1. Mini Grand Challenge Contest for Robot Education Bob Avanzato Associate Professor of Engineering Penn State Abington 1600 Woodland Road Abington PA 19001 RLA5@psu.edu AAAI 2007 Robotics and Education March 27, 2007

2. Objectives Design autonomous outdoor robot contest “Mini Grand Challenge” (MGC) to promote interest in robotics and AI. Partly inspired by DARPA Grand Challenge Include vision and HRI component. Contest should be accessible to advanced high school, lower-division undergrads (Engr, Cmpsci, IST) and beyond. Availability of low-cost robot platform and development environment to improve accessibility to MGC contest.

3. Robot Contests at Abington campus (Phila. PA area) Over 40 robots participating in each contest Over 120 students (K-12 to college and beyond) Over 15 high schools/middle schools represented Over 50 pizzas consumed! Both contests offered annually since 1995. What is next step? Penn State Abington Robot Contests Robo-Hoops Regional Trinity Firefighting

4. Impact on Curriculum & Outreach Robotics supports wide range of educational and outreach goals Robotics Contests EDG 100 Freshman Design CSE 271/275 Digital Electronics Sophomore EE/CSE ENGR 297 Robotics Special Topics Freshmen/Sophomores Comp Sci 201C Intro for Fresh/Soph IST 402 (new!) Emerging Technologies K -12 Outreach Undergraduate Research (ACURA)

5. PSU Abington Robot Platform

6. Mini Grand Challenge (MGC) Autonomous, outdoor, electric ground robot Follow 8-ft wide (unmarked) paths on college campus Reach 6 waypoints (GPS longitude, latitude) Avoid human obstacles on path Entertain human spectators Take off-road detour across field (with obstacles) Payload: 1 gallon of water Robot Speed: 1.5 - 5mph (6) Waypoints disclosed 24 hours prior to contest event.

7. Campus Paths

12. Campus Paths (Field)

13. Campus Paths

15. Sample Path/Waypoint Layout WP #3 WP #5 WP #6 WP #4 Path width = ~8ft Waypoint (WP) diameter = 20ft … orange cones

16. Key Equipment List PowerWheels™ platform$220 GPS (with serial cable)$120 Speaker/amp (15-30 watt)$60 Inverter (DC to AC)$50 Servo (steering)$50 Speed controller$60 USB camera (240 x 320)$50 Camera stand$30 Sonar and servo$50 USB to serial converter$30 Servo controller$50-$150 Battery$50 TOTAL……$850 (approx.) NOTE: Laptop, MATLAB costs not included in above list

17. Robot Block Diagram USB Camera GPS (Garmin eTrex) Sonar Pontech SV203 Controller Speaker (30-watt) Steering Control Drive Motor Speed/Dir Control Laptop Computer Windows XP OS MATLAB

18. PSU Abington Robot

19. Key Software MATLAB with Image Processing Toolbox –Grab image from USB camera –Edge detection –Read GPS text serial output (position, velocity) –Text-to-Speech –Send motor and steering commands to servo controller –Main control loop written in MATLAB Drivers –MS Win32 Speech API (SAPI) (text to speech) –VFM (Video for Windows frame grabber) Any Software/Hardware solution Allowed

20. Pilot Study: Student Solution Background: Sophomore-level EE student with no prior experience in vision Student developed a heuristic, path- tracking algorithm in MATLAB (Image Processing ToolBox; Canny edge detection) within 4 hours (non-optimal). MATLAB environment promotes rapid prototyping and facilitates testing.

21. Big Design Questions Can an outdoor robot platform (hardware and sensors) for MGC be constructed for under $1000? –Answer: Yes (almost) Can an operational outdoor, autonomous, robot prototype (hardware & software) be completed (with minimal testing & performance) for MGC in 40 hours? –Answer: Yes (almost)

22. Results 2005 Mini Grand Challenge (April 2005) –3 participants; no successful robots –Rain limited outdoor event (rescheduled in Dec.) –Robots on display indoors; same day as FF contest –Generated much interest for future events 2006 Mini Grand Challenge (April 1, 2006) –6 participants; one robot manages 50% of course 2007 Mini Grand Challenge (March 31, 2007) –8 robots registered –1 high school team

23. Mini Grand Challenge Event (PSU-Abington PA; April 3, 2005)

24. MGC 2006

25. MGC 2006 (PSU Abington robot)

26. MGC 2006

27. MGC 2006 (PSU University Park robot)

28. MGC 2006 (Spectator Interaction)

29. Conclusions Mini Grand Challenge (MGC) contest successfully promotes interest in robotics and AI for a wide range of participants (freshman college to professional) Low-cost robot platform with MATLAB software allows freshman/sophomore undergrads to participate in sophisticated algorithm development. Spectator friendly; educational & outreach benefits Outdoor contest has risks (example: rain!) Larger robot --> more cumbersome for classroom integration

30. Future Directions Expand student involvement in Mini Grand Challenge (course integration problem). Develop web-based resources and tutorials. Develop K-12 outreach activities based on MGC Assess (survey) student retention and recruitment. Expand spectator-robot interaction (SRI) ArtBots (in Philly, PA) Develop indoor extension to contest to mitigate weather problems. (Example: follow cones in gym) Non-engineering student involvement – IST?