Dimitris Bertsimas
Video

Tariq Samad
Video

Ron Weiss
Video

Duncan Callaway
Video

Marco Campi
Video

Hartmut Geyer
Video

Holger Hermanns
Video

Damon Vander Lind
Video

Glenn Vinnicombe
Video coming soon.

Bo Wahlberg

Colin Jones
Michal Kvasnica
Martin Herceg

Daniel Kuhn
Karthik Natarajan

Ian Hiskens

Alexandre Bayen
Jack Reilly

Henrik Sandberg
Bruno Sinopoli

Heinz Koeppl

Tutorial Overview
(106 KB PDF)

Dimitris Bertsimas:

A computationally tractable theory of performance analysis in stochastic systems

Watch Video

Modern probability theory, whose foundation is based on the axioms set forth by Kolmogorov, is currently the major tool for performance analysis in stochastic systems. While it offers insights in understanding such systems, probability theory is really not a computationally tractable theory. Correspondingly, some of its major areas of application remain unsolved when the underlying systems become multidimensional: Queueing networks, network information theory, pricing multi-dimensional financial contracts, auction design in multi-item, multi-bidder auctions among others.

We propose a new approach to analyze stochastic systems based on robust optimization. The key idea is to replace the Kolmogorov axioms as primitives of probability theory, with some of the asymptotic implications of probability theory: the central limit theorem and law of large numbers and to define appropriate robust optimization problems to perform performance analysis. In this way, the performance analysis questions become highly structured optimization problems (linear, conic, mixed integer) for which there exist efficient, practical algorithms that are capable of solving truly large scale systems.

We demonstrate that the proposed approach achieves computationally tractable methods for (a) analyzing multiclass queueing networks, (b) characterizing the capacity region of network information theory and associated coding and decoding methods generalizing the work of Shannon, (c) pricing multi-dimensional financial contracts generalizing the work of Black, Scholes and Merton, (d) designing multi-item, multi-bidder auctions generalizing the work of Myerson.

This is joint work with my doctoral student at MIT Chaithanya Bandi.

Tariq Samad:

Control for smart grids: Applications and opportunities in the customer domain

Watch Video

One of the principal points of differentiation between smart grids and traditional power systems is the importance the former accord to electricity consumers and customers-residential, commercial, and industrial facilities are an integral part of the smart grid system of systems. Furthermore, the portfolio of energy assets in homes, buildings, and factories is being dramatically enhanced; distributed generation and storage, in addition to heterogeneous loads and in grid-connected implementations, manifest that “demand-side” management is not just about demand any longer.  Realizing the smart grid vision of clean, low-cost, and securely supplied energy depends crucially on the active engagement of facility owners, operators, and assets. With measurement, communication, and computing becoming pervasive in the expanded end-to-end electricity infrastructure, the stage is set for control technologies. Notable progress has been made already. Several innovative and exciting projects are under way worldwide and significant economic and societal benefits have been achieved, especially in areas such as automated demand response, optimal microgrids, and energy analytics. Yet we have just scratched the surface of possibility. Faster dynamics, larger-scale systems, more diverse assets, dynamic pricing, humans in the loop, and other complexities and trends all suggest opportunities for control research with the promise of tremendous impact on society and industry.

Ron Weiss:

Synthetic biology: From parts to modules to therapeutic systems

Watch Video

Synthetic biology is revolutionizing how we conceptualize and approach the engineering of biological systems. Recent advances in the field are allowing us to expand beyond the construction and analysis of small gene networks towards the implementation of complex multicellular systems with a variety of applications. In this talk I will describe our integrated computational / experimental approach to engineering complex behavior in living systems ranging from bacteria to stem cells. In our research, we appropriate design principles from electrical engineering and other established fields. These principles include abstraction, standardization, modularity, and computer aided design. But we also spend considerable effort towards understanding what makes synthetic biology different from all other existing engineering disciplines and discovering new design and construction rules that are effective for this unique discipline. We will briefly describe the implementation of genetic circuits and modules with finely-tuned digital and analog behavior and the use of artificial cell-cell communication to coordinate the behavior of cell populations. The first system to be presented is a genetic circuit that can detect and destroy specific cancer cells based on the presence or absence or specific biomarkers in the cell. We will also discuss preliminary experimental results for obtaining precise spatiotemporal control over stem cell differentiation for tissue engineering applications. We will conclude by discussing the design and preliminary results for creating an artificial tissue homeostasis system where genetically engineered stem cells maintain indefinitely a desired level of pancreatic beta cells despite attacks by the autoimmune response, relevant for diabetes.

Duncan Callaway:

Demand-side modeling, estimation and control in electric power systems

Watch Video

For over a century, one of the central objectives in power system operation and control has been to balance constantly varying electricity demand. Currently, a suite of decentralized and centralized control actions on the supply side is used to manage expected and unexpected changes in demand. Now, however, the supply side is becoming part of the problem, as increasing amounts of wind and solar technologies are added to the grid with variability and uncertainty that easily exceeds that of demand. How will these new challenges be met? An interesting possibility is to make the demand side part of the solution by exploiting latent flexibility in electricity consumption patterns. Quite simply, this means consuming more electricity when wind or solar power is available, and less when unavailable. But, of course, the devil is in the details.

This talk will cover recent efforts to understand if and how system operators can elicit power system flexibility by engaging the demand side. The central control challenge is that end-use function (e.g. thermal comfort, battery state of charge, illumination) is often at odds with system-level objectives to balance real power supply and demand. Balancing these competing objectives first requires models that adequately describe the end-use process but are sufficiently general to be easily parameterized and included in model-based control schemes.  I will discuss these modeling issues and results from my group and others’, with extensive discussion on the role of statistical aggregation. I will then discuss recent research results to highlight the tradeoffs of different control architectures – ranging from completely centralized to fully decentralized – with emphasis on integration into legacy system operations and the cyber-physical systems perspective. Finally I will discuss the importance of models and control system design on the challenge of estimating the size of flexible demand resources.

Marco Campi:

The scenario approach to stochastic optimization

Watch Video

Many design problems in control, identification and signal processing can be expressed as optimization problems. In many cases, the environment in which the optimization is performed contains uncertain elements, and the designer acquires knowledge about uncertainty through experience, that is, by looking at previous cases, or “scenarios”, of the same problem. This is the set-up in which the scenario approach operates. The scenario approach has a practical appeal due to its simplicity. On the other hand, it is also a mathematically solid method whose justification is grounded on a rigorous generalization theory.

In the talk, we shall introduce the scenario approach, and explore some of its potentials and properties.

Hartmut Geyer:

Legged dynamics and control: Basic models, neuromuscular interpretation, and robotic application

Watch Video

Legged mobility forms one of the great challenges and opportunities to scientific understanding and technological development. One day, legged machines will help humans in everyday mobility tasks and powered exoskeletons and prostheses will seamlessly enhance, restore or replace the functionality of human legs. At present however much of the science of legged dynamics and control still needs to be discovered, and much of the technology, to be developed. This talk presents one specific approach toward this goal that emphasizes the inherent connection between the basic physics of legged systems, human neuromuscular control, and technological solutions in robotics.

The talk is organized in three parts. First, we review basic models of legged dynamics and control. We then address how their systematic interpretation with spinal reflexes leads to neuromuscular models which predict human muscle activations and generate a variety of human locomotion behaviors such as walking, running, obstacle avoidance, and stair climbing. Finally, we present work on technological solutions in humanoid and rehabilitation robotics that is inspired by the control strategies identified in the basic and neuromuscular models.

Holger Hermanns:

Power grid stability despite renewable instability

Watch Video

The electric power grids across Europa are evolving towards decentralized structures, rooted in political decisions to counter the worldwide climate change. Especially the German power grid is facing the disruptive challenge to integrate massive decentral infeed of renewable electric power. This implies drastically higher volatility in available electric power, which in turn is a threat to power grid stability across the entirety of continental Europe.

This presentation discusses the challenge as a large-scale distributed control problem. We derive a set of principles that inspire the design of a new generation of distributed and decentralised power grid control appliances. We report on controller designs for masses of microgenerators, and contrast their reliable operation with critical shortcomings of controllers installed on hundreds of thousands of German rooftops to date. While doing so, we discuss a modelling and verification technology revolving around stochastic hybrid automata, employed to validate and evaluate these controllers.

Damon Vander Lind:

The design and control of airborne wind turbines

Watch Video

Wind power has seen tremendous growth over the past decade. However, conventional turbine technology has largely matured. It is unlikely to see substantial cost reductions as turbine cost is largely dictated by the cost of structural materials, and the levelized cost of wind energy has in fact increased over this timeframe. Continued breakout growth for the sector will depend upon truly disruptive innovation that expands the developable resource while significantly reducing cost. Makani's Airborne Wind Turbine offers these advantages, generating 50% more energy, while using 10% of the materials of conventional turbines per unit of capacity. In effect, Makani is replacing steel and fiberglass with computation, both in the way the flight system is controlled, and in the way the system is designed. We will look at the mechanisms through which design and control interact to yield a robust system capable of elegantly responding to high winds, turbulence, and component failures.

Glenn Vinnicombe:

Information processing and control in biological systems; some fundamental limits

Video coming soon.

Many biological processes, from insect vision to gene regulation, are constrained by the effects of delays and small numbers. There are many different ways of attempting to quantify this. In order to rigorously capture the limitations, to keep us honest, we choose to look at their impact on the system's ability to solve causal estimation and control problems. Models of biological systems are, at best, sparsely characterised; with large margins of error even then. Our general approach is to assume that some observable aspects of the sensing or regulatory network are known and to then abstract away the rest of the network by optimising over it. What is the best that nature could be doing, given the constraints it is acting under? For the regulation of copy numbers of molecules in the cell we will show that the limitations imposed by delays in response or by small numbers of signalling molecules are severe, particularly when they occur in combination. For insect vision, an understanding of the fundamental limitations in early visual processing leads to a better understanding of more complex problems. We propose this approach as a new way of studying biological dynamics; where lack of knowledge need not prevent us from saying some things with confidence.