Automated Ball Striker Joseph Black Gina Rophael David Caloccia Paul Savickas
Presentation Outline n n n n Motivation Objective Specifications Design Approach Results Assessment Conclusion
Motivation n n Extension of the Ping Pong Shooter from Team 2 of Spring 2005 Inspiration: to develop training equipment for ping pong players.
Objectives To strike a vertically launched ping pong ball. n n n Launch a ping pong vertically at various heights Visually locate the launched ping pong ball Process data to determine ball’s position and velocity Develop trajectory for paddle’s motion Design a controller to track the desired path
Specifications Step response: n Steady State Error n Percent Undershoot n Rise Time n Settling Time 0. 195% 12. 58% 0. 43 s 1. 43 s Other Specifications: n Range of Motion (Pan) -80°to +100° n Range of Motion (Tilt) 0°to +90° n Payload 200 grams n Noise Tolerance 95% n Weight of the System is ~1 Kg n Commercial Cost $2462. 75 n Speed is 8 rad/sec for pan and for tilt 10 rad/sec n Images processing rate 30 frames/second n Sampling time for the controller 10 micro sec
Design Approach n n n n Modeling Control Design Launcher Vision Physics Model Trajectory Generation Integration
Modeling n General Equation of Motion n Disconnect between model and system n A simpler model was pursued
Modeling n Simplified Equation of Motion n Steady State n n Friction Identification Transient Response n Inertia
Friction Identification Viscous Friction =. 0087 Nm. S/rad Coulomb Friction =. 0141 Nm
Model Validation n Open loop response
Model Validation n Closed loop response n n Differences between model and system The final model was useful in understanding the system, but control design was done on the actual system.
Control Design n Control applied using the FPGA PID block in Lab. VIEW. The PID for set point control with sampling of 100 microseconds, 512 proportional for pan and tilt , 3/128 integration for pan and 1/128 for tilt. Swing controller pan 10 microsecond sampling, proportional 1792 , derivative 7/128, derivative control may not have much affect with such a high sampling rate.
Control Design
Control Design
Launcher n Final Design n Ball Support Multiple Balls Performance n n Launch Height Verticality
Vision n n Collect frames at 30 frames per second, threshold the individual images and filter out small objects, to remove dust and the black balls reflecting light back to camera. Check for ball in frame, first time records the position and time of image after 2 nd image of ball subtracts position and time from previous image to determine velocity
Vision n Used Lab. VIEW Vision Assistant to make calibration information image from a image of a dot matrix in the plane of the launch, with black dots at a known distance from each other fed to Lab. VIEW.
Physics Model n Ball Motion n n 1 D Motion Initial Velocity Position/Time Calculations Air Resistance Time Until Swing
Trajectory Generation n The choice of start and end positions Development in Trajectory Ability to choose between trajectories
Integration n Physical Integration n n Base, Launcher, Webcam, Pan-Tilt Software (Lab. VIEW) n n Physics Model Image Processing
Integration n A laptop is used as a host to connect to the web cam and retrieve images of the ball and calculate position, velocity and time to swing information from physics model. Laptop communicates to begin the swing to the CRIO using VI Server, breaking the CRIO out of a set point loop into the swing signal loop. The CRIO was used as a Real Time system to control the FPGA and to send the swing path and set points to FPGA at 1 k. Hz. The FPGA is used for control of the system and monitoring the position and velocity
Integration n Testing n n n Subsystem Overall System Uncertainty n n Launcher Physics Model Swing Time Strike Zone
Results n Probability of successful striking n Demonstration Video
Probability of Successful Striking
Probability of Successful Striking
Video Demonstration
Assessment n Ball Launching n n Vision/Physics Model n n Reliable for successful launches Trajectory Generation and Tracking n n Inconsistent launch Accurate tracking of trajectory Overall Success
Conclusion n Recommendations for improvement n n n Improvements to launcher would greatly benefit the rest of the system Ability to aim ball toward target zones State-space control
Questions?
Скачать презентацию Automated Ball Striker Joseph Black Gina Rophael David
f99322635bff4283f848b46b973acc68.ppt