Adaptive Latent World Models efficient Inference on autonomous robots

Autonomous robotics requires AI models capable of capturing, understanding, and representing the physical and spatial properties of the environment to effectively predict how the environment will evolve under the influence of actions (causality). These models, known as World Models, can thus simulate a large number of possible action sequences and future states to make optimal decisions. In this context, latent space World Models of the JEPA (Joint-Embedding Predictive Architecture) type have recently stood out by learning to model a representation of the world in a compact latent space and predicting its evolution directly within that same latent space. These models have significantly matured this year, with notable advancements in the learning phase and online adaptation, but their deployability on resource-constrained targets remains largely unexplored in the state of the art. The main objective of the thesis is therefore to optimize the inference of adaptive JEPA models using model compression techniques (quantization, pruning) to meet the constraints of embedded systems in autonomous robotics (real-time constraints, low latency, memory and computational compactness, low power consumption).

Experimental characterization of fast electrons in the WEST tokamak boundary plasma using Langmuir Probe measurements

In magnetic confinement fusion power devices (e.g., tokamaks), one of the primary methods for heating the plasma involves injecting radiofrequency waves that accelerate electrons in the plasma core. Populations of fast electrons can persist to the plasma edge, where they interact with the internal components of the vacuum chamber. These fast electrons can then damage the components by subjecting them to extreme and localized heat fluxes, potentially compromising their lifespan in a fusion power plant.
The goal of this PhD project is to characterize these fast electrons in the WEST tokamak (located at CEA/IRFM) using instruments called Langmuir probes, of which the machine is equipped with a large number. The presence of fast electrons challenges the current interpretation of the signals provided by these probes, and most notably the calculated plasma temperatures.
The PhD will cover, first, a theoretical aspect to improve the physical models required for interpreting probe measurements, accounting for the presence of fast electrons. This work will include collaboration with simulation teams (using codes such as SOLEDGE, VOICE, and other PIC codes) to assess the relevance of the proposed models. Second, it will involve an experimental analysis of all Langmuir probe signals from WEST over recent years to derive their characteristics. Given the large volume of data, this may involve developing machine learning (ML) or artificial intelligence (AI) methods to identify characteristic patterns in the signals. Cross-verification with other diagnostics (X-ray detectors, infrared cameras) will be performed to confirm the findings.
The ultimate aim of the thesis is to develop a generic predictive model for calculating the characteristics of fast electrons impacting the divertor in ITER, DEMO, and future fusion power plants.

Superconducting Electronics for Quantum Computing

Fault-tolerant quantum computing requires thousands of physical qubits for a single logical qubit, presenting major challenges in terms of control and readout. Current solutions, such as room-temperature electronics or cryogenic CMOS, face limitations: cable clutter, thermal dissipation (a few watts at 4 K), and operation fidelity. Single Flux Quantum (SFQ) electronics, based on Josephson junctions, offers a promising alternative: it uses ultra-fast pulses (a few picoseconds) with no static dissipation and extremely low switching energy (~10?¹? J).

This thesis, led by CEA Grenoble, aims to explore the integration of SFQ circuits for qubit control and readout. The goal is to design key functional blocks (ADCs, oscillators, error correction decoders) tailored to cryogenic constraints. Special attention will be given to system architecture and experimental validation under real-world conditions (cryogenic environments).

Expected results:

- Architectural studies demonstrating SFQ's advantage for scaling.
- The design and testing of an SFQ circuit (with or without a qubit) to validate hypotheses.
- Experimental validation in a cryogenic environment, proving the feasibility and energy efficiency of this approach.

An opportunity to contribute to an emerging technology at the interface between superconducting electronics and quantum computing.

Timing Predictability Meets Security: Reasoning About Timing Non-Interference in Out-of-Order Processors

Modern out-of-order processors are difficult to secure against timing leaks because security countermeasures such as flushing, stalling, delaying, or padding execution modify the timing behaviour of the processor itself. This thesis investigates how timing predictability can support security-by-design with quantitative timing guarantees. Building on POP?, our formally verified predictable out-of-order processor, the objective is to derive not only sound bounds on the latency introduced by security mechanisms, but also the processor-state dependencies that determine these bounds. The expected scientific results are a formal framework for deriving and validating latency bounds for timing countermeasures, a method for determining whether this latency remains independent of secret or prior-domain state, and compositional techniques for reusing timing and security proofs after hardware modifications. The broader goal is to make the timing cost of a security mechanism part of the security argument itself.

State Preparation and Output Problem for Multivariate Functions: The Main Remaining Bottlenecks to Process PDEs Using Quantum Algorithms

A fault-tolerant quantum computer could provide an exponential speedup over the best classical algorithms for certain applications. While proposals exist for specific instances, their adaptation to practical cases remains poorly defined. This is particularly true for solving Partial Differential Equations (PDEs), a major challenge in numerical computing, as well as for other applications requiring the preparation of initial states corresponding to discretized multivariate functions. Although such initializations can be achieved with a quadratic advantage using amplitude amplification methods, they raise the following question: Which distributions defined by multivariate functions can be efficiently implemented on exponentially large quantum states?
This theses explores approaches based on basis transformations, i.e., the search for transition matrices that can be efficiently decomposed into quantum gates and provide access to bases where the function's representation is sparse. The developed methods could also address the so-called "output or readout problem", where the expected result is naturally sparse in a function basis.

Frugal and Explainable Open-World Object Detection through Visual Concept Discovery and Transfer

Unlike classical detectors trained on a closed set of categories, and VLMs (vision-language models) that are heavy, non-explainable, and limited to a fixed vocabulary, this thesis proposes an explainable and frugal approach to open-world object detection. Indeed, object detectors in an open world are capable not only of recognizing known objects, but also of signaling unknown or unexpected objects.
The thesis project relies on using VLMs solely as offline teachers during training in order to distill their knowledge into a lightweight, real-time detector. By leveraging learning and reasoning from a hierarchy of shared visual concepts (covering appearance, attributes, and semantic families), the proposed model will be able not only to recognize known objects, but also to signal and characterize unknown objects without prior prompts. This approach will thus provide visual explanations directly grounded in the image, meeting the demands for robustness, transparency, and frugality essential in applications across numerous safety-critical domains (mobility, defense and security, healthcare, and industrial inspection).

Development of mid-infrared sensors and Fourier-transform spectrometers in silicon integrated photonics

Fourier-transform infrared (FTIR) spectrometers have become a reference tool for chemical analysis. One of the next challenges is to integrate them on photonic integrated circuits (PICs), using silicon platforms optimized for the mid-infrared (MIR), in order to improve compactness and scalability. You will simulate key MIR PIC components, with a focus on photonic and thermo-optic modeling, design photolithography masks, follow up on fabrication, and characterize the FTIR spectrometers and sensors. Sensing applications will primarily target the detection and identification of greenhouse gases, but may also extend to plastic pollution and biosensing. Although dedicated engineers will be in charge of silicon wafer fabrication in our cleanrooms, you will be involved in defining specifications and in monitoring in-line quality control to ensure that the fabricated devices perform as intended. You will be responsible for characterizing the FTIR and sensor devices at wafer and chip level on custom mid-infrared optoelectronic test benches. A first generation of non-optimized devices will be available upon your arrival. You will be part of the team upgrading the characterization tools to enable faster, more accurate automated measurements. You will analyze and interpret experimental data to deepen our understanding of the advantages and limitations of integrated FTIR spectrometers compared with their bulkier free-space counterparts. You will be expected to publish research papers and present at international conferences.

Integrated superconducting nanowire single photon detecors on SiN waveguides with photon number resolution capability for photonic quantum computing

Among the various qubit technologies, photonic qubits represent a highly promising approach to the development of quantum computers capable of solving problems beyond the reach of the most powerful supercomputers. The core of photonic quantum computing processors relies on the generation, manipulation, and detection of single photons. In this context, this PhD project at CEA-LETI aims to develop a new generation of high-performance single-photon detectors, offering not only high detection efficiency and low noise, but also photon-number-resolving capabilities, i.e., the ability to discriminate between 0, 1, 2, or more incident photons. This functionality is essential for implementing quantum error correction codes and ultimately enabling the transition from physical qubits to logical qubits. The research will focus on the design of innovative detector circuit architectures, the fabrication of integrated prototypes in CEA-LETI's 200 mm cleanroom, and their experimental characterization at cryogenic temperatures. This work will contribute to the development of a strategic technology for next-generation photonic quantum computers.

Multi-level modeling and accelerated simulation of entropy sources for pre-silicon analysis of TRNGs

This topic focuses on the modeling and accelerated simulation of entropy sources used in True Random Number Generators (TRNGs). The candidate will study how to preserve the precision of analog, netlist, and post-layout simulations while greatly accelerating their execution. The goal is to develop intermediate models that incorporate key physical phenomena such as noise, jitter, variability, coupling effects, and layout dependencies. These models will be based on theoretical foundations from the state of the art and validated against full analog simulations.

An experimental component will also allow the candidate to validate these models using real-world measurements of entropy sources, particularly with the VASCO component. The candidate will aim to link the physical parameters of the circuit to the statistical properties and entropy of the generated sequences.

The work contributes to improving pre-silicon analysis, design, and evaluation of TRNGs. Additionally, the developed approaches may be extended to Ring Oscillator-based Physical Unclonable Functions (PUFs).

Generative AI for the Design of Next-Generation Power Converters, End-of-Life Guaranteed: Accelerated Topology Discovery and Multi-Criteria Optimization Integrating Ageing and Eco-Design

Next-generation wide-bandgap (SiC/GaN) power converters are at the heart of the electrification of our economy, yet their ageing and environmental impact are only assessed a posteriori, on a frozen design: this leads either to costly oversizing margins or, conversely, to nominally optimal designs that no longer meet their specifications at end of life. This thesis aims to remove this obstacle by establishing a closed-loop, generative-AI-based design loop in which end-of-life specification compliance and environmental impact per kWh converted become first-rank objectives, on a par with efficiency, volume, mass, EMC and consumption. The method couples a topology generation constrained by a versioned electrotechnical ontology, frontier LLM models, a propagation of multi-component drift (R_DS(on) and V_th of GaNs, ESR and capacitance of capacitors) up to system-level performance, recalibrated on accelerated tests, a multi-fidelity evaluation (physics-informed surrogates, simulation tools), and a multi-objective Bayesian optimization under probabilistic constraints, integrating LCA impact through a functional unit that depends on the predicted lifetime. The work will be validated experimentally on a converter prototype measured both new and aged, with explainability of the trade-offs. It capitalizes on the CEA's local generative-AI infrastructure (AITIANA Studio), aiming for measurable gains in reliability and environmental footprint, a reusable method, and open data.

Top