RSS feedsUpcoming Event:
Date: Wednesday, November 18, 2009
6:15pm: Networking
6:30pm: Presentation
7:30pm: Adjourn
Location: Cogswell Polytechnical College, Board Room, 1175 Bordeaux Dr., Sunnyvale, CA 94089
Cost: Free
RSVP: (none)
Title
Combining Motion Optimization and System ID for Robotics
Speaker
Dr. Ghulam Mustafa, Crossing Automation
Abstract
The speaker will present a novel approach that combines motion optimization and system identification via Taguchi-type design of experiments (DOE) and Monte Carlo simulation for quantifying the manufacturing variability of substrate-handling robots. Motion control optimization refers to selecting a set of parameters that results in system responses that meets the performance specifications. The parameters of a given controller can have opposing effects on the closed loop response of a system. For example, the overshoot and damping behavior are often conflicting control objectives and can pose a challenge for adjusting the parameters of a PID controller. The PID parameters are often determined using ad-hoc methods, such as those proposed by Ziegler and Nichols in the 1940s. The presented statistical approach addresses this challenge and will cover in-situ tests developed for a robotic application, such as the step and frequency response to optimize PID parameters. Furthermore, the test data are also used to identify system dynamics, which indicate that the robot performance depends on damping and stiffness properties that vary within the population of assembled robot mechanisms. The identified system model is then used in conjunction with Taguchi DOE and Monte Carlo simulation with damping and friction as noise parameters. This results in a statistical distribution of performance metrics, such as over-shoot and settling time, which in turn enables a methodical approach for controller tuning.
Biography
Ghulam manages the Reliability Engineering and System Characterization Group at Crossing Automation, where he is responsible for system reliability and performance. Prior to joining Crossing, he worked at United Technologies Research Center where he developed life-cycle models for new technology development, reliability models for hydrogen fuel cells and noise reduction in jet engines. He has worked on wafer handling robots for the past nine years, and on semi-conductor capital equipment for the past fifteen years. Currently, his is involved in implementing modular design methods for wafer handling systems, of particular interest to him are areas of testing, characterization and system identification for health monitoring, failure prevention and preventive maintenance. His areas of interest include system dynamics, measurement, control and all aspects of product reliability. Ghulam can be contacted at gmustafa@crossinginc.com.