Player/Stage: A Unifying Paradigm to Improve Robotics Education Delivery Monica Anderson, Laurence Thaete, Nathan Wiegand The University of Alabama
Learning to program robots Novice student issues Long learning curve Hardware access “Murphy’s law” Research student issues Porting simulations to hardware Different hardware programming paradigms http://robotics.cs.ua.edu/
Opportunity Use simulation to teach concepts Leverage layering to create a client API that is viable in both simulation and on actual hardware Use lower cost platforms to increase accessibility Cannot eliminate these problems completely but in the classroom, we can mitigate the effects of these issues by regulating when they appear. http://robotics.cs.ua.edu/
Our Approach Provide abstraction and simulation via Player/Stage Use multiple platforms with different strengths Introduce concepts in a controlled manner http://robotics.cs.ua.edu/
Player/Stage Robot control and simulation software [Gerkey01] Open source (easily modified and extended) Supports iCreate roombas Pioneers Khepera Others… http://robotics.cs.ua.edu/
Player Abstracts specific hardware into general control by interpreting client requests into hardware specific commands Player interpreting Roomba commands 1000302 Player interpreting Koala commands C,10,10 TCPIP (6665) Client program C++, Java, Python, etc Player with Stage Simulator Robot.move() http://robotics.cs.ua.edu/
Stage/Gazebo Simulation component for algorithm testing Clients ported to hardware with some “tweaking” http://robotics.cs.ua.edu/
Research Platform K-Team Koala x86 500Mhz controller Proximity sensor ring High fidelity odometry 802.11 networking for off-board control and peer communications http://robotics.cs.ua.edu/
Low-cost Education Platform iRobot iCreate (roomba) - (approx $500) Interfaced with Gumstix SBC Lower fidelity odometry and sensors 802.11 networking for off-board control and communications http://robotics.cs.ua.edu/
Logistics Player/Stage available on virtual Linux machine Simulation assignments completed individually Teams of three students on hardware assignments Hardware included 5 Koala robots and 10 roomba robots http://robotics.cs.ua.edu/
Curriculum Introduce motor control and sensors iteratively in increasing detail over semester Use simulation with and without error coefficients Include self-paced hardware workshops during class time (Students are encouraged to help each other) Move to hardware-based assignments at about 8 weeks http://robotics.cs.ua.edu/
Student Projects Student designed projects to foster creativity PenBot - robotic logo interpreter Topological navigation Life sized Pong “Man vs vision-based player” http://robotics.cs.ua.edu/
Summary/Future Work Only 2 students dropped (18%) for personal reasons Students reported more confidence with hardware after the class Students created player clients for roombas (not included in lectures or class assignments) More advanced class is being planned that incorporates higher level behaviors such as path planning and mapping http://robotics.cs.ua.edu/
Questions http://robotics.cs.ua.edu/