Vinning Vinodh's

Ultimate Guide to Roboteers

Computer Mouse Controlled by Eye

Dated May 5,2004.

Project Members

CONTENTS

Appendix

1. Program Listing
2. Schematic
3. References

1. Introduction

Many physically disabled individuals are deterred from using computers due to their inability to utilize a hand-controlled mouse. However, if directional discrimination of an icon can be achieved, these individuals would be able to take on the functions of a mouse without the use of hands.
We propose to design and build an electro-oculogram (EOG) biopotential amplifier in order to obtain a physiological signal due to eye movements and to use this signal to show directional discrimination. Our design can also be used as a model for future advancements in human-computer interactions.
The EOG biopotential amplifier should be capable of detecting frequencies between dc-10 Hz, the range at which most ocular movements operate. The EOG signal is in the microvolt range (50-3500 ?V). Therefore, when the DC offset is removed, it will be challenging to obtain a strong, usable signal given the minute nature of the recorded signal. Our choice of an EOG over other possible methods was selected based on the ease of usage and the low cost of production. The software choice for data acquisition and display is C, selected for its graphical capabilities and flexibility in programming.

2. Proposal:

Design Considerations

As illustrated in the figure 1, our project has four major subsections, which are discussed below.

The first stage of our design is the electrodes. The electrodes were chosen with the concern of protecting the eyes from hazardous elements. ECG disposable electrodes were used because of their easy availability. Silver/Silver-Chloride electrodes were chosen because the half-cell potential was the closest to zero. Electrodes with the smallest amount of half-cell potential are desirable because they cause the least amount of offset. By definition, the hydrogen electrode has a zero half-cell potential, but due to the gaseous nature, they cannot be feasibly used. Although lead electrodes have a lower half-cell potential than the Ag/Ag-Cl electrodes, lead is hazardous to the health and thus is avoided. Thus our choice of electrodes takes into account a low cost and proper signal pick-up.
Stages 2 and 3 encompass the detection of horizontal and vertical movements of the eye, respectively. The second stage (for horizontal discrimination) detects lateral movements at the periphery of each eye. The hardware in this stage consists of the EOG biopotential amplifier. Similarly, the third stage (for vertical discrimination) consists of another EOG biopotential amplifier, but also includes two summer circuits. EOG biopotential amplifiers were chosen since the alternative Electro-retinalgram (ERG) requires either an electrode in the inner surface of the retina or on the cornea. Moreover, the ERG is used to measure the changes of potential due the stimulation of the retina by a bright flash of light. The EOG is frequently the method of choice for recording eye movement in research because of the proportionality of eye movement to eye position. Refer to figure 2 for electrode placement.
When the eyes look straight ahead, a steady dipole is created between the two electrodes. When the gaze is shifted to the left, the positive cornea becomes closer to the left electrode, which becomes more positive. There is an almost linear relationship between horizontal angle gaze and EOG output up to approximately ?300 of arc. Therefore by placing electrodes to the left and right and above and below the eye, horizontal and vertical movements can be obtained. However, the EOG suffers from a lack of accuracy at the extremes, due to noise compounded from the effects of an EEG, EMG, and the recording equipment equivalent to approximately 10 of eye movement. Thus movements of less than 10 or 20 are difficult to record. In addition, large eye movements of 300 of arc do not produce bioelectric amplitudes that are proportional to eye position. Thus, controlling the mouse will require a moderate movement of the eyes.
The output signal from the final amplifier stage was fed to an 8 bit ADC (AD0808 CCN).

3. Specifications

The following performance specifications were chosen based on the voltage and frequency ranges of the EOG signal.
Frequency Range DC – 10 Hz
Input Voltage Range 50 – 3500 ?Volts
Voltage Gain ~5,000

4. Proposed Block Diagram:

5. Typical EOG waveform

6. Performance Tests
In order for our circuit to perform within specifications, several tests will be made to assure basic functionality. The input to the EOG is approximately in the range of 50 - 3500?V, however, in reality, the signal actually obtained from the body varies slightly. Therefore for our EOG circuit, we are trying to achieve an output in the range from 1 to 10 V, after amplification. The voltage gain needed for this circuit should be in the range of 74 ? 5 dB. However, since the signal obtained from the body is slightly different from person to person, we are not exactly sure what kind of voltage levels we will be picking up from the body. Therefore, the actual voltage gain necessary may possibly be much higher than our initial estimate. We will be testing the output of the EOG to ensure that we have enough voltage gain so that the signal can be properly utilized.

7. Implementation Plan:

1. Hardware:
The hardware for the signal pick up of the EOG consists of an instrumentation amplifier stage followed by a 6 pole filter to restrict the signal to the required Bandwidth. In our Case we used a cutoff frequency of 15Hz for the low pass filter. A six pole active filter was constructed because the Mains supply interference (50 Hz) was very high. The 120dB per decade roll off provided by the six pole filter was sufficient to reject the 50 Hz interference. The final stage consisted of amplifier stage with a gain of 100.The pre amplifier was designed for a gain of 10.The overall gain of the 3 stages of 2nd order LPFs was approximately 5.Hence the total gain provided is 5000.From the final amplifier stage , a diode was used to provide the signal to the unipolar ADC.
DC Offset Compensation:
A significant hardware problem that we faced was the building up of sufficient DC voltage. The DC voltage was identified to be partly caused by the electrodes (improper contact) and the offset currents of the OPAMPs used. Hence we included an offset compensation circuitry for the 741 opamps used. This is a potentiometer arrangement with an end connected to negative supply.741 opamps were used throughout the amplifier chain. The use of 741 opamps caused the size of the circuit to increase but it minimized interference problems. However we intend to use only low offset opamps OP27 in the future because additional compensation circuitry had to be used for the 741 OPAMPS.

Safety Considerations:

? As in any bio potential experiment, the patient is to be isolated from the power supply.
? An important check to be made is to ensure that the supply lines and ground lines are adequately separated. Shocks and sparking were observed due to accidental shorting of negative and ground reference lines.
? Since we used a unipolar ADC, the absolute rating specified a minimum voltage of
-0.3v. Hence, we used a diode to prevent any negative voltage from affecting the ADC.

The final hardware part is the ADC. We used the ADC0808 CCN which was obtained as a free sample from Analog Devices. Since it was a unipolar ADC, we provided a DC offset to the ADC input i.e. the EOG signal with a peak value of 1V was superimposed on a DC voltage of 1.5V.The DC voltage was critical because it decides the positioning of the cursor at the centre of the screen.

8. Observations about the EOG signal

1. The EOG signal has a pulse shape with the pulse duration approximately 200ms on the average.
2. When eye ball is moved one side the voltage remains positive (or negative) and returns to zero when looking straight.
3. When measuring horizontal movement, the potential caused by vertical movement on the horizontal electrodes is less significant compared to horizontal potential.
4. Movement of the patient’s head or body alters the DC level of the signal.
5. The pulse produced by leftward movement is nearly the same as produced by rightward movement in both amplitude and pulse duration.
6. Even with the eye closed, the potential was observed to be the same. This indicates that there is no significant interference from ElectroRetinoGram(ERG).
7. For the two patients under test, the amplitude was approximately the same, but the pulse duration varied slightly.
8. Depending on the angle through which the eyeball was moved, the amplitude of the EOG signal changed.

9. Electrode Placement

10. Requisites

11. Cost and Parts

Electrodes Rs.100
741 op amps Rs.50
Connectors and wires Rs.30
ADC0808CCN --- (available)
Port Interface Rs.50
Resistors, capacitors Rs.25s
----------------------------------------------------------

Total Costs Rs.255

12. Applications:

1. Computer mouse control
2. Identification of squint eyes and degree of squint ness

Appendix
1. C Program Listing

#include<stdio.h>
#include<graphics.h>
#include<dos.h>
#include<math.h>

void *buff;
void main()
{
int gd=DETECT,gm;
int x,y,maxx,maxy,area;
int temp,sens=1,mid,up,low;
int flag=0,val,val1=0,val2=30777;
int i;

//configuring the parallel port as input
temp=inport(0x37a);
outport(0x37a,0xff2b);

//Graphics initialisation
start:
initgraph(&gd,&gm,"");
maxx=getmaxx();maxy=getmaxy();
rectangle(0,10,maxx,maxy);
x=maxx/2;y=maxy/2;
drawptr(x,y);
area=imagesize(x-15,y-15,x+15,y+15);
buff=malloc(sizeof(area));
getimage(x-15,y-15,x+15,y+15,buff);
x=maxx/2;
y=maxy/2;
flag=0;

//Getting the mid value as a mean of 100 samples
//The user is instructed to look at the centre of the screen
//for the first 5 seconds
mid=0;
for(i=0;i<100;i++)
mid=mid+inport(0x378)-30700;

mid=mid/99+30700;

//infinite loop for port monitoring
while(1)
{
val1=inport(0x378);
gotoxy(21,1);
printf("%d",val1);

//Neglecting LSB changes occurring due to noise or interference
if(abs(val1-mid)<35)
{
continue;
}

//Setting the Thresholds for the Right and Left movements
if((val1-mid)>55)
{
if(x>15)
{
putimg(x,y);
x=(x-sens)%maxx;
putimg(x,y);
delay(80);
}
}
else if((val1-mid)<55)
{
if(x<(maxx-15))
{
putimg(x,y);
x=(x+sens)%maxx;
putimg(x,y);
delay(80);
}
}
}
}

putimg(int x,int y)
{
putimage(x-15,y-15,buff,XOR_PUT);
}

//Subroutine for mouse pointer graphic
drawptr(int x,int y)
{
setcolor(RED);
line(x-13,y-13,x,y-13);
line(x-13,y-13,x-13,y);
line(x-13,y,x-7,y-5);
line(x,y-13,x-5,y-7);
line(x-7,y-5,x+11,y+13);
line(x-5,y-7,x+13,y+11);
line(x+11,y+13,x+13,y+11);
setfillstyle(2,RED);
floodfill(x,y,RED);
}

/*
The following changes can be adopted if the program doesnot provide the intended result

1. The mid value can be set as 30850.
2. The Thresholds can be increased or decreased
3. Delays can be modified
*/

2. EOG Schematic

Click here for a Detailed Schematic

3. References:

1. Eagle Eyes - http://www.bc.edu/schools/csom/eagleeyes/
2. Evaluation of the EOG for communication through eye movements E. Lileg1, G. Wiesspeiner2, H. Hutten2 Boltzmann Institut für technische Lebenshilfen,Institut für Biomedizinische Technik,Austria
3. http://www.ee.ualberta.ca/~ee401/archive_fall_2001.html