Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.


Networked Distributed Control for Modular Reconfigurable Printing Systems

This talk is motivated by control problems encountered, while working on the PARC hypermodular high speed parallel printing system, which was designed to be highly modular and reconfigurable, both in hardware and in software, and to serve as a testbed to explore the integration of control and planning in a networked setting. A hierarchical model-based control framework for controlling the printer will be presented. A centralized graph-search based planner handles discrete actions and routing, using discrete models; a centralized coordinator interfaces between discrete & continuous worlds using intermediate models; and distributed controllers handle the tracking of piecewise polynomial trajectories, using physical models. The tracking is effected using two-degree-of-freedom time-optimal controllers, which handle saturation explicitly and are well aligned with overall objective of minimizing job processing time. Basic techniques for networked control were also implemented, such as time stamping & delay equalization to remove randomness from bounded random network delays. Time varying Kalman filter estimation was used to deal with asynchronous distributed sensing. Synchronization techniques were developed for distributed coordination/actuation. Controller state synchronization was accomplished by embedding the dynamic controller in Finite State Machine (FSM), and clock synchronization was implemented using phase locked loop. Every effort was made to keep the individual components as simple as possible, to minimize system integration challenges. Some Golden Rules from “the trenches” will be shared.
Type of Seminar:
Public Seminar
Prof. Haitham Hindi
Palo Alto Research Center, Palo Alto, CA 94304
Feb 23, 2009   17:15 /

ETH Zentrum, Main Building, HG F 33.1
Contact Person:

Prof. H.Bölcskei
No downloadable files available.
Biographical Sketch:
Haitham Hindi has been a member of the research staff at PARC in the Embedded Reasoning Area (ERA) group since June 2003. His research is in control and optimization, and their application to real-world problems. At PARC, he has worked on algorithms for optimal control, synchronization, network flow-based scheduling and routing, system performance analysis, machine learning, systems architecture and protocol design, along with embedded systems software implementation. As of late, he has also been part of PARC's Green Energy Technology Initiative, working on energy resource management problems in buildings, data centers, vehicles and power systems. His previous experience includes system modeling, identification and optimal control for particle accelerators, disk drives and sensor networks. He is recipient of the Hugo Shuck Award, and author of a couple of dozen technical papers on control theory and its application to real world problems. He holds a BSc in Physics from Imperial College in London, and an MS and PhD from Stanford University in Electrical Engineering.