2006-02-Fitts-qe-v2

7/21/2019 2006-02-Fitts-qe-v2

http://slidepdf.com/reader/full/2006-02-fitts-qe-v2 1/61

Human Factors and Fitts’ Law

Ken Goldberg, IEOR and EECS, UC Berkeley

7/21/2019 2006-02-Fitts-qe-v2


What is Ergonomics?

ro!" Wo#ciech $astr%ebo&ski

in oland in '()*+

rom t&o Greek &ordsErgon meaning work

and

Nomos meaning principles or laws

Ergonomics = The Science of Work Ergonomics = The Science of Work

7/21/2019 2006-02-Fitts-qe-v2


Common -e!initions

.Ergonomics is essentially !itting the &ork/lace to

the &orker" 0he better the !it the higher the le1elo! sa!ety and &orker e!!iciency"2 Fitting the Task to

the Human 3 Grand#ean '445

.Ergonomics remo1es barriers to 67ality,/rod7cti1ity and h7man /er!ormance by !itting

/rod7cts, tasks, and en1ironments to /eo/le"2ErgoWeb.com

What is Ergonomics?

7/21/2019 2006-02-Fitts-qe-v2


87man actors

What Is Human Factors? 0he !ollo&ing de!inition &as ado/ted by the

International Ergonomics 9ssociation in 97g7st

:555+

Ergonomics (or human factors) is the scientific

discipline concerned with the understanding of

interactions among humans and other elements of a

system, and the profession that applies theory,

principles, data, and other methods to design in

order to optimie human well!being and o"erall

system performance.

7/21/2019 2006-02-Fitts-qe-v2


87man actors and Ergonomics

; Britain - 0he Ergonomic Society &as !ormed in

'4): &ith /eo/le !rom /sychology, biology,

/hysiology, and design"

; United States - 0he 87man actors Society&as !ormed in '4)*" In the US <h7man !actors

engineering< &as em/hasi%ed by the US military

&ith concentration on h7man engineering and

engineering /sychology"

7/21/2019 2006-02-Fitts-qe-v2


7/21/2019 2006-02-Fitts-qe-v2


!rom =ike =andel, #aking $ood Time >C= B7lletin 1ol" ( no" :,

Cali!ornia =7se7m o! hotogra/hy, UC Cali!ornia, Ri1erside, '4(4

7/21/2019 2006-02-Fitts-qe-v2


Gilbreth @ideo

7/21/2019 2006-02-Fitts-qe-v2


8a&thorne E!!ect

Worker St7dy >'4:* A '4: o! the 8a&thorne lant

o! the Western Electric Com/any in Cicero, Illinois"

ed by 8ar1ard B7siness School /ro!essor Elton

=ayo+ E!!ect o! 1arying light le1els on rod7cti1ity"

7/21/2019 2006-02-Fitts-qe-v2


7/21/2019 2006-02-Fitts-qe-v2


7/21/2019 2006-02-Fitts-qe-v2


Measure of Man, Henry Dreyfuss, !"#

7/21/2019 2006-02-Fitts-qe-v2


Occ7/ational Sa!ety and 8ealth 9dministration,

>OS89, '4*5, &&&"osha"go1

7/21/2019 2006-02-Fitts-qe-v2


De7tral ost7re !or Com/7ter

Use

9d#7st the seat height

so 7//er arms hang

1ertically, elbo&s bent

abo7t 45 degrees,

sho7lders relaed and

&rists !airly straight

osition the monitor abo7t an

armFs length a&ay directly in

!ront o! yo7" 0he to/ o! the

screen no higher than eye

le1el >Unless the 7ser &ears

biA!ocal glasses

Use a doc7ment

holder close to the

monitor rather than

laying /a/ers !lat

=o7se sho7ld be net to keyboard

both at a height e67i1alent to the

7serFs seated elbo& height

Knees com!ortably bent &ith

!eet resting on the !loor" I! the

chair is raised so the keyboard

height e67als elbo& height, 7se

a !ootrest "

9d#7st

the back

rest to

/ro1ide

!irm

s7//ort

to the

small o!

the back

7/21/2019 2006-02-Fitts-qe-v2


a7l =" itts, '4)

itts connected the s/eedAacc7racy

tradeo!! o! choice reaction times to

reaching mo1ement tasks

7/21/2019 2006-02-Fitts-qe-v2


7/21/2019 2006-02-Fitts-qe-v2


ittsF .a&2

; 0 % a & b log ' ( )

W W

**

W W

arameters a, b e/erimentally determined

7/21/2019 2006-02-Fitts-qe-v2


9lternati1e+ S67areAroot a&

; ittsF ogarithmic a& is not deri1ed 7sing

biomechanics and kinematics

; We deri1e a .S67areAroot2 a&+

based on : sim/le ass7m/tions

7/21/2019 2006-02-Fitts-qe-v2


9ss7m/tion '

9cceleration > is /iece&ise constant x

7/21/2019 2006-02-Fitts-qe-v2


9ss7m/tion :

9cceleration is /ro/ortional to target &idth

Wider targets are easier to reach

⇒ larger accelerations /ossible

7/21/2019 2006-02-Fitts-qe-v2


O/timal Control

; Gi1en a bo7nd on ,

astest &ay to reach a target is to 7se

.bangAbang2 control

|| x

T /2 T

7/21/2019 2006-02-Fitts-qe-v2


O/timal BangABang Control @elocity

s = T /2 T

osition at time 0+

7/21/2019 2006-02-Fitts-qe-v2


O/timal BangABang Control osition

A

2

A

s = T /2 T

7/21/2019 2006-02-Fitts-qe-v2


O/timal Binary 9cceleration =odel

; Use 9ss7m/tion : to s/eci!y a single

!orm7la that relates A, W , and T

; 9ss7m/tion : 8y/othesis+

=aimal acceleration set by the h7man is

/ro/ortional to target &idth

>Wider targets /ermit larger accelerations

7/21/2019 2006-02-Fitts-qe-v2



; 9ss7me+

; O/timal bangAbang model+

; 9dd reaction time a+

; arameters a,b set !rom e/erimental data

7/21/2019 2006-02-Fitts-qe-v2


irst =o7se >-o7glas Engelbart and

William English, '4H

7/21/2019 2006-02-Fitts-qe-v2


irst =o7se atent >Engelbart

>Sh7min hai, IB= 9lmaden Research Center

7/21/2019 2006-02-Fitts-qe-v2


=odern In/7t -e1ices

7/21/2019 2006-02-Fitts-qe-v2


ittsF a& $a1a 9//let

7/21/2019 2006-02-Fitts-qe-v2


E/erimental 0ests

ied

Rectangle 0est

@ariable

Rectangle 0est

Circle 0est

8omogeneo7s

C7rsor =otions

8eterogeneo7s

C7rsor =otions

7/21/2019 2006-02-Fitts-qe-v2


91ailable -ata

; Original data set+ J ::: 7sers !or !ied rectangle tests

J :HH 7sers !or 1ariable rectangle tests

J '(4* 7sers !or circle test J User did not com/lete all trials ⇒ Remo1ed

J User has o7tlier /oints ⇒ Remo1ed

; inal data set+

J 'H5 7sers !or !ied rectangle tests J '44H 7sers !or 1ariable rectangle tests

J ')H' 7sers !or circle tests

7/21/2019 2006-02-Fitts-qe-v2


=odel arameters

; arameter set 7sing leastAs67ares linear

regression !or each user

; 91erage /arameters o1er all 7sers+

7/21/2019 2006-02-Fitts-qe-v2


0y/ical User

7/21/2019 2006-02-Fitts-qe-v2


=odels

&ith

o&estR=S Error

7/21/2019 2006-02-Fitts-qe-v2


E!!ect Si%e

; =ean signed di!!erence in R=S errors bet&eenthe S67areAroot a& and ittsF ogarithmic a&,as a /ercent o! the mean R=S error !or ittsFogarithmic a&, &ith 4) con!idence inter1als

S67areAroot

a& better

ogarithmic

a& better

7/21/2019 2006-02-Fitts-qe-v2


WebABased ittsF a& -emo

&&&"teleAactor"netL!ittsL

7/21/2019 2006-02-Fitts-qe-v2


87man actors and Ergonomics

; Britain - 0he Ergonomic Society &as

!ormed in '4):

; United States - 0he 87man actors

Society &as !ormed in '4)*"

7/21/2019 2006-02-Fitts-qe-v2


Human Factors and Fitts’ LawKen Goldberg, IEOR and EECS, UC Berkeley

7/21/2019 2006-02-Fitts-qe-v2


C7/stacking @ideo

7/21/2019 2006-02-Fitts-qe-v2


7/21/2019 2006-02-Fitts-qe-v2


7/21/2019 2006-02-Fitts-qe-v2


O7tline

; ittsF a& Introd7ction

; Kinematics =odels o! ittsF 0ask

J Symmetric Binary 9cceleration =odel

J 9symmetric Binary 9cceleration =odel

; ittsF 0ask in 8CI

; WebAbased E/eriments

7/21/2019 2006-02-Fitts-qe-v2


Choice Reaction 0ime 0ask

$" =erkel, '(()+ Stim7li 1,…, N e67ally likely"

T R = a + b log2 N

1 2 3 4 5 6 7 8

Stim7l7s+ ',M,+ Res/onse+ ',M,+

4

7/21/2019 2006-02-Fitts-qe-v2


In!ormation 0heory

; Base : logarithm o! the n7mber o!

alternati1es is a meas7re o! in!ormation

D7mber o! bits = log2 N

Corres/onds to the a1erage n7mber o! yesLno

67estions re67ired to identi!y correct stim7l7s

; In eam/le+

log2 8 = 3 bits

7/21/2019 2006-02-Fitts-qe-v2


ittsF In!ormation 0heory 9//roach

; -e!ine .in!ormation2 encoded in a reaching

mo1ing task

; In!ormation transmitted I in a res/onse isa meas7re o! the red7ction in 7ncertainty

7/21/2019 2006-02-Fitts-qe-v2


In!ormation 0ransmitted

1 2 3 4 5 6 7 8

7−8

; N /ossibilities be!ore e1ent+ (; N /ossibilities a!ter e1ent+ :

; In!ormation transmitted+ -log2(2/8) = 2 bits

; Uncertainty+ ' bit

555 55' 5'5 5'' '55 '5' ''5 '''

7/21/2019 2006-02-Fitts-qe-v2


-iscrete 1s" Contin7o7s Choice

1 2 3 4 5 6 7 8

555 55' 5'5 5'' '55 '5' ''5 '''

Start

osition

0arget

9m/lit7de A Width W

7/21/2019 2006-02-Fitts-qe-v2


ittsF orm7lation

D7mber o! /ossibilities a!ter res/onse+ W

D7mber o! /ossibilities be!ore res/onse+ 2A

In!ormation transmitted nde of *ifficulty

W b ti l ti ! itt F

7/21/2019 2006-02-Fitts-qe-v2


Weber raction orm7lation o! ittsF

0ask

; Wel!ord, '4H(

; Weber !raction+ W /(A+0.5W)

Start

osition

0arget

9m/lit7de A Width W

7/21/2019 2006-02-Fitts-qe-v2


Shannon orm7lation o! ittsF 0ask

; orm7lation based on ShannonFs

0heorem

PI" Scott =acKen%ie '44:Q

; Shannon orm7lation !or ittsF 0ask+

C In!ormation ca/acity o! comm7nication channel

B channel band&idth

S signal strength

N noise /o&er

7/21/2019 2006-02-Fitts-qe-v2


O7tline





; ittsF 0ask in 8CI


7/21/2019 2006-02-Fitts-qe-v2


O7tline





; ittsF 0ask in 8CI


7/21/2019 2006-02-Fitts-qe-v2


@elocity ro!iles o! ittsF 0ask

'" @elocity /ro!iles are asymmetric

:" 9symmetry increases as target &idth decreases

" 9m/lit7de has relati1ely little e!!ect on asymmetry

C"" =acKen%ie et al,

'4(*[ ]

7/21/2019 2006-02-Fitts-qe-v2


9symmetric Binary 9cceleration =odel

9ss7me+

ercent time accelerating increases &ith W

9symmetric 1elocity /ro!ile+

s T

9cceleration is constant a

-eceleration set so distance

A reached at time T

7/21/2019 2006-02-Fitts-qe-v2


9symmetric @elocity ro!ile

s T

7/21/2019 2006-02-Fitts-qe-v2


9symmetric =odel osition

s T

7/21/2019 2006-02-Fitts-qe-v2



; 9dd reaction time a+

; arameters a,b set !rom e/erimental data

; Same !orm7la as O/timal Binary 9cceleration =odel-i!!erent ass7m/tions and deri1ations

7/21/2019 2006-02-Fitts-qe-v2



@ e l o c i

t y v

=o1ement 0ime T

a

7/21/2019 2006-02-Fitts-qe-v2



@ e l o c i

t y v

=o1ement 0ime T

s

7/21/2019 2006-02-Fitts-qe-v2


O7tline





; ittsF 0ask in 8CI

;WebAbased E/eriments

7/21/2019 2006-02-Fitts-qe-v2


=o7se

; irst mo7se >'4H+

-o7glas Engelbart and William English

2006-02-Fitts-qe-v2

Documents