Special sessions about the following topics will be presented:

Time: 12:15 - 13:30, Wednesday, December 11, 2019
Location: Risso 8
Title: An overview of research activities at MERL (Mitsubishi Electric Research Laboratories), Control and Dynamical Systems Group
Speakers: Karl Berntorp, MERL Principal Research Scientist - Rien Quirynen, MERL Research Scientist

Abstract: Mitsubishi Electric Research Laboratories (MERL) is a leading research organization located in Cambridge, Massachusetts, USA that conducts fundamental research for industrially-motivated problems. In this talk, we will present an overview of research activities at MERL, including fundamental controls research and the application of state-of-the-art control techniques to a variety of products. We will focus on fundamental research topics including model predictive control and the control of constrained systems, estimation and motion planning for autonomous systems, and learning for control. In addition, we will describe how these fundamental research areas have impacted applications such as autonomous vehicles, energy-efficient HVAC systems, high-precision manufacturing, traffic control, and spacecraft guidance and control.

Speakers:Karl’s Berntorp research is on statistical signal processing, motion planning, sensor fusion, and optimization-based control, with applications to automotive, aerospace, transportation, and communication systems. His work includes design and implementation of nonlinear estimation, constrained control, and motion-planning algorithms.
Rien's Quirynen research interests are in model predictive control and moving horizon estimation, numerical algorithms for (nonlinear) dynamic optimization and real-time control applications. His doctoral research was focused on numerical simulation methods with efficient sensitivity propagation for real-time optimal control algorithms.

Time: 12:30 - 13:30, Wednesday, December 11, 2019
Location: Galliéni 5
Title: Secure and efficient with adaptive control - a story of one equation that brought new perspectives for Linux servers and cybersecurity systems
Speaker: Michał Karpowicz, NASK

Abstract: As a National Research Institute executing governmental cybersecurity tasks on one hand and providing commercial IT services on the other, NASK is in constant need of technological solutions that prove to be both secure and efficient. Rapidly changing patterns of cyberattacks and ever-growing demand for computing capacity result in excessive costs of network services. Therefore we are focused on developing solutions for cybersecurity and energy-efficient data center management. Our recent findings show that challenges arising in these areas call for the application of adaptive control theory. And it all started with one equation…

Speaker: Michał Karpowicz is Assistant Professor of Computer Science and Head of IT Systems Engineering Department at NASK National Research Institute for Cybersecurity & AI. He received his B.S., M.S., and Ph.D. from the Institute of Control and Computation Engineering at the Warsaw University of Technology. His research interests include control theory, signal processing, and game theory.

Time: 18:30 - 20:30, Wednesday, December 11, 2019
Organizer and moderator: Antoine Chaillet (Centrale Supelec)
Location: Rhodes exhibition area - Level 2

Building on the success of the past several events, the 2019 CDC will feature the "Meet the Faculty Candidates" poster session. This poster session provides a great opportunity for faculty, search committee members, and recruiters to speak directly with current graduate students and postdoctoral researchers who are seeking faculty positions.

Faculty candidates, registered for the conference, are invited to register for this poster session by completing this online registration form by November 15th, 2019 December 1st, 2019. The session will be held on Wednesday, December 11th, from 6:30pm to 8:30pm at the Acropolis Convention and Exhibition Center. Space will be available on a first-come first-serve basis.

Presenters are asked to bring a poster no larger than 3ft x 4ft (A0 format) along with pushpins to attach the poster. Presenters will likely be more successful providing high-level discussions of their work such as motivation, strategies, unique insights, rather than narrow mathematical detailed discussions, unless asked specifically for those details. Presenters are also encouraged to bring copies of their CV for distribution. If you have questions please contact Antoine Chaillet at antoine.chaillet@centralesupelec.fr

Recruiters: meet great faculty candidates ! This year, more than 40 faculty candidates will introduce themselves through a poster. If you are looking for brilliant and dynamic candidates, feel free to visit! No registration is needed.

Time: 12:15 - 13:45, Thursday, December 12, 2019
Location: Hermès Amphitheater
Title: Reinforcement Learning: Leveraging Deep Learning for Controls
Speaker: Craig Buhr, MathWorks

Abstract: Reinforcement learning is getting a lot of attention lately. People are excited about its potential to solve complex problems in areas such as robotics and automated driving, where traditional control methods can be challenging to use. In addition to deep neural nets to represent the policy, and algorithms to train them, reinforcement learning requires repeated exploration of the environment. As such exploration is time consuming and potentially dangerous when done with the hardware, a simulation model is often used to represent the environment, at least for the initial training.
In this talk, we will discuss reinforcement learning and contrast it with traditional control methods. We will go through the steps needed to set up and solve a reinforcement leaning problem. We will then talk about relevant MathWorks capabilities and resources and will show an example of developing a robot controller using reinforcement learning. Topics include:

• Creating MATLAB and Simulink environment models and provide observation and reward signals for training policies
• Training of policies using various reinforcement learning algorithms
• Parameterizing policy and value functions using deep neural networks, linear basis functions, and look-up tables
• Parallelizing environment simulations and gradient calculations on GPUs and multicore CPUs for policy training
• Deploying trained policies to embedded devices through automatic code generation for CPUs and GPUs
• Implementing controllers using reinforcement learning for automated driving and robotics applications

Speaker: Craig Buhr
Craig Buhr received his B.S., M.S. and Ph.D. from the School of Mechanical Engineering at Purdue University. He joined MathWorks in 2003 as a Senior Developer for the Controls and Identification products. He currently manages the Control Design development team, whose focus is developing tools for the design and analysis of control systems. His research interests include dynamic system modelling, control theory, machine learning, reinforcement learning and computer aided control system design.

Time: 12:15 - 13:15, Thursday, December 12, 2019
Location: Galliéni 5
Speakers: Marios Polycarpou (University of Cyprus), Sandra Hirche (TUM Munich, Germany), Telma Carvalho (ERC Executive Agency)

ERC grants support individual researchers of any nationality and age who wish to pursue frontier research in any field of science. The ERC encourages in particular proposals that cross the disciplinary boundaries, pioneer ideas that address new and emerging fields and applications that introduce unconventional and/or innovative approaches.
The ERC Session presents the current funding opportunities and discusses the evaluation and submission process from the perspective of a grantee and panel member.