G.U.N.D.A.M.

BLAKE
DIDIER LESSAGE
GABRIEL
 What is it?

A robot whose primary function is solving

mazes of varying types while transmitting the
layout of the maze back to a computer/laptop
to display said maze to the user
Maze will be custom built with a layout
capable of being changed to any type
depending on the user’s specifications
Motivation
 Features

Wall Detection
Wireless Communication
Maze Solving
Discover entire maze layout
Accept path inputs from user
Forward and backward movement
Isolated rotation
 Parts Being Used

Two IR Sensors for the sides

Two ultrasonic sensors for the front and back
RFM12B-S2 Wireless Transceiver
Robot Chassis
Java GUI Interface
Wireless Subsystem

-ROBOT MODULE
-COMPUTER MODULE
 Robot Module

PCB layout
 MSP430 Microcontroller
 Interface with Main µC through SPI or I2C
 RFM12B RF Transceiver
 Interface with MSP430 through SPI
Schematic
 Computer Module

PCB layout
 MSP430 Microcontroller
 RF µController
 RFM12B RF Transceiver
 Interface RF µC through SPI
 CP2101 UART to USB
 Interface RF µC through UART to PC USB
 Schematic
 Wireless Protocol

TX node (Robot Module)

 Initially Transmitting to Establish Connection

RX node (Computer Module)

 Initially Listening to Establish Connection
Collision Avoidance Algorithm
 Avoid TX or RX at the same time
 Wireless Protocol

Employ AES-128 on both the Robot and

Computer module
 Data encryption of TX packets on each node
 Data decryption of RX packets on each node
 RF Microcontroller (MSP430)

Encrypt data in AES-128 algorithm

Read data from RF Module for RX
Send data to RF Module for TX
Specifications:
 SPI, I2C, and UART interface
 16-bit Architecture
 RF Transceiver (RFM12B)

Low power: 2.8-3.8V

High data rate: up to 115.2kbps
Programmable TX and RX bandwidth
Automatic Frequency control
SPI interface
16 bit RX FIFO
Two 8 bit TX data registers
 RF Transceiver (RFM12B)
FSK Modulation Scheme
RSSI Strength indicator
Operating Temp -40-85˚C
At 433MHz bandwidth
 Max TX/RX current 24mA/13mA
 Range > 200m
 UART to USB Bridge (CP2101)

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
- U LT R A S O N I C S E N S O R S
 Infrared Range Finder (GP2D120)

Operating Voltage 4.5V to 5.5V

Operating Current 33 to 50mA
Measures 4cm to 30cm
Analog output
IR Range Finder Function
Output Voltage (V) vs. Reflected distance (cm)
 Ultrasonic Range Finder

Measures 15cm to 510cm

Operating Voltage 8-12V
Current consumption 14mA
Ultrasonic Frequency 40kHz
SPI/I2C interface
Onboard ATtiny26 µController
 Physical Maze

Plastic, Wood, Metal, Rubber, and Paper

reflect ultrasonic waves.
Things to consider:
 Cost : Metal > Plastic > Wood
 Easy of Manufacturing: Metal > Plastic > Wood
Lap Joints
Lap Joint
Maze Layout
 Nodes

Nodes will be “placed” at intersections and

turns. These nodes will be stored in a list on
the computer side. The node will have
information on their location, the amount of
neighbors they have if discovered, and the
distance between the neighbors.
Information on how far the robot has traveled
before reaching an intersection or turn will
be stored and sent to the computer to allow
for accurate representation of the maze and
its dimensions
 Walls

Using the information given to it by the robot

itself, the location and length of the walls will
be able to be determined as well as any turns
and openings along these walls. This
information is then used to draw out the
actual maze.
 GUI

Maze will be presented in its own frame

along with options for the user to request
either a maze be solved (if not already), the
maze be explored, or a particular path be
traversed or destination reached.
 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
Maze Solving (Path Finding) Algorithm

Tremaux
 This algorithm assigns values to paths according to
how many times it has been traversed. At a fork in the
road, if there is a path valued at 0, it will take it. If
not, and the current path is a 1, it will backtrack and
take the next smallest path value. If the current path
is a 2, the smallest valued fork will be taken. This
method will be used if it is determined that the maze
cannot be solved with the wall following method
 Entrances and Exits

There will be a single entrance and exit for the

maze
A check will be done to determine if the maze
type can be discovered simply through the
entrance’s characteristics. If surrounded by walls,
it is safe to determine that the robot is inside of
the maze rather than outside of it. This is a signal
to use Tremaux, as emphasized previously.
Exits can be within or outside of the maze. A
check will be done to determine the distances on
all sides. If these sides exceed the pre-determined
dimensions of maze, we have found our exit.
 Format of Information

Upon reaching an intersection, along with

placing a node, the robot will write to the
port how far it has estimated its travel, the
number of turns open to it (left, right,
straight ahead)
The program will read this information and
use it to construct the next set of walls. Each
path drawn out will simply be two lines of a
pre-determined width between them
 Classes

Nodes
 Neighbors
 Location
 Dead ends
 Part of path
 Heuristic
 detDistDest()
 getLocation()
 getHeuristic()
 getDistCurr()
 Classes

Path
 Nodes
 Distance
 Final (bool)
 getNextNode()
 removeNode()
 isDest()
 isSource()
 Classes

Walls
 Location1
 Location2
 Length
 Width
 Stand Alone (Bool)
 getLoc1()
 getLoc2()
 getLength()
 getWidth()
 Classes

Robot
 Location
 State
 Solving
 Exploring
 UserPath
 getState()
 getLoc()
 Explore Maze

The robot will explore and discover any and

all paths within the maze using a hacked
version of the Tremaux algorithm. Instead of
having the sole purpose of discovering an
exit, a list of all nodes and paths along with
their number of times visited will be stored.
The robot will then work through this list and
make its way to each node using a
combination of Tremaux and A star.
 User Input Path

The user will be able to input a path for the

robot to take inside of the maze.
The user can either input an exact path for
the robot to take or….
…the user can simply select a destination and
the robot will use A* path finding to find the
shortest path to the destination.
 A*

Once a destination node is chosen, the algorithm takes

into account only the destination and any neighboring
nodes to the current position of the robot.
Cost (distance) of moving a node is calculated for each
neighboring node not blocked off by a wall
Estimated cost of reaching destination is then
calculated
The smallest calculated distance and the node that has
achieved said distance is then chosen as the next node
to travel to in the pre-path.
Once a path has been chosen, the robot then traverses
the maze using the path
 A*

If user selects a location that is not a node, a

new node will be placed at the user’s desired
location
This is necessary to enable to algorithm to
actually find a path to the destination
 Specified Path

User can also highlight a path of nodes to

traverse for robot
User will select nodes it wishes to be
incorporated within the path itself or simply
draw out a general line to follow
If general line made, the program will
determine any and all nodes that are
sufficiently close to path to incorporate within
the list of nodes needed to traverse it
 Path Completion

Once the path is completed, program will

store the path (simply a list of nodes and the
order they should be traversed)
The program will then write to a serial port
to be read by the robot itself
Upon reaching a node specified by the
program, the robot will request the next
instruction on whether to turn or continue
straight on its path. This check also involves
determining if the node it is currently at is
the destination
 Progress

Research

Design

Hardware Completed
Remaining

Software

Testing

0 20 40 60 80 100 120
 Budget

Part Model Price

Blake Didier Gabriel
QUESTIONS?