Yaneck Gottesman, a metrologist and specialist in the analysis and characterization of components, contributes to the development of the Optics and Photonics laboratory at Télécom SudParis. The lab is equipped with innovative high-performance electronic instruments for measurement, with multiple applications covering the fields of healthcare, telecoms and security.

Why does a component stop working? What happened in its transition from ‘working’ to ‘broken’? What unforeseen physics are at work? This is what Yaneck Gottesman tries to understand; he is interested in “seeking out the flaws in our understanding of everyday objects“. Metrology, or the science of measurement, demands precision and adaptability and is at the heart of the Professor’s research at Télécom SudParis. An example of this is an experiment during which the optical properties of a mirror produced results that were difficult to explain. It took a year of questioning the framework of interpretation to show that the object was affected by vibrations that were almost undetectable because they were of an amplitude of considerably less than a micron. The benefit of this experiment was that it enabled the team to develop their expertise and establish a protocol for characterizing the dynamic properties of the objects studied: “when I carry out an analysis I always have to ask myself what it is I am really measuring and whether I need to reconsider the model. It is a question of being able to disassociate the object being measured from the instrument used to measure it.” Here, it was necessary here to break free of a supposedly static context in order to carry out a full dynamic analysis.

‘Homemade’ instruments to meet specific requirements

This ability to break free of the conventional field is a fundamental quality that allows the development of these specific tools and is a strength of the Optics and Photonics laboratory. Such tools include the OLCI (Optical Low Coherence Interferometry) and the OFDI (Optical Frequency Domain Interferometry). These two instruments have been specially developed for measurement and analysis with micrometric resolution over distances of up to 200 m (depending on the instrument used). Interferometry is a method that uses two signals produced by the same optical source, one of which serves as a reference while the other examines the object to be analyzed. Data is obtained by the superimposition of these two signals, which have undergone different conditions of diffusion.

In the case of research work carried out by various French laboratories on next-generation fiber optics, for example, the tools on the market were not suitable due to the ambiguity of interpretation of results recorded by these instruments. This difficulty led Yaneck and his colleagues to work on controlling the properties of emission from the source, obtaining a flexible interferometric architecture and controlling the full signal processing chain. This approach was a determining factor in demonstrating an unusual property of the fibers studied.

The prototype of the reflectometry bench developed in the field of frequencies: this system is used, among other things, for the spectral, spatial and modal characterization of optoelectronic circuits and components

Quality and variety of the data recorded

What really adds value to the work carried out on the laboratory’s instruments, however, “is the quality and variety of information recorded simultaneously when an object is measured”. One of the initial, very important and difficult challenges is the limit in terms of absolute precision of an instrument. Here, it is assured by a systematic approach that involves the combination of benchmarks, methods of optical referencing and electronic circuits used to digitally compensate for optical fluctuations specific to the environment and the instrument. The second challenge concerns the diversity of the information recorded. The solution proposed consists in developing instruments capable of simultaneously recording all vectoral values of the light collected, such as intensity, temporal and spatial phases and polarization. Thanks to this ‘all in one’ approach, each instrument becomes a spectrum analyzer as well as an ellipsometer, a telemeter or a Doppler scanner. Moreover this diversity provides detailed information on the electromagnetic field that is otherwise unobtainable.

The approaches that have been developed, some of which have been patented by Télécom Sud-Paris, provide extremely powerful observation tools with multiple and specialized uses according to the objects examined, such as fiber optic instruments for optoelectronic components, or instruments for use through free space for OCT (optical coherence tomography) imaging. Potentially there is a wide range of applications. In the healthcare sector, biosensors are very promising (see insert), as is cell imaging. When biologists start to use these instruments a large number of medical applications will emerge. Other fields are also concerned such as telecommunications, whose components would benefit from extremely precise diagnostics. Security will also benefit from unfalsifiable biometric sensors: a property that is a direct product of the variety of different measurements carried out simultaneously by these unique devices.

[box type=”shadow” align=”” class=”” width=””]The OLCI, patented and serving healthcare
Radical transformations are expected in the field of healthcare thanks to innovative measurement instruments. The OLCI, and its use with an OCT – two devices that have both been patented by Institut Mines-Télécom – may allow the early diagnosis of certain illnesses. Imagine a drop of blood on a miniature optical surface made up of specialized zones in which millions of bio-photonic sensors are placed and which react with the molecules present in the blood. These chemical reactions modify the optical nature of the surface, which is analyzed in detail by the OLCI. The results are correlated and interpreted by a physician in order to provide a very precise diagnosis.[/box]

An unrivaled instrumental platform and level of expertise

The laboratory has an instrument base named the VCIS (Versatile Coherence Interferometry Setup), which can be combined and structured according to needs and applications. It relies on high-performance tools, which Yaneck Gottesman hopes can be used by laboratories, in industry, by universities, manufacturers and researchers: “whether they are private or public, the platform is open for them to come and analyze or evaluate objects of interest in detail, test the performance of their components or develop new, more specialized instruments using the existing modular base.”

Beside the economic benefits of this partnership, there is a desire to create a place of open exchange which will allow progress to be made in multiple fields at the same time. Attracting users and being surrounded by them in order to understand their needs and benefit from their expertise, is definitely the best way to remain in contact with the domain in question and anticipate the future.

Editor: Umaps

Using nature as inspiration is certainly the oldest scientific approach and one that still has much to reveal. Bio-inspired robotics is a research topic at Mines Nantes that uses this process. It does not aim to simply mimic, but actually to understand the tricks nature has found to solve problems. Researcher Frédéric Boyer’s work is driven by such meticulous observation. He and his team develop various robots that do not need the use of computer calculations to obtain autonomy, and which explore their environment thanks to the characteristics of their bodies and new senses such as the electrical sense.

Locomotion: when morphology does away with calculation

Underwater robots for deep mining operations, humanoid robots for aiding elderly or dependent people, and drones the size of an insect for passing unnoticed all share the common task of having to reproduce complex human or animal behavior such as vision or locomotion, and to act fully autonomously even before displaying behavior that could be considered intelligent, such as decision-making. Research in artificial intelligence has long provided answers to these questions by allowing itself to be guided by two major lines of thinking: the symbolic paradigm represented by expert systems, based on rules and logic, and the subsymbolic paradigm represented by the neuron networks approach. However, observations of nature have revealed that the brain is not always absolutely necessary. For example, even when dead, a fish continues to extract energy from a series of whirlpools in order to move forward: a form of ‘passive swimming’ that has been modeled at Mines Nantes. Making robots perform complex behavior does not therefore always require computer calculations.

There are a large number of physicians in the robotics community, as Frédéric Boyer reminds us, himself a researcher and professor in robotics, and they do not especially look at the brain. “We try instead to reduce top-level cognitive problems to lowest-level solutions, those that are as closely bound as possible to the body”. This has the advantage of freeing the brain of tasks that it does not need to process, since the body is adapted to perception and action. “Bio-inspired robotics explores a new paradigm for unlocking autonomy by reconsidering intelligence as an attribute emerging from the interactions between the machine-animal’s body and its environment; it is intelligence incarnated, computational morphology.” In 2004 the team in which Frédéric Boyer works at IRCCyN (Institut de recherche en communication et cybernétique de Nantes) created an underwater machine that can move around like an eel in marine environments that are not easily accessible.

The researcher looks at locomotion problems in highly restricted environments, for example snakes in a tree or worms in a pipe. The resulting applications, currently few and far between, may concern maintenance in pipelines. Even more interesting is a gibbon-robot, an assembly of articulated arms ending in magnets, which swings majestically like a pendulum until it releases one of its grips, targeting a higher one, and thus climbs up a wall. The objective here is to propel the machine by sourcing part of the necessary energy from its surroundings, with the movement produced making the most of gravity which is conceived as a resource provided by the environment.

The electrical sense is studied in a pool in which a probe is able to move in three dimensions in a maze.

The bio-inspired process in three stages

Questions have long been asked about the flight of insects, which is theoretically impossible if we take reference from the flight birds or planes. Through close observation, it was eventually understood that the wing’s torsion as it beats supplies the additional lift required for flight. “Without this mechanism that uses morphology, it could never work” the roboticist explained. It has paved the way to the design of drones the size of insects, previously impossible.

“A technical barrier? Think bio-inspiration,” continues the researcher, giving the example of small, autonomous underwater robots dedicated to navigation in very awkward environments in rough waters. “Nature constitutes a wonderful library of inspiration”, he adds. The bio-inspired process occurs ‘naturally’ in three stages. Firstly, looking at nature and observing living matter with the aid of biologists; this observation phase can be carried out at face value, with no set objective because in nature there is still a lot to learn and to record. Next, functions are extracted from these observations. “We don’t copy, we understand. This phase is based on mathematics. It should not be confused with biomimicry,” the researcher stresses. It must be remembered that nature is not always optimized: it remains fairly redundant and precise copying could constitute a source of inefficiency. On the other hand, “nature has found simple tricks, and the mathematization of living matter allows these good ideas to be extracted”. The final step in the process entails implementing these tricks in technological devices that do not belong to nature. Such implementations are a bonus for the biologists who participated at the beginning and an advantage of these multi-disciplinary exchanges, often allowing them to understand their own field of research better.

As Frédéric Boyer explains, systematic exploration of nature within the framework of this approach, with no preconceived search for an application, should be encouraged. Indeed, it quickly becomes clear that “looking at an animal or plant, even the most humble animal that you might squash underfoot, can occupy the life of several researchers”, as the nature lover enthusiastically explains. The subject goes as far as the creation of new senses, such as the electrical sense known to be possessed by certain fish that emit electrical fields which are deformed when they come into contact with obstacles, which the fish then detect thanks to sensors on their skin. This observation was used to enable the robot named Angels, developed in a European project coordinated by Frédéric Boyer, to swim in the dark, its electrical sense giving it “a sort of immaterial body”. Although new and still little-explored, this field of research is proving of interest to industry players. The researcher and his team work with Areva on swimming in contaminated mud and snake-robots in pipes, and with CEA on remotely operated commands using electro-haptic feedback. They are also partners of a new H2020 European project called subCULTron (Submarine cultures perform long-term robotic exploration of unconventional environmental niches) whose aim is to create a bench of cooperative underwater robots with an electrical sense for monitoring the canals in Venice.

The next step: getting out into the air. The human body contrasts greatly with the air in terms of electrical fields. Research on the robot-eel and the electrical sense may create new and profitable forms of approach in terms of cooperation between robots and human beings. Frédéric Boyer concludes with an invitation to his fellow researchers who are aware of the importance of cross-disciplinary work: “bio-inspiration provides a systematic way of ‘shaking up’ your research”.

A Professor at Mines Nantes, Frédéric Boyer is passionate about biology. After completing a thesis in robotics at Paris Diderot University, he began by doing “mathematical calculations all day long in geometrical mechanics for high-distortion structures such as cables and soft robots”. He was then awarded his Accreditation to Lead Research and entered the field of bio-inspired robotics. He won the Michel Monpetit prize in 2006 from the Académie des Sciences for his work in dynamics. More recently, he and his team received the La Recherche 2014 prize in the category for technology, for their work on the electrical sense. These results motivate Frédéric Boyer to do research that is even more applicative, “combining my love of nature with my work”. His team collaborates with the other laboratories at Mines Nantes: Subatech (a joint research unit in nuclear technology) for sensors, and DSEE (Département Systèmes Énergétiques et Environnement) for fluid mechanics, as well as with several European laboratories..

Editor : Nereÿs

Big Data is an issue not only of technology, but also for society. Aware of the value of data, Institut Mines-Télécom has made it a major field of research, because a new approach is needed in order to store, share and use data to achieve optimal use. Machine Learning is one such approach. It is being researched by Stéphan Clémençon, Professor at Télécom ParisTech.

Several zettabytes of data are generated each year, which is the equivalent of several billion billion thousands of octets, and while this supply of data enables the creation of new services it also considerably modifies our needs and requirements. Yesterday’s tools are outdated and new ways of putting this abundance of data to good use must be found. Machine Learning aims to do this. The discipline combines mathematics and information technology in order to create algorithms for processing big data, with a large number of industrial applications.

Télécom ParisTech recruited Stéphan Clémençon, a mathematician who specializes in modeling and statistics, to help develop Machine Learning. As Stéphan explains, “when you start to deal with very large quantities of data, you enter the realm of probability”. This field has long been neglected, especially by engineering schools where mathematics was taught in a very deterministic way, but students are becoming more and more interested in it, which is fortunate, for “questions of large scale raise difficult problems requiring lots of creativity!” New methods are required and the main difference between these and the old methods is that the latter were based on traditional statistics and relied on predetermined models of the data. They were developed principally in the 1930s when the methods and challenges were different, computers had a very limited capacity for calculations and when data was expensive to produce.

Finding the hidden meaning in big data

Nowadays there are sensors everywhere and data is collected automatically, with no pre-defined use but with the notion that it contains valuable information. The idea is to examine the data keenly and make the best use of it. Machine Learning’s objective is to design algorithms suited to dealing with big data. Enabling machines to learn automatically was an idea born from the fact that the data is too large to realistically enable each stage of processing it to be carried out by an expert, as well as from a desire to see the emergence of innovative services and teaching with no a priori.

The question is “how can a machine learn to decide by itself?” How can we compress, represent and predict information from data selected to provide examples? This is the challenge of Machine Learning, which is fuelled by probabilistic modelling based on optimization and a theory of learning which guarantees sound results. The major problem is designing algorithms with good potential for generalization. Using criteria that is too strict may lead to overfitting, meaning the production of models that suit the given examples perfectly but which cannot be generalized. On the other hand, criteria that is not specific enough does not have sufficient predictive capacity. In the case of Machine Learning, the right amount of complexity must of course be deduced automatically from the data.

Designing new services from data

“The multiple applications of Machine Learning are a driving force for research“, says Stéphan Clémençon, giving a number of examples showing the variety of fields in which big data is collected and used: “automated facial recognition in biometrics, risk management in finance, analysis of social networks in viral marketing, improving the relevance of results produced by search engines, security in intelligent buildings or in transport, surveillance of infrastructures and predictive maintenance through on-board systems, etc.”

In Machine Learning, potential applications are found first, and then the math that allows them to be understood and clearly defined, significantly improving the process. A “professional” knowledge of these applications is therefore necessary. It was in view of this that the chair in Machine Learning for Big Data of Télécom ParisTech (see insert) was created with Criteo, PSA Peugeot Citroën, the Safran group and a major French bank. The idea is to work together with industry and academics to produce effective products based on the partners’ projects and notably providing increased knowledge of the state of the art for some and a keener understanding of the challenges of application for others.

Big Data refers at the same time to a specific infrastructure and a list of unresolved problems. Stéphan Clémençon regrets that in France, “we have missed out on the equipment stage” but fortunately, he adds: “we do have a large number of innovative SMEs headed by well-educated students, especially in applied mathematics.” Data engineering is multi-disciplinary by definition and the trump card of a school like Télécom ParisTech, which teaches in a variety of fields, is being able to offer specialized programs, in particular since, as Stéphan Clémençon underlines, “Machine Learning relates to key business challenges and there are lots of potential applications in this field.”