The AI4CODE project brings together 6 research team with strong expertise in the design, decoding and standardization of forward-error-correction codes. The aim is to develop skills in artificial intelligence and machine learning, and to explore how learning techniques can contribute to the improvement of code design methods (by using less parameters, more relevant heuristics, producing stronger codes) and decoders (better performance, reduced complexity or energy consumption), on selected scenarios of practical interest for which a full theoretical understanding is still lacking. The proposed methodology is to augment legacy design methods and decoders with learning capabilities or decision support systems wherever relevant, rather than replacing them by a generic, black-box neural network, so that we can inspect the trained solutions and try to infer why they work better. Our ultimate goal is to obtain new theoretical hindsight that could translate into better codes and decoders.

The VR-MARS project represents a support system for urgent healthcare delivery in isolated environments, based on virtual reality and embodied conversational agents (ECA). We hypothesize that these two technologies enable better situational awareness and care coordination between 3 parties: a care provider in an isolated location, a critically ill patient and the control centre on Earth. VR-MARS explore the scientific fields of emergency medicine, human factors and virtual reality. The use case of VR-MARS will be related to space medicine, in particular emergency care during a manned spaceflight to Mars. During these missions, temporal isolation will add to physical isolation, because of delays in communication between the care provider (on Mars) and ground control (on Earth), which will preclude real-time telemedical support. VR-MARS will be built around two simultaneous decision loops which will allow task assignment and synchronisation between the care provider, the ECA and ground control. The ECA will interact with the care provider via augmented reality. Upon request, it will deliver step-by-step guidance on medical protocols, using reassuring verbal tone and cues in order to mitigate the stress of the care providers. As soon as it is available, ground control on Earth will be made aware of the situation on Mars and of the procedures being undertaken by the care provider. This will improve situational awareness on the ground and enable the most optimal decision making in the mid- to long-term. In return, ground control will deliver its recommendation to the care provider via the ECA. Therefore, the ECA will represent the central hub of communication between the two sites. VR-MARS will be tested on two medical scenarios involving a critically ill patient represented by a high-fidelity simulator. Technical and non-technical skills of the care provider will be assessed at two levels: immediate interactions between the care provider and the ECA (for urgent, life-saving decisions) and delayed interactions between the care provider and ground control (for mid- and long-term decisions). With regards to research output and spinoffs, we anticipate that VR-MARS will improve medical care in remote environments, such as humanitarian missions, the combat environment, medical evacuations, expedition medicine, etc.

The CONTACT project is part of the context of a very significant increase in the integration density of electronic systems for communications, localisation or surveillance equipment. This requires the design of antennas which must be both miniature and multifunctional since they have to be able to meet several communication standards. In addition, many spatial and military applications need for circular polarization, which is also an interesting solution for civil domains to overcome misalignments between transmitter and receiver and to mitigate inherent polarisation loss factors due to multipath problems. More specifically, the CONTACT project aims to demonstrate an antenna array that will be composed of 4 miniature, tri-band (GPS/Galileo) and circularly polarized radiating elements for satellite radionavigation systems. The state-of-the-art shows that obtaining a circular polarization when the antenna needs to be miniature and work on different frequency bands is really challenging. Indeed, circular polarisation can be produced by integrating an electronic circuit to excite two orthogonal current modes with a phase difference of 90°. This kind of solution can be complex, is bulky, lossy and not working on different frequency bands simultaneously. The litterature exhibits that is difficult, if not impossible, to keep a good circular polarization on different working frequencies. In the CONTACT project, we propose to reach these three objectives (miniature, multi-band with circular polarization) by proposing innovative solutions based on ferrimagnetic materials. Indeed, this kind of materials has interesting properties since: - Their non-diagonal anisotropy allows them to naturally generate circularly polarized waves - In addition of a high dielectric permittivity (er=15 typically), their effective permeability can be higher than 1 on some well selected frequency bands. Therefore, it will be possible to reach high miniaturization degrees. - Their permeability is dispersive, i.e. it varies with the frequency. The antenna could operate on its fundamental mode at two different frequencies. We could therefore develop a multi-band antenna with circular polarization (since the antenna will work on the same mode on two different frequency bands). Published studies on ferrites antennas, remain theoretical with an ideal magnetization of the material or carried out using permanent magnets, implying an increase of the global dimensions of the antenna. This problematic is avoided in the framework of the CONTACT project since we propose a solution based on self-polarized ferrites. The development and the realization of self-polarized ferrites used to propose miniature and multi-band antennas while having a circular polarization will allow us to avoid permanent magnets and therefore reduce the size of the global device, its complexity and improve its efficiency. The benefit compared to the state of the art offers a dual approach, the first one on the development of the antenna and the second one on the material. Indeed, by developing new concepts we will simultaneously obtain a miniature, multi-band and circularly polarized antenna with a single feed point. For the material part, improved processes will be developed to reach self-magnetized ferrites with features fitting with RF applications. The consortium complementary and the interdisciplinary approach within the CONTACT project is perfectly adapted to give solutions and implement this high-potential concept.

Covering more than 70% of earth’s surface, the oceans, especially the upper oceans (e.g., the first few hundred meters below the oceans’ surface), play key roles for the regulation of the earth climate (e.g., climate change) as well as for human societies (e.g., marine resources and maritime activities). Despite ever-increasing development of simulation and observation capabilities leading to ocean big data, our ability to understand, reconstruct and forecast upper ocean dynamics and police maritime activities remains limited for key societal and defense challenges (e.g., catastrophe monitoring, global current changes, fishery, energy production, illegal activities, geostrategic issues, tactical oceanography). Building upon the cross-fertilization of the cutting-edge expertise of Ifremer and Univ. of Brest in marine science and technology and IMT Atlantique in engineering/data science, OceaniX aims to explore and develop AI-driven strategies and frameworks for the next-generation of dual self-adaptive multi-platform ocean monitoring and surveillance systems and services with an emphasis on observability and sampling optimality issues for complex dynamics and processes, including extremes and long-term properties. This general objective will rely on bridging model-driven paradigms underlying physical sciences and data-driven learning-based approaches at the core of AI to learn novel computationally-efficient and physically-sound representations of complex dynamical systems. These developments will provide the basis for addressing key topical challenges: i) the design and optimization of smart multi-platform ocean sensing systems, ii) the observation-driven modeling, forecasting and reconstruction of poorly-resolved upper ocean processes (e.g., wave, wind, current, biogeochemical processes), iii) the multi-platform surveillance of maritime activities. The associated training program covers comprehensive curriculum from Msc./Eng. Degrees, PhD program to lifelong training at the interface between ocean science and data science. Based on project-based and active teaching activities, it will strongly promote interdisciplinary interactions among trainees as well as a global awareness of the role of AI technologies w.r.t. societal and environmental issues. The ambition of our research-training track is to establish an internationally-recognized research & training group which largely transcends boundaries of AI, oceanography and climatology with strong academy-industry and science-society interactions. It will contribute to stimulating entrepreneurship and academy-industry cross-fertilization (e.g., PhD co-supervised with SMEs, collaborative workshops open to students and lifelong trainees). Our international attractiveness will benefit from the participation to and coordination of international programs (e.g., scientific leadership of international space oceanography missions) and will further develop through among others our English-taught training program, international academic partnerships for incoming and outgoing visiting scholarships and the organization of international workshops and data challenges. Data challenges, at the core of AI communities, remains to be developed in ocean science and is regarded as a key instrument of our strategy. OceaniX gathers a transdisciplinary team in the framework of EUR Isblue with a recognized expertise in AI, applied statistics, numerical modeling, remote sensing and ocean sciences. Supported by institutional partnerships (CNES, ENSTA Br., Ecole Navale, ESA, Ifremer, IRD, IMT Atl.) and industrial ones (ACRI-ST/ARGANS, CLS, Eodyn, ITE-FEM, Mercator-Ocean, Microsoft, Naval Group, ODL, OceanNext), its total budget amounts to 4M€, which the requested ANR grant covers 15%. Keywords: upper ocean dynamics, maritime activities, tactical oceanography, multi-platform and multi-source data, complex dynamical systems, geophysical extremes, deep learning, data-driven representations, inverse problems

The efficient exploitation by software developers of multi-core architectures is tricky, especially when the specificity of the machine is visible to the application software. To limit the dependencies to the architecture, the generally accepted vision of the parallelism assumes a coherent shared memory and a few, either point to point or collective, synchronization primitives. However, this requires to share information between may if not all the nodes. Unfortunately, as soon as the number of core is around 10 (ten), the communication cannot occur on a shared medium anymore, and designs make use of bus hierarchies or Networks on Chip. This latter solution is clean and efficient, but each core can see only the communications it is the target of, and unlike shared but, cannot spy what is going on between other cores. This is particularly difficult when implementing cache coherence and collective synchronizations, and a possible solution to overcome this issue is to use radio communications on chip. By nature, radio communications provide broadcast capabilities at negligible latency, they have thus the potential to disseminate information very quickly at the scale of a circuit and thus to be an opening for solving these issues. In the RAKES project, we intend to study how RF communication can solve the scalability of the above mentioned problems for architectures with a large number of cores (>256), by using mixed wired/RF NoC. We plan to study several alternatives and to provide (a) & virtual platform for evaluation of the solutions and (a) an actual implementation.