Search

The use of recurrent neural networks to represent the dynamics of unstable systems is difficult due to the need to properly initialize their internal states, which in most of the cases do not have any physical meaning, consequent to the non-smoothness of the optimization problem. For this reason, in this paper focus is placed on mechanical systems characterized by a number of degrees of freedom, each one represented by two states, namely position and velocity. For these systems, a new recurrent neural network is proposed: Tustin-Net. Inspired by second-order dynamics, the network hidden states can be straightforwardly estimated, as their differential relationships with the measured states are hardcoded in the forward pass. The proposed structure is used to model a double inverted pendulum and for model-based Reinforcement Learning, where an adaptive Model Predictive Controller scheme using the Unscented Kalman Filter is proposed to deal with parameter changes in the system., Under review

Purpose To objectively investigate the transfer validity of simulation training using wireless elbow-worn motion sensors intraoperatively to assess whether surgical simulation leads to improvements in intraoperative arthroscopic performance. Methods In this randomized controlled trial, postgraduate year 2 to 3 trainees in nationally approved orthopaedic surgery posts were randomized to standard junior residency training (control group) or standard training plus additional weekly simulation training (intervention group). Both groups performed a supervised real-life diagnostic knee arthroscopy in the operating room at 13 weeks. Performance was measured using wireless elbow-worn motion sensors recording objective surgical performance metrics: number of hand movements, smoothness, and time taken. A participant-supervisor performance ratio was used to adjust for variation in case mix and difficulty. The study took place in a surgical simulation suite and the orthopaedic operating rooms of a university teaching hospital. Results The intervention group objectively outperformed the control group in all outcome metrics. Procedures performed by the intervention group required fewer hand movements (544 [interquartile range (IQR), 465-593] vs 893 [IQR, 747-1,242]; P Conclusions This study uses intraoperative motion-analysis technology to objectively show that surgical simulation training improves actual intraoperative technical skills performance. Clinical Relevance The described wireless objective assessment method complements the subjective observational performance assessments commonly used. Further studies are required to assess how these measures of intraoperative performance correlate to patient outcomes. Intraoperative motion analysis is translatable across surgical specialties, offering potential for objective assessment of progression through competency-based training, revalidation, and talent selection for specialist training.

This paper addresses the problem of maintaining an autonomous robotic vehicle in a moving triangular formation by regulating its position with respect to two leader vehicles. The robotic vehicle has no a priori knowledge of the path described by the leaders and its goal is to follow them by constantly regulating the inter-vehicle distances to a desired fixed value, using range-only measurements. To solve this station keeping problem, we propose a control strategy that estimates the formation speed and heading from the ranges obtained to the two leading vehicles, and uses simple feedback laws for speed and heading commands to drive suitably defined common and differential errors to zero. For straight-line motion, we provide guaranteed conditions under which the proposed control strategy achieves local convergence of the distance errors to zero. We also indicate how our design procedure can be extended to full dynamic models of marine robotic vehicles equipped with inner loops for yaw and speed control. Simulation results using realistic models are described and discussed.

In this paper, the predictor design, for altitude control of a seaweed harvester, is investigated. The harvesting system consists of a vessel and a suspended harvester device, the altitude of which is controlled by a winch. The control approach of Gallieri and Ringwood (2010), including a feedforward action, which requires a single step disturbance prediction, is investigated further, focusing on the disturbance prediction, for noisy sensors. The prediction is performed using AR and ARMA models, identified online, by using the Recursive Least Squared with Forgetting Factor (RLSFF) algorithm and the Kalman Filter (KF). The dependance between the error spectrum and the quality of the control is shown, and the prediction performances are evaluated, using an FFT-based criterion, oriented to the feedforward application. The control performances are then evaluated, and the results are compared to Gallieri and Ringwood (2010).

Roll motion in ships is a major cause for problems to passengers and human operators. The reduction of vessel roll, or roll damping, has been widely addressed in the control literature as surveyed in Perez, Ship motion control: course keeping and roll stabilization using rudder and fins (2005). The considered set-up is shown in Fig. 8.1. The system, controlled by rudder and fins, is fully-actuated.

This chapter addresses the use of \(\ell _{asso}\)-MPC for control scenarios in which actuators can be clearly divided into two groups: preferred actuators and auxiliary actuators. A pre-existing stabilising MPC is assumed to be given, which considers only the former. These preferred actuators are set to be used for most of the time and are meant to stabilise the system. On the other hand, the auxiliary actuators are meant to play a secondary role. In particular, given a pre-existing MPC the chapter provides a set of numerical tools to construct a new \(\ell _{asso}\)-MPC, which includes the auxiliary actuators among its decision variables.

Least Absolute Selection and Shrinking Operator (LASSO) is a powerful regression technique, which aims to induce sparseness in the solution (Tibshirani 1996; Osborne and Presnell 2000; Kim et al. 2007a; Ohlsson 2010; Schmidt 2010; Schuet 2010).

It is becoming standard among researchers to use a terminal constraint in MPC. This constraint generally depends on the particular set-point we want to regulate the plant to. For changing set-points this means changing the terminal constraint, which could make the problem infeasible.

This chapter introduces the first and simpler of the proposed \(\ell _{asso}\)-MPC versions. The cost function is the same as for a standard quadratic MPC, with the addition of a 1-norm penalty on the input channels. Similar to LASSO regression and the elastic-net, it is shown that this cost function can produce (for LTI systems) sparse input predictions under certain conditions. This characteristic is then shown to result in a control law that is nonlinear, even without constraints. This is in clear contrast with LQ-MPC, the unconstrained solution of which is a constant gain. The explicit solution of \(\ell _{asso}\)-MPC for LTI systems is shown to be unique, continuous and piece-wise affine in the control error. The main feature of the control law is an implicit input dead-zone, namely, the regularised actuators being exactly at zero when the control error belongs to a certain neighbourhood of the origin. The shape and extent of this region depend on the cost function as well as the constraints. The explicit solution is computed for a small example to illustrate these properties. The solution path of \(\ell _{asso}\)-MPC is then shown to be also piece-wise affine, and it is computed by means of a single multi-parametric quadratic program.

This chapter presents a modified version of \(\ell _{asso}\)-MPC that guarantees closed-loop asymptotic stability for arbitrary 1-norm input penalties. The approach is based on a new terminal cost and terminal constraint. Appropriate scalings are computed to adjust the proposed former according to the stage cost matrices. In particular, two strategies are presented to compute the required ingredients. The first one makes use of norm inequalities and linear matrix inequalities.

This paper deals with the control of systems for which there is a clear distinction between preferred and auxiliary actuators, the latter to be used only when the control error is large. Explicit MPC and exact penalty functions are used to show how l asso -MPC can implement this idea. Two l asso -MPC versions are reviewed, that allow the designer to impose a certain nominal operations zone, namely, a neighbourhood of the set-point in which the auxiliary actuators are never used. For the sake of brevity, the required procedures are shown only for version 1, but it is also discussed how they can be extended to version 2. Limitations due to the presence of constraints are also formalised. The l asso -MPC version 1 can be used to embed an existing linear quadratic MPC, while l asso -MPC version 2 can be used to obtain multiple levels of priority. The paradigm is demonstrated for version 1 through the control of the linearised lateral dynamics of a Boeing 747. In particular, the approach uses the spoilers only when the control error is larger than a desired threshold.

This paper investigates the robustness of a soft constrained LTI MPC for set-point tracking, using an l1-regularised cost. The MPC using this type of cost (informally dubbed lasso-MPC) is suitable, for instance, for redundantly-actuated systems. This is because of its ability to select a set of preferred actuators, leaving the other ones at rest for most of the time. The proposed approach aims to recover from state constraint violation and to track a changing set-point. Nominal stability is guaranteed for all feasible references and robustness to additive uncertainties is formally characterised, under certain assumptions. In particular, sufficient conditions are given for the feasible region to be robustly invariant, this region being larger than the MPC for regulation. The closed-loop system is input-to-state stable (ISS), and a local ISS gain is computed. All results apply to stabilisable LTI systems. Results hold as well for the more common quadratic MPC, a special case of the proposed controller.

In recent literature, l1-regularised MPC, or lasso-MPC, has been recommended for control tasks involving complex requirements on the control signals, for instance, the simultaneous solution of regulation and sharp control allocation for redundantly-actuated systems. This is due to the implicit thresholding ability of LASSO regression. In this paper, a stabilising terminal cost featuring a mixed l1/l22 penalty is presented. Then, a candidate terminal controller is computed, with the aim of enlarging the region of attraction.

The recently investigated $\ell_{\mathrm{asso}}$ model predictive control (MPC) is applied to the terminal phase of a spacecraft rendezvous and capture mission. The interaction between the cost function and the treatment of minimum impulse bit (MIB) is also investigated. The propellant consumption with $\ell_{\mathrm{asso}}$ MPC for the considered scenario is noticeably less than with a conventional quadratic cost and control actions are sparser in time. Propellant consumption and sparsity are competitive with those achieved using a zone-based $\ell_1$ cost function, whilst requiring fewer decision variables in the optimisation problem than the latter. The $\ell_{\mathrm{asso}}$ MPC is demonstrated to meet tighter specifications on control precision, and also avoids the risk of undesirable behaviours often associated with pure $\ell_1$ stage costs.

Companies are involved in a high competition for reducing the cost of production in order to maintain their market shares. Since the costs of maintenance contribute a substantial portion of the production costs, companies must budget maintenance effectively. Machine deterioration prognosis can decrease the costs of maintenance by minimizing the loss of production due to machine breakdown and avoiding the overstocking of spare parts. This paper gives a review of some fault detection and prognosis methods to diagnose faults and failure on rotating electrical machines. To develop the monitoring system accelerometers have been used to acquire vibration measurements. Performance are studied on a laboratory-scale experimental system.

In this paper, the feasibility of the altitude control of a seaweed harvester is examined. The harvesting system consists of a vessel and a suspended harvester device, the altitude of which is controlled by a winch. The goal of the control action is to maintain the harvester at a constant altitude with respect to the seabed profile. A control strategy is proposed, including a vessel motion feed-forward action, using a motion reference unit (MRU), and an altitude feedback loop, using a sonar device for altitude measurement.