The methodological developments of the SYCOMORE team cover a wide spectrum of theoretical approaches, ranging from the modeling and estimation to the control of complex/uncertain/constrained systems and to advancement of tools for the diagnosis and reconfiguration of these systems. For this purpose, the developed (predictive, robust, nonlinear, etc.) control laws generally use optimization techniques that exploit structural properties or are based on specific heuristics. They are also oriented towards distributed/decentralized structures that are adapted to the control of multi-agent systems, large-scale systems, etc.
The theoretical developments are implemented in several application domains as interactive robotics (telemanipulation, haptics) and human movement sciences, machine tools in the context of the industry of the future (additive manufacturing), aeronautics and space, life sciences, electrical systems/networks, or transportation systems, especially automotive. Specific applications to be mentioned involve interactions among several vehicles/drones in formation flights, satellite control, bioprocesses, wind farms, charge and ageing battery state estimation for electric vehicles, energy/power optimization in multi-source networks, etc.
Moreover, the SYCOMORE team is strongly involved in Control Education, both at the engineering level (“Control Engineering” mention in the “Large Interacting Systems/Grands Systèmes en Interaction” major of the CentraleSupélec engineering program) and at the Master’s level (“Control, Signal and Image Processing/Automatique, Traitement du Signal et des Images” – ATSI, “Engineering and human movement sciences/Ingénierie et Science du Mouvement Humain” – ISMH, “Aéronautique et spatial : mécanique, automatique, énergétique” – AS-MAE), as well as in open science presentation (Fête de la Science, Girls in Control, etc.).
Robust predictive control is a key-actor in SYCOMORE research field, with applications to discrete-time, linear and nonlinear, time-varying, time-delayed and hybrid systems. Over the years, the developed constrained predictive control techniques allowed to take into account robustness with respect to bounded disturbances and/or uncertainties. Decentralized/distributed/hierarchical control has been applied to multi-agent systems, often in a fault-tolerant control context.
Optimization methods are an important tool in the SYCOMORE team. In addition to classical approaches related to convex optimization, which are still relevant for the computation of efficient control laws, significant progress has been made in considering any kind of constraints in the design process of the control law (e.g. metaheuristic methods) or in stability analysis (stability guarantee of multi-model systems, automatic computation of Lyapunov functions, etc.). Results on the utilisation of machine learning techniques in closed-loop analysis have been recently developed by the team.
Set-membership state estimation techniques based on ellipsoids and zonotopes have been developed for different classes of systems: linear time-(in)varying systems or descriptor systems subject to unknown but bounded perturbations, measurement noise and uncertainties. Interval observers have been proposed for time-delayed systems, nonlinear systems, etc. Distributed moving-horizon estimation approaches for linear and nonlinear multi-sensor multi-agent systems, as well as fault detection methods are the subject of current research.
Transportation systems, and in particular automotive, are a key application domain for the SYCOMORE team. This research axis has been further strengthened with the development of electrified vehicles, which has led to a significant number of industrial collaboration and joint PhD theses. The topics addressed in this work cover several research fields as energy management, battery state estimation, electric actuator control (e.g. to reduce vibrations) and robust control, in particular in the context of autonomous vehicles (e.g. robust control at high speed for obstacle avoidance maneuvers).
Several linear and nonlinear guidance laws have been developed for UAVs (winged, helicopter, octorotor, X- or V-rotor) while considering their utilisation for different missions (e.g. radar, cinematography). In this highly-connected domain, predictive control architectures in decentralized, distributed and/or hierarchical frameworks have been considered to control a fleet of UAVs. Moreover, on topics more related to space application, robust moving-horizon estimation methods have been applied to the problem of estimating the trajectory of space debris. Control methods for propulsion systems for small satellites have also been investigated.
The energy domain is a natural application field for the methodologies developed in the SYCOMORE team, e.g. control of nonlinear systems as wind farms, advanced hybrid control methods for power electronics, decentralized or hierarchical control structures in energy networks, etc. The team is involved in several industrial partnerships, and in particular in the RISEGrid institute (Research Institute for Smarter Electric Grids) with EDF or in the RTE chair on the digital transformation of electrical networks. Energy-related aspects are also considered in the context of the Industry of the Future, and particularly in the field of additive manufacturing. Here, energy sources (laser in LSM or electron beams in EBM) lead to different working conditions and performances that need to be analyzed.
Control and estimation methodologies find numerous applications in robotics, especially in robotic manipulation (manipulator arms or robotic hands), human-robot collaboration, and the modeling of human movement properties. The main related research themes are stability analysis of interactive robotic systems subject to uncertainties, development of robust control strategies, as well as fault detection methods. These methodological axes find a direct application in Industrie 4.0, for example for the implementation of collaborative robots in production workshops and all issues related to distribution flows (supply chain management and, in perspective, delivery via a fleet of drones).
Bioprocesses are an interesting application for the theoretical results of the SYCOMORE team. More in details, this research axis concerns the modeling, optimization, estimation and control of bioprocesses. Cultures of microorganisms grown in bioreactors have been studied in the framework of interdisciplinary collaborations, including cell biology and process engineering. Theoretical developments have been corroborated with experimental results (microalgae and bacteria culture) that have been obtained by pilot processes at laboratory scale. Various applications have been considered, mainly for sustainable development (e.g. CO2 biofixation by microalgae, waste valorization by bacteria, anaerobic digestion and biomethane production).
Professor – CentraleSupélec
Automatique et systèmes – SYCOMORE
cristina.stoica@l2s.centralesupelec.fr
Bât. Breguet C3.17
