Pre-conference Workshop on Co-Design of Control and Real-Time Computing

Conference on Decision and Control (December 14, 2010)

Organizers: Vijay Gupta (University of Notre Dame) and Paulo Tabuada (UCLA).
Presenters: Antonio Bicchi, Luca Greco, Luigi Palopoli, Daniele Fontanelli, Manel Velasco, Pau Marti, Fumin Zhang

Registration information is available at PaperPlaza Conference Registration System

Description

Cyberphysical systems (CPS) represent the next generation of engineered systems. Such systems - also known by terms such as networked and embedded control systems - use computation and communication embedded in and interacting with physical processes to add capabilities to physical systems. As such systems become ubiquitous, it will be necessary to evolve a systematic design theory for them. Such a theory would unify diverse branches of systems theory, including estimation and control, networks, information theory, distributed processing, and so on.

While the interaction of communication with control has been studied over the last decade or so, issues that arise at the intersection of control and processor design are less well-understood. This workshop focuses on the co-design of control and processing algorithms. Traditional design approach assumes a separation of concerns between the two domains. Control designs have largely ignored the limitations and possibilities of various software and processor implementations. Due to the deeply embedded nature of CPS, issues such as scheduling of control tasks, anytime and event triggered control algorithms, and battery consumption due to control algorithm execution become extremely important. On the other hand, computational and scheduling models need to be tailored to and be flexible with respect to the demands of control applications.

This workshop will bring together researchers working towards developing a unified theory that integrates process control and real-time computing. In particular, topics related to real time control, event triggered and and anytime control, and battery aware control will be covered.

Brief description of the intended audience

The primary goals of the workshop are to act as a tutorial for young researchers seeking to work in this important area, and to foster conversation among the two different communities of control and real-time computing. The invited speakers are leaders in the field with significant work and experience in this area. Thus, the workshop will be especially useful for graduate students, post-doctoral researchers, and young faculty not only in control, but also in real-time computing.

Tentative Program

The duration will be half-day, on the morning of Dec 14, 2010.

The tentative schedule is as follows: (Slides will be posted following the workshop.)

9:00am-9:10am V. Gupta, P. Tabuada Welcome, Overview
9:10am-9:50am A. Bicchi, L. Greco, “Anytime Control Paradigm for Stochastic Embedded Real-Time Systems
9:50am-10:30am L. Palopoli, D. Fontanelli, “Implementation of Anytime Control for Stochastic Embedded Real-Time Systems
10:30am-11:10am P. Marti, “Feedback Scheduling: Theory and Practice
11:10am-11:20am Break
11:20am-12:00pm M. Velasco, “Sampling in Event-driven Control Systems
12:00pm-12:40pm F. Zhang, “Battery Supported CPS
12:40pm-1:00pm Discussion and Wrap Up

Abstracts

Anytime Control Paradigm for Stochastic Embedded Real-Time Systems: In this talk we consider the problem of designing controllers for linear plants to be implemented in embedded platforms under stringent real-time constraints. These include preemptive scheduling schemes, under which the execution time allowed for control tasks is uncertain. In a conservative design approach, only a control algorithm that is executable (in the worst case) within the minimum time slot guaranteed by the scheduler at each period would be employed. On the opposite, we consider here a more flexible “Anytime Control” design approach, based on a hierarchy of controllers for the same plant. Higher controllers in the hierarchy provide better closed-loop performance, while typically requiring a longer execution time. Stochastic models of the scheduler and of controllers execution times are used to infer probabilities that controllers of different complexity can be executed at different periods. We propose a strategy (in the form of a switching policy) for choosing among executable controllers, maximizing the usage of higher controllers, which affords better exploitation of the computational platform than the conservative design while guaranteeing stability (in a suitable stochastic sense). Simulation results on the control of two prototypical mechanical systems show that performance is substantially enhanced by our anytime control technique w.r.t. worst case-based scheduling.

Implementation of the Anytime Control for Stochastic Embedded Real-Time Systems:In this talk we present a methodology for designing embedded controllers based on the so–called “Anytime Control” paradigm. A control law is split into a sequence of subroutine calls, each one fulfilling a control goal and refining the result produced by the previous one. We propose a design methodology to define a feedback controller structured in accordance with this paradigm and with the ensuing stochastic switching policy for closed loop system stability. The cornerstone of this construction is a stochastic model describing the probability of executing, in each activation of the controller, the different subroutines. We show how to construct such a stochastic scheduler model for realistic real-time task sets and how to let the switching policy be robust with respect to uncertainties on the task model. Since the performance of the closed loop system can be severely impaired by switching between different controllers in the case of reference-tracking tasks, a simple practical bumpless transfer technique to assist in making smooth transitions between controllers is also presented and adapted to the Anytime Control paradigm. Finally, experimental validation of the proposed technique is reported using a mechanical system endowed with an embedded real-time platform.

Feedback Scheduling: Theory and Practice: The most common method to the analysis, design and implementation of networked and embedded control systems consists on assuming the periodic execution of control algorithms. However, for resource-constrained systems, this assumption may be inappropriate because the selection of fixed rates of execution is not an easy task (low rates imply low resource utilization but also imply low control performance, and viceversa) and because the enforcement of a fixed rate may be not suitable in front of changes in the CPU/network load and in the controlled plants. To overcome these periodicity limitations, feedback scheduling aims at applying efficient sampling period selection techniques that account for load and plants dynamics in such a way that the aggregated control performance delivered by the set of control loops is improved. The talk will give an overview of existing work on feedback scheduling for micro-processor and networked control architectures, outlining and discussing main results, while placing an special focus on implementation aspects. The talk will finish discussing open problems

Sampling in Event-driven Control Systems: The standard design of control systems is based on the periodic sampling. Every period the data is read from the input, the control law is computed, and the output is written to the actuators. However the periodicity of the sampling instants is a constraint that arises from the ease of implementation and it is not strictly necessary in the control system. This talk will give an overview of event-driven control systems, whose execution model mandates to sample the input ``when needed'': the controller is activated upon some condition on the system status and not periodically. As a result, controllers resource demands can decrease while stability and acceptable control performance is still guaranteed. The talk will place an emphasis on the impact of event-driven controllers on the computing platform. To this extend, their computational load will be analysed in terms of activation patterns and real-time feasibility analysis. The implementation of this controllers adopting the self triggered approach will be also presented. Finally, open problems will discussed.

Robustness Analysis for Battery Supported Cyber-Physical Systems: We introduce methods to analyze the robustness of battery supported cyber physical systems under co-designed control, scheduling and battery management algorithms. Robustness refers to the ability to maintain system performance under perturbations. Robustness in controller design has been well defined and understood for a large class of systems, yet robustness of schedul- ing and battery management methods are relatively less understood. We analyze robustness of scheduling algorithms by introducing a novel concept of dynamic schedulability. It is then possible to measure robustness of scheduling algorithms through the strength of the perturbations that break the dynamic schedulability. Robustness of battery management al- gorithms is measured by the capability to endure or reject potentially damaging discharge. Utilizing a dynamic nonlinear battery model, we implement a particle filtering algorithm to accurately predict the status of the battery under any possible discharge patterns predicted by the controller and the scheduling algorithms. This procedure allows any battery management algorithm to make proper decisions.

Biographies

Antonio Bicchi is Professor of System Theory and Robotics at the University of Pisa. He graduated at the University of Bologna in 1988 and was a postdoc scholar at M.I.T. A.I. Lab in 1988–1990. His main research interests are in Dynamics, kinematics and control of cplex mechanical systems, including robots, autonomous vehicles, and automotive systems; Haptics and dextrous manipulation; Theory and control of nonlinear systems, in particular hybrid (logic/dynamic, symbol/signal) systems. He has published more than 200 papers on international journals, books, and refereed conferences. He currently serves as the Director of the Interdepartmental Research Center “E. Piaggio” of the University of Pisa, and as Editor in Chief of the Conference Editorial Board for the IEEE Robotics and Automation Society (RAS). Antonio Bicchi is an IEEE Fellow since 2005. He has served as Vice President of IEEE RAS, Distinguished Lecturer, and editor for several scientific journals including Transactions on Robotics and Automation and Int.l J. Robotics Research. He has organized and co-chaired the first WorldHaptics Conference (2005) and Hybrid Systems: Computation and Control (2007).

Luca Greco received the “laureadegree in Computer Engineering in 2001 and the Ph.D. degree in Robotics and Industrial Automation in 2005 from University of Pisa, Italy. From 2005 to 2007 he has been a postdoc at the Interdepartmental Research Center “EPiaggio”, University of Pisa, Italy. From 2007 to 2009 he has been a postdoc at DIIMA, University of Salerno, Italy. Since September 2009 he is postdoc at L2S – Supélec, Gif-sur-Yvette, France. During his Ph.D. he studied stability problems for variable structure and switched systems. Quantized and symbolic control problems have also been considered. His current research interests concern anytime control, sensor deployment and network controlled systems.

Luigi Palopoli received the computer engineering degree from the University of Pisa, Pisa, Italy, in 1992 and the Ph.D. degree in computer engineering from “Scuola Superiore Sant’Anna, Pisa” in 2002. He is an Assistant Professor of computer engineering at the University of Trento, Trento, Italy. His main research activities are in embedded system design with a particular focus on resource–aware control design and adaptive mechanisms for quality-of- service management. He has served on the program committee of different conferences in the area of real-time and control systems.

Daniele Fontanelli received the M.S. degree in Information Engineering in 2001, and the Ph.D. degree in Automation, Robotics and Bioengineering in 2006, both from the University of Pisa, Pisa, Italy. He was a Visiting Scientist with the Vision Lab of the University of California at Los Angeles, Los Angeles, US, from 2006 to 2007. From 2007 to 2008, he has been an Associate Researcher with the Interdepartmental Research Center “E. Piaggio”, University of Pisa. From 2008 he joined as an Associate Researcher the Department of Information Engineering and Computer Science, University of Trento, Trento, Italy. His research interests include robotics and visual servoing, embedded system control, wireless sensor networks, networked and distributed control.

Manel Velasco graduated in maritime engineering in 1999 and received the PhD degree in automatic control in 2006, both from the Technical University of Catalonia, Barcelona, Spain. Since 2002, he has been an assistant professor in the Department of Automatic Control at the Technical University of Catalonia. He has been involved in research on artificial intelligence from 1999 to 2002 and, since 2000, on the impact of real-time systems on control systems. His research interests include artificial intelligence, real-time control systems, and collaborative control systems, especially on redundant controllers and multiple controllers with self-interacting systems.

Pau Marti received the degree in computer science and the PhD degree in automatic control from the Technical University of Catalonia, Barcelona, Spain, in 1996 and 2002, respectively. Since 1996, he has held different teaching positions in the Department of Automatic Control at the Technical University of Catalonia. From 1999 to 2002, he spent several months as a visiting student at Malardalen University, Vasteras, Sweden, working on real- time control systems with Prof. Gerhard Fohler. From 2003 to 2004, he held a research fellow appointment in the Computer Science Department at the University of California at Santa Cruz, working with Prof. Scott A. Brandt in research on soft realtime systems. His research interests are real-time control systems, with emphasis on the interaction and integration of control systems, real-time systems, and communication systems.

Fumin Zhang received the B.S. and M.S. degrees from Tsinghua University, Beijing, China, in1995 and 1998, respectively, and the Ph.D. degree from the Department of Electrical and Computer Engineering, University of Maryland, College Park, in 2004. He has been an Assistant Professor in the School of ECE, Georgia Institute of Technology (Georgia Tech), Atlanta, since 2006. He was a Lecturer and Postdoctoral Research Associate in the Mechanical and Aerospace Engineering Department, Princeton University, Princeton, NJ, from 2004 to 2006. He worked for the Institute for Systems Research, University of Maryland. He founded the research and teaching program in the fields of robotics and control at Georgia Tech Savannah Campus. His major research focus includes design and control of underwater robots and mobile sensor networks, battery modeling and control, and theoretical foundations for cyber-physical systems. He received the NSF CAREER award in September 2009, and the ONR YIP Award in April, 2010.


Registration information is available at PaperPlaza Conference Registration System

back