Transcript Sensing for Robotics and Control

Sensing for Robotics and Control
ME 4135
R. R. Lindeke
General Ideas about Sensors


Sensor are truly systems!
Sensors systems consist of three separable
ideas:



Informational sources: physically measurable data
sources (light beams, audio beams, electrical fields, etc)
Detector areas: Devices that react to changes in the
informational sources
Data Interpreters: devices (hard or soft based) that
convert informational changes into useful information
Sensor Topics

Positional Control Sensing
–
Kinesethic Sensing




Resolvers
Absolute Encoders
Incremental Encoders
Environmental Sensors
–
Contact


–
–
Point
Field Sensors
Proximity – typically single point
Remote


Single Point
Field Sensors
Kinesethic Sensing



These sensors provide feedback information
to the joint/link controllers (servo information)
They use analog or digital informational
responses
We will explore 3 generally used types:
–
–
–
Resolvers
Absolute Encoders
Incremental Encoders
Resolvers


Operating principle is
that a charged rotating
shaft will induce voltage
on stationary coils
Secondary Voltages are
related to Primary
voltage as Sin and
Cos ratios of the
primary field voltage
Resolver Ideas:
Typically we use 2 stators
one (not shown) mounted
normal to an axis that is 90
away from the one thru
Winding A
Resolvers, cont.





Position is determined for
computing stator ratio
Winding A carries Sin
signal
Winding B carries Cos
signal
A/B = tan so
Shaft position
=Atan2(B_Reading,
A_Reading)
 of interest
Resolver Issues



These devices are susceptible to Electrical
Noise – must be highly shielded
Usually use gearing to improve resolution
Typically are expensive but very rugged for
use in harsh “shock motion” environments
Optical Encoder Positional Sensors




Based on Photoelectric source/receiver pairs
Looks for change of state as changing receiver
signal level (binary switching)
Uses a carefully designed disk with clear and
opaque patches to control light falling on a fixed
sensor as disk rotates
Can be made ‘absolute’ with several pairs of
emitters/receivers or Incremental with 2 ‘out of
phase’ photosensors
Optical Servo Measurement Systems

Absolute Encoders
–
–
–
Use Glass Disk marked for positional resolution
Read digital words (0010111011) at receiver to
represent shaft position
Commonly Available with up to 16 bits of
information (216) to convert into positional
resolution
Operating Principle
Zdcvkpsdjfpsdjfgoipsdjf
g’oadgn’oiardfgnd’oikrjg
hdar’okgjnsldkfgknllkknr
esffzsdfsdfkjfksdf;kkjnflll
l
Absolute Encoders – optical disks
Absolute Encoder Variations – 8bit
Comparing Natural Binary to Gray
Code

Natural Binary give actual position when read
–

Gray code is designed so only one bit
changes “at a time”
–

Actual position is known w/o analysis
Where Bit change is subject to positional errors
as light “bleeds” around patch edges
Gray codes are, therefore, less error prone,
but require an ‘intelligent converter’ to give
actual shaft position
Using Absolute Encoders Resolution:
360
 ABS  n
2
here: n is # of 'lines' on disk
Determine resolution if n = 5?
ABS for ‘5 liner’ = 360/25
= 360/32 = 11.25
To Improve Resolution:

Add Gearing to shaft/encoder coupling
–
New Resolution is:
 ABS


360

n
 2
is gear ratio on encoder shaft
Increase # of Lines – this increases complexity and cost of
the encoder (can be a significant cost increase)
Absolute Encoder for 0.18 Resolution
360
 ABS  0.18  n
2
log 360
0.18

n

n  3.301
0.301

log 2
 10.965  11bits
Incremental Encoders


This devices use
3 pairs of
Emitter/receivers
Two are for
positional
resolution, the
third is a
‘calibrator’
marking rotational
start point
Sine wave is observed
due to leakage (light
bleeding) around opaque
patches!
Incremental Encoders

The positional detector
uses what is called
“Quadrature”
techniques to look at
the changing state of
the 2-bits reporting
position for each
opaque/clear patch on
the optical disk
Incremental Encoders

Notice the “square wave”
quadrature signals
–


they are offset by “½ phase”
Each patch resolves into 22
or 4 positions!
Without hardware change,
resolution is a function of the
number of patches – or lines
Incremental Encoders
 INC 
360
C patch
360

2
2
4C patch
Consider a 500 ‘Line’ incremental encoder?
For 500 line Inc. encoders,
resolution = .18 (w/o gearing)
Comparing Absolute and Incremental
Encorders:


Incremental are usually cheaper for same
level of resolution
Absolute are able to provide positional
information at any time under power
–

Incremental must be homed after power loss to
recalibrate count numbers
Compared to resolvers, encoders are fragile
so must be shock protected during operation
Environmental Sensors


These sensors provide ‘code decision making’ power
to the Manipulator
These sensors can be simple



Single point devices,
Simple devices typically trigger yes/no decisions with switch
changes
These sensors can be complex 2-D array (or even 3D field) devices



Typically the receivers are complex arrays
The data interpreters are sophisticated software and hardware
devices
They can add “intelligence” for decision-making by the
manipulator
Contact Sensors – Force and
Deflection Sensing

Force Sensors:


Measure pressure
for gripping –
direct or indirect
Measure
deflection during
contact – typical
of indirect contact
sensing
Contact Sensing



Indirect contact sensors use Strain Gages (and
Hooke’s Law: Stress = E*Strain)
The strain gage is a resistive device that exhibits a
change in resistance due to changes in shape
(length or width)
The Strain Gage is mounted into a carefully built
(and calibrated) Wheatstone bridge



small changes to the strain gages resistance, observed while
using a highly linear voltage source, are calibrated against
observed deflection
This ‘bar’ deflection is strain and multiplying the strain times
the bar’s modulus of elasticity yields stress and hence applied
force!
Stress = Force/Areabar
Contact Sensing





Other contact sensor are “Direct
Reading”
These devices use the piezoelectric
principle (effect) of the sensor material
Piezoelectric effect states that in certain
material (quartz and some silicates)
applied forces (dynamically) will cause a
minute – but measurable – flow of
electrons along the surface of the crystal
based on di-polar disruption due to
shape change
This flow is measure as a “Nanocurrent”
The Current is linearized, amplified and
measured against a calibrated force
Contact Sensing



A second general type would be the class of
“Micro-Switches”
Like at the end of the Conveyor in the S100
cell
Typically, applied forces directly move a
common contact between NC and NO contact
points
Examples of Micro-Switches:

One Directional Reed
Switch:

Omni-Directional Reed
Switch:

Roller Contact Switch:

Etc., etc., etc.!!!
Tactile Sensors – “feeler arrays”

Potential Advantages of Tactile Sensors:
–
–
–
They generate far fewer data bits (compared to
visual arrays) leading to simpler interpretation
analysis
Collection is more readily controlled – we
completely control background and contrast
The properties we measure are very close to
(exactly?!?) the properties we desire
Defining the “Ideal” Tactile Sensor


They must be rugged and compliant to faults in the
manufacturing (operating) environment
They should be “Smart” – That is able to process
most of the data into information for decision making
locally
–


they send only results to the main controller
Resolution should be on the order of about 100 mils
(about 10-4 inch)
Sensors should respond to forces on the order of
about 5 -10 gmforce (0.1 N or 0.022 lbf)
Tactile Arrays:


Machine Equivalent of
Human Skins
Use arrays of microsized switches or other
methods to detect
shapes and sizes due
to contact images of
“made” Switches
Tactile Arrays


This device “measures”
shapes and sizes by
determining which of an
array of target points
have been charged
Targets are “charged”
through contact with the
conductive Elastomer
skin and the PC ‘board’
targets
Tactile Arrays



In this device, a series of thin
rods are pushed into an
object
A “positive” image of the
object is produced by the
displaced rods
In modern sensors,
displacement of each rod is
measured by the
detector/interpreter system –
this might be a vision system
located normal to the
direction of contact
application or an LVDT unit at
each ‘rod’
Tactile Arrays



The Anisotropic conductive
rubber sensor
The ACR and gold contact
surface is separated when
unloaded
As load is applied contact
patches grow indicating
shape and size of external
object and force being
applied
Proximity Sensors:


Devices, including Photocells, Capacitance
sensors and Inductive sensors, that can be
used in areas that are near to but not
directly contacting an object to be sensed
Like all sensors they use structured signal
sources, receive changes of state in their
energy (sensing) fields and interpret these
changes with signal changes to the “outside”
Photo Sensors



The modern photosensor (in the
proximity range) emits
modulated light (at infrared or
near-infrared wavelengths). The
emitters are LED.
The receivers (phototransistors)
are ‘tuned’ to be sensitive to the
wavelength of the source emitter
during the ‘on’ steps in the
modulated output stream
The interpreters are (typically)
transistors that switch the power
(or ground) source on to the
output lead
Diffuse Mode Photosensor



In proximity mode, the device is
looking for its own emitted beam
reflected back to its paired receiver
The level of light falling on the
receiver to trigger positive response
can be ‘tuned’ to the task
The sensors can be tuned to “LightOperate” or “Dark-Operate”


Light operate means positive output
when reflective light is sensed
Dark operate means positive output
when NO reflective light is sensed
Retro-Reflective Photosensors


These devices rely on
“broken beams” to
detect
They are “typically”
dark operate – that is
waiting for the object to
interrupt the light path
to the reflector
Thru-beam or Separated Systems


The Emitter and
Receiver are
separate devices
These again rely
on dark operate
mode (typically) –
that is a broken
beam indicates
objective present
Inductive Sensors
They typically oscillate
In ranges: 3 KHz – 1MHz
Inductive Sensors
Shielded types have slightly longer
range but smaller field of view
Uses:

Inductive Sensors can (only) detect metals
as they draw power by induced surface
currents (eddy currents)

The more magnetic the metal the greater the
sensor’s range
Principles of Capacitive Sensing
Uses And Capabilities

Capacitive Sensors are able to detect any
material that raises the field dielectric in the
vicinity of the sensor
–
In air this is nearly any other material!
Uses of Capacitive Sensors:
When properly calibrated, the sensor can
detect any higher Dielectric Material thru
any lower Dielectric Material
Typical Application of
Capacitive Sensor:
Detecting Liquid
(H2O) levels in bottles
Dielectric Values of
Various Materials: