Ergonomics

1
Ergonomics
Ergonomics Human Factors
Focus Interaction of humans with
devices Objective To understand, evaluate,
and thereby, to improve the interface between
the human and the device
2
Outline
1. Examples of product design related to
ergonomics issues 2. Case Study digital images
and JPEG 3. Methodology and tools useful for HF
3
Example 1. Office desk and Chair
Question How do we decide the height of the desk?
Depends on (a) the height of the chair (b)
the size of the person who will use them
4
Example 1. Chair ergonomics..
(iv) The arm rests height ? elbow height at rest
(v) Backrest lumbar support 15-25 cm above
seat level
(i) Seat pan length calf clearance (gt 5cm) to
95 women
(iii) The seat pan angle 6?
(ii) The chair height contact lower thigh
with both feet on floor
5
Example 1. Chair ergonomics...
(i) Seat pan length calf clearance (gt 5cm) to
95 women
(ii) The chair height contact lower thigh with
both feet on floor
IMPLICATIONS 1. Need for adjustability 2.
Design of a good chair depends on the
statistics of the users
6
Example 1. Chair ergonomics user statistics
Design of a good chair depends on the
statistics of the users
USA Germany Japan Netherlands
Males 175.5 174.5 165.5 182.5
Females 162.5 163.5 153.0 169.6
Problem 1. What statistics are sufficient? Proble
m 2. How to collect such statistics? Problem 3.
Statistics are time dependent e.g. height of
urban Chinese males increased by 6 cm over the
last 20 yrs
7
Example 2. Keyboard design
Extended periods of use of a computer in the
wrong posture ? repetitive stress injury (RSI)
8
Example 2. Keyboards Carpal Tunnel Syndrome
why compression of the median nerve as it
enters the hand. symptoms numbness of thumb
and fingers, pain along the median nerve
including hand, wrist, elbow, weakness of
thumb. treatment rest, surgery main cause
flexed or extended wrists when keying!
9
Example 3. How to turn on the shower
Non-intuitive design ? wasted time/user-discomfort
tub-faucet
Pull down this ring to turn shower on
10
Example 4. Toilet flush (airport)
Non-intuitive design ? discomfort (for next
user?!)
11
Example 5. Is the water too hot? Too cold?
12
Ergonomics
(i) Understanding of human physiology (ii)
Understanding of human psychology (iii)
Statistical data about populations Goal --
Improve design (more efficient) -- Improve
design (safety, comfort)
13
Ergonomics Case Study Improve design
Digital Image Files
Digital Cameras (digicams) ? pictures in a
digital memory
What is the data composed of ? The RGB-pixel
model
14
Digital image files pixels
An image of a lion fish What is the image made
of ?
15
Digital image files pixels
pixel PICture ELements
16
Digital image files The RGB model
What is a color?
? Store the wavelength, intensity at each
pixel Problem ? (Technical how to display?)
The primary color theory any color ?
combination of primary colors (R, G, B) ?
at each pixel, decompose into primary color
values, store R, G, B.
17
Digital image files The RGB model
R Red level 8 bit number 1
byte G Green level 8 bit number 1
byte B Blue level 8 bit number 1
byte
Original lionfish file 1920x2560 4,915,200
pixels ( 5 Megapixel digicam) 1 Byte per color
per pixel ? 4,915,200 x 3 x 1 14,745,600 15
Mbytes
PROBLEMS 1. Large memory requirement 2. Slow
transfer speed
? need for COMPRESSION
18
Digital Image Files compression
Strategy 1. Compress data without losing any
information
Example run-length-encoding
raster model each pixel 0 or 1
run-length-encoding 0203,1403,
203x191 pixels
LOSSLESS compression ? No need to understand
human vision
19
Digital Image Files compression
Strategy 2. Compress data by throwing away parts
that we cannot see ? Needs a good understanding
of human vision
How we compress image files depends on how we
see images ? Understanding of human vision ?
more efficient compression technique
20
Digital Image Files compression
Uncompressed BMP (bitmap) 14 MByte
Lossless compression PNG 7.9 MByte
Lossy compression JPG (JPEG) High quality
3.67 MByte 0.8 quality 0.83 MByte 0.6
quality 0.5 MByte 0.2 quality 0.2
MByte
http//iesu5.ieem.ust.hk/dfaculty/ajay/courses/iee
m101/lecs/hf/lionfish.html
21
JPG How do we see
Do you believe what you see?
The Koffka ring
22
JPG How do we see..
Do you believe what you see?
23
JPG How do we see
Do you believe what you see?
24
JPG How do we see -- the eye
RODS scotopic vision (in dark) only on in
darkness only distinguish lightness
CONES photopic vision
25
JPG How do we see -- the eye..
Trichromacy theory different intensities of R-
G- B- cones allows brain to estimate
frequency of the spectral light striking a
zone
26
JPG How do we see -- the eye...
Hue discrimination ability to distinguish
between two different wavelengths of
light Lightness discrimination ability to
distinguish between two different levels of
lightness Lightness grey level
Lightness discrimination is MUCH more sensitive
than Hue discrimination
Reasons (a) lightness is estimated by (RG)
response of cones, and also from RODS (b) many
more rods than cones
27
JPG How do we see -- the eye.
Webers law Our ability to discriminate levels
of lightness depends n the ratio of lightness
Shades that are in geometric series look
equally spaced in lightness.
geometric
arithmetic
28
JPG How do we see -- the eye..
Hue discrimination vs Lightness discrimination
29
JPG How to eliminate what we cannot see?
1. Intensity changes are much more significant
than hue changes
2. Intensity change steps are logarithmic
PROBLEM Technically, it is easier to handle R-
G- B- shades
Why ? (a) Recording instruments (digicams)
sensors can sense colors (b) Display
instruments can handle RGB values better
30
JPG How to eliminate what we cannot see..
must be invertible mapping
Converting R G B ? Y Cb Cr
Luminance (lightness)
Chrominance (chroma) components
31
JPG How to eliminate what we cannot see
JPEG compression Step 1. Convert RGB data into
YCbCr data Step 2. Sub-sample and quantize Cb and
Cr data Step 3. Compress resulting stream
(run-length encoding)
file-size reduction
Higher compression Step 2 ? sub-sample more,
sub-sample Y also
32
JPG Details -- How to Sample, Sub-sample?
Break the image into tiles of NxN
pixels. Store data of each tile
Example
7
10
7
6
7
7
8
4
2x2 tile 4 values ? average 7 ? combine tiles
into block with value 7.
33
JPG Details -- How to Sample, Sub-sample..
Sub-sampling and quantization basics
How Fourier analysis works for 1-D functions
34
JPG Details -- The Discrete Cosine Transform
DCT function
35
JPG Details -- Quantization and encoding
36
JPG Conclusions
1. Understanding of human sensory system is
important for better product designs 2.
Levels of adjustability useful for variations
among users -- older person with poor sight
might prefer higher compression
NOTE You dont need to know details of DCT, and
the exact mathematics of the transformation Impor
tant ideas sub-sampling ignore some data, or
replace multiple values by the average quantizati
on instead of storing exact value, round up/down
to nearest step
37
Methods and tools in Ergonomics
Product design must consider ease of use,
comfort and safety in use
Optimization for ergonomics ? understanding how
human body works
Design parameters f( physical measurement)
e.g. Chair seat height
The study of measurement of human body is called
anthropometry.
Human size variations ? need to know the
statistics of anthropometric data
38
Ergonomics methodology
1. Optimal product is designed based on
anthropometric measurements
2. Statistical variations of expected users are
estimated
3a. Design is modified to allow critical
parameters to be adjusted by user so as to
fit the individual need
or
3b. Size variations are provided to cover
estimated market (e.g. shoe sizes)
Next Quality control