The Bionic Clicker Mark I & II

In this manuscript, we present two 'Bionic Clicker' systems, the first designed to demonstrate electromyography (EMG) based control systems for educational purposes and the second for research purposes. EMG based control systems pick up electrical signals generated by muscle activation and use these as inputs for controllers. EMG controllers are widely used in prosthetics to control limbs. The Mark I (MK I) clicker allows the wearer to change the slide of a presentation by raising their index finger. It is built around a microcontroller and a bio-signals shield. It generated a lot of interest from both the public and research community. The Mark II (MK II) device presented here was designed to be a cheaper, sleeker, and more customizable system that can be easily modified and directly transmit EMG data. It is built using a wireless capable microcontroller and a muscle sensor.


Introduction
The Aspire Centre for Rehabilitation Engineering and Assistive Technology investigates techniques that are applicable and transferable between different domains in related areas of interest, including but not limited to, stroke, muscular dystrophy, amputation, the ageing population and training of specialized skills. One area of research that the center is involved in is neuroprosthesis. Of the many techniques used for control of neuroprosthetic arms, EMG is one of the most common inputs for the control systems 1,2,3,4,5,6 . This is in a large part due to its ease of use and affordability when compared to other control systems 7 . Recently developed 3D-printed prostheses such as the Ada hand can cost only 1,000 USD when using this type of control 8,9,10 . However, when attempting to demonstrate such systems to the public there is no easy way of doing so without the aid of an amputee.
To raise the awareness of the research activities in this field to members of the public, a bionic clicker demo device was developed. It is very important to use object-based demonstration as it attracts attention and accelerates the learning and understanding of the subject being taught 11 .
Our device not only helps to teach the concept of EMG but also to increase the knowledge of the current development of modern technologies. Moreover, it inspires younger generations to choose studies within Science, Technology, Engineering and Math (STEM) areas.
The Bionic Clicker MK I was made using readily available parts that were already in use within the lab. It consisted of a microcontroller, a biosignals shield 12 , electrodes, a control board, a wireless communication board and a 9 V battery. The device functioned by picking up the activity of the abductor indicis muscle located between the index finger and thumb. It triggers a slide change by mimicking a keyboard and sending a 'right keypress' whenever a preselected threshold was reached. The control board allowed for the manual sending of 'right' and 'left' keypresses (progress slides and retreat slides) and could also be used to override the EMG input if things went wrong during a live demonstration.
As part of the Medical Physics and Biomedical Engineering department public engagement activities, we demonstrated the Bionic Clicker to members of the public. It received an enthusiastic response from the audience and generated interest in starting several collaborations. After the success of the initial device a second version of the device was designed.
The goal for the design of the second device was to produce a device that was cheaper, less cumbersome and more customizable than the first device. The purpose of this device was to design something that could easily be modified for student projects and be cheaply incorporated into existing projects. The main advantage of this device over others available 1,2,3,4,5,6 is its ease of use, small size, and low cost. Although the bionic clicker devices may not have the resolution of other research devices, such as trigger devices, they are more than good enough. The MK II would be an ideal basis for any system that uses an EMG threshold to trigger a device, such as a prosthetic controller or assistive device.

Assembling the Base Components
NOTE: Figure 3 gives a high-level overview of the protocol steps.
1. To build the Bionic Clicker MK I, plug the bio-signals sensor shield into the microcontroller and screw the EMG cables into the E, M and GND screw terminals of the shield (see Figure 4). Continue to step 1.6. 2. To build the MK II, place a row of three header pins into the muscle sensor in the +, -and SIG holes (see Figure 5)  Place the stripped end of the black wire into the R hole, the blue wire into the E hole, and the red wire into the M hole of the muscle sensor ( Figure 5). Solder the wires in place on the underside of the muscle sensor. Move to step 1.5.
4. Clip two electrode pads into the underside of the muscle sensor and one electrode into the connector of the black reference cable.
1. Stick the muscle sensor onto the selected muscle with the electrode pads and place the black reference electrode in an appropriate place.
5. Cut 8 single-core multi-thread wires to length and strip each end: 5 short (7 cm) wires to run from the microcontroller to the control board (red, black, green, white and blue) and 3 longer (approximately 12 cm but dependent on wrist size) wires (red, black and green) to run from one side of the wrist to the other. NOTE: If placing the muscle sensor on a different muscle make sure that the longer wires will run from the muscle sensor site to the wireless microcontroller site. 1. Place the wires into the microcontroller ready for soldering: 2 red wires into the 3V hole, 2 black wires into the GND hole, the long green wire into the A0 hole, the short blue wire into the 2 hole, the long white wire into the 3 hole, and the short green wire into the 5 hole. Solder the wires in place on the underside of the microcontroller. 2. Solder the other end of the 3 long wires to 3 header sockets in the order: red, black, green. See Figure 5. If not using the abductor indicis muscle proceed on to step 2.
6. Place the EMG sensor pads on the hand, as shown in Figure 6, with two of the electrodes at either end of the abductor indicis muscle and one EMG sensor pad on the middle of the back of the hand. 1. Clip the electrode pads into the connector end of the muscle sensor cables (push fit). The blue and red electrodes clip above the muscle, the black electrode clips on the back of the hand. the EMG will now be displayed. 6. Move the index finger from side to side and move the hand around without moving the index finger. Write down the values displayed in each case. NOTE: When using the MK II make sure that the cables do not move as it is extremely sensitive to noise generated this way. 7. Select a value that is above what is seen when the hand is moved around, but below what is seen when the finger is moved from side to side.

Test EMG Output
Write down this value. NOTE: The value is selected so that the device will only be activated by a purposeful movement of the finger. This is the Threshold Trigger Value, the value at which the device will be activated. The muscle sensor has a gain setting that can be manually altered if the threshold value is hard to find. The electrodes might need to be replaced. If using the abductor indicis muscle, set the gain to minimum as the starting

3D Print the MK II Case
1. If building the MK II, download the stl files for all 5 components of the case (see Figure 7 for all 5 parts). Print the parts of the case by any preferred method. Proceed to step 5.2. If building the MK I, move on to section 5. NOTE: The case has been successfully printed by both fused deposition modelling 16 (FDM) and photolithography printers 17 .

Solder the Control Board
NOTE: If building the MK II, proceed to step 5.2.
1. Place a row of two header pins, five 10 KΩ resistors, a sliding switch and two push button switches for the components as shown in Figure 8A; then solder them in place on the underside of the board.
1. Break the copper tracks on the strip board by slicing through with a craft knife, following the grey lines on Figure 8A. This allows for individual tracks to have multiple functions across the board. 2. Cut 7 wires (black, red, blue, orange, white, brown and yellow) of the correct length with a wire cutter so that they will run from the forearm to the upper arm (around 30 cm). Cut a red wire of 7 cm, a black wire of 3 cm and an orange and a blue wire of 4 cm. 3. Strip both ends of the wires with a wire stripper. 4. Place the wires in the control board, following the circuit diagram shown in Figure 9; solder the wires in place on the underside. 5. Solder the long red and black wires to a pair of header pins, and then solder the other long wires to a strip of header pins in the order: blue, orange, white, brown, yellow. 6. Solder the 5V and GND pins of the wireless module to the header pins on the control board. 7. Solder the short orange wire to pin 2 of the wireless communication module and the short blue wire to pin 3.
2. Place three 10 KΩ resistors, a sliding switch and two push button switches as shown in Figure 10A and solder them into place on the underside of the board. 1. Break the copper tracks on the strip board by slicing through with a craft knife, following the grey lines on Figure 10A. This allows for the track to have multiple functions on the board. 2. Cut the wires that were previously soldered to the microcontroller with a wire cutter so that they can run through the mid layer of the microcontroller case to the control board without stopping the case from closing ( Figure 10B). 3. Place the wires in the control board, following the circuit diagram (Figure 11). Solder the wires in place. Proceed to step 6.2.

Assemble Clicker and Update Microcontroller
1. Re-assemble the Bionic Clicker, connecting the header connectors from the control board wires to the microcontroller and bio-signals shield (5V and GND on the MK I, pin 22-30 on the MKII). Connect the battery to the microcontroller. See Figure 12. Move on to step 6.3. 2. Re-assemble the Bionic Clicker, connecting the header connector from the microcontroller to the muscle sensor (green wire to SIG). See

Connect the Device to a Computer
1. If using the MK I, follow the instructions to pair the wireless module to the device by following the manufacturer's guide 18 . If using the MK II, connect to the device wirelessly following the procedure to connect a wireless keyboard to the computer being used.

Test the Clicker
1. Open some typing software and enter some text, such as 'Lorem ipsum dolor sit amet'. This allows the presses to be perceived to test whether these commands are sent and received. NOTE: If the battery is low the device may give erratic behavior; always use a fresh battery. 1. Press the manual forward button to see the cursor move forward and the manual backwards button to see the cursor move backwards. Raise the index finger to also move forward.
2. To test the clicker with the presentation software, raise the index finger to progress the slides. NOTE: The override switch turns the EMG function on and off, and the manual forward and backwards buttons progress and retreat the slides in both scenarios.

Representative Results
The MK II is more affordable, customizable and less cumbersome than the MK I device. The entire MK II costs only slightly more than the biosignals shield alone (75 USD). The device is significantly smaller sitting on the wrist rather than the arm and the wireless microcontroller could potentially simultaneously support inputs from 6 muscle sensors. The functional battery life of the MK I device is just under an hour using a 9 V 550 mAh battery and the functional battery life of the MK II device (when used as a clicker) is around 8 hours using a 3.7 V 150 mAh battery; see Table 1 for a comparison of the devices.
The Bionic Clicker MK II can have an issue when used on the abductor indicis: the amplifier can saturate and take more than a second to discharge (see Figure 15). Careful placement of the electrodes and correctly setting the gain can overcome this issue. This does not happen with the Bionic Clicker MK I or on any other commonly used muscles for EMG.
Whilst calibrating the devices to find the Threshold Trigger Value, many different values can be observed. They fall into three ranges: the values when the hand is stationary, the values when the hand is moving, and the values when the finger is moved. Table 2 shows recorded values in each range; for the stationary and hand moving ranges, the maximum values are shown and for the finger tensing range the minimum value is shown. The threshold value is selected to lie above the hand moving value and below the finger tensing value. A value nearer to the hand moving range increases the chance of false positive and reduces the chance of false negatives, whilst a value closer to the finger tensing range has the opposite effect.
Both devices where tested for false negatives and false positives when tensing the abductor indices muscle. A false negative was recorded when the device did not trigger a change of slide upon tensing of the muscle and a false positive was recorded if the slide changed when no tensing occurred. Neither device had an issue with false positives, though the MK II device experienced the occasional false negative (less than 5% of the time). The MK I device experiences no false positives or negatives during the first 45 minutes of operation, though the number of false negatives increases rapidly until total device failure between 50 minutes and an hour (see Table 3).
These results show that the device succeeded in its stated aims. Table 1 shows that the MK II is cheaper and has more flexibility than the MK I. Table 2 and Table 3 show that the device functions as intended and can be used as an EMG-based trigger device. Figure 15 shows the issues that can occur if using the abductor indices muscle: this is not a problem that occurs with most muscles and can be fixed by altering the gain. Although the devices have some issues, they are sufficient for the intended use.

Discussion
The saturation of the MK II when used on the abductor indicis is less of an issue than it may first appear. Careful placement of the electrodes and correct gain setting stops this from being an issue when the device is used as a clicker. Unless interested in accurately recording the activity of the abductor indices, this is unlikely to be a problem at all. No over-saturation has been seen on any other muscle after the gain has been set. The false negatives with the MK II are due to the difficulty of selecting the proper threshold value when using the abductor indicis. With larger muscles the difference between the magnitude of non-purposeful activation of the muscle and purposeful tensing of the muscle is greater, allowing for the selection of a threshold point that is further from both the false-positive and false-negative points. On particularly small hands the abductor indicis muscle may be too small for the electrodes to be correctly placed (though with smaller electrode pads this could potentially be solved).
The considerably longer battery life for the MK II is useful for a variety of reasons. Firstly, the MK I device started to act erratically after 45 minutes of use, so it cannot be used for longer demonstrations. Secondly with a multi-hour battery life, the MK II can be considered as a control input for a useful device, and with only a small increase in physical battery size, it could be used as an all-day monitoring device. The wireless microcontroller has 6 analogue inputs and 13 digital inputs; this means that the device could accept signals from multiple muscle sensors to create a device with more degrees of freedom in the control inputs. It should also be noted that the muscle sensor could be replaced by any biosensor with an analogue output to create a device that uses other biological signals as the input. The code of the device can also be easily modified to change its functionality. Changes to the software and hardware of the device allow for simple and varied modifications to the device.
One limitation of the device as it currently stands is that the EMG output cannot be sent wirelessly at a high data rate as this can overload the wireless microcontroller buffer. Another limitation is that the technique uses the abductor indices as the input, and as the muscle is very small, the spacing of the electrodes on the hand almost overlap; if someone has particularly small hands, it may be impossible to place the electrodes correctly over this muscle.
The device has several advantages over the more expensive devices when it comes to flexibility in potential research projects. It is low cost: the device costs 80 USD and additional EMG channels only cost 35 USD, making it ideal for smaller or student projects. It is easy to customize, the software can be easily edited, and the inputs changed for other hardware. It has a small size, so a person wearing it does not need to carry heavy or bulky equipment. It also appears as a wireless keyboard to other devices, so it can be easily integrated with any compatible wireless device. The device has already been incorporated into an assistive device that will be published in the near-future.
Due to the size and ease of customization of the MK II, it is already being considered for incorporation into several research projects as a wireless EMG module and as a wireless trigger mechanism. It is also being used as the foundation of one of the lab sessions on a master student's course. The main improvement we would like to make to the device is to increase the wireless transmission rate; the goal is to achieve 10 Hz, and whether this will be done through hardware or software is yet to be determined.
The most critical steps within the protocol are steps 2.6 and 2.7: the selection of the Threshold Trigger Value. In step 2.6, special attention needs to be paid to the movement of the EMG cables, as these can act as antenna and generate motion artifacts; however, if these are kept Copyright © 2017 Creative Commons Attribution 3.0 License August 2017 | 126 | e55705 | Page 12 of 12 stationary this is not an issue. In step 2.7, if the selected value is too high, this results in false negatives. If this value is too low, this results in false positives. In the case of the abductor indicis, it is very difficult to find a value that does not result in the occasional false negative, although with larger muscles this does not appear to be an issue. If finding the correct value is still an issue, the gain can be corrected by setting it to the minimum value and increasing it until a large difference between non-activation and activation is seen through the serial readout, with the values staying below the point of saturation.
Overall the MK II is a considerable improvement over the MK I as a potential research device, although because the MK I has a stronger visual impact, it is likely to still be used in future public engagement events.

Disclosures
The authors have nothing to disclose.