-Po) battery with the prosthetic arm creates the opportunity of long duration use without recharging. We strongly believe that, the amputees of the developing countries can afford the cost effective prosthetic arm.

II. MATERIALS AND METHODS

A. Design Procedure

As mentioned before, we designed the prosthetic arm using SolidWorks simulator. Figure 1 shows the basic model of the proposed EMG sensor actuated prosthetic arm for integrating with the cleft arm.

The design of the prosthetic arm depends on the level of disability and size & shape of the cleft arm of any amputee. For proper functioning, the cleft arm is put inside the prosthetic arm and integrated with it by means of flexible belt. The schematic diagram of the EMG sensor actuated prosthetic arm and its integration process with the cleft arm is illustrated in Fig. 1(a). The design of the hand and fingers of the proposed prosthetic arm is depicted in Fig. 1(b). There are three joints in each finger where the movements of the fingers are controlled by pulling-up/down the artificial tendons by means of a servo motor. Depending on the requirement of the amputee, we need to modify the design of the prosthetic arm, especially, the length of the fingers & location of their joints, palm, and other parts of the prosthetic arm separately before 3D printing. Since the proposed model is same for all amputees, designing and fabricating person-specific prosthetic arm is not a matter of big concern.

B. Simulation Details

The simulation of the electrical and electronic circuits was conducted using Proteus simulator. The model of the prosthetic arm was designed using commercial SolidWorks Simulator. We examined the movement smoothness and strength of the prosthetic arm during simulation. Depending on the simulation results, necessary changes were made in our design before finalizing the model. Figure 2 represents the initially designed and finalized models of the prosthetic arm.

After we received acceptable simulation results, we went for fabricating small pieces of the prosthetic arm using 3D printing technology. Afterwards, we combined all small pieces to develop the cohesive prosthetic arm. At first, we designed a model of the thumb-less prosthetic

arm, as depicted in Fig. 1(a). Although the simulation results were good, the thumb-less prosthetic arm showed poor performance during closing and holding goods. As a result, we decided to add a thumb with the prosthetic arm and achieved the final model shown in Figs. 2(b)–2(f). The electronic control module of the prosthetic arm was installed inside the prosthetic arm, as illustrated in Fig. 2(c). During simulation, we examined different kinds of movement capability of the prosthetic arm to analyze the performance. Simulation results confirmed excellent movements of fingers with high degree of freedom in the final model of the prosthetic arm (Figs. 2(d)–2(f)).

C. Construction Materials

We fabricated the fingers, palm, and forearm of the prosthetic arm using polymer filaments (diameter: 1.75 mm ± 0.02 mm; tensile strength: ³ 55 Mpa; density: 1.25 g ± 0.05 g/cm³) by means of a 3D printer (company: AnyCubic; model: i3 Mega). Artificial tendons were utilized to connect the fingers of the prosthetic arm with a metal gear servo motor (radius: 1.5 cm; rotation time (full-load): 0.15 s/60°). The rotation of servo motor enabled the pull-up/down of tendons to control the movement of the fingers.

III. SYSTEM ARCHITECTURE OF THE PROS-THETIC ARM

The proposed prosthetic arm consists of fundamental three modules: EMG signal acquisition module, electronic control module, and mechanical module. Figure 3(a) illustrates the system architecture of the EEG sensor actuated prosthetic arm.

A. EMG Signal Acquisition Module

Evaluating and recording electrical activity of skeletal muscles can be referred to as electromyography. EMG signal acquisition was achieved by connecting the three electrodes (Middle, End, and Reference) of the EMG sensor (company: MyoWare; Model: At-04-001) in the target muscle. The Middle electrode of the EMG sensor was connected at the middle of the target muscle body (bicep or forearm). The End electrode was placed adjacent to the Middle electrode towards the end of the muscle body. The Reference electrode should be placed on a separate section of the body such as the bony portion of the elbow or a nonadjacent muscle. Position and orientation of the electrodes have significant impact on the strength of the received signal. Placement location (forearm or bicep) of the EMG sensor depends on the size of the cleft arm. The primary output of the EMG sensor was not the raw EMG signal, rather an amplified, rectified, and integrated EMG signal, as illustrated in Fig. 3(b).

B. Electronic Control Module

The EMG sensor, we considered for our prosthetic arm, can directly send the amplified, rectified, and integrated EMG data to the microcontroller. After receiving the rectified & integrated EMG signal from the sensor, the microcontroller converted the EMG signal to digital signal by means of an analog-to-digital converter (ADC). The converted digital EMG signal was represented by a 10 bit data indicating an integer value varied from 0 to 1023.

The integer value signified the voltage level of the EMG signal ranging between 0 to +5 V. The EMG signal voltage was different for different muscles of the arm. Depending on the muscle strength data of the forearms or biceps, received from the EMG sensor, the microcontroller sent a pulse width modulated (PWM) pulse to the metal gear servo motor for the desired movement of the prosthetic arm, as shown in Fig. 3(c). A rechargeable high current rating Li-Po battery (current rating: 1 Ah) was integrated with the electronic control module to support as a power supply unit of the prosthetic arm and other electronic devices integrated with it. Because of the high current rating battery, the prosthetic arm can function uninterruptedly for long duration without recharging the battery.

C. Mechanical Module

During designing the physical model of the prosthetic arm, we considered the conventional theories of designing bionic arm. We fabricated all the fingers, palm, and forearm of the prosthetic arm separately using a 3D printer. Afterwards, the small components were integrated to achieve the complete prosthetic arm. Using artificial tendons, we connected a metal gear servo motor with the fingers of the prosthetic arm for precise movement control of the fingers (i.e. pull up/down). Based on the muscle strength data, the microcontroller sent a pulse to the servo motor, which in turn controlled the rotation angle of the servo motor and suitable movement of the prosthetic arm. The technical specification of various components required to fabricate the EMG sensor actuated prosthetic arm is summarized in Table 1.

IV. RESULTS AND DISCUSSION

A. EMG Signal Acquisition and Analysis

During designing the system architecture of the proposed prosthetic arm, we considered muscle strength data as the desire of the amputee. Before real-life performance analysis, we tested the movement of the servo motor by controlling the EMG signal. As mentioned before, the EMG sensor provided us a rectified, amplified, and integrated EMG signal, as represented in Fig. 3(b). The integrated EMG signal was converted to digital EMG signal by means of an ADC. The digital EMG signal was represented by an integer value ranging from 0 to 1023, indicating the signal voltage level varying from 0 to +5 V. This value varied with the amount of muscle strength sensed by the EMG sensor. Based on the muscle strength data, we categorized three different states of the prosthetic arm: rest state (RS), semi-open/close state (SOCS), and full-open/close state (FOCS). The person-specific threshold value, required to reach each state, was selected based on the EMG signal data collected from the EMG sensor attached to the forearms or biceps. The experiment was conducted at least 10 times before finalizing the person-specific threshold values for different operating states of the prosthetic arm. Depending on the EMG signal voltage level and predefined threshold values of different states, the microcontroller generated a PWM pulse and sent it to the servo motor for appropriate rotation to reach any particular state. The rotation angle is dependent on the width of the PWM pulse, which is a function of the EMG signal voltage level and threshold values of the three predefined operating states.

We investigated the EMG signals of various muscles (bicep and forearm) of the left and right arms coming out from the EMG sensor by connecting the microcontroller output using a computer. Figure 4 illustrates the conversion of integrated EMG signal into movement commands. Figure 4(a) shows the integration process of the EMG sensor with the target muscle of the arm (in this case, left forearm). The integrated EMG data was converted to digital EMG data and sent to the microcontroller, as shown in Fig. 4(b). The output of the microcontroller was connected with a computer for monitoring the EMG signal voltage and connected with the servo motor for examining the movement operation. For stable operation, the microcontroller periodically processed the EMG signal data at an interval of 5 seconds rather than continuous data processing. When the forearm muscle was in the rest state, the EMG signal voltage didn’t cross the threshold value required to move to the SOCS or FOCS state. As a result, no PWM pulse was generated by the microcontroller and the servo motor remained in the rest state (rotation angle: 0°), as depicted in Fig, 4(c). As soon as the stress on the muscle crossed the predefined threshold value of semi-open/close state, the microcontroller sent a short PWM pulse to the servo motor and it was rotated by an angle of 65°, as shown in Fig. 4(d).

Table 1. Technical specification of various components of the prosthetic arm

Component Name	Technical Specification
EMG Sensor	• Operating Voltage: +5 V • Data Type: Amplified, rectified, and Integrated EEG data
Li-Po Battery	• Capacity: 1 Ah • Nominal Voltage: 7.4 V
Li-Po Battery Charger	• Operating Voltage: 3.7 – 22.2 V • Current Range: 1 – 2 A (Capacity: 1 Ah)
Digital Metal Gear Servo	• Speed (4.8 V no load): 0.17 sec/60° • Speed (6.0 V no load): 0.13 sec/60° • Stall Torque (4.8 V): (11 kg/cm) (180 oz/in.) • Stall Torque (6.0 V): (13 kg/cm) (208 oz/in.)
Kevler Tendon Cord	• Strength: 100 lb • Diameter: 1 mm
Microcontroller	• Operating Voltage: 5 V • Frequency (Clock Speed): 16 MHz
3D Printer	• Positioning Accuracy: X/Y = 0.0125 mm; Z = 0.002 mm • Ambient Operating Temperature: 8 ºC – 40 ºC
PLA Filament	• Diameter: 1.75 mm ± 0.02 mm • Tensile strength: ³ 55 Mpa • Density: 1.25 g ± 0.05g/cm³

Figure 4: Conversion of integrated EMG signals into corresponding movement commands. (a) Integration of the EMG sensor with the forearm; (b) processing of the EMG signals into movement commands by the microcontroller; (c) position of the servo motor when the forearm was in the rest state; (d) 65° rotation of the servo motor when the forearm was in the semi-open/close state; (e) 95° rotation of the servo motor when the forearm was in the full-open/close state.

Figure 5: Oscilloscope images of the integrated EMG signals collected from the forearms and biceps of a human being. (a) Right forearm; (b) left forearm; (c) right bicep; (d) left bicep.

When the muscle strength was too strong, the EMG signal voltage level crossed the threshold value of FOCS state and the microcontroller sent a long PWM pulse to the servo motor. As a consequence, the servo motor was rotated by an angle of 95° and moved to the full-open/close state, as depicted in Fig. 4(e).

In order to investigate the integrated EMG signal, we sent the integrated EMG signal coming out from the SIG pin of the EMG sensor (Fig. 3(c)) with a digital oscilloscope. Figure 5 represents the oscilloscope images of the integrated EMG signal received from the forearms and biceps of both arms of a human being. From the oscilloscope images it is evident that, the voltage level of the EMG signals received from different muscles of human arms such as forearms and biceps are different. After several times signal acquisitions from the muscles of the cleft arm, it is easy to estimate appropriate person-specific threshold values to reach different operating states of the prosthetic arm.

B. Fabrication of the Prosthetic Arm and It’s Operating States

As mentioned before, we modified our design several times and finally achieved the proposed model of the prosthetic arm. Figure 6 represents the finally developed prosthetic arm and its different operating states. At first, we fabricated small pieces of the arm (e.g. fingers, palm, and forearm) using 3D printing technology. Afterwards, all small pieces were integrated to achieve the complete prosthetic arm. After successful transformation of the integrated EMG signals to corresponding mechanical movements, we installed the metal gear servo motor inside the prosthetic arm (Figs. 6(a) and 6(b)). In order to pull-up/down the fingers of the prosthetic arm, the fingers and the servo motor were connected to artificial tendons.

Figure 6(a) depicts the image of the prosthetic arm operating at the rest state where no stress has been detected on the bicep muscle by the EMG sensor. When the person wants to hold something, the muscle is stressed moderately and the EMG signal voltage level crosses the threshold value required for the SOCS state and the prosthetic arm moves to the semi-open/close state, as illustrated in Fig. 6(b). The side view of the SOCS state is shown in Fig. 6(c). To operate in the full-open/close state, strong muscle stress was required where the EMG signal voltage level crossed the required threshold value of the FOCS state. After laboratory test, we spray coated the prosthetic arm in white color.

C. Real-life Performance Analysis

We accomplished real-life performance analysis over 10 people from different age groups (15 years to 55 years) including 3 amputees by wearing the EMG sensor based prosthetic arm. Both males and females were considered during real-life performance analysis. EMG signals from forearms and biceps of both arms were collected during the experiment. Table 2 summarizes the real-life performance analysis data collected from the forearms and biceps of 10 people. For right-handed people, the voltage level of the EMG signals collected from forearms and biceps from the right hand was higher compared to the left hand data and vice versa. Figure 7 represents the line plots of the EMG sensor readings collected from various muscles of 10 people under rest state, semi-open/close state, and full-open/close state during real-life perfor-

enough to modify the threshold values and associated programming during installation of the prosthetic arm.

Figure 8 demonstrates the holding and weight lifting capability of the prosthetic arm observed during real-life performance analysis. The experiments were conducted with several people. All of them were successful in holding or carrying a large variety of products by pressing their EMG sensor attached muscle (forearm or bicep) and controlling the operation states. Experimental results confirmed the holding and carrying products of different sizes and shapes, which in turn exemplified high degree of freedom of the developed prosthetic arm. However, we should keep one thing in mind that, the requirement of size and shape of the prosthetic arm might be different for different people, depending on the size and shape of the cleft arm and the level of disability. Since we used solid works software for designing the prosthetic arm and utilized 3D printing technology, the modification of the design and printing of the newly designed prosthetic arm is easy.

D. Response Time of the Prosthetic Arm

Furthermore, we examined the response time of the prosthetic arm under SOCS and FOCS state different operating states. We utilized Arc Length Formula (2021) to calculate the response time of the prosthetic arm, which can be represented as follows:

(1)

where is the arc length, i.e. length of the tendon required to reach the SOCS state ( cm), is the rotation angle to reach the SOCS state from the rest state, and is the radius of the servo arm ( cm). Equation (1) confirms that, it is required to rotate the servo arm by an angle of 65° to reach the SOCS state, which agrees with the value measured during real-life performance analysis. The time required to reach the SOCS state can be calculated as follows:

(2)

where represents the operating time of the servo motor under load condition (0.15 sec/60°). Using Eq. (2), we determined the time required to reach the SOCS state was 0.16 sec. Similarly from Eq. (1), the rotation angle to reach the FOCS state for the tendon length of L = 2.5 cm was 95.5°, which is also similar to the value measured during real-life performance analysis (rotation angle for FOCS state: 95°). From Eq. (2) we found that, the time required to reach the FOCS state was ~ 0.24 sec. The

highest response time measured during carrying a shopping bag of 2 kilograms to reach the SOCS and FOCS states were 0.19 sec and 0.3 sec. During carrying small product such as smart phone, the response time agrees well with the mathematical modeling. The amount of payload might have significant impact on the response time. Still the response time of the prosthetic arm was excellent. The overall cost of fabricating the proposed prosthetic arm including the 3D printing cost was estimated as 450 US dollars, which is within the capacity of the people of developing countries.

V. CONCLUSION

In summary, we designed and developed an EMG sensor based prosthetic arm. The EMG sensor was connected with the bicep or forearm of the sample people of different age groups, genders (male & female), and amputees. By processing and analyzing the EMG signals, we can control the movements of the prosthetic arm. Initially, the analog EMG signals were converted to digital EMG signals where the signal was represented by 10 bit binary date. Afterwards, the binary data was converted to appropriate voltage varying between 0 and +5 volt. The robust prosthetic arm was made of light weight polymer resin that was developed by means of 3D printing technology. Based on the EMG data, we designed the control unit of the prosthetic arm. Furthermore, we investigated the performance of the EMG sensor actuated prosthetic arm. The proposed prosthetic arm showed excellent results during real-life performance analysis by which an amputee can perform almost like a normal people. The proposed prosthetic arm has high degree of freedom and can be considered for commercial applications.

ACKNOWLEDGEMENTS

We would like to acknowledge the financial support of Khulna University Research Cell, Khulna University, Bangladesh

REFERENCES

Arc Length Formula (2021) https://byjus.com/arc-length-formula/ [Accessed: June 07, 2021].

Bright, D., Nair, A., Salvekar, D. and Bhisikar, S. (2016) EEG-based brain controlled prosthetic arm. Proc. Conf. Adv. Sig. Proc. Pune, India, 479-483.

Beyrouthy, T., Kork, S.K.A., Korbane, J.A. and Abdulmonem, A. (2016) EEG mind controlled smart prosthetic arm. Proc. IEEE Int. Conf. Emerg. Technol. Inno. Bus. Pract. Transfor. Soc., Balaclava, Mauritius, 404-409.

Heger, H., Millstein, S. and Hunter, G.A. (1985) Electrically powered prosthesis for the adult with an upper limb amputation. J. Bone Joint Surg. 67, 278-281.

Islam, M.R., Haque , A.N.M., Amin, S.N. and Rabbani, K.S. (2011) Design and development of an EMG driven microcontroller based prosthetic leg. Bangladesh J. Med. Phys. 4, 107-114.

Li, G. (2011) Electromyography Pattern-Recognition-Based Control of Powered Multifunctional Upper-Limb Prostheses. Advances in Applied Electromyography. Intechopen, Rijeka.

Mavani, R., Rank, D.H. and Sheth, H.N. (2014) Design and working of myoelectric prosthetic arm. Int. J. Eng. Devlop. Res. 2, 3324-3333.

McDonald, C.L., McCoy, S.W., Weaver, M.R., Haags-ma, J. and Kartin, D. (2020) Global prevalence of traumatic non-fatal limb amputation. Prosthe. Orthot. Int. 1-12.

Neogi, B., Mukherjee, S., Ghosal, S., Das, A. and Tibarewala, D.N. (2011) Design and implementation of prosthetic arm using gear motor control technique with appropriate testing. Int. J. Comput. Appl. Eng. Technol. Sci. 3, 281-286.

Parajuli, N., Sreenivasan, N., Bifulco, P., Cesarelli, M., Savino, S., Niola, V., Esposito, D., Hamilton, T.J., Naik, G.R., Gunawardana, U. and Gargiulo, G.D. (2019) Real-time EMG based pattern recognition control for hand prostheses: a review on existing methods, challenges and future implementation. Sensors. 19, 4596.

Quraishi, M.A.A., Elamvazuthi, I., Daud, S.A., Parasuraman, S. and Borboni, A. (2018) EEG-based control for upper and lower limb exoskeletons and ptosthesis: a systematic review. Sensors. 18, 3342.

Sawake, N., Gupta, S., Ghatge, A., Khatri, A. and Bhurchandi, K.M. (2014) EMG-based prosthetic leg for above-knee amputee. Proc. Texas Instru. India Educa. Conf. Bengalore, India, 69-72.

Selvan, M.P., Raj, R., Sai, R.G., Jancy, S. and Mary, V.A. (2021) Prosthetic hand using Emg. J. Physics: Conf. Series. 1770, 012018.

Sudarsan, S. and Sekaran, E.C. (2012) Design and development of EMG controlled prosthetic limb. Procedia Eng. 38, 3547-3551.

Verma, S., Chatterjee, A., Singh, H.P., Kumar, A., Garg, K. and Chattopadhyay, K. (2013) A design approach for myoelectric arm with hand and wrist motions using single actuator. Int. J. Adv. Manuf. Technol. 67, 1899-1907.

World Health Organization (2005) Guidelines for training personnel in developing countries for prosthetics and orthotics services, Scotland. https://apps.who.int/iris/handle/10665/58401

Xu, K., Guo, W., Hua, L., Sheng, X. and Zhu, X. (2016) A prosthetic arm based on EMG pattern recognition. Proc. IEEE Int. Conf. Robot. Biomet. Qingdo, China, 1179-1184.

Received: January 7, 2021

Sent to Subject Editor: September 16, 2021

Accepted: December 22, 2021

Recommended by Subject Editor Jose E Guivant