University
of Pennsylvania
Electrical
and System Engineering
ESE 519 Real Time Embedded
Systems
2013 Fall
Instructor: Prof. Rahul Mangharam
Final Project
REAL WORLD “Tank CITY”
Final Report for Group:
Wei-Ting Lo
Peng Hao
Yun Yuan
Date: Dec 20, 2013
Github link:https://github.com/ESE519/TankCity
Part 1 Motivation and Goals
Most of us have played a game called “Tank
city” on PC when we were little kids. Our team is going to build the whole
scenes in the reality world.
For this real game, we will have enemy tanks,
user tank, base flag, and a real map.
There is one player that is controlled by a
user. It can moves around and shoot the enemy tanks. We use white LEDs light to
simulate the “shoot” and use light sensors equipped on the enemy tanks to
detect the when the tank is being shot. If enemy tanks get shot, they will lose
HP value. If all HP Value of them are gone, they dead. If all enemy tanks are
dead, player win the game.
There are going to be two enemy tanks, which
would run automatically. Their goal is to go to the base and attack it. If they
arrive the base and hit it, the user lost the game. It is like a tower defense
game. User need to try their best to stop the tanks.
We design a base and some lanes for the map. We
decorated and made walls, which are made of cardboards, as barrier.
This is an integrated project. We utilized many
techniques here, including microcontroller, sensor system, analog and digital circuit,
and network technology.
Part 2 System
Design
The whole system mainly consists of three
parts: Player Tank Control System, Enemy Tanks Automatic System, and Location and
Map Navigation System. All three parts has their own necessary features and
collaborate to get the whole system work.
 |
|
Figure 1.The whole system
|
Figure 2. The network for the location
|
There is a network between
Enemy Tanks Automatic System, and Location System.
 |
 |
Figure 3. The whole map for the project
|
Figure 4. The tank family
|
The location system has 2 IR transmitter
blocks, let’s just call them A and B, on the ceiling. Since there is a distance
limitation for the IR location system, we set up two blocks in order to cover
the whole map. The A system covers the left part of the map, and B covers the
right side of the map.
If the enemy tank is in area A, it uses block A
to navigate the tank. When the tank reaches the limit of area A, the location
system switches to B and shut down A.
The Hardware we used in
this project:
m3pi robot (3);
mbed (5);
joystick (1);
Xbee (2);
White LEDs;
ZigBee (4);
Light sensors;
Resistors with different values;
Relay (4);
Batteries;
Solenoid (2);
IR camera (2);
5 V power supply;
IR LEDs;
Speaker;
firefly sensor board;
green LEDs;
wires.
The software IDE we use
in this project:
Keil uvison4
Putty as terminal and input for the XBEE.
Part 3 Hardware and Software Detail
1. User Tank Control System
The tanks we are using are m3pi robots from
pololu.
We designed a remote controller, which has a joystick and a button, for
user to control the user tank. Joystick as moving the tank
and button as shooting.
The idea of controlling the tank is:
Control joystick or button->send analog data
to mbed->mbed determine the data and output digital signals to Xbee->Xbee
tells the Xbee on the tank which digital signals it has received from
mbed->Xbee on the tank tell the mbed on the tank to move according to the
digital signal it has received.
By moving the joystick leftwards, rightwards,
forwards and backwards, it will output different voltage to a microcontroller,
mbed from NXP.
The mbed processes this input and send data to
the user tank via XBee.
The user tank, has a mbed and Xbee on it. We
use mbed to control every decision on the tank, and use XBee as communication.
 |
 |
Figure 5. Joy stick control panel
|
Figure 6. User tank with “cannon”
|
2.
Location System
A. Hardware summary
The central control system consists of one
mbed, one MRF24J40 (ie ZigBee RF Transceiver Module (2.4 GHz), to communicate
with the whole system), two relays, and one speaker. This system is mainly
responsible for GPS localization tiles switch and communication with the enemy
tanks.
For the tiles switch part, we have 2 IR
transmitter blocks on the ceiling with 4 IR LED spots each with 5V power
supply. We use two relays to control power supply for different IR transceivers
and controlled by mbed. And the MRF24J40 is used by both the central control
system and enemy tanks to communicate to know when the tiles need to switch.
When the central control system is told to switch a tile, it sets the
respective pin to high and so that the corresponding relay will supply for that
tile.
The central control system also has another
functions that it will alert the speaker when the enemy tanks attacks the tank
base. This is implemented by the communication between the tank base board and
the central control system.
 |
 |
Figure 7. IR blocks on the ceiling
|
Figure 8. Design block for location system
|
As mentioned before, the enemy tank would
request IR block switch when it arrives at the boarder of area under current IR
coverage.
Then the enemy tank would send a request to the
location control system and wait. After receiving the request, the control
system switches the working IR Block to another with relays, and sends an ACK
to the request enemy tank. After receiving the ACK, the enemy tank knew that
the IR Blocks have switched and continued to move under the new coordinate.
 |
 |
Figure 9. The switch system
|
Figure 10. The base circuit
|
The tank base system consists of one mbed, one
MRF24J40, and one firefly sensor board. For the firefly sensor board, we used
the sound sensor to determine whether the base is being attacked or not. Since
our enemy tank will shoot a rubber bank and when it hits the base, it would
have sound for the sound sensor. After the sensor detects a sound signal, it
would send a signal via ZigBee to the central control system to alert the
speaker.
Once he speaker sounds, the user lose the game.
B. Software summary
On the Central Control System:
For the enemy tank, when it arrives at the
boundary of two tiles, it will call a function, calibrate, to switch the
location system. It includes the command in the transmitting package and send
it to central control system.
The central control system will listen signals
all the time, when it receives a signal, it will parse the command. When it
receives the “switch” command, it will set the corresponding pin high to switch
different tiles. And when it receives a “Finish” signal that means the game is
over and it alerts the speaker.
On Base:
Determinate
whether the base is being attacked or not. If yes, send a signal to central
control system and ask it to announce the game is over. If no, stays the same.
3.
Enemy Tank Automatic System
A. Hardware summary
We have built a sophisticate
Enemy tank automatic system.
 |
 |
Figure 11. The enemy tank with full function
|
Figure 12. Design blocks for automatic enemy tank
|
The enemy tank use a mbed as its main
controller.
How does the tank moves:
A Wii IR camera equipped on the enemy tanks is
used to analysis the relative x and y coordinate according to the IR location
system.
Since we have four group of IRs in a set, the tank will receive four
different xy coordinate.
These 4 groups of data can be used to calculate
the absolute location of the tank on the map. The enemy tank use this
information to navigate and guide the tank. We use two of the IR groups as
angle reference. Let’s say we pick group 3 and 4. The calculation to find the
reference angle is: atn2:((x3-x4),(y3-y4)). Once we have xy coordinate and
angle, moving the tank to a desired destination is convenient.
How does the enemy tanks
know they are being attacked:
There are light sensors equipped on the sides
of enemy tanks, so they can detect “shoots” from the user tank. Once the light sensors
detect the lights, it would reduce the health point of the tank.
How does it attacks:
There is a physical shooting unit on the enemy
tank to attack the base. We designed a simple rubber band gun and install in on
the tank. If the enemy tank arrives the destination, in front of the base, it
will start shooting the rubber bank by using solenoid as trigger.
B. Software summary
On Enemy tanks:
We used nested if statement to program the
tanks.
Basically, each if statement is like a check
point for the tanks. Once the robots reach the destination, it will close the
current loop and goes to the next one. At the mean time, the tanks would also
constantly check their current angles to make sure they are moving straight.
For the first tank, when it reaches the center of the map and switch tile, we
uses new tile systems as reference to move. Hence, It needs to use totally
different angle and xy coordinate as reference to move.
For different tanks, their status change as
following.
Tank 1: Starts first -> move to the center
of the map -> use ZigBee to call for switching tile -> move to the
base->attack the base.
Tank 2: Starts second (after the tile switch)
-> move to the base -> Attack the base.
Part 4 Discussion and Conclusion
After we decided to do the project, we used
top-down design idea and used it to direct our whole development process.
Separate the whole system into 3 different modules
at the first and integrate them together in the end, which makes our design
more reasonable.
For each modules, we first build a test system on breadboards. On the
breadboard, we adjust parameters and circuits. Once we were sure everything was
correct and good to go, we transfer the test system to the real device. After
done transfer, we test each of them and see if they were working fine as they
did on the test system. In the end, we integrated each of the modules together,
and tested all of them. We found this method is good for our project. By using
top-down idea, we keep checking everything before implement them in the real
devices. In addition, it is a good way to debug as well.
