AutonomousRobot Navigation Using Optical Mouse-based Odometry: Autonomous Robot Navigation Using Optical Mouse-based Odometry Craig Putnam October 2004


About The FIRST Competition: About The FIRST Competition FIRST = For Inspiration and Recognition of Science & Technology FIRST was started by Dean Kamen in 1992 as a way to encourage high school students to get excited about engineering The first competition in 1992 had 28 teams 20,000+ students from over 900 teams representing 7 countries competed in 2004


About The FIRST Competition: About The FIRST Competition Adult mentors from industry and academia work with college and high school students Each year we build a new competition robot from scratch The robots are radio-controlled and must be driven by the high school students The matches are 2 minutes in length and (in recent years) have been “2 on 2”


About The Competition: About The Competition The robots have to be able to: Maneuver around on a 24’ × 48’ playing field Drive over / under / around various obstacles (platforms, ramps, metal bars, teeter-totters, etc.) Manipulate objects (balls, “floppies”, tote bins, etc.) Interact with various field elements (move mobile goals, hang from bars, balance on a teeter-totter, …) Meet various design criteria (weight, size, allowable materials, costs, …) Be rugged (able to survive frequent “interactions” with other robots or field elements)


About The Competition: About The Competition Most importantly, the robot has to be able to go from a design concept to the shipping dock in 6 weeks


DWC and FIRST: DWC and FIRST DWC has supported the Alvirne H.S. FIRST team (P.A.R.T.S.) for the past 10 years DWC was one of the first colleges in the country to offer a FIRST scholarship Nick Bertozzi has incorporated work on FIRST robots into the Engineering Design courses as a possible project component Engineering students are given responsibility for the design, manufacture and testing of various robot subsystems Virtually all of the 2004 robot was manufactured in the DW106 shop


Project Background: Project Background 2 years ago FIRST added a new wrinkle to the challenge: During the first 15 seconds of the match the robot must perform autonomously Some ways to navigate autonomously: Dead Reckoning Line Tracking Beacon Tracking Inertial Navigation We expect autonomous mode operation to become even more important in the future


Autonomous Mode Challenges: Autonomous Mode Challenges Where am I? How am I oriented? How should I get to where I need to go? What should I do if I get off course? What else do I need to do along the way?


Autonomous Mode Challenges: Autonomous Mode Challenges Where am I? How am I oriented? How should I get to where I need to go? What should I do if I get off course? What else do I need to do along the way? This project specifically addresses only the first four items


High-Level Project Goals: High-Level Project Goals Address the problems of autonomously measuring, controlling and correcting the travel path of a robot in a FIRST competition environment Design a rugged, small, lightweight, inexpensive and “cool” navigation system Create a project that will engage and challenge DWC Engineering / CS students


Prior Research: Prior Research Considerable work has been done over the last 20 years on robot navigation Existing solutions: Odometric sensors Active beacon triangulation Geomagnetic compass Ultrasonic or laser ranging from known landmarks GPS (“regular” as well as “differential”) Inertial navigation using accelerometers & gyros Vision systems Usually done via wheel rotation measurement


Prior Research: Prior Research While using computer mice to provide odometric path/error feedback has been done in the past, to my knowledge it has never been done using an optical mouse It may also have never been done using a single mouse in conjunction with a gyro It certainly has not been done in a FIRST competition


Challenges: Challenges Using an optical mouse for robot navigation – particularly in the FIRST competition environment – presents some unique challenges


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration Varied surfaces: Carpet Tape Painted wood Aluminum plate Wire mesh Slick white plastic


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration Surface Transitions: Carpet wrinkles, seams & tears Steps


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration Debris: Dirt Robot parts Collisions: Sometimes one robot will ride up over part of another robot


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration Optical mice have a maximum tracking speed of about 12 – 15 ips They may or may not report when they are in an “overflow” state


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration Optical mice are designed to be in contact with a work surface (desk, etc.) But the playing field surface is a hostile environment – so we don’t want the mouse down there…


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration Move the mouse up into the body of the robot for protection To do so we need to change the focal length of the optics Doing this also mitigates the tracking speed problem


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration Computer mice are not very sensitive to rotation about their optical axis The solution to this is to use a solid-state rotation rate sensor (gyro) in addition to the mouse


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration Mounting the O.M. circuit board in the robot’s chassis means the embedded LED does not work properly Additional illumination of the floor is needed


Slide30: Also insert picture of the mouse board…


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration The mouse speaks PS/2 protocol – which means it wants to generate the clock signal The PIC also wants to generate the clock Some “glue logic” is needed to manage & buffer the signals


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration Initialization of the PAK-VIa chip & optical mouse Interrupts vs. polling for mouse communications Inner-loop vs. outer-loop timing


Challenges: Challenges Playing field surface Tracking speed Optics Robot rotation Illumination Electrical interface Programming Calibration Changes in the focal length of the optics will change the size of a “Mickey” We need to make sure that we know the correct number of Mickeys per unit distance travelled


Recent Work: Recent Work Steve Jackson (’04), using last year’s robot, developed a crude “proportional” feedback loop navigation system based only on input from the gyro The robot was able to drive the length of DWH with a maximum deviation from the desired straight-line path of about 6”


Current Work: Current Work Jennifer MacDonald (‘05) is working on developing a proportional, integral, derivative (PID) feedback loop based navigation system This system will combine inputs from the optical mouse & gyro and is being prototyped on a miniature robot


Current Work: Current Work An EG101 student project group will develop a fixture for the modified optical mouse system We will mount the modified optical mouse & electronics in the ’04 full-size robot for calibration and testing once the code is done The goal is to have the ’04 robot navigating autonomously using the modified optical mouse & gyro by the end of the term We want to see how close we can come to the theoretical ¼” position resolution


Current Work: Current Work Modify the ’03 robot to navigate using an infrared beacon tracking system We also hope to have the waypoint generation program (written as a CS project in the fall of ’03) modified to generate C code instead of PBASIC and to generate spline curve paths in addition to straight line paths


Questions?: Questions? http://faculty.dwc.edu/putnam/