ChArm- An Affordable Prosthetic Arm (VIT University, Vellore)

Chandan Dhal, Akshat Wahi,  Jaya Jain, Karthik Ramaswamy, Pawan Reddy, Pragya, Sutharshan K R, Shruti Arya, Spondon Kundu, Uma Balakrishnan, Viswanath Vaidyanathan

Team ChArm

Team ChArm

ABSTRACT

As of 2008, there were 1.7 million Americans with missing limbs[1]. (Reliable global statistics are unavailable.) While prosthetic solutions are available in the market for people with “below-elbow” amputations and for those with shoulder disarticulation, functional prosthetic units are extremely expensive: a single artificial “wrist” joint with the rotational movement costs roughly $3,000[2], while a full myoelectrically-controlled prosthetic arm can cost upwards of $30,000[3], which makes these arms affordable only to a limited number of people. Barriers imposed by high prices are often exacerbated by limited employability in the period following trauma.

The aim of our project is to design and fabricate an affordable, functional prosthetic arm, with elbow and fingers all under myoelectric control. Potential beneficiaries include – irrespective of socioeconomic status – adults and children, military veterans and civilian casualties as well as people who have lost their arms in accidents and in medical procedures.

INTRODUCTION/BACKGROUND

We aim to create an affordable prosthetic arm that is lightweight, easy to use, and performs as similar as possible to a natural arm. Our first task was to detect myoelectric signals from muscle contractions, process them and convert them to a form suitable as input to a microcontroller. We made analog circuitry to process the electromyographic (EMG) signal by bandpass filtering, rectification and amplification. We also made a functional elbow joint and a palm with fingers capable of independent movement, minimizing weight and construction cost while maintaining structural strength.

Our teammate, Chandan Dhal, is an electronics student with a shoulder disarticulation. With his cooperation, we tested the control, functionality and utility of our prosthetic arm. Constraints like weight, dimensions, sensitivity of the arm are verified at each stage of development.

PROBLEM STATEMENT

We aim to develop a prosthetic arm with target production cost under $500, which can be controlled by signals from a set of muscles in the body (i.e., EMG signals). This arm will have a functional elbow and palm/fingers with independent functionality. The arm will be lightweight and strong enough for daily activities.

In the video below, Chandan talks about the project and his expectations. We also give an overview of ChArm.

DESIGN/DEVELOPMENT

Prosthetic hands, although resembling real hands in some aesthetic qualities and basic functions like lifting small objects, generally lack in giving the individual with the amputated arm precise simultaneous control of individual joints like the elbow and the palm. We wanted to make something that resembled a real hand.

The team was divided into two sub-teams: the myoelectric team and the mechanical team.

The myoelectric team focuses on how to make control of the hand easy and user-friendly and the mechanical team focuses on how to make the arm robust, lightweight and more affordable.

Electronics

Electromyographic (EMG) signals are measured as the electric potential of the skin when the muscles contract.

The prosthesis is divided into 3 parts:

1) Pre-processing of EMG signals

2) Processing in microcontroller

3) Mechanical prosthetic mechanism.

The concept is that when muscle groups are activated (contract/relax), there is a voltage difference in the order of a few millivolts. This difference can be easily detected using surface electrodes. We used Ag/AgCl surface electrodes with hydrogel.

EMG signals generally lie in the amplitude range of 20-2000µV[4] depending on the muscle unit selected and the electrodes – but they are extremely noisy. To reduce this noise, we have used three surface electrodes, namely E1, E2 and E3. E1 and E2 are used to record muscle activity from two separate groups of muscles, such as biceps and triceps. E3 is a reference electrode placed on a muscle group isolated from these muscle sets. Electrodes are generally passive or active, needle based or surface based. Here, we use Ag/AgCl surface electrodes with adhesive conductive hydrogel for easy removability. This makes them conform easily to the skin and ensure good trace quality, leading to good EMG signal acquisition.

The potential levels of E1 and E2 are fed to an Instrumentation Amplifier (INA), which amplifies the difference of both the input signals with respect to E3, eliminating noise due to surface impedance. Thus, the three-electrode channel is more efficient than the two-electrode channel with respect to overall noise removal and cancellation.

Most of the EMG activation signals lie in the range of 10-500 Hz[5]. The amplified signal is then  fed through an 80 Hz high pass filter, so as to remove noise caused by the 50 Hz power supply. The signal is passed through a low-pass filter of 500 Hz after high-pass filtering to restrict it within a band of 80-500 Hz. We used Audacity [6] for characterizing the EMG signals (See figures 1 and 2).

Figure 1. Various stages from amplification, high-pass and then low-pass filters.

Figure 1. Various stages from amplification, high-pass and then low-pass filters.

In order to enable the microprocessor to read these signals, we rectify them to remove negative half cycles from the signals. The rectified signals are then smoothened so that only the high peaks are selected for thresholding the voltage (setting a minimum value of voltage for the microcontroller to read it as HIGH) and fed into the Arduino analog pins in the microcontroller. The signal is smoothened by a 25 Hz low-pass filter.

Figure 2. The pre-processing executed in Audacity. Here, the filtered wave is rectified to remove the negative half cycles of the wave.

Figure 2. The pre-processing executed in Audacity. Here, the filtered wave is rectified to remove the negative half cycles of the wave.

For practicality, the entire process is carried out in an analog circuit. Raw EMG signals are processed by an Instrumentation Amplifier (INA128P) with a high common mode rejection ratio (CMRR) of 20 dB and high gain of 106-107 adjusted by a 1k potentiometer, and can hence be adjusted depending on varying levels of responsiveness of the amputee. Differential inputs will lead to self-removal of the DC component from the electrodes. Furthermore, the amplified signals are passed through a bandpass filter (gain=150) which filters the signals  over a range of frequencies (80 to 480 Hz) and reduces any undesirable signals like electrocardiogram (ECG) signals[4]. The cumulative gain of approximately 100 x150 =15000 is enough to amplify the raw EMG signals before processing in the microcontroller. After filtering, the signal passes through a simple diode (IN4001) in reverse bias, culminating in another 2.3 Hz low-pass filter which acts as an envelope detector and removes any adjacent noise spikes, and stabilizes the multiple peaks to a constant value.

Figure 3. The smoothing low-pass filter has a cut-off of 2.3 hz so that we get sharp spikes at the output.

Figure 3. The smoothing low-pass filter has a cut-off of 2.3 hz so that we get sharp spikes at the output.

Analog EMG signals are then fed to the designated analog input pins of the Arduino after which they undergo A/D conversion and are converted to time restricted pulses which are then thresholded and fed to the servo motors for actuating the different joints.

Figure 4. The oscilloscope shows the signals after processing. The peak voltage, as shown, is 2.18V.

Figure 4. The oscilloscope shows the signals after processing. The peak voltage, as shown, is 2.18V.

Mechanics

For our prototype, string actuated fingers made up of three sections (See figure 4) were used. These sections were composed of Styrofoam affixed to cardboard sheets. Strings guided by thin plastic cylinder sections (attached to the underside of the cardboard sheets) actuate finger movement by pulling on section I. This, in turn causes sections II and III to curl towards the root of the finger. Finger retraction is assisted by an elastic string affixed to the back of each section. When the finger curls inwards, the elastic string tends to pull it back to place.

Figure 5. The design of the fingers

Figure 5. The design of the fingers

We used servos to activate finger movement by pulling the connecting string. Elastic strings attached to the back of the fingers allow them to return to their original positions with ease. The thumb and the index finger were made separately while the middle, ring and little finger form a single unit. These three units are connected to three separate motors. The entire structure is mounted on a wooden base and support assembly, which makes it highly compact.

Two PVC pipes hinged at a common end form the basic structure of the arm, while providing a rigid support for the entire hand. The elbow movement is controlled by a servo, which transmits torque with the help of a cable. This cable winds around a disc/spool as the servo moves, thus turning the “forearm” in a mechanism similar to that of the finger.

Depending on the combination of motors activated, different fingers move to enable the user to perform a number of functions. An electrode sensor setup, as described earlier, senses muscular impulses at the bicep, trapezius(shoulder blade) and pectoralis major (chest). Each muscle movement controls a servo, thus moving the corresponding  finger. Different combinations of muscle movements lead to seven different functional finger movements, as shown below:

Figure 6. Let the three finger units be f1, f2 and f3, the following combinations can be achieved: Default (all open), f1, f2, f3, f1f2, f2f3, f3f1, f1f2f3

Figure 6. Let the three finger units be f1, f2 and f3, the following combinations can be achieved: Default (all open), f1, f2, f3, f1f2, f2f3, f3f1, f1f2f3

COST

The prototype was fabricated in less than three months. A rough estimate of the total cost of the prototype has been made in the tale below.

Sr. No.

Component

Number of components

Cost per component

(INR)

Total  cost

(INR)

1

DC Servo Motors

4

750

3000

2

Microcontroller (Arduino)

1

1500

1500

3

Integrated circuitry: Op amp LF-351

9

15

135

4

Integrated circuitry: INA 128p

3

450

1350

5

Surface electrodes

7

20

140

6

Mechanical components: PVC supports

2

7

Mechanical components: Finger materials

3

TOTAL

6125(less than 125USD)

EVALUATION

The design was evaluated for strength, functionality, durability, ease of use and safety. The prosthetic hand was found to be capable of holding and maintaining a firm grip on various objects such as cups, phones and keys. The elbow is also capable of rotating through an angle of 120 degrees. The arm was found to be lightweight, easily controllable and extremely useful for basic needs.

Our teammate Chandan tried using the arm and expressed comfort in controlling the device. He also confirms that our prototype weighed considerably lesser than those available in the market. Sensitivity of the EMG signal sensors (electrodes), band pass frequency and threshold amplitude were carefully chosen to ensure ease of utilization and prevention of accidental movements.

Figure 7. Our prototype weighs less than 800 grams

Figure 7. Our prototype weighs less than 800 grams

DISCUSSION/CONCLUSION

As a conclusion, we can say that our prosthetic arm has the following features:

  • Affordable (approximately 125 USD)

  • Light (weighing approximately 800 grams)

  • Independently moving fingers and thumb with firm grip

  • A movable elbow with 120 degree clearance

  • Complete EMG control with 3 surface electrodes

All this enabled us to achieve the overall objective of our project, which was to make a prosthetic arm affordable to a larger number of people, irrespective of their socio-economic status. Our product will give individuals with an above-elbow amputation an easily controllable, functional and affordable prosthetic arm.

When we began, we were faced with a variety of design and technical challenges. Detection and characterization of EMG signals from various muscle groups was the first challenge we faced. Processing the signal to eliminate noise while retaining maximum information was another formidable task.  Design and development of the fingers and arm to optimize functionality and strength proved difficult. Extensive materials research (still ongoing) was and is being carried on to improve the durability without affecting the affordability and weight.

With research, brainstorming and many trials, we managed to optimize these parameters. However, we still have scope for further research and development.  A controllable wrist, a lightweight shoulder casting, better sensors, aesthetic development and a rechargeable battery back up are just a few of the many areas we intend to work on. We want to give affordable prosthetics a kick start, with the hope that it opens many more avenues for research in the near future.

ACKNOWLEDGEMENTS

Firstly, we would like to thank our guide and mentor, Dr. Theodore Moallem, who has been a source of constant support, knowledge and inspiration to our team. We would also like to express our gratitude to Dr. Partha Sarathi Mallick, Dean, School of Electrical Sciences, for providing us with lab facilities to carry out our work. Lastly, a special mention to the management at VIT University, for providing us with a platform to work, and co-operating with us at every step.

Team ChArm with the Prototype

Team ChArm with the Prototype

REFERENCE

1 http://www.amputee-coalition.org/fact_sheets/amp_stats_cause.html

2  Personal communication to Chandan Dhal from experts at Vimaans Hospital, New Delhi

3 http://www.disabled-world.com/assistivedevices/prostheses/prosthetics-costs.php

4 Pan, T. -T., Fan, Ping-Lin, Chiang, H.K., Chang, Rong-Seng, Joe-Air Jiang. (2004) Mechatronic experiments course design: a myoelectric controlled partial-hand prosthesis project. Education, IEEE Transactions, vol.47, no.3, 348-355, doi: 10.1109/TE.2004.825528

5 Shinde, Chandrashekhar P. (2012) Design of Myoelectric Prosthetic arm. International Journal of Advanced Science, Engineering and Technology, Vol 1, Issue 1, 21-22 ISSN 2319-5924

 6 http://audacity.sourceforge.net/about/

 

, , , ,

Powered by WordPress. Designed by WooThemes