Economic Model Predictive Control: State of the Art and Open Problems
Organizers: Timm Faulwasser, Matthias A. Muller, Lars Gruene

Distributed Parameter Systems: from theory to applications
Organizers: Michael A. Demetriou, Orest V. Iftime

Scenario-based optimization for stochastic optimal power flow problems
Organizers: Ian Hiskens, John Lygeros, Maria Vrakopoulou

Detectability for Time-Varying Systems
Organizers: Ying Tan, Ti-Chung Lee, Iven Mareels

30 years of the Ramadge-Wonham Theory of Supervisory Control: A Retrospective and Future Perspectives
Organizers: Stephane Lafortune, Karen Rudie, Stavros Tripakis

Gain Scheduled Model Predictive Control System Design and Implementation
Organizers: Liuping Wang, John Anthony Rossiter
THIS WORKSHOP HAS BEEN CANCELED

Model-Based and Model-Guided Approaches for Automotive and Maritime Powertrain Calibration and Control
Organizers: Guoming Zhu, Xiang Chen, Chris Manzie, Ying Tan, Noam Olshina, Gokul Siva Sankar Gowri Sankar
THIS WORKSHOP HAS BEEN CANCELED

Deterministic and Stochastic Hybrid Methods for Robust Learning and Optimization in Dynamical Systems
Organizers: Jorge I. Poveda, Andrew R. Teel, Mouhacine Benosman, Kyriakos G. Vamvoudakis, Chris Manzie, Martin Guay

What's the big deal with Deep Learning?
Organizers: Girish Chowdhary, Alexander Schwing, Hassan Kingravi, Chinmay Soman

Economic Model Predictive Control: State of the Art and Open Problems

Workshop overview: Over the last years, the development of tailored optimization methods and increased computational power have led to a considerable speed-up of Nonlinear Model Predictive Control (NMPC) algorithms such that new areas of application besides classical process control can be targeted. Alongside this development, one can observe a trend to consider more flexible formulations in NMPC with respect to the considered cost functions. In other words, a lot of recent research efforts have focused on considering so-called economic objective functions in NMPC, which gave rise to the term Economic MPC (EMPC) for such approaches [1-21]. In EMPC, compared to classical stabilizing (tracking) MPC, the primary control objective is not stabilization of an a priori given setpoint (or trajectory), but rather optimization of some general performance criterion. As a consequence, the resulting closed-loop system might not converge to an equilibrium setpoint, since a different operating behavior (e.g., periodic) might result in a better performance. In recent years, various results on different aspects of economic MPC have been presented in the literature, and many of the crucial (theoretical) questions have been answered. Furthermore, successful implementations of economic MPC have been reported in various fields of application such as energy and process systems. In conclusion, the field of economic MPC is, by now, in a quite mature state. On the other hand, there are still various open issues and problems. As detailed below, the proposed workshop is intended to give an overview of the main developments in economic MPC obtained in recent years, and will also point the participants to interesting open problems.

Target audience: The objective of the proposed half-day workshop is to provide the participants a tutorial overview of recent developments on economic MPC. In particular, the workshop will cover a wide range of different EMPC approaches and topics, including closed-loop performance guarantees, the classification of the optimal operating behavior, closed-loop convergence analysis with and without dissipativity assumptions, economic MPC approaches with and without terminal constraints and terminal penalties, time invariant and time-varying problem data. The intended audience includes both PhD students as well as experienced researchers working in the areas of optimal and model predictive control. Considering the relevance of economic MPC for many application areas – such as energy, productions, and process systems – and the ongoing scientific interest in the topic, the proposed workshop is likely to attract a large number of participants.

Organizers and speakers: Timm Faulwasser, Lars Grüne, Matthias A. Müller

Distributed Parameter Systems: from theory to applications

Workshop overview: Using a series of four interrelated lectures, this half-day workshop will demonstrate the basic theoretical underpinnings of DPS via a series of engineering examples and subsequently will discuss the practical aspects of controller design for DPS. The interplay between the sensor technology limitations will be presented in the context of mobile sensor guidance in distributed estimation of DPS and an overview of data assimilation in the estimator design for DPS will be given. At the conclusion of the half-day workshop, the participants are expected to have an understanding of the modelling and system theoretic properties (well posedness, controllability, observability, stability) of representative examples of infinite dimensional systems including the diffusion equation, the wave and beam equations, delay differential equations and platoon-type systems. Additionally, the participants are expected to have an understanding for the practical aspects of integrated actuator-sensor-controller design, an appreciation of the sensor technology limitations and their effects on the distributed estimation of DPS and to have a familiarity of the data assimilation in the use of estimation of DPS. The proposed half-day workshop is divided into four lectures addressing various aspects for the control and estimation of DPS. The starting point is the revised book An Introduction to Infinite-Dimensional Linear Systems Theory, Springer-Verlag, New York (1995), by Ruth Curtain and Hans Zwart. The point of departure will be an introduction to Lyapunov stability and colocated systems with boundary control. This introductory talk will be given by Ruth Curtain. Subsequently, Kirsten Morris will extensively discuss the modeling of the actuators and sensors in control design and how this affects system behaviour. Physical intuition does not always lead to the best choice of locations. The incorporation of the sensor technology limitations into the design of distributed state estimation with mobile sensor networks will be presented and discussed by Michael A. Demetriou. A performance-based guidance and spatial repositioning of mobile sensors that considers sensor technology limitations, the computational model used and physical properties of the process will be discussed. The ultimate goal for simulation and control synthesis, is the application of the distributed-parameter systems theory in practice. Thus the final lesson, given by Sergiy Zhuk, will focus on experimental studies which have been done by the speaker in collaboration with T. Tchrakian, A. Akhriev, S. Thirupati, S. McKenna (IBM Research-Ireland), S. Lu and H. Hamann (IBM T.J.Watson Research Center) and S. Moore (IBM Research-Melbourne).

Target audience: Covering both theory and applications, the workshop is of interest for a broad audience.

Organizers: Michael A. Demetriou, Orest V. Iftime

Speakers: Ruth F. Curtain, Kirsten Morris, Michael A. Demetriou, Sergiy Zhuk

Scenario-based optimization for stochastic optimal power flow problems

Workshop overview: This workshop aims to provide systems and control researchers with, 1) an understanding of the challenges of power system operation in the presence of generation uncertainty arising from renewable energy sources, and 2) methodologies for achieving optimal and reliable operation. Generation uncertainty may significantly increase operational costs due to the resulting stochastic formulation and the additional reserve requirements. Existing optimal power flow (OPF) and reserve scheduling algorithms need to be reformulated to provide solutions that are reliable with respect to uncertainty. The workshop will initially focus on a specific class of stochastic optimization tools that can provide solutions with probabilistic performance guarantees. Those tools build on scenario-based optimization and may be applied to both convex and non- convex problems. The workshop will then provide an overview of power flow models and establish optimal power flow formulations. Finally, the workshop will present stochastic formulations of the optimal power flow and reserve scheduling problems, and demonstrate how scenario-based optimization techniques can be applied to achieve reliable and cost-effective operation. Various forms of policy-based feedback control will be presented for generator reserve deployment, and extensions to other forms of controllable resources will be considered. A discussion of the effects of policies on economic performance will be provided.

Target audience: The workshop is intended for systems and control researchers who have interests in the large-scale integration of renewable energy sources in power systems. The expected audience includes engineers and scientists from industry and academia. The workshop will be self-contained so that it is suitable for participants who may not have prior familiarity with the topics.

Organizers and speakers: Ian Hiskens, John Lygeros, Maria Vrakopoulou

Detectability for Time-Varying Systems

Workshop overview: Similar to observability, detectability indicates the ability of a dynamic system to use its output signals from sensors and the model to “observe” the behaviour of the state that can fully capture the characteristic of the system. Thus it plays an important role in many engineering applications in order to design a feedback control law using the measured output signals. For a time-invariant system, the well-defined detectability also indicates its uniform attractivity under mild assumptions. However, it is usually much harder to define such detectability for complex time-varying dynamic systems such as switching systems and hybrid systems. The main objective of this tutorial is to present a few different definitions of well-defined detectability for different types of systems; to reveal the link among detectability, the persistent excitation (PE) and uniform attractivity; and to demonstrate the power of detectability in analysing stability properties for switched nonlinear time-varying systems to researchers in the area of systems and control who may not be familiar with the concept. This tutorial will

• revisit the concept of detectability and its link with uniform attractivity for linear- time-invariant systems;
• demonstrate the difficulty of the well-defined detectability for time-varying systems and revisit various detectability definitions of different dynamic systems in literature;
• introduce the concept of Signal Set, which can serve as a unified framework to characterize a large class of dynamic systems;
• introduce a well-defined detectability, weak detectability (WD) in the Signal Set framework;
• reveal the link among WD, PE condition and uniform attractivity;
• provide sufficient conditions to check WD;
• demonstrate the power of WD in stability analysis of switched nonlinear time-varying systems including arbitrarily switching cases and constrained switching cases with new stability results;
• consider a few examples from the literature to illustrate the strength of the proposed framework.

Target audience: the expected audience includes engineers, scientists, postgraduate students, and academic researchers.

Organizers and speakers: Ying Tan, Ti-Chung Lee, Iven Mareels

30 years of the Ramadge-Wonham Theory of Supervisory Control: A Retrospective and Future Perspectives

Workshop overview: 2017 marks the 30-year anniversary of the publication of the two seminal papers of Ramadge and Wonham on supervisory control of discrete event systems, in the January and May issues of the SIAM Journal on Control and Optimization (SICOPT) in 1987:

• [RW1] P.J. Ramadge and W.M. Wonham, "Supervisory Control of a Class of Discrete Event Pro- cesses," SIAM J. Control Optim., 25(1), 206-230.
• [RW2] W.M. Wonham and P.J. Ramadge, "On the Supremal Controllable Sublanguage of a Given Language," SIAM J. Control Optim., 25(3), 637-659.

Target audience: The speakers will give talks that will be accessible to a wide audience of CDC attendees. The objective will be to combine the "retrospective" theme with the "future perspectives" theme, in presentations that will have tutorial value. Given the emergence of cyber-physical systems as an important area of research and given that there are typically multiple CDC sessions on discrete event systems, formal methods in control, and hybrid systems, we expect that a large number of CDC attendees, in particular students, will be interested in this workshop.

Organizers: Stéphane Lafortune, Karen Rudie, Stavros Tripakis

Speakers: Peter Ramadge, Calin Belta, Kai Cai, José Cury, Martin Fabian, Alessandro Giua, Hervé Marchand, Richard Murray, Necmiye Ozay, George Pappas, Sanjit Seshia

For further information, please go to the workshop webpage.

Gain Scheduled Model Predictive Control System Design and Implementation
THIS WORKSHOP HAS BEEN CANCELED

Workshop overview: Model Predictive Control (MPC) has a long history in control engineering. It is one of the few areas that has received on-going interest from researchers in both the industrial and academic communities. Three major aspects of model predictive control make the design methodology attractive. The first is the design formulation, which uses a completely multivariable system framework where the performance parameters of the multivariable control system are related to the engineering aspects of the system; hence, they can be understood and "tuned" by engineers. The second aspect is the ability to handle "soft" constraints and hard constraints in a multivariable control framework. This is particularly attractive to industry where tight profit margins and limits on the process operation are present. The third aspect is the ability to perform process on-line optimization. MPC systems are designed using linear models unless a nonlinear model is explicitly stated. Nonlinear MPC is conceptually similar to its linear counterpart except that nonlinear models are deployed for the prediction and optimization. However, because of its computational intensity and complexity, nonlinear MPC is not widely applied. Instead, gain scheduled control system techniques have found success in the area of predictive control of nonlinear plants. This one-day short-course will begin by introducing and reinforcing the basic concepts in the design of an effective linear predictive controller. This is followed by the design of a gain scheduled predictive controller: (i) linearization of a nonlinear plant model according to operating conditions; (ii) design of linear predictive controllers using the family of linear models; (iii) gain scheduled predictive control law that will optimize a multiple model objective function with constraints and ensure smooth transitions (i.e. bumpless transfer); (iv) simulation and experimental validation of the gain scheduled MPC with constraints using MATLAB and Simulink as a platform.

Target audience: The course is suitable for engineers, students and researchers who wish to gain basic knowledge about linear and gain scheduled MPC of nonlinear plant, as well as understand how to perform real time simulation and implementation using MATLAB and Simulink tools. It is partially based on the speakers books' entitled "Model Predictive Control System Design and Implementation Using MATLAB" (Wang, 2009), "Model based predictive control: a practical approach" (Rossiter, 2003) [New edition in preparation], and "PID and Predictive Control of Electrical Drives and Power Converters" (Wang, et.al, 2015).

Organizers and speakers: Liuping Wang, J.A Rossiter,

Model-Based and Model-Guided Approaches for Automotive and Maritime Powertrain Calibration and Control
THIS WORKSHOP HAS BEEN CANCELED

Workshop overview: Efficiency and emission associated with internal combustion powertrain systems have been the focus of research efforts in both academic and industrial stakeholders due to more and more stringent governmental regulations and public concerns about the environment and the energy resources. Yet the complexities of internal combustion with both fossil and alternative fuels pose significant challenges to the development of effective and efficient with satisfied performance. In recent years, both model-based and data-driven optimization approaches have been explored to address automotive and maritime powertrain combustion control and calibration in order to squeeze the potential over the entire operation ranges. These are in addition to the traditionally empirical calibration based method and also motivate the newly developed model-guided datadriven optimization approach which essentially emerges as a process learning oriented design. This full-day workshop intends to introduce the motivation and application of model-based and model-guided data-driven optimization with sample studies in both automotive powertrain and maritime platform calibration and control. In particular, discussion will be presented in detail on the control-oriented modeling approach for automotive powertrain with initialization and validation in real-time hardware-in-loop simulations, the concept of model-guided data-driven optimization, as well as the model-based and model-guided calibration and control examples for automotive powertrain and maritime platform, including control of mechatronic actuation subsystem and internal combustion engine systems.

Target audience: This workshop is designed to facilitate both control researchers from academic communities and engineers from relevant industries to conduct model-based and modelguided innovative design for powertrain systems. It is intended to share with audiences the motivation, rationale, challenges, and achievements in the model-based and model-guided powertrain (internal combustion) control approach. Those who are interested in the model-based and model-data integrated approaches for control and optimization would also find the workshop beneficial in terms of both concepts and applications.

Organizers: Guoming Zhu, Xiang Chen, Chris Manzie

Speakers: Guoming Zhu, Xiang Chen, Noam Olshina, Gokul Siva Sankar Gowri Sankar, Ying Tan

Deterministic and Stochastic Hybrid Methods for Robust Learning and Optimization in Dynamical Systems

Workshop overview: The goal of this workshop is to present novel results in the area of deterministic and stochastic hybrid systems applied to robust optimization and learning in dynamical systems. These include, but are not limited to, hybrid extremum seeking control, event-triggered sampled-data optimization, stochastic learning in asynchronous sampleddata games, robust set-based estimation, and event-triggered Q-learning optimal control. The workshop will present constructive approaches for the design of novel robust hybrid algorithms for black-box and grey-box optimization, as well as some engineering applications that motivate the study of hybrid optimization and learning. After finishing the workshop attendees will be familiar with a broad class of hybrid dynamics in the area of black-box and grey-box optimization, control, and estimation, as well as a basic understanding on how to design and analyze some of these hybrid algorithms. The merits of working with hybrid dynamics will be clearly illustrated throughout the workshop.

Target audience: The workshop is intended to be a brief course on recent analysis and design tools for robust algorithms for modelfree optimization and learning using deterministic and stochastic hybrid dynamics. It targets a broad audience in academia and industry, including graduate students looking for an introduction to a new and active area of research; control practitioners interested in novel design techniques and applications; and researchers in dynamical systems, optimization, and control. The workshop audience is not expected to have any advanced background in hybrid systems or optimization. A basic knowledge of linear/nonlinear system theory and probability theory is useful.

Organizers: Jorge I. Poveda, Andrew R. Teel, Mouhacine Benosman

Speakers: Jorge I. Poveda, Andrew R. Teel, Mouhacine Benosman, Kyriakos Vamvoudakis, Chris Manzie, Martin Guay

What’s the big deal with Deep Learning?
Bringing together data-driven learning and feedback control

Workshop overview: The last few years have seen multiple major successes of deep learning. From beating professionals at games like Go, to fast detection of cancer, classification of complex images, and generation of captions for images with incomplete information. In classification and reinforcement learning, deep learning indeed has outperformed all existing machine learning and model-based methods. Fueled by rapid advances in parallel computing and GPUs, new impressive results are achieved almost on a daily basis by deep learning methods in a vast array of applications from finance to self-driving cars. In the midst of this excitement however, there are fundamental questions that remain unanswered: For example, recent results have also shown that deep learning can be highly susceptible to noise and can misclassify images with very high confidence that are very easy for humans to process. If deep learning is going to make its way into self-driving cars and other autonomous systems, it is justifiable to ask just how stable and robust can deep learning based control and decision making algorithms be? Indeed, one could ask a even deeper question: What is the role of feedback in an increasingly data-driven world? This we believe is a question that the controls community can significantly contribute to answering. In the view of this vision, the objective of this workshop is to:

1. Introduce the theory, algorithms, and implementation of Deep Learning to the audience, including conducting tutorials on using deep learning software (Google Tensor Flow). Embedded in this introduction, is a compact introduction to machine learning in general
2. Have a candid conversation about the connection of deep learning with the Neural Network theory that has been developed in the controls community in the late 90s-early 2000s, and the larger controls efforts in system identification, adaptive control, and Kalman filtering
3. Bring together experts from deep learning and machine learning with control theory experts
4. Hold a discussions on what critical roles feedback control and estimation play in a world dominated by data-driven learning

Target audience: The target audience for this workshop are control theorists and practicing controls engineers. We will assume only familiarity with MATLAB and some familiarity with Python programming language (we will circulate material beforehand to the registrants to prime them in Python). Our presentations will be self-contained and leverage material that instructors have developed in teaching courses on related topics UIUC or their organizations. In addition, we will provide a repository of relevant links that help the audience navigate the wide variety of literature and resources on the internet on deep learning.

Organizers and speakers: Girish Chowdhary​, Alexander Schwing​, Hassan Kingravi, Chinmay Soman​

For further information, please go to the workshop webpage.