In order to enhance human-machine interaction, usual metrics include questionnaires and performance measures which do not allow for a continuous and online assessment of the quality of interaction, nor for a direct cognitive state monitoring of the human operator. In recent years, the development of physiological computing methods including that of brain-computer interfaces has enabled the rise of symbiotic systems that adapt the interaction using involuntary user inputs. Yet, to our knowledge, this technology has never been applied to human-robot interaction (HRI) in the context of mobile and collaborative robotics. This might be due to several challenges that need to be overcome, including the impact of user physical activity on the acquired metrics. This project will provide the first evaluation of the usability of electrophysiological metrics from wearable sensors for a rich, out-of-the-lab and online quality of interaction (QoI) assessment for collaborative robotics. The main objectives will be to: i) characterize the users’ cognitive state -i.e. cognitive effort and automation surprise- during collaborative and mobile HRI using involuntary electrophysiological features elicited by two standard collaborative robotic tasks -i.e. a joint navigation task, and an interactive manipulation task; ii) create an enriched QoI index that takes as inputs these features processed through a mental state estimation pipeline; iii) adapt the HRI thanks to this new QoI index; as well as iv) promote good scientific practices including data sharing.

The project FLOCCON is dedicated to the development and validation of an innovative strategy to accelerate solvers for fluid mechanics. The project focuses on incompressible solvers, containing two parts: (1) a linear Poisson equation, and (2) a non-linear advection equation. The key idea of this project is to use deep learning to train neural networks based on solutions of these two equations. To go further, the project will examine learning methods which can guarantee a target accuracy. To do so, physical-based and long-term 'loss functions' will be introduced, in order to ensure a limited error accumulation in time. Moreover, an hybrid strategy will be proposed to obtain a robust solver. finally, an optimisation of this new network-based solver will be carried out on CPUs/GPUs. In addition to classical validation cases, a target application will be simulated on the pollutant dispersion in a large city, which is a challenging case for classical HPC solvers.

Recent research in neuroergonomics and security reports in high risk systems have shown that operators’ decision making impairments can be observed in critical situations. In addition it has been suggested that these impairments can originate from a lack of mental flexibility. External factors, such as time pressure, conflicts or workload can interfere with executive functions, the cognitive processes that maintain mental flexibility. These interferences have been proven to be the main cause of numerous accidents in high risk systems with consequent financial and human losses as a result. This kind of impairments is also observed in patients with frontal lobe lesions, albeit with chronic and serious deficits in executive functions without any external pressure. In order to improve mental flexibility, patients can benefit from reeducation based on regular training of executive functions along with a non-invasive brain stimulation. One of the key areas for such processing is the prefrontal cortex. Previous studies have shown however that maintaining a good level of mental flexibility in complex tasks involves a dynamic integration of several brain areas distributed in large networks. These areas in cooperation can be solicited by a multi-modal intervention such as motor-cognitive tasks associated with a transcranial stimulation. These tools could therefore be transposed to the field of high risk systems to help operators maintaining an efficient level of mental flexibility when facing contextual interferences. The goal of the present project is to develop a new training program and to assess its impact in emergency management, in comparison with current training programs for high risk systems operators. We will develop training programs inspired by the aeronautical and clinical fields. We will also develop new methods to measure long term cerebral modifications induced by these programs. This goal will be reached by: 1. The identification of the brain networks of mental flexibility. We will compare the functional networks at rest and at work between dysexecutive patients and healthy participants. 2. The assessment of multimodal task effects associated with or without a transcranial brain stimulation. Behavioral (performance in executive tasks) and functional (parameters of cerebral functional connectivity networks at rest and during trainings) measures will be used. 3. The evaluation of the learning transfer to the management of high risk systems. Pilot activity in critical situations will be used as an experimental paradigm. 4. The identification of mental flexibility predictors. The goal will be to find which initial individual’s parameters of brain functional connectivity at rest or during activity could predict an optimal level of mental flexibility after training. This project aims at improving safety in high risk systems and the quality of life in the brain-damaged population; impact that could eventually extend to the other parts of the population

Systems using gigahertz electromagnetic radiations are nowadays everywhere providing communication, localization, or detection functionalities. Economic and defense stakes concerning their operation are high even though they are particularly vulnerable to accidental or deliberate electromagnetic perturbation. Existing protection solutions still need to be improved. Their resistance to pulsed power is satisfying and so their time response, but their insertion loss tends to increase the overall noise figure and reduce the sensitivity of the receiver. Moreover, high power quickly breaks operation. The DIOMEDE project aims at addressing scientific and technologic issues in order to provide innovative microwave power limiters based on plasma microdischarges. The project consortium gathers three partners: two belongs to Université Fédérale Toulouse Midi Pyrénées, namely LAPLACE laboratory and ISAE, and the third one is the CEA Gramat. Complementarity of the partners comes from their respective expertise in microwave engineering, plasma microdischarges, and electromagnetic perturbations. The first part of the DIOMEDE project deals with improving scientific knowledge of interaction mechanisms between a microwave and a microplasma discharge. Both experiments (hot S parameters measurements, CCD imagery) and numerical computation with ANR funded MACOPA program will be used. One of the main goals is to identify key parameters to optimize plasma igniting time. Then, several microwave power limiters prototypes in both conduction (wire protection) and radiation (antenna protection) will be developed to assess their behavior and optimize their performance. A parametric study will be conducted considering different topologies (above and below the circuit ground plane), preionization currents, gas composition, and pressure. Regarding pressure, two specific solutions will be considered: one close to atmospheric pressure and the other at lower pressure. Specific work will also be carried out to achieve an original instrumentation dedicated to these characterizations. A vacuum chamber will thus be designed, produced and operated. The joint identification of measurement protocols dedicated to diversity and the microwave plasma approach will be a major concern of the project.It is well known that the effectiveness of the protection is highly dependent on the waveform of the electromagnetic threat. The second part of this project proposes a series of studies to evaluate the performance of these innovative devices in a realistic context. The methodology is well identified and controlled by the CEA Gramat. It can be implemented with a framework for GPS reception system. Finally, keeping the same experimental setup, a receiving system including state-of-the-art diode-based microwave power limiters will be tested. This experiment will effectively compare our innovative system to the dominant technology today. At the end of DIOMEDE project, partners will be in capacity to support industrial transfer in order to make available, either for other defense or critical civilian applications, high performance antennas and circuits hardened against electromagnetic interferences.

The IMPACTS project aims at an ever-increasing integration of modeling, numerics and control design for complex multi-physical implicit systems described by both ordinary and partial differential equations. This integration is achieved considering the novel class of Implicit port Hamiltonian (PH) Systems, analyzing their system properties and developing new dedicated methods for numerical simulation and control design. Implicit PH Systems arise from the modeling of systems with non-local constitutive relations, implicit geometric discretization in time and space or control by interconnection. The methodological contributions of this project will concern the modeling and control of implicit PH systems using irreversible Thermodynamics, geometric numerical methods for space-time discretization and order reduction, canonical implicit discrete-time PH systems and energy-based control design, and in domain/boundary control of distributed parameter systems under implicit interconnections.