Final Presentation
Download
Report
Transcript Final Presentation
G.U.N.D.A.M.
GROUND UTILITY NETWORK DECIPHERING
AUTOMATED MACHINE
GROUP 10
BLAKE SIMONINI
DIDIER LESSAGE
GABRIEL RODRIGUEZ
What is it?
A robot whose primary function is solving mazes of
varying types using the wall follower method of maze
solving
Maze will be custom built with a layout capable of
being changed to any type depending on the user’s
specifications
Motivation
Provide a system for exploring locations others
cannot.
One of the main advantages of using a robot to
traverse unexplored territory becomes evident in
places where humans can’t go due to size or potential
hazards, such a cave.
Parts Being Used
Two ultrasonic sensors for the sides
One ultrasonic sensors for the front
One MSP430 microcontrollers
RC Car Base
Two XBee Modules
Failed Design
DIDIER LESSAGE
GUNDAM 1.0
Chipped gears
Solutions?
Attach plastic base to front for
reduced friction after lost front
wheels
Attach a ball to base to allow
for smoother movement
Turns now made virtually
impossible on carpeted surface
Blown out motors
Solutions?
New motors, resulting in
changed values for the software
Low weight support
Solutions?
New Chassis
GUNDAM 2.0
Four ultrasonic sensors
Two on sides to detect
angles
Auto correction
RC Car Base
Two motors control front
and back
Issue?
Blown out sensor during
construction
Turning proved difficult
without isolation
GUNDAM 3.0/Controller
BLAKE SIMONINI
GUNDAM 3.0
• Added a connector to
interface with the
MSP430 development
board without having to
remove the Processor
• Cleaned up the wires and
used colored wires for
what each line was
• Clamps to keep
everything organized
• Wires don’t interfere
with sensors now
Motor Microcontroller
Deals with collision detection
Calibration of the motors
Directly controls whether the robot is turning left,
turning right, or going forward
Interfaces with the Ultrasonic sensors in order to
solve a particular kind of maze (which will be
explained later)
Motor Microcontroller
H-Bridge
Motor Microcontroller
Input 1
Input 2
Front Motor Output (Steers
left and right)
Low
Low
Steer Straight
High
Low
Turn Wheels Left
Low
High
Turn Wheels Right
High
High
Steer Straight – Motor Off
Input 1
Input 2
Back Motor Output
(Drives forward and
backward)
Low
Low
Nothing Happens
High
High
Nothing Happens
Low
High
Drive forward
High
Low
Drive backward
Motor Microcontroller
Battery Options
AA Batteries
Wouldn’t supply enough current and when they got hot they
would supply less current
D Batteries
Supplied enough current, but are too heavy.
Lithium Polymer
Supplies enough current and is lightweight
Wireless Subsystem
-
ROBOT MODULE
-
-
Mounted on the robot to send and receive
information from the MSP430
COMPUTER MODULE
-
Connected through a computer’s USB port
to send and receive data from the Java GUI.
Robot Module
PCB layout
XBee Wireless Transceiver Module
Interface with MSP430 through UART
Computer Module
PCB layout
XBee Wireless Transceiver Module
CP2101 UART to USB
Interface XBee through UART to PC USB
XBee Wireless Module
Encrypt data in AES-128 algorithm
Specifications:
3.3V operating voltage
2.4GHz operating frequency
Adjustable Power output 1.25-2mW
Range 120m
Data rate: 250kbps
UART interface
UART to USB Bridge (CP2102)
USB Bus powered powered: 4.0-5.25V
Baud rate up to 921.6kbps
On chip voltage regulator
Virtual COM port for GUI
Range Finder Subsystem
-INFRARED SENSORS
•
The initial plan was to mount one on the left
side and another on the right side of the robot.
-ULTRASONIC SENSORS
•
The initial plan was to mount one sensor on the
front and another on the back of the robot.
Infrared Range Finder (GP2D120)
Operating Voltage 4.5V to 5.5V
Operating Current 33 to 50mA
Measures 4cm to 30cm
Analog output
Output Voltage (V) vs. Reflected distance (cm)
IR Sensor Malfunction
During the construction of GUNDAM 2.o, an IR
sensor was damaged
Solution?
Two ultrasonic sensors were then used to replace the IR
sensors that were originally planned to be used to auto correct
and detect paths
GUNDAM 3.0 modified this even further by only using one
sensor on each side
Final Ultrasonic Layout
One ultrasonic sensor mounted on the left, the other
on the right
Used to auto correct the GUNDAM while traveling straight
Used to detect paths
One ultrasonic sensor in front
Used to detect obstacles
Used to detect forward paths
Ultrasonic Range Finder
Measures 2cm to 3m
Operating Voltage 5V
Current consumption 20mA
Ultrasonic Frequency 40kHz
Communication
Positive TTL Pulse
Physical Maze
Plastic, Wood, Metal, Rubber, and Paper reflect
ultrasonic waves.
Things to consider:
Cost : Metal > Plastic > Wood
Ease of Manufacturing: Metal > Plastic > Wood
Wooden planks, each one foot in length
Maze Layout
Maze Solving (Path Finding) Algorithms
Wall Follower
Simple maze solving solution that involves following the left
side of the maze, including any turns that may follow. Will be
the default maze solving method
This solution is only valuable in certain maze situations. If the
entrance of the maze happens to lie in the center and not on
the outside edge, or if a wall happens to lie on its own with no
connections, it will fail
GUNDAM Maze Solving Walkthrough
GUNDAM continues forward until a path is detected
on one of its sides. It will try to center itself with the
walls while it is moving forward.
If path detected:
Check for a left_path();
Check for a front_path();
Check for a right_path();
• When each path is open, we know we have solved the
maze.
How Turning is Done
The Ultrasonic Sensors become inaccurate after they are angled at
an object past about 45 degrees.
To determine a turn, we use a wall around the robot for the turn and turn
the robot until that wall distance is giving very large and inaccurate values
from the sensor.
This way we can get an idea of how far we have turned, even though the
sensors aren’t that accurate for turning.
Drawback is that we can’t have a full intersection, since not having
at least one wall on a turn would break this idea.
Turning has to be three point turns.
How To Perform Proper Turns
Side Feature: Maze Drawing
Draws out maze as GUNDAM solves using the wall
follower method
Upon reaching each node, present paths are sent to
laptop
The same algorithmic idea on GUNDAM is applied to
path to predict next stretch of maze pathways to
draw out as the GUNDAM’s next move
Packets may be lost or resent, causing software to
not be reliable
Progress
Research
Design
Hardware
Completed
Remaining
Software
Testing
0
20
40
60
80
100
120
QUESTIONS?