senses, sensing, and sensors
with applications to
ICT for
sustainable development
mel siegel
robotics institute
school of computer science
starting points
• sensing is often the problematic element in
the sense-think-act-communicate paradigm
– it is all well-and-good to say, e.g., “what we need
to halt the spread of AIDS is a 50-cent battery
operated pen-sized sensor for HIV infection ...”,
but where is that sensor?
• the unavoidable tradeoff between sensitivity
and specificity threatens system robustness
• sensor fabrication technology lags – so can
ride the coat tails of – VLSI technology
what senses do we want?
at least the human ones ...
•
•
•
•
•
•
vision: eyes (optics, light)
hearing: ears (acoustics, sound)
touch: skin (mechanics, temperature)
odor: nose (vapor-phase chemistry)
taste: tongue (liquid-phase chemistry)
sixth sense (proprioception): joint angles
human senses are dual-use
•
balance
–
•
ears (in addition to hearing)
acceleration
–
•
stomach (in addition to digestive senses)
sound
–
•
chest cavity (in addition to breathing senses)
touch
–
–
touch: tongue (in addition to taste)
temperature: skin (in addition to force)
a blessing and a curse
•
artificial sensors also respond to
multiple stimuli
–
–
–
–
•
almost all respond to temperature
many respond to acceleration and pressure
many respond to light and other radiation
sensors are “physics experiments that failed”
these can often be “compensated”
using differential techniques
however noise is always present
•
–
–
environmental stimuli of same sort as signal
thermal and other noise internally generated
we can’t escape it ...
• it’s the Joe BFSTPLK effect …
but with clever techniques ...
• ... we can extract very small signals
from overwhelming quantities of noise
military communications hide
small signals in overwhelming
environmental noise, so your adversary
doesn’t even know you are talking –
encryption is a backup in case he
somehow manages to actually hear you
we would also like to have
extended ranges and modalities
• vision outside the RGB spectrum
• active vision
– radar and laser range measurements
• hearing outside 20 Hz – 20 kHz range
– ultrasonic range measurement
• chemical senses beyond taste and smell
• radiation: a, b, g-rays, neutrons, etc
and sensors for modalities that
humans might have ...
•
•
•
•
•
•
electric fields?
magnetic fields?
radio (electromagnetic) waves?
pheromones? (probably yes)
“weather”?
what else??
we need artificial sensing for …
• everything the human senses can do …
• without handicaps (specs, hearing aids, …)
• able to use human sensory aids
– telescopes, microscopes, etc
– stethoscopes, sound amplifiers, etc
– micromanipulators w/ haptic feedback, etc
• plus all the senses that we don’t have
(or we don’t have with good cognition)
but which our applications demand
here’s the textbook I recommend
• Handbook of Modern Sensors
Physics, Designs, and Applications
Jacob Fraden
3rd ed., 2004, XVII,
589 p. 403 illus., Hardcover
ISBN: 0-387-00750-4
$89.95
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Ch1 Data Acquisition
Ch2 Sensor Characteristics
Ch3 Physical Principles of Sensing
Ch4 Optical Components of Sensors
Ch5 Interface Electronic Circuits
Ch6 Occupancy and Motion Detectors
Ch7 Position, Displacement, and Level
Ch8 Velocity and Acceleration
Ch9 Force, Strain and Tactile Sensors
Ch10 Pressure Sensors
Ch11 Flow Sensors
Ch12 Acoustic Sensors
Ch13 Humidity and Moisture Sensors
Ch14 Light Detectors
Ch15 Radiation Detectors
Ch16 Temperature Sensors
Ch17 Chemical Sensors
Ch18 Sensor Technologies
(3) digital interface
(1) raw signals
(0) materials science
(0) optical science
(2) analog interface
is someone there?
where is something?
how fast?
how heavy?
how heavy (fluid)?
how fast (fluid)?
how loud?
how wet?
how bright?
how radioactive?
how hot or cold?
what is it made of?
(0) sensor fabrication
old: transduce to human vision
•
•
•
•
•
•
•
thermometers: temperature-to-length
barometers: air pressure-to-length
scales: weight-to-angle
humidity: hair curl-to-angle
indicator dyes: chemistry-to-color
photo film: light/radiation-to-silver density
speedometers: velocity-to-angle
new: transduce to electronics
•
•
•
•
•
•
•
thermistor: temperature-to-resistance
electrochemical: chemistry-to-voltage
photocurrent: light intensity-to-current
pyroelectric: thermal radiation-to-voltage
humidity: absorbed water-to-capacitance
length (LVDT): displacement-to-inductance
microphone: sound pressure-to-<anything>
sense-think-act loop
m
e
a
s
u
r
a
n
d
measure in
volts, amps, ohms,
henrys, farads, etc.
transduce perception
to electrical signal
SENSOR
ADC
convert from
signal to symbol
ACTUATOR
transduce signal to
heat, displacement,
illumination, etc
DAC
compute control action
convert from
symbol to signal
sense-think-act loop
m
e
a
s
u
r
a
n
d
measure in
volts, amps, ohms,
henrys, farads, etc.
transduce perception
to electrical signal
SENSOR
ADC
convert from
signal to symbol
environment
ACTUATOR
transduce signal to
heat, displacement,
illumination, etc
DAC
compute control action
convert from
symbol to signal
measurand & measurement
system immersed in environment
• nature confounds measurement with:
– temperature
– quantization
– chaos
• environment confounds measurement with:
– complex universe: 3 K radiation, weather, ...
– interfering signals: power lines, radio/TV, ...
– engineering and materials limitations, ...
distribution of replicated
measurements
9
8
Number of Measurements
7
6
5
4
3
2
1
0
998.0
998.5
999.0
999.5
1000.0
Measured Height [mm]
1000.5
1001.0
1001.5
1002.0
sensor fusion: people & robotics
• Barfogenesis : Does Virtual Reality Make You Sick?
–
... I
was getting nauseous from the film. My eyes wanted me to
"stop flying," while my body said, "you're not moving at all" ...
– http://serendip.brynmawr.edu/bb/neuro/neuro00/web3/Pili.html
• ... the only things consistently real about Virtual Reality
were headaches and motion sickness ... (Steve Ellis)
– http://www.firstscience.com/site/articles/virtual_reality.asp
• some corresponding problems in “artificial” sensing:
– multi-modality image registration (e.g., light + ultrasonic)
– multi-media synchronization (e.g., sound + video)
– one modality, several measurements, how to combine them?
– several measurements contribute to a calculated result;
given component confidences, how to calculate overall confidence?
combining
multiple measurements
• three thermometers give measurements
T1 = 19.1 ± 2.0 C
T2 = 18.6 ± 2.5 C
T3 = 19.3 ± 1.0 C
• what is your best estimate of the actual
temperature, and what is your estimate of the
error in your estimate of the actual temperature?
• weight by reciprocals of respective uncertainties
(actually weight by reciprocals squared)
• this is what we mean by “sensor fusion”
combining
multiple errors
• you compute the volume of a box by
multiplying together measurements of its
height, width, and depth:
V = (h ± Δh) (w ± Δw) (d ± Δd)
• what is your estimate of the error ΔV?
• it is the (quadrature) sum of
{∂V/∂h,∂V/∂w,∂V/∂d}
weighted by {Δh, Δw, Δd}
• this is what we mean by “error propagation”
electronic sensors
• many sensors are just resistors, capacitors,
or inductors (sometimes containing “unusual”
materials) whose parameters depend on
some feature of the environment:
–
–
–
–
–
thermistors: R (temperature T)
humidity sensors: C (absorbed water vapor)
proximity sensors: L (distance to a surface)
magneto-resistive sensors: R (magnetic field B)
photo-conductors: R (incoming light intensity)
• other sensors are fundamentally voltage
sources:
– electrochemical sensors: V (chemistry)
– photovoltaic sensors: V (light intensity)
– magnetic pickup loops: V (B(t))
• still other sensors are fundamentally
current or charge sources:
–
–
–
–
Faraday cup (e.g., solar wind collector)
photocell (e.g., “electric eye”)
CCD camera sensor
many kinds of radiation detectors
• for example, smoke detectors in which vapors
accompanying the smoke affect the charge
collected in a radioactive environment
• and other sensors fundamentally extract
power from radiated fields:
– antennas: directed radio frequency energy
– microphones: directed acoustic energy
• many kinds of sensors are essentially
perturbed communication devices:
– a signal transmitter
– a signal receiver
– the nature of the path between them
disturbs (or sometimes enhances) the
communication between them
– from which disturbance an interesting
property of the medium is deduced
• for example, smoke detectors in which
the smoke just attenuates a light beam
instruments vs. sensors
• many devices thought of as “sensors” at an
applications level are very complex
instrument systems at an engineering level
analog-to-digital
conversion
scan generation
background subtraction
curve fitting
library searching
pattern recognition
sensing needs in context
of the WEHAB agenda
•
•
•
•
•
Water and Sanitation
Energy
Health and Environment
Agriculture
Biodiversity and Ecosystem Management
water & sanitation
• good clean water essential to all WEHAB:
energy, health, agriculture, biodiversity
• humans not good sensors of water quality:
awful seeming can be safe, & vice versa
• complex instruments, e.g., GC-MS, too
big, expensive, hard to use and maintain
• semiconductor, MEMS, organic polymer
sensors all promising
– but require IT support for linearization, library
searching, pattern recognition, alarms, etc
• see: http://www.geocities.com/RainForest/5161/lab2.htm
– separation techniques
– measurement techniques
•
•
•
•
•
•
gravimetric
electrochemical
colorimetry / spectrophotometry
titration
chromatrography
mass spectrometry
alternative 3rd world appropriate
water quality sensing
• arrays of discrete sensors 
• integrated sensor arrays / chemical imagers 
• hand-held smart instruments
(Cyrano Inc.)
energy (& off-grid capability)
• reliable IT requires reliable hardware even
in the absence of a reliable power grid
– sensors needed for controls that provide
local backup and stability to compensate
for global unreliability and instability
• IT can ensure reliability of the power grid
and power generation infrastructures
– sensors needed to detect faults (and theft!),
compensate for environmental perturbations,
monitor and correct power quality, etc
• see: http://www.electrotek.com/seminars/hirel.htm
– living with an unreliable power grid
• better batteries: modern batteries include complex circuitry
to sense and regulate changing and discharging
• local generation means: a lot of low-grade energy is available,
but it takes smart adaptable systems to utilize it
• uninterruptible power supplies: batteries + inverters + means
to intelligently shut down when batteries run down and intelligently
re-start when power becomes available again
– making the power grid reliable
•
•
•
•
rapidly sensing fault, diagnosing problem, taking corrective action
remote sensing to detect illegitimate usage (e.g., theft)
measuring, assessing, and improving power quality
integrating local and national generating capacity by allowing users
to become suppliers when they have excess local capacity
power availability and quality
http://portland.indymedia.org/en/2004/08/294881.shtml
health & environment
• instruments for rapid, reliable, sensitive
detection of illness – or impending illness
– diabetes: disease of modern diet and lifestyle
– identification of infectious agents (viruses etc)
• protecting employees, residents, and their
environments from industrial pollution
• detecting bad side effects of good projects,
e.g., arsenic in ground water
• detecting side effects of war, e.g., mines
• see:
http://www.sciencedaily.com/releases/200
3/09/030924054754.htm
Livestock Health Sensors And Wireless
Data Storage In The Works
• see:
http://www.cis.rit.edu/~rlkpci/urssra_Krem
ens.pdf
Low Cost Autonomous Field-Deployable
Environment Sensors
health & environmental sensors
http://www.yenra.com/
glucose-monitor/
glucose monitor +
insulin pump
http://www.weathertools.com/oregon4.html
weather station w/ wireless sensor module
transmitters
agriculture
• many sensing requirements intimately tied
to health (e.g., animal health), weather
sensing and prediction, water availability
• many chemical sensing requirements: soil
and water pH and nutrient content, oxygen
availability and demand
• many physical sensing requirements:
water availability, retention, evaporation
• for advanced technology, need the full
robotic arsenal for navigation and work
• see
http://www.agrotechnology.kvl.dk/teaching/p
hddanetpft/pdf/04_sensingsystems.pdf
Sensor Technology in Agriculture
• see
http://www.kuleuven.ac.be/onderwijs/aanbo
d/syllabi/I0F33AE.htm
syllabus of a course (in Dutch) on physical
sensors, chemical sensors and bio-sensors
for agricultural and food applications
sensors for agriculture
http://www.ntechindustries.com/
mapping.html
greenseeker optical plant health
monitor
http://www.trimble.com/
AgGps_autopilot.html
automatically steers
tractor perfectly straight,
center pivot, curves or
headlands
http://www.sentek.com.au/
products/products.asp?lang=en
portable soil monitoring system
biodiversity & ecosystems
• large- and small-scale characterization
and monitoring of animal and plant species
• evaluation and exploitation of medicinals,
both unknown and used traditionally
• control of poaching, encroachment, etc,
while preserving indigenous cultures
• tracking animal movements
• tagging legitimately collected specimens
• enabling transport of more fragile species
• see
http://www.digitalgovernment.org/search/
projects/project.jsp?ID=108
Biodiversity and Ecosystem Informatics Wireless Sensor Networks for Dense
Spatio-Temporal Environmental Monitoring
• see
http://investigate.conservation.org/xp/IB/ex
peditions/pantanal/day8/day8_tools.xml
neat night pictures of animals in jungle
sensors for biodiversity &
ecosystems
http://www.esa.int/export/esaEO/
SEMPMB0XDYD_environment_1.html
pressure map above hurricane Frances
http://www.landcareresearch.co.nz/
services/ecosat/presentations/
wetland.pps#7
wetland mapping project in New Zealand
parting points
• sensing is often the problematic element in
the sense-think-act-communicate paradigm
– it is all well-and-good to say, e.g., “what we need
to halt the spread of AIDS is a 50-cent battery
operated pen-sized sensor for HIV infection ...”,
but where is that sensor?
• the unavoidable tradeoff between sensitivity
and specificity threatens system robustness
• sensor fabrication technology lags – so can
ride the coat tails of – VLSI technology