|Personal data||Research themes||Ongoing teaching||Publications|
Department of control engineering and system analysis
Person in charge of the Unit : Oui
The research activities of the SAAS group are essentially oriented in two directions: - Supervision and fault diagnosis of technical systems based on mathematical models - Advanced control of nonlinear dynamic system subject to constraints, with emphasis on reference governor methods The developed methods are applied to a wide range of areas including: - Electric energy production, distribution and storage - Mechatronics (UAVs and robotics) - Industrial processes
Optimization Free Constrained Control
In the present times, Model Predictive Control (MPC) is the standard for constrained control. This scheme computes the prediction of the control action over a sufficiently long horizon by solving an optimization problem online. Its strong point is also its weakest: to have the certainty that the problem will be solved on time, the system MPC is applied to must be slow enough – especially when dealing with nonlinear systems. The goal of this project is to develop and extend constrained control schemes that do not rely on optimization in the loop. In this project, we focus in the recently proposed Explicit Reference Governor. In essence, this scheme modifies the applied reference to the system in a way that makes the controlled variables fulfill the constraints. All of this is done by computing a certain safety index, the Dynamic Safety Margin. Up to now, computing a suitable Dynamic Safety Margin is not straightforward, except for linear systems subject to linear constraints. This project aims to extend the ERG with new ways of computing the Dynamic Safety Margin for wider ranges of constraints and with definitions of the Dynamic Safety Margin. For more information, visit the website of the project: https://saasofcc.wordpress.com
PANTHEON : Precision farming of hazelnut orchards
Project PANTHEON is a research project funded by the European Commission within the Horizon 2020 framework programme. Inspired by the real needs of the partner Ferrero, the project aims at defining a new paradigm for the precision farming of hazelnut orchards. The core idea is to build a system able to monitor the phytosanitary status of each single plant of the orchard. This will result in focused interventions, yielding an increase in the overall orchard production while being more cost-effective and environmentally friendly. The PANTHEON consortium is coordinated by Prof. A. Gasparri and consists of four universities (Roma Tre, ULB, Trier, and Tuscia) and two industrial partners (Sigma Consulting, and Ferrero), which provide the complementary expertise and technologies (ranging from robotics and control theory, to agronomy, remote sensing, and big data) needed for the success of the project. For more information : https://www.facebook.com/ProjectPantheon/ and https://twitter.com/ProjectPantheon
Navigation and trajectory control of a robotic hummingbird
This project is a part of the COLIBRI project, which aims at building a robot mimicking the flight of a natural hummingbird. The objective of the research is to integrate appropriate sensors (such as vision-based sensors) to the COLIBRI control board in order to, in one hand, increase the accuracy of the hovering flight and, on the other hand, utilize it as a navigation system to perform desired trajectories. Then a control algorithm will be developed to process the real-time signal obtained by the integrated sensor in order to steer the robot through the desired trajectory. The performance of the proposed control algorithm including sensor integration will be evaluated by flight experiments.
Knee Advanced Rehabilitation Device
Nowadays, knee surgery is a relatively common procedure used to treat cartilage defects and/or soft tissue lesions, osteoarthritis with total or unicondylar knee arthroplasty, and cruciate ligaments (ACL/PCL) injuries. After surgery, each patient undergoes a long period of rehabilitation (typically from 6 weeks to 6 months) consisting of long sessions of physiotherapy and medical training therapy carried out by qualified personnel. This procedure is long and expensive, and may cause work-related pathologies to physiotherapists because of the significant workload it implies. These considerations highlight the high potential benefits that robotic solutions could bring in this field. As a matter of fact, if technology could provide an effective tool to assist the physiotherapist, the rehabilitation time and cost could be reduced, with important benefits for both the patients and the operators. Although some knee rehabilitation devices are already available, they have proved to be not very effective and, in some situations, could even harm the patient. The main reason of this is related to the over simplistic and non-physiological assumptions used for the device design: the first one is to consider the knee as a 1 Degree of Freedom joint enabling only flexion-extension. Second, these devices constrain the knee to “blindly” perform movements with very few adaptations to the patients’ anatomy and muscular reaction. This project aims to solve this issue by investigating it from a biomechanical and robotic perspective, considering human physiology patient-specific knee kinematics and kinetics as the starting point of the design.
Robotic Bricklayer: a multi-robot system for sand-lime blocks masonry
In this project we will design and develop an innovative concept for the automatic bricklaying of sand-lime bricks. The robotization of this activity poses a series of fundamental methodological challenges. The first challenge is that the weight of the blocks and the typical distance at which the blocks must be placed makes the classical robotic assumption of “rigid-manipulator” unreasonable. Accordingly we will study a “soft-robotic” mechanical design, together with a suitable sensing equipment and control law. The second challenge in this project is the design of safe and effective control solutions to perform the construction task. The correct placement and alignment of each block will be done through a combination of position and force control. A prototype will be developed for laboratory tests.
Optimal Constrained Set Point Regulation for Networks of Systems
The project deals with complex systems arising from the composition of many locally controlled subsystems. These systems are interacting and they must be coordinated to achieve a common goal, namely the optimization of a common cost function, while meeting operational constraints defined in terms of control and state variables. The overall dynamic model for such systems is typically known only partially. Besides, the subsystems can be influenced by unpredictable external factors and they can be subject to faults. The aim of the project is to study and develop a management scheme that is able to set the set-point (or the reference trajectory) of each subsystem in order to meet the optimality conditions and the constraints mentioned above, despite the modelling uncertainties and the external disturbances. As an example, let us mention the case of a wind farm. The wake effects induce a coupling between the wind turbines. The aim is to determine the set point of each of them so as to maximize the produced power while limiting fatigue loads, despite the restricted knowledge on the wind velocity field within the wind farm. The work will be divided in three lines of research: – set point optimization for constrained and uncertain systems subject to external disturbances through a centralized approach, – set point reconfiguration upon occurrence of faults, in order to pursue the operation, possibly in a degraded mode, – development of an approach for set point management able to solve the problems mentioned in the first two lines of research in a decentralized way. The developed methodologies will be validated on a realistic wind farm simulator on the one hand, and on a cooperating multi-robot system on the other hand.