1. Final Presentation EE 396 – Micromouse Spring 2008 Friday, May 9, 2008 Donald Kim Lab - POST 214

2. Team Mighty Mouse  Mark Fujihara  Team Leader, Hardware Designer  Bryant Komo  Hardware Designer  Programmer  David Ota  Programmer

3. Project Overview Build an autonomous robot mouse that can find the center of a 16 x 16 maze, then calculate the shortest/fastest path to the center and make multiple runs.

4. Initial Team Goals Implement the side looking sensor technology Build a light and fast mouse Build a “smart” mouse that can efficiently find the center of any maze Right, Left, Random Wall Huggers Map the maze Follow the flood fill algorithm Solve the maze Make 45° turns Bring a mouse to the mainland and win!

5. Our Design Features

6. The Design – Where We Left Off The Not-So Final Cuts Four sensors for 45° turns Sensors mounted on plexiglas brackets screwed in to base, allow changes More space and rounded corners for sensor mounting Design more powerful “American Muscle Car” type mouse 7.4 V Lithium-Polymer battery

7. The Chassis – Problems Time Fluctuating sensor values, yielding unreliable results in maze Sensor angles not consistent Not good for correcting and turning Very noisy, too much vibrations

8. The Chassis – Changes Three Sensors – no 45’s, no time, “broken” sensor Balsa wood sensor mount - fluctuating sensor values Angles and directions change too much, makes sensors unreliable 11.1 V battery More speed and more torque on motors Lock washers for all nuts and bolts

9. The Circuit – Initial Design Prototype on breadboard For testing while we learned to etch a PCB Prototype on self-etched PCB To run at Mini-Competition I ExpressPCB – PCB For final mouse

10. The Circuit - Problems Holes not big enough Headers, Regulator Switch not implemented in circuit Too much current for thin lines Too much low frequency noise in the high voltage signals Too much heat on voltage regulator and motor drivers Melted solder, broke pins, shorted out components 4 regulators, 3 Rabbits Poor set-up for debugging

11. The Circuit – Final Design Re-drilled larger holes Ran thicker wire for high voltage and ground lines Added 100μf capacitors to source voltage pins Testing heat sinks, larger resistors Added in sensor LEDs

12. Award Winning Mouse Design

13. The Code – First Take Separate functions for motor movement, sensor reading Co-states, delay milisecond for clock rate Read in port by port, convert binary to decimal Plans Mapping Bellman-Ford Flooding Algorithm

14. Take One - Problems Clock rate only had three usable speeds Not enough definition Correcting was too slow on the left Left turn did not work

15. The Code – Take Two No co-states at all For loops instead; fixes speed problem K-controlling Left or right wall hugger ADC ports changed to read in and calculate binary to decimal automatically Plans Mapping Bellman-Ford Flooding Algorithm

16. Take Two - Problems No independent motor control Proportionality controlling did not work with for loops Both motors sped up at same time Flood Fill simulation did not work

17. The Code – Final Algorithm Inspired by Jay Abel with assistance from Alex Zamora Created own clock with control to toggle clock signal up or down Flags tell mouse of openings ahead, mouse moves one cell before turning Reliant on step count Will not turn till exact step count is reached Set up for a solving algorithm Proportionality control correcting Random wall hugger code

18. Final Algorithm - Problems Code is messy Magic numbers Global variables Randomly commented out “stuff” Right wall flag would not go down after seeing right wall and moving straight Turned right into dead end corner Very complex

19. Unfulfilled Goals  Build a “smart” mouse that can efficiently find the center of any maze  Follow the flood fill algorithm  Solve the maze  Make 45° turns  Stay within budget  Over $800

20. Fulfilled Goals Implement the side looking sensor technology Build a light and fast mouse (relatively speaking)  Build a “smart” mouse that can efficiently find the center of any maze Right, Left, Random Wall Huggers Mapping code Bring a mouse to the mainland Win!

21. Improvements Connectors more forward or back, not right over wheels Chip sockets for everything Thicker PCB traces with less right angles Wider wheels for more traction Mapping and solving code implemented

22. Design Suggestions Read the Spec. Sheets FIRST! They are gold and should always be on hand Be NEAT! Use a PCB, but make sure it is right. Watch out for pad size and trace thickness Send it out and get a double layer design, the circuit design is a lot easier to do. Keep wires neat and use connectors Use chip sockets for easy replacement of ALL parts Design for easy debugging LEDs work very well especially for sensors Don’t be afraid to change the design

23. Completion Suggestions Do NOT waste time, the semester is not as long as you think…The project is also not as easy as you think. Work in parallel (half code, half hardware) Make sure EVERYONE understands function of each component, and how mouse will sense and move first Don’t be afraid to seek help Work, Work, Work. If you plan to go to competition, every bit of free time needs to go into the mouse. Nights, weekends, breaks, etc.

24. Thanks To…  Mitchell aka “Light Weight” LaPuente  Malcom aka “Double Digit” Menor  John aka “John” Miyajima Thanks for helping to make the trip fun, in between all the work! It was great getting to know you guys.

25. Special Thanks To… Dr. Tep Dobry Thank you for believing in us and taking us to the Regional Competition Thank you for the wonderful trip, it was a great experience for us all

26. Lifetime Thanks To… Thank you for your debugging skills, hardware tips and lessons and “killer” code ideas. We would like to present you with this Commemorative Green Rabbit so you never forget how much your help meant to us Mr. Alexis “Savior” Zamora Thank you for spending all of those long, sleepless all-nighters with us, leading up to the competition.

27. Last, but not least, a demonstration…  Disclaimer: In case our step count is off and it screws up, here is proof that our mouse did actually move at the competition Project Demonstration

28. Thank you! Any Questions??