A5072322964- Adaptive Control of Nonlinear Systems
- Advanced Control Systems Optimization
- Control Systems and Identification
- Stability and Control of Uncertain Systems
- Control and Dynamics of Mobile Robots
- Polynomial and algebraic computation
- Fault Detection and Control Systems
- Advanced Optimization Algorithms Research
- Numerical methods for differential equations
- Adaptive Dynamic Programming Control
- Advanced Differential Equations and Dynamical Systems
- Distributed Control Multi-Agent Systems
- Robotic Mechanisms and Dynamics
- Robotic Path Planning Algorithms
- Stability and Controllability of Differential Equations
- COVID-19 epidemiological studies
- Formal Methods in Verification
- Gene Regulatory Network Analysis
- Numerical Methods and Algorithms
- Neural Networks and Applications
- Magnetic confinement fusion research
- Game Theory and Applications
- Nonlinear Dynamics and Pattern Formation
- Control and Stability of Dynamical Systems
- Dynamics and Control of Mechanical Systems
University of Rome Tor Vergata
2014-2024
Istituto di Analisi dei Sistemi ed Informatica Antonio Ruberti
2019-2024
National Research Council
2023
Istituto Nazionale di Fisica Nucleare, Roma Tor Vergata
2022
Polytechnic University of Turin
2018-2019
Istituto Nazionale di Fisica Nucleare
2018
RWTH Aachen University
2016
Abstract In this paper, mechanical systems subject to impacts and contacts, that would be not controllable if the were absent (usually called jugglers ), are considered. On basis of an algorithm taken from literature a new procedure determine reference trajectory for such class systems, fully algorithmic procedure, able compute control input achieves dead–beat regulation “uncontrollable” subsystem just by using impacts, is given. Such exploits some tools borrowed algebraic geometry allow...
The purpose of this work is to give a contribution the understanding COVID-19 contagion in Italy. To end, we developed modified Susceptible-Infected-Recovered (SIR) model for contagion, and used official data pandemic up March 30th, 2020 identifying parameters model. non standard part our approach resides fact that considered as also initial number susceptible individuals, well proportionality factor relating detected positives with actual (and unknown) infected individuals. Identifying...
In this paper, we show that a one-layer feedforward neural network with exponential activation functions in the inner layer and logarithmic output neuron is universal approximator of convex functions. Such represents family scaled log-sum functions, here named log-sum-exp (LSE <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> ). Under suitable transformation, class LSE maps to generalized posynomials GPOS , which similarly be approximators...
We show that a neural network whose output is obtained as the difference of outputs two feedforward networks with exponential activation function in hidden layer and logarithmic node, referred to log-sum-exp (LSE) network, smooth universal approximator continuous functions over convex, compact sets. By using transform, this class maps family subtraction-free ratios generalized posynomials (GPOS), which we also be approximators positive log-convex, subsets orthant. The main advantage...
In this paper, we deal with the problem of output feedback stabilization for a class linear hybrid systems. This is addressed by characterizing structural properties such Namely, reachability, controllability, stabilizability, observability, constructibility, and detectability are framed in terms algebraic geometric conditions on data system. Two canonical forms, recalling classical Kalman decompositions respect to reachability given. By taking advantage characterization, duality between...
Post-disruption runaway electron (RE) beams in tokamaks with large current can cause deep melting of the vessel and are one major concerns for ITER operations. Consequently, a considerable effort is provided by scientific community order to test RE mitigation strategies. We present an overview results obtained at FTU TCV controlling position improve safety repeatability studies such as massive gas (MGI) shattered pellet injections (SPI). show that proposed beam controller (REB-C) implemented...
The goal of this technical note is to design a switching signal estimator for class elementary continuous-time or switched systems. First, the system recast into polynomial form and, secondly, some tools borrowed from Algebraic Geometry are used express as function time derivatives output and input.
In presence of indeterminate lesions by fine needle aspiration (FNA), thyroid cancer cannot always be easily diagnosed conventional cytology. As a consequence, unnecessary removal gland is performed in patients without based on the lack optimized diagnostic criteria. Aim this study identifying molecular profile long noncoding RNAs (lncRNAs) expression capable to discriminate between benign and malignant nodules.Patients were subjected surgery (n = 19) for cytologic suspicious nodules or FNA...
We employ Difference of Log-Sum-Exp neural networks to generate a data-driven feedback controller based on Model Predictive Control (MPC) track given reference trajectory. By using this class approximate the MPC-related cost function subject system dynamics and input constraint, we avoid two main bottlenecks classical MPC: availability an accurate model for being controlled, computational solving MPC-induced optimization problem. The former is tackled by exploiting universal approximation...
A data-driven strategy to estimate the optimal feedback and value function in an infinite-horizon, continuous-time, linear-quadratic control problem for unknown system is proposed. The method permits construction of policy without any knowledge model, requiring that time derivatives state are available design, even assuming initial stabilizing available. Two alternative architectures discussed: first scheme revolves around periodic computation some matrix inversions involving Q-function,...
In this paper, Linear-Quadratic (LQ) differential games are studied, focusing on the notion of solution provided by linear feedback Nash equilibria. It is well-known that such strategies related to coupled algebraic Riccati equations, associated each player. Herein, we propose an algorithm that, borrowing techniques from geometry, allows recast problem computing all stabilizing into finding zeros a single polynomial function in scalar variable, regardless number players and dimension state...
This paper presents three novel methods for the multiobjective design of controllers with tunable parameters and fixed structure linear systems. These techniques exploit geometric properties varieties envelopes to obtain a closed-form expression set candidate Pareto optimal values. The objectives span from regional pole placement more general quadratic cost functionals, thus tackling wide variety control tasks. Examples applications proposed both (possibly, nonconvex) classical benchmark...
This paper deals with the problem of computing a state feedback optimizing quadratic cost function for class linear hybrid systems. A solution to finite-horizon and infinite-horizon Linear Quadratic optimal control is found through an extension classical Differential Difference Riccati Equations. Necessary sufficient conditions, guaranteeing that stabilizes closed loop system, are stated. physically motivated example reported.
The main goal of this paper is to design a state observer for class affine switched or switching dynamical systems, without requiring the knowledge signal. To reach such some tools, taken from Algebraic Geometry, are used express signal as function output and its time derivatives. Then, similar tools an estimate both system. A physically motivated example application reported.
The main goal of this paper is to compute a class polynomial vector fields, whose associated dynamical system has given affine variety as attractive and invariant set, point in such an another solving the application technique robotic area. This objective reached by using some tools taken from algebraic geometry. Practical examples how these fields can be computed are reported. Moreover, techniques, two feedback control laws, respectively, for unicycle-like mobile robot car-like robot, which...
Summary The main goal of this paper is to design a compensator able restore the nominal behavior planar system, which rendered chaotic by an unmeasurable sinusoidal disturbance input. To reach such goal, some instruments, taken from algebraic geometry, are used estimate time derivatives output system and control Copyright © 2015 John Wiley & Sons, Ltd.
In this paper, a "practical" observer for nonlinear systems is given. Namely, it shown that if the output and its time derivatives up to sufficiently high finite order are bounded initial estimation error small, then proposed such can be made arbitrarily small all large times. Differently from other observers presented in literature, does not require either global injectivity of observability map or knowledge local inverse.
The objective of this article is to introduce a novel data-driven iterative linear quadratic (LQ) control method for solving class nonlinear optimal tracking problems. Specifically, an algorithm proposed approximate the Q-factors arising from LQ stochastic This then coupled with LQ-methods determining local solutions problems in purely setting. Simulation results highlight potential field applications.