Transcript AICIP 1 Chris Beall Design of a Mobile Sensor Platform and

AICIP
Design of a Mobile Sensor Platform and Localization
Design of a Mobile Sensor Platform
and Localization
Chris Beall
1
Design of a Mobile Sensor Platform and Localization
AICIP
Overview
•
•
•
•
•
•
•
•
Chris Beall
Motivation and System Requirements
Drive System and H-Bridge
Sensing capabilities
Processing capabilities
System Facts
Platform Improvements
Introduction to Localization Methods
Localization using ARToolkit
2
Design of a Mobile Sensor Platform and Localization
AICIP
Motivation
• The goal of this project is to construct a mobile
autonomous sensor paltform which is able to
“sense” the environment and act upon it through
both local decision making as well as
collaboration among networked mobile sensor
platforms
Chris Beall
3
Design of a Mobile Sensor Platform and Localization
AICIP
System Requirements
• System must be small in size for portability
• It must be low cost for large volume platform
deployment
• Tether-free outdoor deployment
• Energy-efficient for longer battery life
• On-board processing capability for autonomous
local decision making
Chris Beall
4
Design of a Mobile Sensor Platform and Localization
AICIP
Drive System
• The drive system uses two gearhead motors
attached on either side at the back of the mobile
sensor platform
• The motors can be driven with up to 12VDC.
However, to limit the speed of the mobile sensor
platform only 5VDC is used
• The platform uses an omni-directional third wheel
to allow for differential steering
Chris Beall
5
Design of a Mobile Sensor Platform and Localization
AICIP
H-Bridge
•
•
•
•
•
Chris Beall
The H-Bridge is used to control the motors
Separate inputs for logic and motor power
Enable/Disable Input
Four data inputs
Four motor outputs (2 per motor)
6
Design of a Mobile Sensor Platform and Localization
AICIP
Controlling the H-Bridge
• The H-Bridge is controlled with the parallel port
on the computer
• Pulse width modulation is used to control the
speed of the platform as it moves along
• The H-Bridge can be enabled/disabled to conserve
power and to prevent overheating
Chris Beall
7
Design of a Mobile Sensor Platform and Localization
AICIP
Sensing
• The mobile platform is currently equipped with
two kinds of sensors
• Imaging sensor
• Range sensor
Chris Beall
8
Design of a Mobile Sensor Platform and Localization
AICIP
Imaging Sensor
• The imaging sensor is a Logitech
Quickcam Pro 4000, used for target
recognition through advanced
image processing. It is able to take
pictures at a resolution of up to 1.3
megapixels, with a default setting at
640x480pixels.
Chris Beall
9
Design of a Mobile Sensor Platform and Localization
AICIP
Range Sensor
• The range sensor works by transmitting
a pulse of sound outside the range of
human hearing. The sound reflects back
to the ranger from any object in the
path of its sonic wave. The distance of
the object is obtained by measuring the
time it takes for the sonic wave to
bouce back. The range sensor has an
approximate beam pattern that is 60 in
width, and less than 10 feet deep.
Chris Beall
10
Design of a Mobile Sensor Platform and Localization
AICIP
Processing
• The mobile sensor platform uses a
Mini-ITX (7x7in) form factor
motherboard, the VIA EPIA-M10000
• The board has a 1Ghz processor,
room for one memory module,
integrated video, audio, ethernet, and
one PCI expansion slot (used for
wireless)
• It is excellent for this application
because it is powerful, yet
very compact
Chris Beall
11
Design of a Mobile Sensor Platform and Localization
AICIP
System Facts
•
•
•
•
•
•
Dimension 12x15.5x10 inches
Cost: Less than $600
Battery Life: 3hours with moderate use
Voltage regulator shuts down itself when overheated
Fast processor & 512MB of memory
Switches allow to toggle between internal (battery)
and external power
• Software: Latest version of Fedora Core and
open source drivers
Chris Beall
12
Design of a Mobile Sensor Platform and Localization
AICIP
Platform Improvements
• Place additional range sensors along the sides and the
back of the mobile platform as the current design only
allows it to detect objects directly in front of it.
• Reduce the weight of the platform by replacing the
heavy, although low-cost, sealed lead acid (SLA)
battery. One option is the much lighter Li-Ion battery,
which maintains a flatter discharge curve than other
kinds of batteries. However, this option is far more
expensive.
Chris Beall
13
Design of a Mobile Sensor Platform and Localization
AICIP
Research Highlights
• Self-deploying sensor networks for robust
reconnaissance and surveillance
• Self-organizing sensor networks for adaptive target
tracking
• Distributed coordination with or without
communication
• The mobile platforms will orient themselves and
collaboratively achieve complex tasks like tracking of
moving targets, target identification,
multiple target detection, etc.
Chris Beall
14
Design of a Mobile Sensor Platform and Localization
AICIP
Localization
• It is necessary for the mobile sensor platforms to be
able to determine where they are located
• Localization can be relative: A robot knows where it
is located relative to another robot
• Localization can be absolute: A robot knows where it
is located in reference to the building it is in
• Ideally each platform would be aware of both its
absolute position, and its position relative to other
mobile sensors
Chris Beall
15
Design of a Mobile Sensor Platform and Localization
AICIP
Localization Methods
•
•
•
•
•
Chris Beall
Global Positioning System (GPS)
Laser Tracking
Sonar
Radio – UT wireless network (triangulation)
Visual markers
16
Design of a Mobile Sensor Platform and Localization
AICIP
Localization Using Visual Markers
• There are several different ways of using visual
markers for localization:
– Use a color coded cylinder mounted on top of each
mobile sensor platform to allow them to identify
each other. This method provides only
identification and relative location. Distance and
exact location are difficult to derive.
– Use pattern recognition to identify patterns
attached to platforms or landmarks. Based on the
pattern the orientation, distance, and
angle can be determined.
Chris Beall
17
Design of a Mobile Sensor Platform and Localization
AICIP
Localization Using Pattern Recognition
• There is open-source software which is widely used
to create Augmented Reality (AR) applications
• The ARToolkit is able to localize targets in live video
without any noticeable delay
• The first step is to threshold the image, then it detects
where there is a square marker
• Once the marker is found, it calculates the distance,
position and orientation
Chris Beall
18
AICIP
Design of a Mobile Sensor Platform and Localization
Using the ARToolkit
• The software detects the
marker and displays the
coordinate system on
top of it using OpenGL
• The coordinates can be
used to determine
where the platform is
located
Chris Beall
19
AICIP
Design of a Mobile Sensor Platform and Localization
Using the ARToolkit
• In this example the
software actually superimposes a 3D cube on
the live video
Chris Beall
20
AICIP
Design of a Mobile Sensor Platform and Localization
Questions?
Chris Beall
21
Design of a Mobile Sensor Platform and Localization
AICIP
Thank You
Chris Beall
22