Search

International audience; A new design of output-feedback repetitive control scheme for nonlinear minimumphase systems with arbitrary relative degree and globally Lipschitz nonlinearities is proposed. This work extends the recent results in Astolfi et al. [2021]. The delay of the repetitive control scheme is represented by a transport equation. A high-gain observer and a forwarding-based feedback law are employed to steer the desired output to zero in presence of periodic signals (references and/or perturbations) and model uncertainties.

This paper exploits the so-called congelation of variables method to design an adaptive controller for nonlinear systems with time-varying parameters. Two motivating examples describing scalar systems are discussed to illustrate the flexibility of the congelation of variables method to deal with the cases in which the time-varying parameters are coupled with the state and with the input, respectively. Interpretations from a passivity perspective are also provided. Then design procedures are derived for general nonlinear systems in parametric strict-feedback form, and it is shown that the state of the underlying system converges to the origin and all signals of the closed-loop system remain bounded. Simulations show that, in the presence of parameter variations, the performance of the proposed controller is superior to that of the classical adaptive controller designed for time-invariant systems.

This paper presents a project aimed at the optimization of a water distribution network located in Trento (Italy). Several in-depth hydraulic studies have been conducted in order to perform hardware modifications through sectorization procedures. Advanced Process Control methods have been designed in order to optimally and automatically manage the net pressure and the scheduling of the involved pumping stations. Net pressure has been minimized through two-layer Model Predictive Control techniques, while advanced logics have been designed for the pumps scheduling. The developed Advanced Process Control system has been successfully installed on the considered network and the achieved results are here illustrated.

International audience; High-gain observers and sliding mode observers are two of the most common techniques to design observers (or differentiators) for lower triangular nonlinear dynamics. While sliding mode observers can handle bounded nonlinearities, high-gain linear techniques can handle global Lipschitz nonlinearities. In this preliminary paper, we present a novel observer design for second and third order systems which benefits from both techniques. More precisely, the proposed observer converges in finite-time and handles nonlinearities satisfying an incremental affine bound.

This work investigates the position control problem for a soft continuum manipulator in Cartesian space intended for minimally invasive surgery. Soft continuum manipulators have a large number of degrees-of-freedom and are particularly susceptible to external forces because of their compliance. This, in conjunction with the limited number of sensors typically available, results in uncertain kinematics, which further complicates the control problem. We have designed a partial state feedback that compensates the effects of external forces employing a rigid-link model and a port-Hamiltonian approach and we have investigated in detail the use of integral action to achieve position regulation in Cartesian space. Local stability conditions are discussed with a Lyapunov approach. The performance of the controller is compared with that achieved with a radial-basis-functions neural network by means of simulations and experiments on two prototypes.

International audience; We combine the low-power high-gain observer recently proposed in Astolfi and Marconi (2015) with the updated-gain technique used in Andrieu et al. (2009). The resulting adaptive low-power high-gain observer inherits the advantages of both techniques and can be used to address the state-estimation problem for Lipschitz systems in lower triangular form with nonlinearities having a Lipschitz constant that depends on a known external input.

We introduce a method for model reduction of linear time-varying (LTV) systems by extending the Loewner framework developed for linear time-invariant (LTI) systems. This extension is accomplished by utilizing a state-space interpretation of the Loewner matrices previously developed by the authors. New time-varying Loewner functions are defined, and a Loewner equivalent model is produced using these functions.

We investigate output regulation for linear networked control systems using the emulation approach. We consider a regulator solving the problem for linear systems in the absence of the network, by following the Francis-Wonham framework. Next, the network is taken into consideration, in particular the induced time-varying inter-transmission intervals and scheduling constraints, and we model the overall system dynamics as a hybrid system. We show that only practical regulation can be achieved in general since the steady-state control input steering the output to zero cannot be generated in the networked context (in general). Explicit upper bounds on the output asymptotic gain and on the maximum allowable transmission interval guaranteeing boundedness of the closed-loop trajectories are provided, which are shown to be related.

We propose a technique to modify a given discrete-time (nonlinear) observer so that the state estimate remains in a given convex set, without altering the observer performances in terms of convergence and robustness to external disturbances. The proposed approach can be used to remove the peaking phenomenon or to attenuate the effect of impulsive outliers in the measures. It assumes that it is possible to execute a certain number of computations between any two sampling times in order to refine the current estimate and bring it back into the prescribed set. The proposed technique can be applied to any class of nonlinear observers for which a quadratic Lyapunov function is used to prove stability.

Field robotics is a fast developing research field, in particular precision agriculture is gaining in popularity due to the high return in productivity and the reduced pollution impact on the environment. The GRAPE project is an ECHORD++ robotic experiment aimed at the use of a mobile robot for automatic pheromone dispenser distribution in vineyards, to reduce pesticide use thanks to pheromone mate disruption. This work describes the autonomous navigation system of such robot. For the specific scenario a real state of the art does not exists, so we adapted techniques that have been designed for different problems, in particular classical methods for navigation and mapping in indoor environments. The vineyard environment is challenging because of many variability factors such as weather, soil and vegetation. These factors hinder the indoor methods introducing noise in the robot perceptions. To solve this problem, we propose a specific navigation system that takes advantage of multiple sensors: wheel encoders, inertial measurement unit (IMU), and GPS to filter the environment noise and accurately estimate the robot odometry. In addition, the system exploits a LIDAR sensor to localize the robot, through the Adaptive Monte Carlo Localization (AMCL) algorithm, and to map the vineyard. We tested the system in simulation with very good results which have been confirmed during field tests in a real vineyard.

The model reduction problem by moment matching for continuous-time, single-input, single-output, linear, time-invariant systems is studied at isolated singularities (in particular, at poles). The notion of moment at a pole of the transfer function is defined. Exploiting this notion a one-to-one correspondence between moments at a pole of the transfer function and the “limit solution” of a family of Sylvester equations is established. Finally, a family of reduced order models is defined. A simple example illustrates the theory.

The problem of the stability analysis for nonlinear differential-algebraic systems is addressed using tools from classical control theory Exploiting Lyapunov Direct Method we provide linear matrix inequalities to establish stability properties of this class of systems. In addition, interpreting the differential-algebraic system as the feedback interconnection of a dynamical system and an algebraic system, a sufficient stability condition has been derived using the small-gain theorem. The proposed techniques are illustrated by means of simple examples.

This paper studies the trajectory tracking problem for the nonlinear model of a scooter and presents a robust H ∞ controller based on measurements of the tracking errors, the roll angle, the yaw angle and the steering angle. The study first introduces the full nonlinear model developed in Autosim which has 12 degrees of freedom. This is far more complex than a simple bicycle model and provides a good description of the scooter. Then a robust H ∞ controller based on the linearization of the nonlinear model is designed. Finally, the effectiveness of the controller is verified by means of two case studies.

This paper deals with the characterisation of the sensitivity to high-frequency measurement noise of nonlinear high-gain observers. The proposed tool provides bounds for the steady-state estimate in presence of noise when the high-gain parameter characterising the speed of convergence of the observer is fixed. The nonlinear analysis captures the effect of the noise frequency showing the "low-pass" filtering properties of the observer.

With the aim of designing controllers to simultaneously ensure robustness and optimality properties, the mixed H 2 / H ∞ control problem is considered. A class of input-affine nonlinear systems is considered and the problem is formulated as a nonzero-sum differential game, similar to what has been done in the 1990s by Limebeer et al. for linear systems. A heuristic algorithm for obtaining solutions for the coupled algebraic Riccati equations which are characteristic of the linear quadratic problem is provided together with a systematic method for constructing approximate solutions for the general, nonlinear problem. A few numerical examples are provided.

The persistence of excitation of signals generated by time-invariant, continuous-time, autonomous linear and nonlinear systems is studied. The notion of persistence of excitation is characterised as a rank condition which is reminiscent of a geometric condition used to study the reachability properties of a control system. The notions and tools introduced are illustrated by means of simple examples and of an application in system identification.

A high-fidelity, low-cost ultrasound simulator has been developed in order to improve the efficiency of ultrasound training, particularly in Focus Assessment by Sonography in Trauma (FAST). This examination is rapidly increasing its diffusion in Emergency Medicine, as it represents a quick tool for the detection of free fluids. The probe is the main component of the simulator as it provides the on-line tracking of its pose relative to the manikin representing the virtual patient. The proposed algorithm for the estimation of the probe orientation, a nonlinear complementary filter, is based on the fusion of information coming from inertial sensors. It is tested in static and dynamic conditions and its performances are compared with another traditional algorithm: the Kalman filter. Moreover, the software for the processing of medical dataset as images, videos and 3D volumes, has been developed in order to display the appropriate visualization on the simulator monitor. Finally, the simulator has been tested in a training session.

We address the problem of output stabilization for the class of nonlinear minimumphase systems with a well-defined relative degree whose output is affected by measurement noise. In the proposed control structure the derivatives of the output are estimated by means of a new high-gain observer proposed in Astolfi and Marconi [2015], which, with respect to high-gain observers conventionally used in literature, has limited gain power. The advantage in using the proposed observer is in its implementation for systems with high relative degree and in the sensitivity to the sensor noise. Simulation results are presented to show the effectiveness of the proposed control structure.