Information Processing and Systems

The TIS scientific domain

Philippe Bidaud

Scientific Director
of the Information Processing and Systems scientific domain

Highlights

Creation of the AI4AE virtual research center with DLR (2019) The two French and German research centers have initiated cooperation in "AI and applications in aerospace engineering", by creating a virtual research center that will be dedicated to contributing to the digital transformation of the aerospace industry. See www.onera.fr/fr/ai4ae

ONERA gets closer to the Nouvelle Aquitaine region (2019) Signing of a partnership agreement, aimed at fostering and intensifying collaborations in digital transformation for the region's aeronautics, space and defense industry. ONERA commits its research potential to topics such as digital twins, maintenance, agile production...

TIS in brief

This field covers what is usually referred to as sciences and technologies of information and communication, the main components of which are automatic control, signal and image processing, robotics, digital simulation, system and process design and optimization, computer systems engineering, knowledge engineering and cognitive processes.

Much of the research is aimed at the design of aeronautical and aerospace systems, defense and surveillance systems and security systems, as well as the deployment of these systems in complex missions, the processing of massive data from on-board sensors and the development of knowledge about large-scale phenomena. The generic nature of this work means that it has considerable diversification potential, particularly in transport and production systems.

Through innovative digital methods and techniques, and the convergence of modeling, simulation and data, this work aims to make advances that will enable new intrinsic performances to be achieved for the systems in question. It also addresses issues relating to their integration into information and communication systems, and to the need for interactivity with users, as well as safety, security, autonomy and system coordination.

Many of these projects give rise to experimental developments (mini-UAVs and UAVs, land and submarine robots, piloting aid interfaces, intelligent sensor networks, etc.). Others involve the development of software platforms (air traffic, battle lab, MDO design, etc.) to create digital models and evaluate operational performance through physical simulation.

Scientific Officers

Hélène Piet-Lahanier

DTIS
Information Processing and Systems Department

TIS Themes

The TIS domain comprises 8 scientific themes, mainly spread across the DTIS Information Processing and Systems department (and, for one theme, the DAAA Aerodynamics, Aeroelasticity, Acoustics department).

System identification and control DTIS
Perception and information processing DTIS
Robotics and autonomy DTIS
Safety and security of cyber-physical systems DTIS
Systems and software engineering DTIS
Artificial intelligence and decision-making DTIS
System design and optimization DTIS DAAA
Cognitive engineering and human-system interaction DTIS

System identification and control

DTIS Sylvain Bertrand, Clément Roos (@ onera.fr).

Optimal control
Robust adaptive control
Predictive control
Distributed control
Hybrid system analysis
Multicriteria hybridization identification
State estimation and trajectory reconstruction
Fault detection and reconfiguration
LFT modeling for robustness analysis
Robust and self-sequencing structured Hinf control
Approximation of high-dimensional systems
Sensor-referenced control
Fault-tolerant control
Distributed control and estimation for multi-agent systems

Examples

Toolboxes SMAC, MORE and GENETIC SMAC - Robust synthesis and robustness analysis; MORE - Reduction and approximation of dynamic models; GENETIC - Optimization for identification

Mission PICARD Attitude control device Control rules are adapted as a function of deviation from pointing position

ATTOL Project Autonomous "cab" navigation and landing

Perception and information processing

DTIS Luc Meyer, Olivier Herscovici (@ onera.fr)

Massive heterogeneous data processing
Data fusion using Bayesian networks
Learning for image analysis and image-based measurement
Tools for certified hybrid navigation
Near-sensor processing and sensor-processing co-design
Imaging measurement for experimental physics (PIV-BOS)
Tracking, data fusion, sensor networks and situation awareness
Non-linear filtering and data fusion for navigation
Exploitation of airborne or satellite remote sensing images for environmental observation
On-board vision for vehicle autonomy

Examples

BOS3D wind tunnel measurement 6 to 12 synchronized cameras in front of textured backgrounds. Tomographic system for instant quantitative 3D density measurement.

Interpretation of multi-source images and data fusion Comparison of different learning-based automatic classification methods based on multi-source optical data in urban areas.

Robotics and autonomy

DTIS Christophe Grand, Julien Marzat (@ onera.fr).

Multi-robot decision architectures
Resilience and robustness of robotic systems
Cooperative robot fleets
Reinforcement learning
Processing Programming and simulation architecture
Autonomous navigation
Sensor(s)-referenced control
Coordination of multi-robot systems
Autonomy and decision-making

Examples

PRI DROSOFILES Embedded geometric vision and state estimation for 3D modeling of the observed environment. Guidance-steering, stabilization and avoidance of unknown or dynamic obstacles.

*PRI: Research-industry partnership

ACTION/MUST A decision-making architecture to achieve a high level of resilience in the deployment of autonomous multi-robot systems, based on the HiPOP (Hierarchical Partial Order Planning) hybrid planner, which organizes actions in a hierarchical form of execution and repair of action plans.

COPERNIC Lab - Control and perception for autonomous navigation and inter-vehicles cooperation (web site)

Safety and security of cyber-physical systems

DTIS Christel Seguin, Loïc Brevault (@ onera.fr)

Formal modeling and evaluation of system dependability
Probabilistic estimation of risks and system reliability
Fault tolerance mechanisms, health prognosis and fault diagnosis
Dependability argumentation and certification

Examples

DGAC PHYDIAS agreement on drone safety analysis methods https://w3.onera.fr/PHYDIAS/

DROSERA tools for analyzing drone fallout zones after accidents

Analysis of air traffic data and collision risks

Health prognosis for launchers

Book: Estimation of Rare Event Probabilities in Complex Aerospace and Other Systems : A practical Approach. Jérôme Morio, Mathieu Balesdent

Ouvrage Model-Based Safety and Assessment - Conference proceedings, Editors: Christel Seguin, Marc Zeller, Tatiana Prosvirnova

Systems and software engineering

DTIS Frédéric Boniol, Romain Kervarc (@ onera.fr) .

Formal methods for verification/certification of artificial intelligence
Artificial intelligence to accelerate formal methods
Verification of VHDL codes
Engineering large computational codes
Embedded wireless networks
Methods and languages for systems and software development
Formal languages for requirements modeling and specification
Control system modeling
Embedded verification & certification
Real-time systems analysis
Avionics network analysis
Many-core architectures
Simulation architectures

Examples

Controller verification Development of a high-level formal imperative language (pseudo-code), called PySil, for MPC controllers, using the interior-point algorithm.

Electrum language A language for specification in bounded relational logic (a first-order logic with relational terms and whose interpretation domains are bounded by the user) for automated verification of properties and code behavior by translation into a satisfiability problem for propositional logic.

Artificial intelligence and decision-making

DTIS Stéphane Herbin, Gauthier Picard (@ onera.fr) .

Trusted AI for robustness and certification
AI, physical models and knowledge
Interactive learning
Dynamic planning
Knowledge representation for reasoning and decision-making
Planning under uncertainty (applications to space, ATM, industry)
Distributed decision making
Machine learning on heterogeneous data

Examples

DELTA A software environment designed to facilitate the use of deep learning (CNN, RNN, GAN) in ONERA's business activities. DELTA has been used for point cloud semantization, weather prediction (lightning risk and electronic content of the ionosphere), propellant combustion studies, crack counting in composite materials, and fluid mechanics. DELTA is mainly based on Pytorch (although a Caffe version still exists), and relies on an ONERA cluster of new-generation Nvidia GPUs.

InCell is a Constraint Programming tool for managing several types of renewable resources. It considers disjunctive resources, cumulative resources and resources with an appropriation period. Conflicts over resource use are resolved by prioritizing activities for scheduling. In addition, the tool includes an invariant for constraint-based differentiable local search, which transforms a priority order into a complete schedule and incrementally manages this schedule in the event of a change in the order. On this basis, several neighborhoods and search strategies can be used, and new best upper bounds have been obtained for several scheduling benchmarks. The spatial domain is a key application of InCELL.
Five-day simulation of InCELL's scheduling of ground data acquisition and transfer for an Earth observation mission.

Simulation sur cinq jours d'une planification par InCELL d’acquisitions et de transfert d'informations au sol pour une mission d'observation de la Terre.

System design and optimization

DTIS DAAA Mathieu Balesdent, Pierre-Marie Basset (@ onera.fr) .

Uncertainty propagation and multi-fidelity management
Hybrid models and data assimilation
Model reduction
Co-design (aircraft-operations)- Modeling and multidisciplinary optimization techniques for aerospace systems
Methods and tools for system design and performance analysis
Exploration of new aerospace concepts and performance evaluation

Examples

CICAV Design of new flying wing architectures with BLI propulsion using MDAO techniques.

Eole An airborne launch system for small low-Earth orbit satellites. This launcher uses hybrid propulsion, a high-performance composite structure and avionics with autonomous flight functions.

Cognitive engineering and human-system interaction

DTIS Bruno Berberian, Thomas Rakotomamonjy (@ onera.fr) .

Control through the haptic sense
Analysis of operator performance when supervising highly automated agents
Determinants of performance in emergency situations
Determining the neural correlates of conscious perception mechanisms
Multimodal integration involved in the perception and control of self-motion- Analysis of acceptability, readability and, by extension, usability of automated system

Exemples

Spatial disorientation in aeronautics Studies of compensation mechanisms for disorientation phenomena using visual (synthetic) and haptic aids. Visual and physical stimulation - Analysis of pilot reactions through posture capture, head orientation, oculomotor activity, etc.

Pilot assistance Force feedback on haptic interface (coupling stability - effect of delays) - Virtual guides through real-time simulation of aircraft-environment behavior