Speakers

This page lists the confirmed speakers at the iCODE final congress. Abstracts and biographies are provided below.

Plenary speakers:

• Filippo Santambrogio, Univ. Paris Sud: Microscopic and macroscopic modeling of passive and active crowds
• Rodolphe Sepulchre, Univ. Cambridge (UK): Neurons as feedback circuits
• André de Palma, ENS Cachan: Static and Dynamic transportation models of traffic congestion

Invited speakers:

• Control and Neuroscience:
  • Thierry Bal, CNRS - UNIC: Closed-loop interactions between neurons and computer models through dynamic-clamp
  • Bastien Berret, Univ. Paris Sud: Why don't we move slower? The cost of time in the neural control of movement
  • Valérie Ego-Stengel, CNRS - UNIC: Probing neuronal mechanisms of motor control with closed-loop Brain-Machine Interfaces
• Behavioral economics:
  
  • François Pannequin, ENS Cachan: Leaving the market or reducing the coverage? An experimental analysis of the demand for insurance
  • Nicolas Drouhin, ENS Cachan: Optimal control and time consistency
    
  • Guillaume Hollard, X: Gender Differences: Evidence from Field Tournaments (with José de Sousa)
• Large-scale systems and Smart grids:
  • Samer Alfayad, UVSQ: Humanoid platform with Integrated Hydraulic Actuation
  • Ugo Boscain, CNRS:  Anthropomorphic image reconstruction via optimal control and hypoelliptic diffusion
  • Sami Tliba, Univ. Paris Sud: Contribution to the development of an experimental device for physical human-robot interaction in robotics comanipulation

Samer Alfayad: HYDROïD: Humanoid platform with Integrated Hydraulic Actuation

Abstract: HYDROïD (HYDraulic andROïD) is a full-size humanoid robot under development that aims at improving our understanding of the phenomena of locomotion and manipulation of humans. Humanoid with hydraulic actuation are able to achieve hard and useful tasks and replace human in disaster environment. In hydraulic actuation, pistons are used to produce motion. At least one hydraulic piston is implemented for each degree of freedom (DoF). The main difficulty is to bring hydraulic energy to each piston. Hydraulic pipes are usually employed to drive hydraulic power from the control unit to pistons. Each piston needs two pipes, one to drive fluid “in” and the other to drive it “out”. In robotic applications, flexible hydraulic pipes are used to drive the oil in parallel to the joints. This solution suffers from three main disadvantages: I) The hydraulic pipes connected in parallel to joint will give a spring effect. II) The more the number of pipes, the more of the leakage probabilities. III) External pipes increase dramatically the robot size and decrease its anthropomorphic aspects. To answer all these questions, a new “integrated hydraulic actuation” method was proposed and implemented on HYDROïD. The goal is to eliminate all external pipes and replace them with integrated hydraulic passages. Fluid paths is integrated internally through the mechanical structure and not externally through pipes. In other words, “arteries” and “veins” were built inside the HYDROïD body to drive hydraulic fluid like blood in human body. This presentation will focus on two research areas. First, we will focus in two innovative hybrid mechanisms, each consisting of a rotating actuator carrying a parallel structure with two active DoF. The first type has been dedicated to the modules of the hip, shoulder and torso. The second type, actuation of parallel structure with cables was chosen for the ankle, the wrist and the neck modules. The second part of this presentation will be dedicated to the actuation of the HYDROïD robot for which a new highly integrated actuator has been proposed. The actuation principle will be detailed and the benefits of the proposed solution will be shown. Very interesting performance of the realized prototype will be presented.

Bio: Samer Alfayad received his master diploma in sciences and technology from Ecole Nationale Supérieure d’Arts et Métiers (ENSAM-Paris) in 2005. His Ph.d was received in Robotic development from Versailles University (UVSQ) in 2009. Awarded of the best Ph.D Thesis in Robotics for 2010 by the French CNRS. Awarded of the best Ph.D Thesis in Robotics for the 20 years anniversary of UVSQ. From 2009 to 2010, he was a Psot-doc at UVSQ. From 2010 to 2011 he was a Post-doc at Technische Universitat Munich (TUM-Germany) with a Scholarship from the Alexander Von Humboldt foundation. In 2011 he has been appointed as Associated Professor in Humanoid robotic design at Versailles University. Since 2012, he holds an industrial excellence chair about hydraulic domestication at UVSQ. He has been investigator in several French National projects. He is the leader of the Humanoid research group at LISV. He is currently leading the team in charge of HYDROïD (anthropomorphic biped robot) development.  

Thierry Bal: Closed-loop interactions between neurons and computer models through dynamic-clamp

Abstract: The dynamic-clamp electrophysiological technique allows the mimicking of the electrical effects of arbitrary ion channels, controlled by the experimentalist, activating and inactivating into the membrane of a biological cell intracellularly recorded in vitro or in vivo. Dynamic clamp relies on the establishing of a loop between the injected current and the recorded membrane potential. We applied the technique to a number of questions: these include the manipulation of intrinsic ion channels and of single or large number of synaptic inputs to a cell, as well as the construction of hybrid networks in which the biological cell interacts with model cells simulated in real time using a digital system. To illustrate this, I will summarize our findings in the thalamocortical system in vitro (Behuret et al. Front. Neural Circuits, 2015), and propose the hypothesis that top-down corticothalamic synaptic activity is adapted to modulate the transfer efficiency of thalamocortical neurons during selective attention at three different levels: First, on ionic channels by interacting with intrinsic membrane properties, second at the neuron level by impacting on the input-output gain, and third even more effectively at the cell assembly level by boosting the information transfer of sensory features encoded in thalamic subnetworks. 

Bio: Thierry Bal is biologist and tenured research director at CNRS in experimental neuroscience. After a PhD on neuronal networks in invertebrates in Bordeaux, France, he held postdoc and research scientist positions at the Yale school of medicine, USA, with Pr. D.A. McCormick. He is leader of a workgroup in the CNRS UNIC lab in Gif-sur-Yvette. His team has pioneered experimental strategies using hybrid biological-artificial neuronal networks (cell-machine interface based on dynamic-clamp) published in Nature and Nature Neuroscience. They recently showed that the computer-based cell-machine interface could control in real time biological synaptic plasticity homeostasis in the neuromuscular synapse (eLife, in press). In vitro techniques include the preservation of spontaneous rhythmic in vivo-like neuronal activities in slices, and a novel “cellular” voltage sensitive dye technique (collaboration with Yale) with unprecedented temporal resolution to explore voltage transients in dendrites and axons in neurons.

Bastien Berret: Why don't we move slower? The cost of time in the neural control of movement

Abstract: Most daily life actions are executed at a speed which is neither too fast nor too slow. While movement duration (or averaged speed) is a fundamental characteristic of motor control, the principles underlying its formation, be they neural or computational, remain in practice little understood. This talk will address that issue within the inverse optimal control framework where the challenge is to uncover what optimality criterion underlies a system's behavior. Here we rely on the “cost of time” theory that finds its roots into the brain's tendency to discount the actual value of future reward. It asserts that the time elapsed until action completion entails a cost, thereby making slow moves non-optimal. By means of a thorough theoretical analysis, we will show it is actually possible to sample the infinitesimal values of the time cost without making any prior assumption about its hypothetical nature and resorting to parametric fitting. Our results will emphasize its sigmoidal shape for arm reaching, thereby ruling out linear as well as purely concave or convex time costs for limb movement control. These findings may offer a unique opportunity to study how the brain values the passage of time in healthy and pathological motor control, and thus shed new light on the processes underlying action invigoration in human movement.

Bio: Bastien Berret was educated in Dijon (France), where he studied mathematics and computer sciences at the Université de Bourgogne. From 2005 to 2008, he was a PhD student in applied mathematics and movement neuroscience under the supervision of Prof. Thierry Pozzo and Prof. Jean-Paul Gauthier. His work was supported by the French spatial center (CNES) and aimed at investigating the role of gravity during the planning and execution of human movement by mixing computational and experimental approaches. In 2009 he joined the Fondazione Istituto Italiano di Tecnologia (IIT, Genoa) as a post-doctoral fellow where he pursued his theoretical work about neural motor control with a humanoid robotic perspective in collaboration with Dr. Francesco Nori. Since 2012 he is Assistant Professor at the Université Paris-Sud in Orsay (at CIAMS laboratory) and still contributes to develop models to better understand how the brain controls voluntary movement.

Ugo Boscain: Anthropomorphic image reconstruction via optimal control and hypoelliptic diffusion

Abstract: In this talk I will present a model of geometry of vision due to Petitot, Citti Sarti and our research group. One of the main features of this model is that the primary visual cortex V1 lifts an image from R^2 to the bundle of directions of the plane. Neurons are grouped into orientation columns, each of them corresponding to a point of this bundle. In this model a corrupted image is reconstructed by minimizing the energy necessary for the activation of the orientation columns corresponding to regions in which the image is corrupted. The minimization process intrinsically defines an hypoelliptic heat equation on the bundle of directions of the plane. The numerical integration of this equation is difficult and require techniques of non-commutative Fourier analysis. The purpose of this research is to validate the biological model and to obtain an algorithm of image inpainting going beyond the state of the art.

Bio: Ugo Boscain is a Directeur de Recherche CNRS at the CMAP (Centre de Mathématiques Appliquées) de L'Ecole Polytechnique. He is deputy team leader of the team Inria GECO. He graduated in theoretical physics at the University of Torino (1996) and got PhD in mathematics (control theory) at SISSA in Trieste (2000) under the supervision of Benedetto Piccoli. After a post doc with Jean-Paul Gauthier at the Bourgogne University, he was assistant professor in SISSA (2001-2006) and then he joined CNRS in 2006. His domain of research is geometric control theory with ramifications in quantum mechanics (quantum control), Riemannian and sub-Riemannian geometry, and bio-mimetic image processing . He is associate editor of SIAM Journal of Control and Optimization, associate editor of COCV, and managing editor of the Journal of dynamical and Control Systems. He is an ERC laureate 2009.

André de Palma

Abstract: The mathematical formulation of static network equilibrium (for private transport), dates back to 1956 and has been implemented about 20 years later. In dynamic models congestion depends on the time of the day. Such models have been introduced in Economics for one route by William Vickrey, and rediscovered and extended in many dimensions a few years later. Surprisingly, these models have not been very successful in practical applications, and the profession still mainly uses static models. One of the reason, is that the dynamic formulation is very hard to solve mathematically (theoretically and from the computational point of view). Intermediary formulations (stable dynamics) have been proposed. METROPOLIS was the first attempt to compute dynamic traffic equilibrium solutions for large networks.

Bio: André de Palma holds a PhD in Physics (1981), under the supervision of Nobel Prize Y. Prigogine, (1981, Free University of Brussels), and a PhD in Economics (1988, University of Bourgogne). He is specialized in Transportation and Urban Economics, Industrial Organization, Decision theory. With Simon Anderson and Jacques Thisse, he is one of the father of the new industrial organization combined with discrete choice theory. He has introduced dynamic models in Transportation with Richard Arnott, Robin Lindsey and Moshe Ben-Akiva, and developed, with Yurii Nesterov, a fully dynamic congestion software: METROPOLIS. André de Palma is currently Professor at Ecole Normale Supérieure de Cachan / University Paris-Saclay. He has published more than 250 articles and 7 books. He belongs to the top 1% of economists’ world-wide and is ranked #3 in Transportation Economics, according to RePEc. With Nathalie Picard, he has developed two websites to study experimentally risk attitude. Together, they have developed software, based on advanced econometric methods, marketed by the company RiskDesign and used by major banks. He is a founding member of ITEA (International Transport Economic Association), and in the editorial board on several international Journals.

Nicolas Drouhin: Optimal control and time consistency

Abstract : For a long time, the “exponential discounting” model has been the main tool for modeling intertemporal choice of consumption and savings in economics and finance. However since thirty years many evidences have been accumulated against the capacity of this model to describe the actual behavior of economic agents. Alternative model exists, but they are conventionally assumed to be incompatible with a fundamental criteria for rationality, time constancy. My research uses optimal control to prove that this last statement is not true. Many models exist that are different from the “exponential discounting” one and still compatible with time consistency. On the one hand, I characterize rigorously the general properties of the functional belonging to the set of all the time consistent utility functions, and provide some example. I emphasize particularly the case of uncertain lifetime. On the other hand, I also characterize theoretically , for the first time, the observable consumption path of a time inconsistent agent.

Valérie Ego-Stengel: Probing neuronal mechanisms of motor control with closed-loop Brain-Machine Interfaces

Abstract: Brain-machine interfaces use neuronal activity to control prostheses, with the long term goal of restoring motor abilities to impaired subjects. In our recent studies, several motor cortex neurons were recorded simultaneously in head-fixed awake rats and were trained, one at a time, to modulate their firing rate up and down in order to control a one-dimensional actuator carrying a water bottle. The goal was to maintain the bottle in front of the mouth, allowing the rat to drink. All conditioned neurons adapted their firing rate to the instantaneous bottle position so that the drinking time was increased relative to chance. The mean firing rate averaged over bottle trajectories depended on position. The conditioned neuron reacted faster and led to a better control than if bottle trajectories were simulated using the activity of simultaneously recorded surrounding neurons. These studies demonstrated that conditioning single neurons is a suitable approach to control a prosthesis in real-time.
In an ongoing project, we aim to develop a more efficient motor brain-machine interface by including a sensory feedback. Using optogenetic tools in the mice, we will deliver the feedback directly into the somatosensory cortex (SI). This sensory-motor loop will be constrained into time windows similar to those observed in the sensory-motor system of the animal. Our main objective will be to assert the usefulness of a fast, closed-loop BMI to drive adaptive behavior in a rapidly changing environment.

Bio: Valérie Ego-Stengel received a PhD in neuroscience in 2002 from Univ. Paris VI on the subject "Neuromodulation and plasticity of functional properties of cortical neurons: study in somatosensory and primary visual cortex". In 2002-2007, she was a post-doctoral fellow at the Picower Institute for Learning and Memory, MIT, Cambridge, USA, where she studied the role of reactivation, during sleep, of wake-up hippocampal activity. Since 2009, she is a CNRS researcher at UNIC, in the team "Sensory processing, Neuromodulation, and Plasticity".

Olivier Gossner: Crisis and coordinations

Abstract: Crisis are macroeconomic phenomena often triggered by the sum of individual behaviors. At the heart of crisis are the questions of coordinations and anticipations. They are a dynamic phenomenon since individual behavior at any moment depends on the agents’ anticipations of crisis in the future. We present a model of dynamic crisis under private information that leads to a unique equilibrium and allows to derive comparative statics on the game’s fundamentals.

Bio: Olivier Gossner graduated from ENS Ulm in 1993. He holds a CNRS researcher position since 1998. He had visiting positions abroad in Israel, Belgium, USA, and UK. He is currently research director at CNRS and Head of the Economics Department at Ecole Polytechnique. His main scientific interest are in game theory, repeated games, bounded rationality and the role of information in strategic interactions. 

Guillaume Hollard: Gender Differences: Evidence from Field Tournaments (with José de Sousa)

Abstract: Women are under-represented in top positions, such as in business or in politics. Traditional explanations, like differences in productivity and discrimination, are now complemented by psychological explanations based on lab experiments. We provide an attempt to assess the comparative importance of psychological and traditional explanations in a natural field experiment, namely chess competitions. Controlling for discrimination and productivity, we find that women are suffering a systematic handicap when playing against men. This "psychological" effect is further amplified through the tournament structure, preventing women from reaching top positions in the chess hierarchy. The effect is only marginally smaller when we consider the most experienced individuals or the most women-friendly countries.

Bio: Senior researcher at CNRS and Associate professor at Ecole Polytechnique. PhD EHESS (1997). Research interests: Public and individual decision making, experimental economics and neuroeconomics.

François Pannequin: Leaving the market or reducing the coverage? An experimental analysis of the demand for insurance

Abstract: This study develops an experimental analysis addressing the premium sensitivity of the demand for insurance accounting for risk attitudes, including risk-loving. Our contribution disentangles the conditional demand (the non-null demand for insurance) from the propensity to buy insurance. Our research shows that the contraction of the global demand for insurance induced by the raise in unit prices and fixed cost is primarily due to policyholders exiting the insurance market rather than reducing their levels of coverage. However, contrary to the theoretical predictions, an increase in the fixed cost has effects only on the risk lovers’ behavior. The stability of the conditional demand is robust to changes in insurance contracts and individuals’risk attitude. These results suggest that the decision about insurance may boil down to an “all or nothing” choice.

Bio: François Pannequin is Head of the Department of Economics and Management at ENS-Cachan, member of CES-Cachan (Centre d’Economie de la Sorbonne-Cachan) and CREST. He is the coordinator of the Research Initiative in Behavioral Economics of iCODE project. His main research interests are: Insurance Economics, Contract Theory and Experimental Economics.

Filippo Santambrogio: Microscopic and macroscopic modeling of passive and active crowds

Abstract: In the talk, I will start from recent researches, performed in the PDE group at LMO, on crowd motion, both in a microscopic and in a macroscopic setting. In the former case individuals are represented as small particles with non-zero size, subject to non-overlapping constraints, and in the latter a continuous model is considered, using PDEs and transforming the constraints into a bound on the maximal density. In these models, the desired movement of each individual is supposed to be known, in the terms of a given velocity field which has to adapted to the constraints. The key idea is that ``the velocity fields that actually advects the motion is the projection of the desired one on the set of admissible fields (i.e. those which preserve the constraint)'', and this projection lets a pressure appear. Then, I will move to a strategic framework (where the crowd is supposed to be active, in contrast to the previous setting), where individuals try to anticipate the choices of the others, thus giving raise to equilibrium questions typical of congestion games. I will present two models to describe these phenomena arising from the theory of Mean Field Games, introduced by Lasry and Lions some ten years ago.

Bio: Filippo Santambrogio, 35 yo, is a French-Italian scholar working in the fields of optimal transportation and calculus of variations. He studied at the Scuola Normale Superiore in Pisa, where he also prepared a PhD thesis supervised bu G. Buttazzo. He moved to France for a post-doc at ENS Cachan and gor a Maître de Conférences position in Paris-Dauphine in 2007. In 2010 he moved to Université Paris-Sud in Orsay, as a professor. He carries on research in applied mathematics, using tools from mathematical analysis, PDEs and optimization to study different models related to networks, traffic congestion, population dynamics, economical equilibrium...

Rodolphe Sepulchre: Neurons as feedback circuits

Abstract: The foundation modeling paper of neurodynamics by Hodgkin and Huxley in 1952 contained two mathematical representations of neurons: an electrical circuit and a state-space representation. Sixty years later, the latter largely dominates. In the age of neuroscience in silico, novel quantitative biological knowledge is easily added to computational models in the form of extra differential equations. Our talk will illustrate why the most basic questions of experimental neurophysiology are intractable in this formalism. Control theory suggests that questions pertaining to robustness, sensitivity, and (neuro)modulation must be addressed in the input-output framework, regarding neurons as feedback interconnections of electrical circuits. We will discuss a novel methodological framework to quantify the sensitivity of neuronal activity to changes in ion channel densities, either from a detailed conductance-based model or from voltage-clamped experimental data. We will illustrate the generality of the framework and its potential to improve our understanding of the regulation of brain functions and to help in the design of new neuromodulatory treatments.

Bio: Rodolphe Sepulchre is Professor of Engineering at Cambridge University and a fellow of Sidney and Sussex College. His research interests are in nonlinear dynamics, control and optimization, with a current focus on neuronal behaviors. He is currently Editor-in-Chief of Systems and Control Letters and an Associate Editor for IEEE Transactions on Networks and Systems and for Annual Reviews in Control, Robotics, and autonomous systems. In 2008, he was awarded the IEEE Control Systems Society Antonio Ruberti Young Researcher Prize. He is a fellow of IEEE (2010) and SIAM (2015) and an IEEE CSS distinguished lecturer since 2010. He is a member of the Royal Academy of Belgium (class science and technology) since 2013.

Sami Tliba: Contribution to the development of an experimental device for physical human-robot interaction in robotics comanipulation