Documente Academic
Documente Profesional
Documente Cultură
Abstract—This paper describes the motivation for and the propriate to a particular user’s needs or by dwelling on a key
design considerations of a low-cost head-operated joystick. The for a period of time. There are a number of commercially avail-
paper briefly summarizes the requirements of head-operated able devices that act as ‘Head-Operated Mice’. Some are specif-
mouse pointer control for people with disabilities before discussing
a set of technological approaches that can be used to satisfy these ically designed for people with disabilities, others have recently
requirements. The paper focuses on the design of a head-operated been developed for the games market; examples of both types
joystick that uses infrared light emitting diodes (LED’s) and are described below. This paper describes the design of another
photodetectors to determine head position, which is subsequently Head-Operated Mouse. An important question that we must ad-
converted into signals that emulate a Microsoft mouse. There are dress is, if there are commercially available devices, why is the
two significant findings. The first is that, while nonideal device
characteristics might appear to make the joystick difficult to use, development of another device desirable?
users naturally compensate for nonlinearities, in a transparent The first answer to the question is the cost of those systems
manner, because of visual feedback of mouse pointer position. specifically aimed for people with disabilities. In the U.K.,
The second finding, from relatively informal, independent trials, these devices retail at around £1000 (equivalent to around
indicates that disabled users prefer a head-operated device that
$1500 US). Therefore, a system that costs a fraction of this
has the characteristics of a joystick (a relative pointing device) to
those of a mouse (an absolute pointing device). price is highly desirable. Those devices that are aimed by the
games market might address this. This is a high volume market
Index Terms—Head-operated mouse, human–computer interac-
tion, people with motor disabilities.
and prices reflect that. The devices in this area retail for around
£100 (around $150 US) in the U.K., a tenth of the cost of
specialist devices. However, it is not clear as yet whether these
I. INTRODUCTION devices will have acceptable operational and aesthetic charac-
A. Motivation for Work teristics for people with disabilities. Current devices aimed at
games users require the user to wear a considerable amount
• The “mouse” should work as a standard mouse and require video picture and to track the movement of the ocular sphere
no custom driver software to run on the computer. and hence the movement of the head relative to the camera.
• The “mouse” should work with both standard desktop Note that the system tracks the whole of the eye and not the
computers and laptops. pupil. It is the movement of the head that is being tracked, not
• The “mouse” should be reliable. More specifically it the pupil’s movement relative to the head (see Section II-A4
should be unaffected by, or at the very least recover from, below). Unfortunately, the processing requirements for this
alterations in the physical environment such as: changes type of system are large and building practical systems that
in the level or spectrum of ambient light; changes to run in real time is expensive [2]. However, it is very attractive
the local electromagnetic field; changes in the distance in that it does not require the user to wear any equipment and
between the user and the computer; and alterations in the that the mouse button depression may be integrated into the
orientation of a user and his/her computer approach by detecting eye blinks. Processing requirements may
• There should be no physical connection between the user be reduced if optic flow techniques are adopted although that
and his/her computer, i.e., it should be wireless. would probably rule out the chance of eye blinks being used
• Ideally the system should not require the user to have to as mouse button selections. A further significant disadvantage
“wear” any equipment. However, a number of practical of analysing in this way is that lighting conditions need to
approaches, including our own, do have this requirement. be controlled within fairly tight constraints for the system to
If the user has to “wear” some equipment this should be work reliably. One solution to this problem, at the expense of
aesthetically pleasing, small, light, and comfortable. additional system cost, is to “illuminate” the user with nonvis-
• It is desirable that the clicking of mouse buttons be ible (i.e., infrared) light. With such a light source illumination
integrated with the operational approach. However, in can be controlled, although processing overheads are still just
many cases this is not practical and some other form of as high. A practical implementation of this sort of system is
mouse button activation either by appropriate switches or described in [3].
by “dwelling” can be provided. If switches are used, the Another approach requires the user to wear a small, reflec-
device should permit these switches to be connected to it tive target on either his/her forehead or on a pair of glasses. A
and the switch depression mapped to mouse-button clicks device connected to the computer and situated close to the com-
appropriately. puter’s monitor, illuminates the target with nonvisible (infrared
• In addition, a “head-operated mouse” will be subject to or near infrared) light. The device monitors the position of the
cost requirements, such as low initial purchase cost and a reflected signal and its position is used to control the mouse
low cost of ownership. pointer over a standard interface. This approach significantly
reduces the high processing overheads of the video-based ap-
II. APPROACHES TO HEAD-OPERATED MOUSE POINTER proach at the small expense of the user having to wear the re-
CONTROL flective target. However, the processing requirements are still
not trivial. There are a number of commercial systems that use
There have been several approaches to the control of mouse
this approach, for example, Origin Instruments’ HeadMouse [4]
pointer position through head movement based on different op-
and Prentke Romish’s Head Master [5].
erating principles. In this section, we briefly discuss some of
2) Wireless with Significant Amounts of User-Mounted
these approaches.
Equipment: In this approach the equipment mounted on a
To begin with we introduce a simple categorization scheme
user’s head transmits signals that can be detected by a receiver
for the general approach, and for the operating mode of the de-
unit mounted adjacent to the computer’s monitor. As far as
vice.
the computer is concerned the receiver provides signals that
emulate a Microsoft mouse. The head-mounted equipment may
A. Categorization of Approaches
be connected to other electronics and a power source local to
One of the requirements outlined above stated that there the user, mounted, say, on the user’s belt or in his/her pocket.
should be no physical connection between the user and the This style of approach has been used for some time, although
computer he/she was controlling and, although the necessity for not necessarily for human-computer interaction. Head mounted
the user to wear some equipment is deemed to be acceptable, light sources have been used as communication aids for certain
it is not necessarily desirable. We classify the approaches as user groups. See, for example, [6] and [7]. It is also used by at
being the following: least one commercial device, e.g., Madenta’s Tracker [8].
• wireless with little or no user-mounted equipment; The approach that we describe in this paper is based on a
• wireless with significant amounts of user-mounted equip- similar idea, except that we convert the transmitted signal into
ment; mouse control signals.
• connected with significant amounts of user-mounted 3) Connected with Significant Amounts of User-Mounted
equipment. Equipment: In this approach, the user wears some head
1) Wireless with Little or No User-Mounted Equip- mounted equipment that is connected to a local device, which
ment: One approach in this category is to analyze video is, in turn, connected by a one or more cables, to a further
images, taken from a camera typically close to the computer unit placed near the computer. In all cases, the signals that
monitor, to determine the user’s head position. One method are transmitted from the user’s local device to the computer’s
of determining head position is to find the user’s eyes in the device could be transmitted by wireless means, either by using
EVANS et al.: CONTROLLING MOUSE POINTER POSITION USING AN INFRARED HEAD-OPERATED JOYSTICK 109
licence free telemetry radio modules or by using infrared trans- intensive and also have a significant problem to overcome,
mission. Wireless transmission would not in any way alter the that is that the user’s head must remain fixed in relation to the
operation of the devices. However, it would add to system cost camera position. One way to ensure that the head remains fixed
and introduce some additional power supply considerations to is get the user to hold his/her head in a fixed position, this is
the design of the user’s local device. generally accomplished by using a bite bar. This is not very
Apparently, the simplest approach is to detect the user’s head comfortable during sustained use.
position relative to the Earth’s gravitational field. The very sim- Another approach is to fix the camera relative to the user’s
plest way of doing this is to use mercury switches to detect when eye by mounting it on the user’s head [9]. In small quantities
a user’s head is moved from a standard level position. Unfortu- such devices are relatively expensive. However, it is thought
nately the characteristics of these devices make the implemen- that these devices may gain wide market acceptance through
tation of a reliable practical system very difficult. Despite this the games industry. If they do become mass consumer products,
one commercially available system exploits this principle and the cost will fall substantially, the target price is £100 sterling,
is both reliable and usable. A related approach is to mount a making them a very attractive option.
pair of orthogonal potentiometers on the user’s head and using Other approaches do not use video images but detect eye
free-swinging weights to determine the head’s position. As re- movement by electroocular potential. An example of such a
ported in [1] these approaches have a number of drawbacks. system is the Eagle Eyes system [10]. While it is not really a
Perhaps the most significant is that in order to move the mouse mouse, it is sufficiently accurate for an experienced user to be
pointer left or right, users have to bend their heads toward their able to write his/her name in cursive script. A similar approach,
shoulders, rather than rotating their heads to the left or right. geared toward mouse pointer control is described in [11].
This style of movement can be uncomfortable.
Gyroscopes can be used to determine movement and these B. Operating Modes
could be the basis of a practical system based on head move-
ment. Indeed, there is a commercially available device based on A mouse pointer can be controlled in two ways, namely as an
this technology. Unfortunately, the component costs for small absolute pointing device or as a relative pointing device. For an
production runs using this approach are high; the cost of a pair absolute pointing device, the position of the device corresponds
of gyroscopes being more than double the cost of all the parts to the position of the mouse pointer. For example, if the mouse
used in the system we describe later. cursor is in the centre of the screen and it is desired that the
Another approach, also described in [1], is to detect the user’s mouse cursor be moved to the top left of the screen, the device
head position relative to the Earth’s magnetic field. This ap- is moved up and to the left until the desired position is reached
proach works reliably. However, it does have some defects. Like and held in that position until further movement is required. In
the device described earlier it requires users to bend their necks short, the mouse cursor moves, when the device moves. Most of
to move left and right. It is also subject to fluctuations in the the conventional hand-operated computer mice and trackerballs
local magnetic field. These can occur when large metallic ob- operate in this way. Although in the case of the former, there
jects, like tables, chairs and filling cabinet drawers, are moved is the opportunity to pick up the mouse and re-position it on a
close to the device. The approach is also not suited to users desktop without altering the mouse cursor’s position.
in wheelchairs who may dynamically change their, and their Relative pointing devices work by displacing the device until
laptop’s, orientation with respect to the Earth’s magnetic field. the desired position of the mouse cursor is reached, at which
A further approach is to use infrared light emitting diodes point, the device is moved back into a neutral area (a “dead-
(LED’s) and photodetectors to determine head position. For ex- band”). The mouse cursor will only move when the device is dis-
ample an array of infrared LED’s can be mounted adjacent to placed from this deadband. Standard joysticks and a keyboard’s
the computer monitor and a photodetector on the user’s head. cursor keys work on this principle.
If the LED array is organized so that the detected irradiance of Head-operated mice can use either principle, although abso-
each at the photodetector in some way encodes the relative posi- lute positioning requires typically more precise head position
tion of the user’s head to the monitor, a mouse can be developed. control and suffers from low immunity to interference in its en-
Indeed, this is the basis of at least one commercial head-oper- vironment. The device that we describe in this paper is com-
ated mouse, which is targeted at the games market. Moreover, monly used as a relative pointing device. However, it can be
it is the basis of our approach and is described in detail in Sec- enabled to work as an absolute positioning device if desired,
tion III. but the resolution is severely reduced (see later). Section VI in-
4) Other Approaches: The approaches described above all cludes results of a preliminary evaluation of the two operating
rely on the user having full control over his/her head move- modes.
ments. While the approach that we introduce in this paper also 1) Discussion: Video-based approaches would seem to be
has the same requirements, we note that there are other ap- the ideal solution to head-operated control of mouse pointer po-
proaches that are based simply on eye movements relative to sition. However, there are few practical systems, and these are of
the head. relatively high cost. Where cheaper alternatives have been con-
One approach is similar to the video analysis techniques structed they have tended to have significant operational diffi-
described in Section II-A except that it is concerned with the culties. For example, the device based on the Earth’s magnetic
pupil’s movement with respect to the face, rather than relative field requires users to adopt unnatural, tiring positions for left
to a fixed camera. Such approaches are relatively processor and right movement.
110 IEEE TRANSACTIONS ON REHABILITATION ENGINEERING, VOL. 8, NO. 1, MARCH 2000
the receiver, see below) to determine how far it is away form the
deadband and control mouse speed so that greater deviations
from the deadband cause greater speed. Secondly, and most sig-
nificantly, there is a feedback loop. The user monitors the posi-
tion of the mouse pointer on the screen and moves his/her head
accordingly, effectively compensating for the nonideal irradia-
tion patterns. This makes the system very resilient to manufac-
turing variations in the LED’s and also to the alignment of the
Fig. 9. The mounting of the LED’s. LED’s.
practical devices deviate from this. Experience shows that their IV. SYSTEM DESIGN
irradiation patterns may be asymmetric and they may also have
patterns that have ‘flat tops’ or even irradiation patterns that A. Transmitter Unit Design
have maxima that are not at the centre of the beam. The patterns The TxU consists of two parts, an array of LED’s that are
often have very sharp cut-offs outside the centre of the beam. mounted on the user’s head and a small control box that can be
Moreover, between devices of the same type, there is consider- clipped to the user’s belt or rest on their lap. Only the LED’s are
able variation in irradiation pattern. These characteristics are mounted on the head.
thought to be due to the lenses of the LED’s, which have short The TxU signals to the RxU, from which the RxU determines
focal length and may be subject to manufacturing variations. the angular displacement of the head and also the state of the
These problems compromise the accuracy of the measurement two mouse buttons. The latter are physically implemented as
of angular displacement and moreover may affect the optimal sockets on the Control Box into which users can connect stan-
alignment for any given pair of LED’s. dard switches according to their preferences.
The problems are not as serious as one might suppose. This The operation of the TxU is explained using the schematic
is for two reasons. Firstly, we chose to operate primarily as a diagram shown in Fig. 10. The output of each of the LED’s and
relative pointing device. This means that, in the simplest case, the signal received by the RxU are shown in Fig. 11. Fig. 11
we only need to detect when head position has moved from the shows each LED is modulated by an 11-KHz signal at a 25%
central deadband and the direction in which it has moved, al- duty cycle. With an appropriate receiver (see Section IV-B
though we may chose (by setting appropriate dip-switches in below), the modulation removes interference from other light
EVANS et al.: CONTROLLING MOUSE POINTER POSITION USING AN INFRARED HEAD-OPERATED JOYSTICK 113
sources. The modulation frequency must be chosen so that it cost and power consumption compared to the discrete design
away from the line frequency of a computer monitor (generally, implied by Fig. 10.
30–35 KHz) and modern electronic ballast fluorescent lights
(around 30 KHz), otherwise interference may be picked up at
the receiver. B. Receiver Unit Design
The TxU uses a time-division multiplexed protocol to
transfer the necessary data (the four LED positions and two The RxU is responsible for processing the signals from the
mouse button states) to the RxU, as shown in Fig. 11. First, HMU and encoding these as standard mouse control signals. A
all LED’s are illuminated. The RxU uses this pulse as a simple block diagram of the receiver is shown in Fig. 12.
synchronization pulse. There then follows a gap after which The light is received by a photodiode that is configured with a
the right LED and then the left LED’s are illuminated. Next bandpass filter. The bandpass filter’s design is intimately linked
the state of the right mouse button is sent. If it is depressed, all to the characteristics of the photodiode and it is present to re-
LED’s are illuminated, as in the synchronization pulse. The up ject extraneous infrared light. The filter chosen is a single LC
and down LED’s are illuminated in turn, before the state of the resonator. This is low noise, has zero-power consumption, pro-
left mouse button is sent. To ensure reliable synchronization vides a high impedance load for the photodiode giving a good
of the receiver, the transmitter waits for a period as long as the signal at the modulated frequency, and gives excellent rejection
original symbol stream, restarting the transmission with the of unwanted light.
synchronization pulse. This gives 16 transmissions per second The next stage, an HF preamplifier, has a maximum gain of
and gives acceptable operation. around 1000. It is controlled using an AGC element. The AGC
One important design consideration is the amplitude of the is generally controlled by the “slow AGC” signal from the de-
synchronization and mouse button state pulses. As these are the modulator. This gives an AGC with a fast attack and a slow
sum of all four LED outputs, they will, if full illumination of delay characteristic that minimizes the changes in gain between
the LED’s is used, be considerably larger than the output for a sampling. However, if there are significant changes in the re-
single LED. As automatic gain control (AGC) is used at the RxU ceived signal, due, for example, to movement of the user toward
and is controlled by the synchronization pulse, this would mean or away from the receiver. There is a possibility that this AGC
that very low amplitudes are detected for the single LED’s. Thus may not recover in time. Therefore, a fast AGC is also provided.
the LED’s are not illuminated to their normal levels during the This is triggered by the microcontroller, in the gap between the
synchronization and mouse button periods. The required level is synchronization pulse and the other signals (see Fig. 11), when
difficult to calculate analytically and has been set empirically. the received synchronization signal is outside a specified range.
Occasionally this means that the signal received for a single This ensures that the signals from the single LED’s are mea-
LED exceeds that of the synchronization pulse. In this case the sured consistently.
AGC simply adjusts and the system operates normally. The only A linear rectifier with a minimum voltage gain of approxi-
requirement is that the synchronization pulse is detected. mately 3 then demodulates the signal. The signal is then am-
The design of the final system is slightly different to that plified by a low frequency dc amplifier and passes through an
discussed above, although the principles remain the same. The active filter that removes the residual carrier component. This
transmitter is implemented using a PIC 16C54 microcontroller filter is designed to compensate for the slow rise time of the LC
to control the illumination of the LED’s; this reduces system resonator.
114 IEEE TRANSACTIONS ON REHABILITATION ENGINEERING, VOL. 8, NO. 1, MARCH 2000
First, as noted in Section III, the signal derived from the left
and right LED’s is also dependent on the movement in the ver-
tical plane. However, one can easily scale the difference
by dividing by . In practice, however, this approach has
problems, because, as becomes small (toward
the margins of operation) and thus noise and quantization errors
start to dominate. We therefore need to multiply the expression
by some factor. Experimentation shows that, if Fig. 13. The mounting of the photodiodes for wide user angle.
one uses a factor of , i.e., we use only the difference func-
tion , the system works acceptably well, provided that the
result is processed in the correct way, as discussed below.
Second, we need to consider the operational characteristics
of the device as perceived by the user. It is thought that it is de-
sirable that, as the head moves further away from the deadband
position, the mouse pointer should move more quickly. Thus,
close to the deadband the user has fine control, further away
from the deadband, the mouse pointer should move much faster.
This requires a nonlinear response. An associated problem is
that we require a deadband. If we directly use the result of
, head positioning becomes critical and there is signif-
icant “mouse pointer creep.” These problems can be overcome
if the result of is used as an index into a look-up table,
which effectively encodes the rules for the deadband and mouse
speed. Indeed, the system contains a number of rule sets that
Fig. 14. Transmitter unit.
govern the size of the deadband and the speed of the mouse.
These can be selected by DIP switches in the receiver unit.
While this arrangement may not seem ideal, in that movement evaluation shows that the user adapts in an unconscious way, as
in the vertical plane is not factored out of the determination of far as he/she is concerned the system works perfectly. The de-
horizontal position, and vice versa, the system works very well. vice can be configured as an absolute pointing device. Here the
This is another example of the user’s ability to use visual feed- resolution is much less than that achieved by commercial sys-
back of the mouse pointer position to correct for defects in the tems aimed at people with disabilities. The maximum resolution
design of the system and its components (see Section III-C). The of our device is and at this level one requires good
user does not realize that he/she is compensating. control of head position to maintain selection of a given area.
Moreover, in this mode, the device cannot emulate the signals
produced by a Microsoft mouse and a special driver has to be
V. PERFORMANCE AND RESULTS loaded to interpret the serial input.
The system has an effective transmitter to receiver range of One of the motivations for developing this device was to pro-
15 cm to 10 m and is tolerant of changes in transmitter to re- duce a usable, low cost device. The device is commercially
ceiver distance when the system is in use. If an array of three available in the U.K. The device retails for £200 (around $300
photodiodes is used at the receiver, as shown in Fig. 13, the ef- US), this is roughly a quarter of the commercially available
fective angle of use is better than 120 in the horizontal plane. devices aimed at people with disabilities. Of course, as noted
This gives considerable freedom of seating position and move- above our device is not as sophisticated or accurate (as mea-
ment. sured by the resolution of absolute position) as these devices.
The TxU requires 2 mA at 3 V. For two alkaline bat- But the real question concerns whether the device is as least as
teries this gives a working life of 1300 h, which is around 120 usable as the more expensive devices. This issue is addressed in
eight-hour days. The RxU is powered directly from the PC’s se- Section VI.
rial port and requires no other power source. The system will
work with laptop computers, but obviously puts a small addi-
VI. EVALUATION
tional load on their batteries.
One interesting question is how accurate is the system at mea- Around 40 users, of whom nine were people with disabilities
suring angular displacement? Of course, as discussed earlier belonging to target user group, have used the device. All users
this depends on the uniformity of the irradiation patterns of the have successfully used the device. One significant result con-
LED’s, which we find, by experience, rather than detailed and cerns the way in which left-right movement is accomplished.
systematic study, to be rather variable. The interesting result of Devices that use certain types of technology for determining
the work, is that, when working as a relative pointing device this head position (for example those based on gravitational or mag-
does not much matter. The user readily compensates for nonlin- netic fields) require the user to accomplish left or right move-
earities by monitoring the position of the mouse pointer. One ment by bending their necks, so that their ears are moved to-
might suppose that this would be annoying to the user. But our ward their shoulders. All of our earlier devices worked in this
116 IEEE TRANSACTIONS ON REHABILITATION ENGINEERING, VOL. 8, NO. 1, MARCH 2000
operation was much easier to use. The reasons for this are not
clear, but we speculate that “joystick-style” devices require less
precise head positioning than “mouse-style” devices, because,
when the desired mouse pointer position has been reached the
joystick user can rest his or her head in the relatively large dead-
band position. Whereas, the absolute mouse user, has to keep his
or her head at the desired position. Any slight movement from
this position will cause the mouse pointer to move.
VII. STATUS
As noted above the Head-Operated Joystick is available as a
commercial product in the U.K. Fig. 14 shows the Transmitter
Unit, Fig. 15 the Transmitter Unit being worn by one of the
authors, and Fig. 16 shows the receiver unit.
ACKNOWLEDGMENT
The authors would like to thank P. Hawes of the Foundation
for the Communication for the Disabled for conducting the eval-
uation and making a number of helpful suggestions to the de-
velopment of the device. The authors wish to acknowledge the
Fig. 15. Transmitter unit worn by a user. contribution of S. Pettitt to the development of this and earlier
head-operated devices.
REFERENCES
[1] D. G. Evans, S. Pettit, and P. Blenkhorn, “A head operated ‘Joystick’,”
in Proc. 5th Int. Conf. Comput. Helping People Special Needs, R. Old-
enbourg, Ed., 1996, pp. 85–91.
[2] M. V. Rigol, “Determining head position for mouse emulation,” M.Sc.
thesis, UMIST, 1996.
[3] J. Colineau et al., “Laser mouse,” in Proc. 2nd TIDE Congr., P. Porrero
and R. Puig del la Bellacase, Eds., Amsterdam, The Netherlands, 1995,
pp. 290–293.
[4] (1999, Apr. 12). [Online]. Available: http://www.orin.com/access/
[5] (1999, Apr. 12). [Online]. Available: http://www.prentrom.com/ac-
cess/hm2000.html
[6] W. A. Hyman, G. E. Muller, and J. S. Neigut, “Laser diodes for head
pointing and environmental control,” in Proc. 15th RESNA Int. Conf.,
1992, pp. 377–379.
[7] D. Field, “A better mount for light pointers,” in Proc. 14th RESNA
Int.Conf., 1991, pp. 373–374.
[8] (1999, Apr. 12) Available: http://www.madenta.com/rel_tracker.htm
[Online]
[9] (1999, Apr. 12). [Online]. Available: http://www.visioncs.com/
[10] J. Gips, P. DiMattia, F. X. Curran, and P. Olivieri, “Using eagle eyes—An
electrodes based device for controlling computers with your eyes—To
help people with special needs,” in Proc. 5th Int. Conf. Comput. Helping
People Special Needs, R. Oldenbourg, Ed., 1996, pp. 77–84.
Fig. 16. Receiver unit.
[11] G. Norris and E. Wilson, “The eye mouse: An ocular prosthesis,” Circuit
Cellar INK, no. 59, pp. 20–27, 1995.
[12] T. Nielsen, “Opto-electronics by design,” Electron.Wireless World, pp.
way. Users reported that this type of movement was uncomfort- 364–360, May 1994.
able and very tiring after even relatively short periods of use
[1]. The new device has a much more natural left to right move-
ment, in which users move their head to look to their left or right D. Gareth Evans received the B.Sc. degree in elec-
twisting, rather than bending, their necks. In effect, the mouse trical and electronic engineering from the University
pointer’s position follows the position of a user’s nose. This is of Manchester, U.K., in 1985 and the Ph.D. degree
from the University of Manchester Institute of Sci-
a much more natural movement and is reported as being much ence and Technology, U.K., in 1995.
less tiring and operation can be sustained for long periods. Rel- He has been a lecturer in the Department of
atively informal comparative trials have been undertaken with 9 Computation at UMIST for the past ten years. He
is a Co-Director of the Technology for Disabled
disabled users. First, they used a commercially available, abso- People Unit at the University of Manchester Institute
lute positioning, “mouse-style” device to interact with a number of Science and Technology (UMIST), Manchester,
of standard Windows applications. They then used our mouse, a U.K. His research interests include alternative
interfaces to computers for people with disabilities, navigation and orientation
relative positioning “joystick-style” device to carry out the same systems for blind people, speech synthesis and training, and assistive devices
set of tasks. All nine users reported that the “joystick-style” of for people with a range of disabilities.
EVANS et al.: CONTROLLING MOUSE POINTER POSITION USING AN INFRARED HEAD-OPERATED JOYSTICK 117
Roger Drew received the M.Eng. degree in micro- Paul Blenkhorn received the B.Sc. degree in mathe-
electronic systems engineering from University of matics from the University of Manchester, U.K.
Manchester Institute of Science and Technology He has been an active developer of systems for
(UMIST), Manchester, U.K., in 1998 following a people with disabilities for the past 18 years. He
career in electronics engineering. is currently a Senior Lecturer in the Department of
The work described in this paper is based on his Computation at University of Manchester Institute
third-year undergraduate project and subsequent em- of Science and Technology (UMIST), Manchester,
ployment with the Technology for Disabled People U.K., and is Co-Director of the Technology for Dis-
Unit at UMIST. abled People Unit. He has broad research interests in
the area of technology and people with disabilities.