Transcript Diapositiva 1

Measurement Systems and
Applications
Mariolino De Cecco
Antonio Selmo
Michele Confalonieri
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Force Panel
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
The instrument employed
Design and
purpose of the FP
[a couple of
frontal lessons]
How its
electronics work?
[Antonio Selmo]
Measurements for
Diagnostics
[study and identification
of the HTF]
Principles of human
diagnosis and
rehabilitation
[doct Guandalini
and Tomasi]
Rehabilitation and its
quantification (measure of
indexes of performance)
[development of serious
games with processing]
Achronims:
Fprce Panel – FP
Human Transfer Function - HTF
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Logic of the Course
Application: identification of
human transfer function for
dexterity assessment and
serious games for
rehabilitation
Course:
- frontal lessons (few)
- exercitation with matlab
- building of the application
with Processing
- visit to a clinic
Method:
- study of touch technology
- study of signal analog
elaboration and conversion
- acquisition of human data
- signal processing
Exam:
- oral discussion
- homework
Material of the course can be found at http://www.miro.ing.unitn.it
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Organization and final examination
• Project number: 247765
• Project acronym: VERITAS
• Project full title: Virtual and Augmented
Environments and Realistic User Interactions
To achieve Embedded Accessibility DesignS
• Site: http://www.veritas-project.eu/
• Starting date: 1 January 2010
• Duration: 48 Months
VERITAS is an Integrated Project (IP) within
the 7th Framework Programme, Theme FP7ICT-2009.7.2, Accessible and Assistive ICT
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
VERITAS Project
Virtual and Augmented Environments and Realistic User Interactions To
achieve Embedded Accessibility DesignS
OBJECTIVE develop virtual humans for accessibility
design in virtual simulation of ICT and non ICT products
Method
SP1: development
of virtual users
SP2: development
of the simulation
platform
SP3: validation
UNITN activities
 Modeling
 Measurement
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Design and purpose of the FP - VERITAS Project
• Systems developed:
– Garment based Motion
Capture: Ga-MoCap
– Multi-axis load cell
– Force Panel
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
VERITAS Project - measurement systems developed
• Diagnostics by human dexterity estimation
• Requirements:
–
–
–
–
Measure position of the touch
Measure the applied force
Store the above for path and trajectories
Accuracy and sampling frequency able to reveal
human dexterity
– Visual feedback
Idea: “Force Panel”
– Display LCD
– Touch panel
– Force transducers
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Design and purpose of the FP - requirements
Stato
dell’arte
Touch screen
LCD and cameras
Graph tablets
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Design and purpose of the FP - state of the art
Stato
dell’arte
joystick
Aptics devices and virtual
reality
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Design and purpose of the FP - state of the art
Force Panel
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Design and purpose of the FP - Design
Schema
della
struttura
Horizontal use
Vertical use
(disposizione dei vincoli)
= perpendicular to LCD
= parallel to LCD
3 load cells
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Design and purpose of the FP - Design
Tipologie di vincolo considerate
– Spherical joints
– wires
– wires
 2 solutions !
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Design and purpose of the FP - Design
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Design and purpose of the FP - Design
Variante con teste a snodo
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
mechanical design: spherical joint
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
mechanical design: wires
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
mechanical design: wires constraints
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
mechanical design: wires constraints
Requirements:
• No buckling
• No finger force through the blues
• Diameter 1 mm
F
 Wires with load cell
– Length 10 mm (due to physical constraints)
 wires
– Length 30 mm (FEM analysis → finger force < 0.3 %)
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
mechanical design: wires length
horizontal
Vertical
Worst case for buckling
Worst case for
force leakage
Lunghezza dei fili
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
mechanical design: wires length
• Critical length ≈ 65 mm
• Force lost:
Caso
Errore lettura forza [%]
Worst case horizontal, l = 10 mm
2.41
Worst case vertical, l = 10 mm
2.50
Worst case for buckling, l = 10 mm
5.32
Worst case for buckling, l = 20 mm
0.73
Worst case for buckling, l = 30 mm
0.22
Worst case for buckling, l = 30 mm, e = 0.5 mm
0.23
Worst case for buckling, l = 30 mm, e = 1 mm
0.27
• Moments ≈ 10-3 – 10-6 Nm
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
mechanical design: wires length
Touch panel
LCD
VGA
Force sensors
PC
2 x DIGITAL OUT
Force
Conditio
ning
USB
µController
2 x ANALOG IN
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
System design: overall layout
3 x ANALOG IN
A touch screen is a 2-dimensional sensing device constructed of 2 sheets of material
separated slightly by spacers
A common construction is a sheet of glass providing a stable bottom layer and a
sheet of Polyethylene (PET) as a flexible top layer
The 2 sheets are coated with a resistive substrate, usually a metal compound called
Indium Tin Oxide (ITO). The ITO is thinly and uniformly sputtered onto both the
glass and the PET layer
Tiny spacers are placed between the 2 sheets in order to prevent false touch
When the PET film is pressed the two resistive surfaces meet. This position can be
read as illustrated in next slides
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Touch position measurement circuit
Indium tin oxide (ITO, or tin-doped indium oxide) is a
solid solution of indium(III) oxide (In2O3) and tin(IV)
oxide (SnO2), typically 90% In2O3, 10% SnO2 by
weight. It is transparent and colorless in thin layers while
in bulk form it is yellowish to grey. In the infrared region
of the spectrum it is a metal-like mirror.
Indium tin oxide is one of the most widely used
transparent conducting oxides because of its two chief
properties:
- its electrical conductivity
- optical transparency
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Touch position measurement circuit
5V
y+
L
R0
Ly
L
R0
y
L
VS - sense
x
R0 
RT
L
GND


VS  5
R0

R0
y
L
y
Ly
 R0
L
L

5
y
L
R0 is the total resistance of the substrate
RT is the touch resistance
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Touch position measurement circuit
Reading the x coordinate is similar
Note that in the equivalent circuit there are
capacitive effects that lead to a certain
delay that usually is less than 10 ms
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Touch position measurement circuit
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Electronics scheme
Circuito di condizionamento
Output
cella
Cella di
carico
Alim.
cella
PreAmplificazione e
Filtro analogico
Amplificazione
passa basso
Bessel V ordine
Alim. condiz.
(± 15 V DC)
DC / DC
Alimentazione
(12 V DC)
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Conditioning unit
Shift del segnale
(μC legge solo
tensioni positive)
μC
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Conditioning unit
Full Wheatstone bridge
Bessel lowpass 5°
order
Gain and offset
adjustment: Arduino
reads only positive
voltages (0 – 5 V)
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Conditioning unit
No sovraelongation
50
s
Tempo di assestamento
500 s
20s
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Conditioning unit
RISPOSTA
TEMPORALE AL
GRADINO DEL
FILTRO PASSA BASSO
DEL 5° ORDINE
(Risposta di Bessel
con frequenza di
taglio 1 kHz)
Inizializzazioni
(costanti, variabili e pin μC)
Alimentazione lungo X
• Arduino UNO
Attesa assestamento circuito RC
Lettura valore X
X
Alimentazione lungo Y
Attesa assestamento circuito RC
Lettura valore Y
Lettura valore F1
Lettura valore F2
Lettura valore F3
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
sw design: embedded system
Y
F1
F2
F3
PC
t [ms]
TASK 1 – Arduino
TASK 2 – PC
TASK 3 – PC
0
PIN per lettura X
Gestione delle
immagini in
relazione alle
risposte di chi
sta di fronte al
pannello
t assest. RC
5
5.1
8
10.1
10.2
15.2
Read X
PIN per lettura Y
Read F1, F2, F3
nel t assest. RC
Read Y
Invio dati
via RS-232
Ricezione dati ed
applicazione
modello di
taratura
25.2
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
sw design: task analysis
2 task paralleli su PC:
- uno segue l’Arduino,
- uno è indipendente
Acquisition
•
•
•
•
•
Inizializzazioni
Lettura RS-232
Riconoscimento stringa
Distinzione segnali
Eventuali operazioni di
filtraggio
• Operazioni di taratura
Elaboration and graphical
output
• Inizializzazioni
• Attesa inizio
acquisizione
• Calcolo della tara del
sistema
• Elaborazione (gestione
delle immagini in
relazione all’interazione
con l’utente)
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
sw design: two independent tasks on the PC
1. Touch position (resistive touch)
2. Force (load cell)
3. Touch position (load cell)
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Calibration
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Force Panel - calibration
Measurement model:
• Linear behaviour on both directions
• No interactions
 uncoupled modelling
xa [bit]
ya [bit]
xm = f (xa)
ym = f (ya)
xm [pixel]
xm  mx xa  qx
ym [pixel]
y m  m y ya  q y
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
calibration: position
• Non-linear behaviour for:
– Touch position
– Force
3
FT   Fi
i1
FT [bit]
xm [pixel]
ym [pixel]
Fm[N] = f (Fa , xm , ym)
Fm [N]
2
2
F

a

bx

cy

dx

ex
y

fy
 m
m
m
m
m m
m  gFT  hFT xm  iFT ym  jFT xm ym
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
calibration: force
Fm  a  bxm  cym  dxm2  exm ym  fym2  gFT  hFT xm  iFT ym  jFT xm ym
For N measurements made in
different positions and with different
forces it is possible to build the
following linear relation
a 
 
b 
c 
 
d 
2

1 xm1 ym1 xm1

Fm1 
... ... ... ... ... FT1 xm1 ym1

 e   
...
... ... ... ... ... ...
...
... ...
 f   ... 
2

1
x
y
x
... ... ... ... ... FTN xmN ymN 
FmN 


mN
mN
mN
 g 
 
h 
 
Than, with a pseudoinverse computation, the 10
i 
coefficients of the polynomial model are estimated
l 
 
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
WORK IN CLASS Force Panel - calibration: force
Fm  a  bxm  cym  dxm2  exm ym  fym2  gFT  hFT xm  iFT ym  jFT xm ym
1 xm1

... ...

1 xmN
ym1
...
ymN
2
xm1
...
2
xmN
a 
 
b 
c 
 
d

... ... ... ... ... FT1 xm1 ym1   Fm1 
 e   
... ... ... ... ...
...
 f   ... 
... ... ... ... ... FTN xmN ymN 
FmN 

 g 
 
h 
 
i 
l 
 
AT A  x  AT b
1 T
1 T
T
T

A A A A  x  A A A b

The pseudoinverse computation gives
the 10 coefficients of the polynomial
model:

A x b
x  A  A AT  b
T
1
NOTE: ATA is generally a square and full
rank matrix

M. De Cecco - Lucidi del corso di Measurement Systems and Applications
WORK IN CLASS Force Panel - calibration: force
Fm  a  bxm  cym  dxm2  exm ym  fym2  gFT  hFT xm  iFT ym  jFT xm ym
Files given:
- read_float.m
How to use read_float :
jj = 1 ; % numero file
file = ['dati_celle_armon_', num2str(jj), '.txt'] ;
dati = read_float(file, ',', ':') ;
clc
Dati = cell2mat(dati);
% grammi -> Newton
Dati(:,1) = Dati(:,1)*(9.81/1000) ;
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
WORK IN CLASS Force Panel - calibration: force
% Data are saved on 9 files “dati_celle_armon_1…9.txt”
As follows on columns:
% 1 -> calibration load [g]
% 2 -> x position [pixel]
% 3 -> y position [pixel]
% 4 -> load cell 1 (in alto) [N]
% 5 -> load cell 2 (in basso a sx) [N]
% 6 -> load cell 1 (in basso a dx) [N]
% 7 -> load cell sum [N]
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
WORK IN CLASS Force Panel - calibration: force
Fm  a  bxm  cym  dxm2  exm ym  fym2  gFT  hFT xm  iFT ym  jFT xm ym
1 x m1

... ...

1 x mN
y m1
...
2
x m1
...
y mN
2
x mN
a 
 
b 
c 
 
d
... ... ... ... ... FT1 x m1 y m1    Fm1  1 
 e     
... ... ... ... ...
...
 f   ...  ... 
  F 
 
... ... ... ... ... FTN x mN y mN 
mN   N 
 g  
h 
 
i 
l 
 
chi2gof Chi-square goodness-of-fit test on the vector 
Performs a chi-square goodness-of-fit test for discrete or continuous distributions.
The test is performed by grouping the data into bins, calculating the observed and
expected counts for those bins, and computing the chi-square test statistic SUM((OE).^2./E), where O is the observed counts and E is the expected counts. This test
statistic has an approximate chi-square distribution when the counts are large.
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
WORK IN CLASS Force Panel - calibration: force
Procedure:
1.With the pseudoinverse computation estimate the 10
coefficients of the proposed polynomial model taking one
data file at your choice
2.Apply the estimated coefficients to the model to predict
(validate) the model behaviour using a different data file
3.Analyse the residuals and the validation behaviour
4.Apply the chi2gof test to verify the randomness of the
residuals
5.Try to choose different mathematical models and see
how the chi2gof test performs
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
WORK IN CLASS Force Panel - calibration: force
F1a [bit]
F2a [bit]
xm [pixel]
xm = f (F1a , F2a , F3a)
ym = f (F1a , F2a , F3a)
F3a [bit]
xs
 F1m
ys   
 FTm
F2 m
FTm
F3m
FTm
ym [pixel]
F1m F2 m
FTm FTm
3
FT   Fi
i 1
1
2
3
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Force Panel – calibration position
F1m
FTm
ax
b
x


F3m
 c x
FTm 
d x
 ex

ay 
by 
cy 

dy 
e y 
Resolution
Accuracy
Position X [mm]
< 0.40
± 1.80 (al 95%)
Position Y [mm]
< 0.30
± 1.80 (al 95%)
Force [N]
< 0.05
± 0.10 (al 95%)
Time [ms]
10
±5
Nail
Finger
Yellow: resistive touch
Blue:
load cell reconstruction
M. De Cecco - Lucidi del corso di Measurement Systems and Applications
Force Panel – calibration position
finger horizontal
M. De Cecco - Lucidi del corso di Measurement Systems and Applications