1
Sensors
CSCI545 Introduction to Robotics Hadi Moradi
Previous Lecture
DC motors DC motors
Inefficient Operating voltage Operating current Stall current Stall torque Stall torque Gearing up and down Gear ratios PWM Servo motors vs. stepper motors
Sensors CSCI545 Introduction to Robotics Hadi Moradi Previous - - PDF document
Sensors CSCI545 Introduction to Robotics Hadi Moradi Previous Lecture DC motors DC motors Inefficient Operating voltage Operating current Stall current Stall torque Stall torque Gearing up and down Gear
DC motors DC motors
Inefficient Operating voltage Operating current Stall current Stall torque Stall torque Gearing up and down Gear ratios PWM Servo motors vs. stepper motors
Perception through sensors Perception through sensors
Contact: bump, switch Distance: Ultrasound, radar, infra red Light level: photo cells, cameras Sound level: microphone
Perception through sensors
Strain: strain gauge Rotation: encoders Magnetism: compasses Smell: chemical
Perception through sensors Perception through sensors
Temperature: thermal, infra red Inclination: inclinometers, gyroscopes
Pressure: pressure gauges Altitude: altimeters …
Simple
Contact switch
Given the sensory reading what was the
Example: Skin Example: Skin
A switch:
Closed = 5 volts
A digital scale:
Microphone: Camera:
Sensing information Sensing information
Proprioception: Exteroception:
Examples of proprioception
Combining multiple sensors Difficulties:
Example: Human brain
http://news.nationalgeographic.com/news/ 2001/10/1012_TVanimalnavigation.html
http://faculty.washington.edu /chudler/magtur.html
http://www.unc.edu/depts/oceanweb/turtles/ Research by Dr. Kenneth Lohmann
V= IR
V = voltage
I = current
R = resistance
Open vs. closed
A variable resistor
Note: Shielding, position, and
Originally
Volume control in your stereo Typically called pots
Emitter and detector Emitter:
LED
Detector:
Photodiode Photodiode Phototransistor
Photoresistor: Photodiode/phototransistor:
Phototransistor vs. Photodiode:
surface feature detection
wall/boundary tracking wall/boundary tracking rotational shaft encoding
Light reflectivity:
Surface color Texture
Ambient light: How to overcome the ambient light? Sensor calibration
= > Partially observable
detector devices can be used to make a break-beam sensor
Measure angular rotation Measure angular rotation Example:
Speedometer: speed of rotation
Odometer: number of rotations
Q: What happens if there is only one
Clockwise rotation signal
Problem: Ambient light Problem: Ambient light Solution: Example: Home remote control Usage:
The distance to a nearby object The distance to a nearby object
Just the return of signal
Infra red part of the spectrum Used like break beam and reflectance
Advantage
Emitter: send a
Collector:
Elapsed time
1.12 feet/ms
Called
Used to map
Which reflection
Which reflection
Graze the surface
Problem: determine the objects in the Problem: determine the objects in the
Example: RoboCup
Light goes through the iris Impinges retina
Reconstruction: what was the world like Reconstruction: what was the world like
pixels: picture cells
Each picture divided into small cells
Typical camera: 512 X 512 pixels Human eye:
120 x 10^ 6 rods 6 x 10^ 6 cones
Brightness: proportional to the amount of
Brightness depends on:
The brightness depends on:
specular (bounce off the surface) diffuse (re-emitted)
Example: Example:
b&w camera
512 x 512 pixel image plane.
intensity level between white and black
Question:
Do we know if there is an object? How do we find an object in the image?
Edges: curves in the image plane with
A simple approach: to look for sharp
Problem:
Noise: Small picks in differentiated image Noise: Small picks in differentiated image. Eliminating noise:
Step 2: Find objects among all those edges. Segmentation:
Questions:
How do we know which lines correspond to
What makes an object?
Use clues to detect Use clues to detect
Use stored models (model-based vision) Use stored models (model based vision)
Use motion (motion vision) Use motion (motion vision)
Use binocular stereopsis
Use texture Use shading
Use shading
recover shape in a similar way as from texture
Reconstruction: If no need for reconstruction:
Simplify vision processing
Q: What are some ways of doing that?
Use a smaller image plane (e.g., a line) Use other sensors to complement vision Use other sensors to complement vision Use task-specific information
Project a light on a Project a light on a
You may avoid
Images courtesy of http://www.caligari.com/
Any object in the Any object in the
Images courtesy of http://www.caligari.com/
A) The whole scene, B) The object w/o laser light, C) the difference
Images courtesy of http://www.caligari.com/
laser on the image plane
camera
you calculate r?
Images courtesy of http://www.caligari.com/