Human interface design verlag

Of course, by appropriateuse of a keyboard with cursor keys which move a cursor around a display,many of the these tasks are possible, but the advtanage of pointing devices isthat they p

MSc Information Systems 1999

Human Computer Interaction

The physical level

Adapted from the book Human Interface Design (Thomas, 1999 forthcoming,

Springer-Verlag) for the MSc Information Systems Human Computer

Interaction course © Peter Thomas 1999

Introduction

In chapter 1 we developed a definition of HCI which suggested that it was

the investigation of interfaces (physical components of a

system which allow the control and manipulation of a

system, by exploiting the user’s cognitive abilities, and

allowing the uses to get an understanding of the system,

for the purpose of performing some task in a context)

with the aim of designing user technology for ease of use

and effectiveness

In chapers 2 and 3 we looked at the phenomena of the user’s cognitive abilities(in terms of the Model Human Processor), and the phenomena of users’understandings (in terms of mental models) This chapter looks in detail at

The chapter is divided into two sections Section 2 deals with input devices –those physical system components which are available for the user to controland manipulate information in a computer system We will be looking both atthe various devices that are available and also at some frameworks withinwhich to consider the phenomomon of ‘input’ more generally.

The second section looks at output devices This is much shorter section, andlooks primarily at visual output This section briefly looks at what technology

is currently available (the most widely-used being the converstional CRTattached to most personal computers) and also at displays which are justbecoming available (such as the flat-panel displays used in portable and

handheld computers) We will also look at some of the issues relating to thehazards of displays and some of the regulations designed to lessen thosehazards

Input Devices

This section is essentially a catalogue of input devices Several types of deviceare discussed along with some of their adavantages and disadvantages, andsome of the applications for which they are most appropriate However, thepurpose of this section is not to provide a complete catagogue; firstly because

to do so would mean several chapters in itself; secondly because there aremany other books on user interface design which provide straightforwardcatagogues of devices; and thirdly because many of you will be all too familarwith the basic characteristics of many available input devices (and will haveused many of them)

The second part of this section discusses some of the more fundamental issuesrelating to input devices, particularly on examining a framework of conceptsfor the evaluation and development of input devices and ways of using them.Many of the concepts we will look at are the work of a researcher called

William Buxton, whose excellent state-of-the art review book (Buxton, 1994)both gives a more complete catalogue of input devices, and explains the

concepts we will look at in greater detail

Input devices: some general considerations

Interaction tasks and interaction techniques

We can consider the input devices we will look at in two broad classes:

text-entry devices (keyboards) and pointing deevices (touchscreens, tablets, mice,

josticks, trackballs, lightpens and some others) We will discuss keyboardslater in this section, where the issues are primarily those of keyboard layouts tooptimise the speed and accuracy of text entry

In terms of pointing devices, there are many more issues we might consider.This is largely bacause the range of activities the user can perform with

pointing devices is much broader We can suggest define several types of

interaction tasks which are possible with pointing devices: selection (choice

from several items on a diplay), positioning (choosing a point in a 2, 3 or more dimensional space), orienting (choice of a direction in a 2, 3 or more

directional space), pathing (the combination of position and orientation actions

to suggest a trajectory for an object, or to trace a path), and quantifying

(specifying a numerical value) (A further task, which requirtes a keyboard in

addition to a pointing device, is textentry – entering, moving and editing text in

a 2-dimensional space, as would be possible in a mouse-driven text and

graphics editor)

We could reduce these intertaction tasks further to the basic tasks of position (specifying a position), text (entering text), select (choosing from a set of items) and quantify (specifying a numeric value) Of course, by appropriate

use of a keyboard with cursor keys which move a cursor around a display,many of the these tasks are possible, but the advtanage of pointing devices isthat they provide users with the possibility of much more effective and efficientperformance of these tasks, with fewer errors and greater speed This fact,that these basic interaction tasks could be performed with different devices,suggests a further distinction between interaction tasks (what the user needs to

do) and the interaction tecehniques (how the user does it) We will return to

this notion later in the chapter, but for now we can note that there is not onlyone way in which an interaction task can be performed, and not only onedevice which can be used to perform it, although there may be more effectivetechniuqes and devices for different tasks

Levels and devices

We can also make another useful set of distinctions We can also look at

interaction devices in terms of several levels: the device level (the physical characteristics of the device), the task level (in terms of the four basic tasks and the use of different techniques to realise them), and finally the dialogue level

(the ways in which sequences of tasks are linked together) In this chapter wewill be looking primarily at the device level and the task level: how devices areconstructed and how they are used for specific tasks We will look at the

dialogue level in the next chapter when we look at ‘interaction styles’

Finally we might note, as we will see later in this chapter, that we can add an

additional level, the pragmatic level, which considers the ways in which the

device, task and dialogue levels tie together and create the overall

characteristics of the device and its suitability for different tasks and contexts ofuse

levels in considering interaction devices

Further classifications of devices

Further classifications of input devices is also possible In terms of pointing

devices we can distingush between indirect devices (where the user controls a

screen representation such as a cursor without using the display directly – such

as a mouse, trackball or joystick) and direct devices (where thea user

manipulates objects directly on the display – such as a lightpen or

(which allow the user to input a single value) and choice devices (which allow

the user to indicate some choice from a number of alternatives) A further

category is that of dimensional devices (and devices of more than

3-dimensions, which allow the user to provide input along several dimensions)

The class of Locator devices can be further subdivided in a number of ways For example, absolute locator devices (such as a talbet or touchscreen)

provide input with respect to a particular frame of reference, where the userindicates a position using the device which is the same position on the display

In contrast, relative locator devices (such as mice, trackballs, tablets and some

joysticks), provide input which indicates not absolute positions but changesfrom a previous position Here a user can, for example, move a mouse along adesktop and then pick the mouse up, reposition it to the starting point andmove it again a similar distance Further distinctions can also be made which

combine properties Direct locator devices (such as a lightpen, or touchscreen)

allow the user to point directly at an object on a display; in contrast with

indirect locator devices, such as a mouse or joystick, the user manipoulates a

screen cursor or other object which allows the user to point indirectly at the

object Continuous locator devices (again such as the mouse or trackball)

allow the user to control the cursor smoothly through smooth motion of the

device, whereas discrete locator devices (such as cursor keys) only provide

stepped movements of a screen cursor (thorugh a text document, line by line,

for example) The class of keyboard devices are also, of course, discrete

devices which provide the user with the possibility of input of discrete

information usually, in the case of the QWERTY keyboard, alphanumericcharacters

Valuator devices, such as knobs (or ‘potentiometers’) allow the user to specify

single values; a bounded valuator device (such as a rotary volume control on a

TV) can be turned only so far before the maximum value is reached when theuser encounters a stop Such a device inputs an absolute value In contrast, a

knob which turns continuously, or an unbounded valuator device, can be

turned any number of times in any direction and can therefore input relativevalues

Choice devices can be of several types, but the most familar are function keys

of the types nortmally found on many personal computer keyboards Choicedevices are used to signal one from a fixed range of choices

Pointing devices such as the mouse are essentially two-dimnensional devices,providing the user with the possibility of (in the case of locator devices)

indicating objects and specifying choices in 2 dimensions Some of these

devices can be modified to allow the user to input information in

3-dimensions For example, a joystick can be equipped with a shaft that rotates

to provide a third dimension of input A more sophisticated 3D input device is

a ‘spaceball’ which is a ball mounted in a solid base and which the user pulls,pushes and twists without actually moving the ball itself The ball contains anumber of strain guages which sense the user’s attempts to move it

A more complex, and very comprehesive, scheme for classifying the

properties of inpout devices is proposed by Lipscomb and Pique (1993) They

distinguish between multi-axis physical devices, which can control or

manipulate objects along multiple dimensions, and single-axis physical devices

which allow the user to manipulate objects in only one axis at a time

Accirding to their classification these devices can be of several types:

1D, 2D, 3D (etc) devices: these devices have a number of

‘degrees of freedom’ (directions in which the user canmanipulate the device) for each hand or finger For

example a mouse with buttons has similar up-down andleft-right axes, which make it a 2D device

Free Devices: these can move in any direction on more

than one axis but require considerable skills to

manipulate accurately, for example a mouse, a puck used

on a tablet, or a trackball

Sticky Devices: these have a barrier that prevents

inadvertent movement along one axis when another isused

Rotation devices: these respond when twisted, for

example a rotoary potentiometer

Translation devices: these respond when pushed, such as

mouse or a slide potentiometer

Unbounded devices: these move without limit (for

example a rotary potentiometer)

Bounded devices: these have physical limits of motion,

such as a one turn rotary potentiometer

Homogenous devices: these cannot be set to a

remembered physical position such as a trackball).

Held-up devices: such as a pen, or a user’s finger on a

touch pad

Body-mounted devices: which are attached to the user

(such as a dataglove)

Some of these classifications will be used in the following sections, which

discuss each device in turn In particular the notions of absolure/relative and

bounded/unbounded devices will be used to distinguish the properties and uses

of the devices

A brief cataogue of input devices

In general terms, all of the discussions of input devices in the following

sections cover the following themes: the physical device itself (mouse, joystick, tablet, pen) the software interpreters (which accept input from the device and

translate it into an appropriate representation for the application software and

services) and the representation of the input from the device on a computer

display There are issues for human interface design in all of these themes Interms of the physical device there are many low-level ‘ergonomic’

considerations which are important, such as the shape of the device and itssuitability for the user’s hands (for example the size, shape, contouring, weightand resistance of a mouse) For software interpterers, issues exist in terms ofthe ways in which the software maps the input from the device into a formsuitable for the application For example, as we will see in the discussion ofgraphics tablets, the movement of the pen or puck across the tablet can be

translated differently with quite different resilts Finally, in terms of the

representation of the input, this interpretation must be fed back to the user in aform which provides ways of most effectively using the device for an

interaction task In the following sections we will look at all three issues, butfocus particularly on the final two, since it is here where it is possible to

influence the design of human interfaces most directly These two issues willalso lead us to to a more detailed discussion of a particular framework forconsidering input devices

Touchscreens

Touchscreens allow users to provide direct input by touching or moving afinger on the surface of a display Two main forms of operation are used in

touchscreens: either the user’s finger contacts with an overlay on the display,

or her finger interrupts beams projected obver the surface of the display For overlay touchscreens there are several types of technology used: conductive

(conductive layers which carry electrical signals which are bridged by the

user’s finger), capacitative (the user’s body capacitance causes the generation

of an electrical signal on the display) or crosswire (a grid of wires set in the

display generates input when specific wires are contacted togther) For beam

touchscreens, there are infrared (the user’s finger interrupts light beams

projectred over the surface of the diaplay) or acoustic (ultrasonic waves are

projected over the surface of the screen are are again interrupted by the user’sfinger)

These different types of touchscreens have different properties For example,

different types of screen provide different resolutions (the possible number of

touch points on the display that can be mapped by software to correspond tousers’ selections) with conductive screens having the highest resolution, and

infrared or acoustic screens the lowest), different durability (how these

displays react to different types of usaage, with the conductive screens’

membranes being easily damaged), and environmental limitations(in dirty or

smoky environments devices based on infrared can be falsely activated, andthe capacitative touchscreen canot be used by users wearing gloves, for

example)

Greenstein and Arnaut’s (1988) review of touchscreen technology suggeststhat even though touhscreens are intuitively ‘easy to use’, careful design isrequired to ensure that they are used appropriately For example, touchsreenswith low resolution may be frustrating for users since they may have difficlty inselecting appropriate regions of the screen, and conductive screens whichrequire the user to press hard on the surface of the diplay may be

uncomfortable for the user over long periods Also, as you might expect, thesize and organisation of the visual ‘keys’ (areas that the user can select

represented on the display of touchscreens) has a great deal of impact onusability (Potter, Berman and Shneiderman, 1989) Generally, larger keyswhich are more separated tend to result in more effective selections, and theprovision of feedback to the user (perhaps in the form of providing highlightsaround the user’s selected ‘key) increases users’ accuracy and perception ofthe ease of use of touchscreens

A further issue is that of how the user’s contact with the screen is interpreted

by the software There are many possibilities here, for example a single-touch

strategy , in which the user’s input is taken to be the first touch on the screen.

The disadvantage here is that the user is given little feedback on a selection andmay need to perform several actions to correct an error Another approach is

the lift-off strategy (Shneiderman, 1992), where the user first touches the

screen to reveal a cursor which can then be manipulated and the user’s choice

is only signalled when her finger is lifted from the display A third strategy is

the touch-in touch-out strategy, where the user confirms a choice generated by

a touch by a second touch on the display Here the user’s input effectivelysignals different modes in which input are to be interpreted (‘select mode’ and

‘choice mode’)

There are some extremely novel designs of touchscreen For example, the

Perex Touchmate is a cross between a tablet and a tocuscreen This strange

peice of technology is effectively a touchsensitive pad which is placed under a

PC monitor and detects not direct contract with the screen, but the

movements genertated by the user touching the screen Software which

accompanies Touchmate allows the user to calibrate the device so that

touches on the screen will be translated into the correct positional

information

Applications of touchscreens are varied They are perhaps most appropriatefor use where there is a fixed and predetermined set of options for users tochoose from, for example in public-access computer systems in buildingswhich display maps or general information, or in applications where the user’sattention needs to be centered on the display itself rather than on the use ofother input devices (for example in tasks such as air-traffic control wheresustained attention to the display is necessary) For the same reasons,

touchcreens are less appropriate for applications where alphanumeric dataneeds to be entered: the use of a standard keyboard displayed on the

touchscreen has been found to be to use since the position of the screen makes

a normal typing posture impossible and would cause discomfort and possiblyinjury to the user, in addition to the discomfort of reaching out with the arm

that is a problem with all touchscreens.

Tablets

Tablets are flat panels which are placed on the user’s work desk There are

two basic types Graphics tablets are large tablets which the user inputs

information using a stylus or pen, or a graphics puck Touch tablets are

smaller versions which are primarily intended to accept input from the user’sfingers Graphics tablets are primaily used for applications which involve themanipulation of visual information in the form of diagrams, plans, blueprintsand sketches As with touchscreens there are a variety of technologies used fortablets (conductive or acoustic, for example)

Since the tablet is effectively a representation of the display on which the user

is manipulating and viewing information, one of the prime issues in the use oftablets is the mapping between the display and the tablet The issues are those

of relative vs absolute modes of operation (the user’s finger, pen or puck input

is mapped in an absolute way to the corresponding point on the computer

display, but when the finger, pen or puck is moving it is mapped relatively to the original position) and control/display ratio – often known as the CD ratio –

(the amount of movement of the input device - in this case the pen, puck orfinger on the tablet - and how it is represented by the movement of a cursorobject on the display)

These two issues – absolute/relative mode and CD ratio – are common to allinput devices which control a cursor object on a diplay In the case of thetablet, these issues are impottant since tablets can be of varying sizes, none ofwhich need be exactly the same as the display If the tablet is smaller than thedisplay then the user needs only to make a small movement on the tablet to

move the cursor on the display a large distance This means that a smallertablet may be better configured in relative mod, since the movement of thedisplay cursor is not dictated by the size of the tablet Conversely, when used

in absolute mode (the position of the finger or pen on the tablet is mappeddierctly to a display location) this effectively conditions the size of the tablet to

be that of the display Now, if we were to alter the CD ratio so that, for

example, a movement on the tablet is magnified by a factor of 2, then thedisplay would have to be twice the size of the tablet (or alternatively that thetablet be half the size of the display)

In addition, it is possible to manipulate not only the CD ratio but also to add a

velocity component to the configuration of the tablet which allows a rapid

movement of the finger, pen or puck on the tablet to be interpreted as a largermovement on the display (Becker and Greenstein, 1986) This feature is useful

to allow users to move across a large display with a small tablet, or to assistusers with disabilities in moving objects on the display without large arm orhand movements

This ability to reconfigure the input device may not only be in terms of thesoftware interpreters, but in terms of physical configuration Since the tablet is

a flat device used away from the computer display, it can be overlayed with a

variety of templates which can be used to mark out different areas of the

tablet One idea is that the tablet can be reconfigured in terms of several other

input devices These virtual input devices can provide a variety of inputs

without the need for the user to have each device attached to the computer(Brown et al., 1990)

As with the touchscreen, the fact that the user may use her finger as an input

mechanism brings several problems which conterbalance the advatanages ofnot having extra devices such as a puck, pen or stylus to lose, break or

otherwise mishandle For example, a finger, in comparison to the point of apen, is large and irregular, and differences in finger pressure and how the userlifts her finger from the tablet may cause unintended or inaccurate movemnent

of the screen cursor (Parng and Ellingstrand 1987) Again, as with the

touchscreen, the use of the tablet would seem to be intuitive and natural,especially for drawing tasks which would be undertaken in exactly the samemanner - using a pad and pen - without the computer

In practice however, most tablets are used with a separate device, usually astylus or pen (which may have pressure sensitive tips, switches on the body ofthe stylus which can can sense orientation, velocity and tilt to provide extrainformation, or which can be used in combination with an additional keypadfor data entry) or a graphics puck, similar to a mouse There are also wirelesspucks and styli which remove the need for cabled connactions between puck,tablet and computer, and also transparent tablets which can overlay

diargrams, maps and charts that user is manipulating and/or amending

Tablets also come in portable form, such as the Acecad Acecat portable tablet

which has 1,000 point-per inch resolution and detects proximity of a pointingdevice within 0.25 inches of the surface The tablet’s controller software alsoallows a switch betwen absolute and relative modes, and can be used with astylus with built-in mouse button, or with a digitising puck

LightPens

Lightpens provide positional information when the pen is pointed at a displayand the pen is activated by the light output of display itself Lightpens can be

used for both selection (the user points to an object or region on the screen andpresses a button on the pen to select), or tracking (the user moves the pen overthe surface of the display whilst activating the pen) Obviously users must bewithin the immediate region of the display to use the pen, which is physicallycabled to the computer, and this can be tiring for users, in addition to theaction of picking up and putting down the pen In general, lightpens are usefulfor tasks requiring simple selections of large objects, or for tasks which

require slow and steady tracking movements

Portable pen/digitisers

Although both styli and light pens have been used for some time, the use ofpens with small digitising pads such as those found in small palmtop

computers is growing Pen/digitisers work using the same kinds of technology

as tablets: direct contact resistance or overlay digitisers (where the pen

completes a circuit), conductive or underlay digitisers (where the pen tip forces two conductive layers togther), infraredbeams (where the pen tip breaks

criss-crossing light beams) , or electromagnetic digitisers (which sense the

presence of the pen tip near the surface of the screen)

overlay and underlay technology for portable pen/digitisers.

As a conseuquence, the pens that are used in these systems are either tethered

(as with overlay digitisers which require the circuit to be completed and

therefore the pen is wired to the computer) free/non-electronic (where, as

with conductive underlay digitisers, the pen merely connects the two surfaces

and so can be a piece of plastic with a small metal tip) or free/electronic (where

the pen contains batteries since it needs to interact with the digitiser, as withelectromagnetic digitiser systems) Since users have input devices that more

PCB electronics

LCDWire grid sensor

Clear glass

PenPCB electronics

Glass sensorLCD

Pen

clsoely match pens that they would use with real paper, there have been avariety of designs for the display glass on which the user writes, with somesystems using etched glass which approximates the drag of a pen on a piece ofreal paper.

Many developers have also been active in developing the underlying

technology for pen/digitiser interfaces with the development of digital ink (the

user’s inputs are stored as data structures which can be manipulated andrecalled – for example maps, diagrams or sketches), and the develoopment of

a language of gestures with which the user can use to issue commands Digital

ink in particular is an extremely powerful way of recording the user’s input: itstores not only the x,y coordibates of the inputs so that the sketches or

diagrams can be redisplayed, but also the origins of the penstrokes, strokethickness, the pressure with which a stroke was created, and the relative scale

and position of a stroke One standard for digital ink, the JOT standard (a

collaboration between the developers Microsoft, Go Corporation, AppleComputer, Lotus and General Magic) is meant to be platform- and

application-independant This means that information in JOT format can be

displayed even when the application doesn’t have pen-input

Mice

Most people who have used a PC will be familiar with a mouse, a box-shapeddevice which is used to control a screen cursor for pointing, selecting anddragging objects around the display The mouse, along with the trackball, isthe most common form of pointing device in the human interfaces to currentcomputer systems

The mouse can almost be considered to be an inversion of a trackball: wheras

the user moves the mouse across a surface to control a cursor, the trackballrepresents the surface being moved across the trackball in order to move thecursor The most common form of mouse design uses a freely-moving ballwhich rolls in response to movement on a surface The forerunner of thismodern mechanical mouse was invented by Douglas Englebart of the UScompany SRI in 1964, and variations on the mechanical mouse have been used

ever since However there are other forms of mouse, such as optical (where

senors in the mouse underside detect movement on reflective mousepads,

invented in 1984 by Steve Kirsh of Mouse Systems International) and acoustic.

There are also many alternative designs of mouse at the ergonomic level whichclaim to better fit the contours of the user’s hand, help prevent injury andincrease the usability of the mouse

One of the most important issues in mouse design is the shape, number,

placement and use of mouse buttons The advantage of the mouse is that itrests under the user’s fingers and thus buttons can be placed on the mouse’ssurface to allow the user to signal choices (for example selecting and dragging

an object by holding down a mouse button) Various studies of mouse designand use have suggested that buttons should be mounted on the front surface ofthe mouse rather than the top surface (as this is a more natural resting placefor the fingers), that the buttons should be resistive enough to support theresting fingers without accidental activation, and that the entire mouse should

be textured to provide a comfortable and effective gripping surface

At the software interpreter level, the issues we have seen of absolute/relativemode, CD ratio, and velocity also apply to the mouse However, as should beobvious, since the mouse can be picked up and put down it can only operate inrelative mode: the user can move the screen cursor for a distance, and (if she is

running out of space on the mousepad or desk surface) reposition the mouseand continue the movement This means that devices such as the mouse areless than useful for applications where space is limited (as on most people’sdesks), space is unavailable (as in the use of portable, laptop or palmtopcomputers) or where both the user’s hands are required to operate the

computer It is for this reason that many touch-typists find the mouse anunusual and unacceptable device since it requires off-keyboard movementwhich disrupts typing performance

In general however, as Milner (1988) reports, mice prove to have significantadvantages over other devices for tasks such as the positioning of a cursor andthe selecting of text In a study by Card, English and Burr, 1978) the use ofcursor keys, function keys, a joystick and a mouse were compared on a textselection task The results suggested that the mouse was the faster device forthe positioning part of the task and produced fewer errors, even when thedistance to the target on the screen increased

distance and positioning time for four devices (adapted from Card,

English and Burr, 1978)

1 2 3 4

0 5

Cursor

Function Keys

Joystic Mous

Distanc

As with some of the other input devices discussed in this chapter, there are also

3-dimensional variants For example the roller mouse (Venolia, 1993) allows

the user to input the normal 2D positional information by moving the

mechanical ball mounted underneath the mouse but also, by the placement of

a pair of wheels at the front of the mouse mounted on a common axle, allowsmovement in a third dimension – away from and toward an object

the roller mouse From Venolia (1993: 32) and its construction.

An alternative to the physical redesign of the mouse to provide 3D input to usethe software interpreter to map the movement of the mouse in 2 dimensions to

a 3D object This approach, using virtual controllers allows users to control

3D objects using a variety of mappings between mouse movements in 2D andthe 3D rotation of an object Rotations are done with respect to the computerscreen’s frame of reference, with the x-axis pointing to the right, y-axis

pointing upward and z-axis pointing at the user So, rotations in x, y and z

Whee

Ball Axle

correspond to rotating the object up and down, left and right and clockwisecounter clockwise, respectively.

three graphical controllers (Courtesty Michael Chen,

chen@apple.com, Apple Computer 20525 Mariani Ave, MS 76-3H,

Cupertino, CA 95014).

The graphical dliders controller (a) uses a sliders to simulate ‘treadmills’ and therefore

provides relative control over the amount of rotation of the object A full sweep across

a slider provides 180 degrees of rotation about an independent axis As long as the mouse button is initially depressed inside one slider, the user can rotate about the

corresponding axis even if accidentally crossing into another slider The overlapping

sliders controller(b) is a modification of the conventional slider approach The three

squares in the middle column of the 9-square grid represent a vertical (x) slider The three squares in the middle row represent a horizontal (y) slider The outside eight squares represent a circular (z) slider A full sweep of the vertical or horizontal slider rotates the object 180 degrees about the x or y axis respectively A full circle around the outside squares rotates the object 360 degrees about z The difference between this controller and conventional sliders is that he direction of movement of the mouse more closely corresponds with the direction of rotation In addition, superimposing the controller on the object is intended to give the user more of a sense of directly

manipulating the object The continuous XY with added Z controller (c) operates in

two modes If the mouse button is depressed while the mouse cursor is inside the circle, left-and-right and up-and-down movement of the mouse will rotate the object left-and right and up-and-down on the screen Diagonal movement will rotate the object the proportional amount about the x and y-axis ( i.e the axis of rotation is on the x-y plane and is perpendicular to the direction of mouse movement) If the mouse

(c)(b)

(a)

button is depressed while the mouse cursor is outside the circle, the user can rotate the whole object clockwise by going around the outside of the circle Thus this controller provides either 1) continuous rotation on the x-y plane, or 2) exact rotation about the z-axis A full sweep of the mouse across the circle rotates the object 180 degrees about the corresponding axis in the x-y plane A full circle around the outside rotates the object 360 degrees about z.

Trackballs

As we have suggested, the trackball can be viewed as an inverted mechanicalmouse: this means that, since the user effectively moves the surface over thetrackball (rather than the trackball over the surface) trackballs can be used inapplications where space is limited, such as with portable computers Since thetrackball can be rotated freely it is effectively a relative device – the user canrotate the trackball even when the cursor will no longer move past the edges

of the display Again, as you would expect of this device, the issues of CD ratioand velocity also apply: it is possible to reconfigure the trackball to, for

example, move the cursor a greater distance in response to a greater speed ofrotation of the ball Allowing the user to configure the trackball’s behaviour is

a common feature of many systems such as the Apple Macintosh

Apple Macintosh Control Panel which allows the user to configure

the behaviour of a mouse or trackball (Courtesy Apple Computer).

The combination of gross movment and fine accuracy possible with a trackballmeans that it can be used for a range of positioning tasks, but to provide morethan positional information many trackballs are also fitted with buttons nearthe trackball itself

In general, users appear to be able to use the trackball effectively very quicklyand to adapt to the various differences in feel (different trackballs’ damping ofmovement, size, and intertia) very well In terms of the trackball’s

effectiveness in comparison with other devices, Milner (1988) reports thatstudies have suggested that is one of the most accurate input devices

There are also range of novel designs for trackballs, for example the Dextra

DexraPoint hand-held trackball, which allows the user to hold the trackball in

her hand whilst manipulating the trackball with her thumb , the Microsoft

Ballpoint mouse/trackball, which can be mounted on the edge of a portable

PC, and the MSI PC stylus, which is a cross between a trackball, a pen and a

mouse – the user can rotate the ball with her thumb or move the device across

a surface

Joysticks

If trackballs are upside down mice, then joysticks might be considered to betrackballs with a stick inserted in them Joysticks broadly fall into two types:

displacement joysticks (which use potentiometers to sense movments of the

joystick and where the displacement generates an output indicating the amount

of the displacement) and binary or switch-activated joysticks (also called

joyswitches, which contain a number of switches in the base of the joystick that

can be on or off, depending on the position of the jostick) With these secondtype of joysticks a movement of the joystick produdes not a variable output,but a steady one: when the joystick is moved to a particular psoition, a swich isclosed and a screen cursor (for example) starts to move at a constant rate untilthe switch is opened by the joysrick returning to the centre position A third

type of design is an isometric joystick, which does not move in its base, but

senses force or strain applied to the stick This joystick can sense force in anydirection and the screen cursor moves in proportion to the force applied by theuser

Again the issues of absolute/relative mode and CD ratio which we saw in

relation to mice and trackballs also apply to the design and use of joysticks.One finding reported by Greenstein and Arnaut (1988) is that joysticks may be

more appropriate for use in the form of a rate-controlled joystick in which the

displacement of the joystick merely controls the rate of movement of a screen

cursor rather than moving the cursor to a specific position A greater

displacement of a displacement joystick, or a greater force applied to an

isometric joystick, makes the cursor move faster, and allowing the joystick tocentre stops the cursor movement This corresonds to the velocity feature formice and trackballs we saw earlier, and also means that the joystick cannotoperate in absolute mode, although in general joysticks can be configured tooperate in an absolute mode where a displacement of the joystick correspondsdirectly to the position of a screen cursor It is also possible to configure thejoystick’s operation so that the relationship between a displacement (or force)can increase the velocity of movement of the screen cursor

As with the trackball, the joystick is effectively the user moving a surface

across the device rather than (as with the mouse) the user moving the deviceacross a surface Joysticks share the space-saving properties of the trackballand can be mounted alongide other devices such a keyboards (and, as we areall familiar with, on games consoles) In fact, joysticks are most appropriatefor tracking tasks such as those found in many game applications which do notrequire a great deal of precision, since (depending on how the joystick is

configured) it is difficult to make small, accurate movements of the stick Ingeneral, users appear to find the joystick easy and intuitive to use and thisdevice requires limited learning time Many of the ‘shoot-em-up’ computergames also use devices like the joystick to control the flight of jets or spacecraft

or to control fast-moving objects, and it is clear that the joystick is useful forthese types of fast-tracking tasks There are now some joysticks produced foruse with computer games which are extremely sophisticated, such as the

SpectraVideo Logipad This joystick has six buttons and an eight-way

directional thumbpad in addition to the normal movements of a joystick

There are also small joysticks, mounted in the keyboards of some portable

computers, such as that in IBM’s ThinkPad notebook computer The joystick

is advertised as “the world’s smallest stick shift”

Finally, similar to both the joystick and the trackball, are isometric spaceballs

which allow the user to provide input for 3D navigation, which are particularlysuited to interactive graphics manipulation The spaceball contains strain

sensors to allow the user to move a screen object in 3D and a set of selectionbuttons mounted on the front surface

Potentiometers

Potentiometers are the kinds of knobs we are all familiar with from homeelectronics devices (although more of these are being replaced by buttonswhich control graphic displays) These potentiometers increase the size of asignal as the knob is turned in a particular direction The major distinction here

is between bounded potentiometers, (which have a stop which doesn’t not

allow the knob to be turned beyond a particular point and are thus absolute

controllers) and unbounded potentiometers (which the user can turn without a

stop, and are thus relative controllers) Potentiometers are often mounted in

banks of dial boxes which are used by some specialised graphics systems to

allow the user to input data on rotation of objects on the screen

Function Keys

Many keyboards, in addition to the usual QWERTY keyboard layout for typing

characters, often contain specialised function keys which activate special

functions or are used in applications for common operations (such as cut or

paste in word processors, for example) We can distinguish between hard

function keys (such as the keys labelled F1 F12 on most keyboards) and soft function keys which are those, such as ALT, CONTROL and ESC, which are

used in combination with other keypresses In general, there is little

information about the performance of users with function keys This is partlybecause their use is not really dependant on issues in input device design (theyare merely ‘extra’ keys which are configured for special functions and so some

of their design issues are the same as for the keys on a keyboard), and becausetheir use is dependant on the ways in which various applications configurethem: the usual approach is to use function keys to reduce memory load andthe need to remember sequences of operation for specific commands

However, the Card, English and Burr (1978) study we saw earlier comparedfunction keys to other devices in a cursor positioning and selection task andsuggested that they were in fact usable for positioning a cursor in a block oftext and selecting some text They were, as one might expect, slower in thepositioning phase of the task when compared to devices such as the mouse orjoystick

Cursor Keys

Many applications allow a set of cursor keys to be used to control the position

of the screen cursor, especially in text-editing tasks Cursor keys, usuallyconfigured as below, move the cursor in one of four directions However, asMilner (1988) reports, there are several other cursor-key layout patterns, inparticular the inverted ‘T’ layout which allows user to perform a range of tasksmore effectively

(a) notmal cursor key layout and (b) a more effective inverted ‘T’

One of the latest 3D devices, which is used with Virtual Reality (VR)

applications is the dataglove The dataglove is a lycra glove worn on the user’s

hand(s) and equipped with a variety of sensors running along the fingers whichdetect flexion of the joints Some of these sensors detect when joints bend andothers detect the positional orientation of the hand The dataglove, although inits infancy, has a number of potenial applications, not only in VR systemswhich expoloit the glove’s ability to return 3D information, but also in thedevelopment of systems for users to communicate in the ASA sign languageused by deaf users Although initially expensive, the dataglove is becomingextremely cheap to manufacture (some gloves for computer games

applications retail at under $50) and this form of input will be used more andmore in future human interface designs

Another form of input device relies not on the user’s hands but on her eyes

Eyetrackers allow the user to signal positions on a display by measuring the

deflection of a beam shone into the user’s retina and reflected back from theeye as it moves The applications of this device for non-able-bodied users areobvious but other applications, for example for tasks where the user’s handsare occupoied, are possible Such an input device involves a set of very

different issues than the ‘haptic’ (or hand-operated devices) we have seen inthis chapter, and devices such as the eyetracker often require considertable

training to operate effectively An example of such a system is the EyeGaze

Development System : EyeGaze consists of an infrared video camera, an

infrared light source, an adjustable monitor with a bracket for attaching thecamera, and some speciased harware and software

Keyboards

The first typewriter was designed by William Austin Burt in the 1820’s Thedevice was a box made almost entirely of wood with type mounted on a metalwheel which the user operated by turning a knob wheel until the particularletter appeared, and then pulling a lever to imprint the letter on paper.

Devices such as these were large, unusable, and couldn’t really speed up theprocess of writing beyond the 30wpm which was possible by the fastest writersusing pens The first development toward the modern typewriter was that bythe inventors Sholes, Gidden and Soule in the 1860’s

The problem with this typewriter however, was that users could hit each keytoo quickly and as a result the keys jammed as the typist’s speed increased As

a result Sholes redesigned the keyboard of the typewriter to slow the typist

down, which produced in the now universally familiar QWERTY key layout(named after the left-hand third row keys) Although the new design certainlyremoved the key-jamming problem it wasn’t maximally convenient for typists,since to achive any speed with the keyboard, users were forced to move awayfrom the two-finger ‘hunt and peck’ style of typing to a method of typing usingall the fingers.

The ‘touch method’ as became to be known is attributed to Frank E

McGurrin, who taught himeslf how to use all of his fingers whilst not looking

at the keyboard and could thus increase his typing speed The development ofalternative ‘touch’ methods provided the basis for touch-typing contests in the1880’s, increased the sales of typewriters produced by manufacturers such asRemington, and as a result between 1905 and 1915 over 100 different

typewriter manufacturers appeared and by 1898, there were some 60,000female typists in the US, trained in many newly established typewriting

schools

This history of the typwriter is also the history of the modern keyboard

attached to an interactive computer which forms one of the main parts of thehuman interface to many systems Unfortunately, the QWERTY key layouthas been a source of considerable peoblems due to the need to learn a touch-typing method and due to the problems caused by the uncomfortable posturesthat users have to adopt when using the keyboard Accordingly, the majoreffort in terms of keyboard design has been to redesign the key layout and also

to redesign the physical features of the keyboard itself, in particular its shape.The remainder of this section looks at some of these alternative keyboarddesigns

The Dvorak Keyboard. In the 1930’sAugust Dvorak examined the Engishlanguage with the aim of redesigning the keyboard to speed typing up Hiskeyboard design increased accuracy by 50% and speed by 15-20% The

Dvorak keyboard removed the requirement that the user uses her weaker lefthand and weaker little fingers to perform most of the typing work whilst

underusing the stronger right hand and middle and index fingers The Dvorakkeyboard allows 70% of all typing to be performed using the middle or

‘home’ row – some 3,000 words can be typed on the home row as opposed to

120 on the home row of the QWERTY keyboard Most studies (Potosnak1986) confirm that the Dvorak keyboard is indeed faster than the QWERTYlayout, but some studies do not confirm the 15-20% speed increase figure,suggesting as small incease as 2%

The Alphabetic Keyboard A different keyboard layout is to place the keys in

alphabetical order (with a variety of ways of arranging the alphabet on thekeyboard) Studies of the use of alphabetic layouts verus QWERTY suggestthat in general those users experienced with a QWERTY keyboard were

significantly faster on the QWERTY layout, whilst inexperienced typists

showed little difference in speed on either after some initial training Studiessuch as these suggest that “other than as an academic exercise, the redesign ofthe QWERTY layout appears to be a fruitless effort (Potosnak, 1986: 479)

One-handed and portable keyboards Normally, typists use two hands as in

the QWERTY and Dvorak designs However there have been several keyboard

designs which allow the user to type using only one hand by chording

sequences of keys together using a small number of keys The MicroWriter is

one of the most popular versions of the one-handed chord keyboard.

The MicroWriter chord keyboard.

Another vesrion, The Data Egg (created by the Jet Propulsion Laboratory in

the US) is a small keyboard for one-handed text entry The device is similar to

the MicroWriter since it fits in the palm of the hand and does not need a

supporting surface It is claimed that users can reach speeds of up to 30 wpm

using this device Other one-handed input devices for text include the Twiddler

which has 18 keys which can be translated by chording into the full inputs

possible from a standard 101-key keyboard The Twiddler also has an internal

tilt sensor so that by changing the orientation of the device it can be used tocontrol a screen cursor

Virtual Keyboards There are also suggestions for keyboards which have no

keys, thus potentially alleviating the problems of configuring keyboards, strain

Display

Keyboard

Command Key

on the user’s hands and the problems of the wear of components Researchers

at IBM in Germany have suggested that it is possible to video-record users’hands as they type on a flat surface with a template and use software to

interpret the placement of fingers and to display the approriate character theuser typed

Concept Keyboards. Concept keyboards use flexible membranes under whichare touch-sensitive switches,and over which is mounted a template whichdefines regions of the keyboard as active for particular functions Conceptkeyboards are useful where there are small number of options to select

Ergonomic Keyboards It is now being recognised that both the QWERTY

keyboard layour and the flat, square design of the entire keyboard can bechanged to accomodate new knowledge about the most effective typing

methods and knowledge about the possible dangers of sustained use of

keyboards We will be looking one of the most common dangers associatedwith keyboard use, Repetitive Strain Injury in a later section, but there havebeen a number of keyboard redesigns which aim to to provide comfortableand safe keyboards for users

The reactive keyboard One approach to accelerating the speed at which a

keyboard can be used is not to redesign the keyboard layout itself, but to

provide software aids which allow a system to predict the user’s likely next

input via the keyboard The reactive keyboard (Darragh and Witten, 1992) is

essentially a sophisticated algorithm for predicting text entry which buildhistories of user’s previous entries and consults tables to determine the

likelyhood of some entry apperaring again This approach is not useful for fast,

skilled typists, but is likely to be useful for novice or reluctant typists, or fornon-able bodied users.

Other factors in keyboard design There a number of other factors in kayboard

design, in addition to keyboard layout, which we can briefly mention, sauch as

keyboard height and slope (a slope of around 15o is suggested to be the most

effective to reduce muscular strain); keyboard size (smaller keyboards are

more difficult to type on effectively due to reduced key size, and larger

keyboards with more keys provide the difficulty of inceased search time for a

particular key); keyboard profile (the relative angles of different keys on the keyboard – key arrangements can be dished, sloped or stepped); key size and

shape (most keys are best sized around 0.5in2, and the key centres distanced at

0.75 inches); key force and travel (most studies suggest a confortable value for

the force required to puish a key is 1-5 ounces with 0.05 to 0.25 inches oftravel and in general keyboards with low values for key travel and force arepreferred, but keys should provide some tactile feedback in the form of

‘breakaway’ - a gradual increase in force required to activate the key followed

by a sharp decease in force to complete the key travel movement); feedback

(some keyboards and controllers allow for auditory clicks with each keypress,which can reduce errors when the user cannot see the keyboard) Other

variables in keyboard design (reviewed by Potosnak, 1988) include rollover

(the ability of a keyboard to store each keypress in the correct sequence –without rollover rapid typing will result in the keypresses being procesed in the

wrong order); buffer length (the ‘typeahead’ feature on mosy keyboards) and

key-repeat (the ability to hold a key down to insert several of the same

characters) These features can all be manipulated in oredr to change thecharacteristics of the keyboard

Other Input Devices

Bar-code readers. Bar-code readers are hand-held or surface-mounted deviceswhich take reflectivity mesurements from paper baessd bar codes now found

an almost every manufactured object The device is essentially a wand sewiped

across the printed code which is translated into data fed to a host programsuch as a database Although quite unlike some of the input devices we havelooked at in this chapter, the wand of the bar-coide reader shares many

characteristics with lightpens, although bar-code readers are obviously notused for the input of user-defined data, and not used for direct text-entry orpositioning tasks

Fingerprint readers. Although not strictly an input device, fingerprint readersoperate in the same way as bar-code readers by inputting predefined data – inthis case the data is a user’s fingerprints These devices, which are based ondigitization of images taken of a finger placed on a reading window, includesoftware which takes the digitsed image of a fingerprint and extracts several

‘feature points’ which are stored as a unique identity Systems such as theStartek FC100 when used for tasks where the user needs to be identified as alegitimate person, there is a false rejection rate of less than 1%

Neural Control One of the more unusual methods for input involves not a

physical device, but the monitoring of the user’s mental actions Although inits early stages, and at the centre of some controversy, several research teamshave developed systems that allow users to move a cursor by mental actionalone, or which allow users to type (although very slowly) by spelling outwords in their minds Techniques have also been developed to determineusers’ likely movements of a joystick by monitoring brainwaves These

techniques work by monitoring electrical bain activity using electrodes to the

scalp Attempts to develop ‘thought input’ for computers began in the 1970'swith the ‘biocybernetics’ program financed by the United States Defense

Department: these developments had the goal of enabling a computer to

determine the state of mind of a fighter pilot Another approach is to

‘discipline’ the brain to emit signals which can be interpreted by a system in the

form of mu waves, rhythmic signals emitted by the brain Large mu wave

amplitudes are translated into upward movements of a display cursor and lowamplitudes into a downward movement In many cases, thinking aboiut

images (weightlifting for example) helps in moving the cursor (Wolpaw et al,1991) At present, it is not clear whether this form of input is at all useable inhuman interfaces

Some comparisons between input devices

We have commented on some of the results of comparative studiues of inputdevices in the individual sections for each type of device However Milner(1988), in a survey of input devices, revealed that research studies of the useand applicability of devices provided contradictory results, not only in terms ofthe use of different devices for the same experimental task, but between

different varieties of the same device (for example directional versus isometricjoysticks)

In general experimental studies have considered the following generic tasks:data entry, object selection, object manipulation and drawing/tracking tasks.Milner’s survey suggested a number of things Firstly, selecting the fastestinput devucde is dependant on the task for which it is to be used For highresolution applications (where there are many small objects represented on adisplay) using direct devices such as the touchscreen may be problematic sincethe user may need to extremely accurate, something generally difficult with

the end of a finger In low-resolution applications (where there are few objectswhich are spread out), the touchscreen is a very fast means of input However,since speed is not the only, and often not the most important, issue in choosing

an appriorate input devuce, other factors play an important part One of these

is accuracy, since it is clear that users may prefer, and ultimately work moreeffectively, when the device they use allows accurate positioning, selection andmanipulation of screen objects Yet, as Milner suggests, a device may be both

the most accurate and the fastest on a particular task, and this underlines the

importance of considering in some detail the task for which an input device is

to be used For example, direct input devices such as the touchscreen are bothfastest and most accurate for some applications where the task is short (thusnot fatiguing the users), and low-resolution (there are few objects to

manipulate or select); indirect input devices on the other hand are most

suitable for quick and accurate secetion and manipulation in high-resolutionapplications over extended periods of time (devices such as the mouse andtrackball are minimally fatiguing since the user’s hands rest on the desktop or

on the device)

Other attempts to draw comparisons beween input devices have found similarresults For example, the survey in Greenstein and Arnaut (1988) compared

devices in terms of criteria such as eye-hand coordination requirements, input

resolution, and flexibility of placement .

In general they suggest that for tasks such as target acquisition (where a user must place a screen cursor on or inside a stationary screen object), menu

selection (where a number of target are presented for choice) or text selection,

direct pointing devices such as the light pen or touchscreen are fastest, due tothe high degree of eye-hand coordination posessed by users and users’

familiarity with the notion of pointing at objects Among indirect pointingdevices, the trackball, mouse and the graphics tablet (in absolute mode) do notdiffer in positioning speed or accuracy, but the mouse may be slightly fasterdue to lack of resistance, whilst the trackball may be more accurate due to itsgood tactile feedback, users’ abilities to spin the trackball towards a target,and the abilities of users to manipulate the fingers of the hand accurately In

tasks such as moving target tracking, devices such as the trackball and joystick

are good since they again exploit users’ hand coordination and ability to

respond quickly to changes in direction

Simple guidelines for input devices

We can, looking at the kinds of research findings and issues discussed in thischapter, suggest some simple guidelines for input devices (1) In general anyselection of an input device needs to be based on not only the characteristics ofthe device itself, but on the intended application, the users, and the tasks

which the users will perform (2) The situation becomes more complex whenthere is not a clear choice of a single input devuce, and one reccomendationhere would be to consider the use of several input devices for the differenttasks the user would perform with a single application For examople,

keyboards used with computers such as the Apple Macintosh provide bothmouse input and cursor keys, even though a mouse can be used to perform allthe neccessary operations on both graphical objects and and on text Here thedesign provides the user with the widest possible chopice of input methods (3)Even more problematic is the case where there are several varieties of oneinput device, for example a joystick and its variants, and the recommendationhere is to perform detailed evaluations of different input devices against thetasks to be performed (4) Devices such as the touchscreen might be preferredwhere the users do not have computer expertise, where users will use the

application infrequently, where the tasks requires no text input and wherethere are a small number of choices represented by large tartgets on the

display, and where implementation constraints permit General guidlines such

as this can be supplemented by specific advice drawn from empirical studies oftouchscreens, which suggest that the applications needs to provide largespread out targets and provides visual feedback (perhaps by highlighting

choices on the screen) (5) Finally, it would seem desirable to minimise theamount of work a user has to do to operate input devices This is most

apparent in the use of input device combinations such as the mouse and

keyboard: users often find the switch between using the mouse to navigatearound the display and the use of the keyboard for typing tasks irritating andconfusing Although little can be done about this (since it would be impossible

to integrate a mouse and keyboard effectively) the software design can takeinto account the need for minimal switching between devices For example, inthe dialogue box below the applilcation, a wordprocessing package called

Nisus, provides a set of key equivalents for performing the task of selecting

buttons which might usually have required the user to move her hands keyboard to use the mouse Providing this kind of alternative in the design canincrease both accuracy and speed