This paper proposes a Robust Stochastic Model Predictive Control for Autonomous Underwater Vehicles that adapts controller settings online using a Golden Section search optimization method. The baseline control system employs a tube-based Stochastic Model Predictive approach within a Linear-Parameter-Varying framework to manage complex, unpredictable underwater disturbances. The proposed adaptation policy consists of two phases. In the offline stage a bank of nominal controllers is designed by exploiting Voronoi tessellation and Lloyd's algorithm to represent the Linear-Parameter-Varying model polytope. In the online stage, the control system switches between these controllers while simultaneously tuning ancillary policy settings via Golden Section search. Simulation tests under disturbance scenarios demonstrate the effectiveness of this approach compared to different robust Model Predictive Control techniques. Finally, a computational complexity analysis highlights the performance trade-off, confirming the algorithm's suitability for real-time applications.

In the last years, several robotics platforms have been proposed and applied in different scenarios to evaluate solutions and test application aimed to solve problems and issues related to old and impaired people. In this paper, we present the design of the control software to a new robotic platform conceived for an Ambient Assisted Living use case. The robot considered in this work is the Mercury X1 from Elephant Robotics, a mobile robot equipped with two robotics arms and a set of modern video and audio sensors and actuators. The interesting advanced features characterizing this robot are related to a simple and effective control software based on Python APIs, a powerful computational unit, and the native possibility to be controlled by combining the the baseline Python software with the preinstalled Robot Operating System framework. In this paper, the capabilities of the Mercury X1 are presented and exploited for developing the control system for a real-world Ambient Assisted Living problem. In this paper, we show how the robot can be used to create an application for bedridden people. In particular, in the considered case study, the robot is used to be remotely controlled by the patient, who thus has the possibility to interact with the surrounding environment more or less remotely, reducing the need for support in everyday life. This is achieved by developing a remote control system for the robot through its control exoskeleton, which is characterized by the particular type of application. The paper explores and evaluates the possibilities and limitations of this innovative robotics platform. The reported results illustrate the performance of the robotic arms control using the baseline robot control system developed by the developer.

This paper presents the design and development of the proof of concept of a driver classification and coaching system for Hybrid Electric Vehicles. Modern electrified vehicles require a change of behavior of drivers to achieve expected economic performance. The design of the system aimed to detect and train a driver to adjust his driving style is approached by exploiting capabilities of modern data-driven Machine Learning paradigm and combining Supervised and Unsupervised Learning techniques. A simulation model of a hybrid vehicle is considered to generate the dataset needed for Artificial Intelligence training and to test the performance of the developed driver classification and coaching system. The performance of the developed system is validated over a wide set of simulation tests performed by using the considered hybrid vehicle simulation system.

The design and development of a data-driven algorithm to estimate the State-of-Health of a battery is presented. The approach is based on a data-driven Least-Squares Support Vector Machine approach. By combining the data-driven method with a dataset pruning procedure and nonlinear optimization technique, the computational complexity of the estimator is reduced whilst maintaining the performance of the estimator. The design approach was validated in simulation testing by considering the simulated model of a battery. An Estimator Design Tool was developed within the MATLAB environment. It provides a user-friendly interface for the different algorithms that may be used in the estimator design. The approach and tool is quite general and is suitable for a wide range of other estimation applications.

This paper presents the preliminary results of a Linear Parameter Varying-Autoregressive eXogenous model, identified through Least Squares Support Vector Machines, able to optimally drive a robotic arm system by emulating the performance of a Nonlinear Model Predictive Control (NMPC) policy. The support vector machine framework is employed to replicate the control performance of a computationally demanding NMPC. Due to the nonlinear characteristics of the robotic arm, the NMPC is suitable to guarantee expected control performance. However, its application in real-time systems with fast dynamics is limited by high memory and computational demands required at each sampling instant. In this work, the linear parameter varying model is trained using a data-driven approach to imitate the control actions of the NMPC across different scenarios. The proposed controller and the original NMPC are evaluated in simulation, considering multiple operating conditions of the robotic arm. The control performance of both approaches is then compared to assess the effectiveness of the proposed method.

The performance of Wave Energy Converters (WECs) depends on the capability of the control system to effectively predict the force of excitation, caused by the dynamics of sea waves acting on the system. This is of particular importance in the case of advanced control policies, as for constrained and predictive control algorithms, that makes explicit use of the predicted dynamics of controlled system and related disturbances acting on it for developing the control law. This paper proposes a prediction algorithm, developed within the Support Vector Machine framework, able to provide an effective prediction of the excitation forces acting on WECs. The proposed data-driven algorithm can be designed by off-line training but, due to the unpredictable long-term dynamics variability of sea conditions, pre-trained data-driven algorithms cannot effectively consider such a varying conditions. To overcome this limit, the proposed approach is featured by the capability to adapt the prediction to unknown dynamics by learning from on-line measured or estimated data. This feature also allows to limit the computational complexity of the algorithm while its prediction capabilities are adapted to time-varying sea state conditions evaluated in real-time. The proposed approach is tested on simulated data generated from a high-fidelity WEC simulator.

Linear Parameter-Varying Model Predictive Control has been shown to provide an effective design approach for developing an Energy Management System for Hybrid Electric Vehicles. However, despite the good performance achieved, modern data-driven Artificial Intelligence methods can improve the performance due to the approximations involved in generating the models. An approach is described for reducing the sub-optimality due to the modelling problem in predictive control using an AI algorithm belonging to the class of data-driven Machine Learning techniques. This provides more effective vehicle speed and driver torque demand predictions that are used within the predictive controller. The proposed combined policy is compared with a baseline control design developed using the well-known Equivalent Consumption Minimization Strategy and an MPC neglecting the use of AI predictors.

This article presents a method to detect and classify series arc faults affecting domestic AC electrical circuits by the analysis of electric current time series data, based on the HYDRA (HYbrid Dictionary-Rocket Architecture) algorithm, a fast dictionary method for time series classification employing competing convolutional kernels. The key novel contributions are twofold: Competing convolutional kernels are suitable to effectively extract features representing an effective set of arc fault detection indicators, and the classification performed in this way is feasible to be executed in real time. The proposed method is validated using a public database, where data from 13 different types of loads is collected according to the IEC 62606 standard. To reduce inference time and optimize the algorithm for embedded control units, a feature reduction strategy is employed. The effectiveness of the proposed method is demonstrated through experimental tests conducted under both arcing and non-arcing conditions and across different load types. Moreover, its accuracy is also tested in case of transients caused by operational changes in common electrical appliances. Achieved results show a detection accuracy of approximately 99%, with appliance classification performance around 98%, with inference times ranging from 2.8 to 172.0 ms while executing the algorithm on an ARM Cortex-based board.

In this article, we propose a time-varying model predictive control (MPC)-based scheme to enhance the dynamic performance of dc-dc converters. The proposed approach employs MPC as a reference governor (RG), addressing industrial certification constraints that may limit modifications to the low-level controller. To accommodate the computational limitations of conventional control boards, we introduce a highly efficient real-time optimization algorithm for solving equality-constrained quadratic programming (QP) problems. The algorithm is based on a tailored QR factorization that outperforms well-known linear algebra libraries, and it is shown to be superior to condensing with state elimination. Furthermore, we implement an efficient recursive least-squares (RLS) method to provide a linear-time varying model for the adaptive MPC-based RG. No information regarding the topology of the converter nor the structure of the low-level controller is required for such adaptation, making the proposed method self-tuning and eliminating the need for prior identification steps. The proposed control scheme has been tested on various simulated and real dc-dc converters, demonstrating its computational and memory efficiency, as well as its versatility across different converter topologies.

The results of a study that investigated the use of Digital Twin technologies for Electric Vehicle propulsion system state of health monitoring is considered. Modern vehicles can share large amounts of data in the cloud through wireless connections. Digital twins represent an effective approach to exploit data-sharing and modern data-driven Artificial Intelligence and Machine Learning technologies, including vehicle monitoring or driving scenarios analysis. This study describes the design and development of a proof-of-concept digital twin demonstrator, that can detect fault/fault-free conditions in electric motor components. It can be used to assess the overall electric drive Failure Rate and to estimate the Remaining Useful Lifetime of the motor. The demonstrator developed within a simulation environment has been validated over a wide set of simulated operating scenarios demonstrating the effectiveness of the proposed approach.

The decarbonization of the commercial transport sector is a crucial part on the pathway to a fully green economy and the use of zero-emission Heavy Duty Vehicles (HDVs) is a major aim. Hydrogen Fuel Cells will probably represent the main technology to provide an alternative future fuel source to replace fossil fuels. This will involve combining fuel cells with batteries in HDVs, exploiting the full potential of these technologies in an economically effective way. To guarantee expected performance from fuel cell-based powertrains, these should be controlled by an appropriate Energy Management System to optimize the performance of the vehicle. This paper illustrates the performance achieved by applying optimal predictive control to the zero-emission Heavy-Duty Vehicles' power management problem.

Input saturation and actuator faults are common issues in control system engineering. This paper proposes an Error Governor (EG) policy that dynamically manages the feedback error to enhance the tracking error in Multiple Input-Multiple Output (MIMO) closed-loop system controlled by Proportional-Integral-Derivative (PID) regulators. By integrating an Adaptive Kalman Filter (AKF) for optimal fault estimation with the EG scheme, the solution enables fault-tolerant control without requiring modifications to the baseline controller, which can be impractical or unsuitable in certain applications. Furthermore, the proposed control scheme completely avoids windup, thereby replacing conventional Anti-Windup (AW) schemes for PIDs, and is computationally cheap and easy to implement, needing only the same inputs as conventional AW algorithms and the actuator fault estimation. Simulation results on the MIMO model of the Zagi flying-wing aerial vehicle show that the EG reduces the tracking error in presence of actuator faults by $ -49.92\% $ -49.92% in terms of Integral Absolute Error (IAE), outperforming conventional AW methods.

This paper presents an attack strategy for autonomous unmanned aerial vehicles. Unmanned aerial vehicles are driven by two-layer control systems composed of an inner loop driving the vehicle’s altitude and attitude and an outer loop managing the position. In this study, it is assumed that the attacker can access the lower control loop and modify the control signal driving the actuators. While the vehicle is under attack, an optimal policy oriented to drive the vehicle in a failure condition is applied to compute the hacked control signals, thus overriding the inner-loop controller. This optimal policy is an Antagonistic Controller based on the Model Predictive Control paradigm. The antagonistic controller iteratively evaluates the effect of a possible attack, within the available attack time interval, to identify the suitable operating condition to initiate the attack. This evaluation is performed by the analysis of a performance index related to the Antagonistic Control state predictions to damage the vehicle. The proposed approach has been tested in simulation using a detailed nonlinear quadrotor model to show the effectiveness of the proposed approach.

The design and development of a data-driven algorithm for battery State-of-Charge estimation is presented. The estimation of battery SoC is important in the development of Battery Management Systems. The proposed approach exploits the Least-Squares Support Vector Machine data-driven estimation paradigm and statistical methods. The algorithm’s computational complexity is reduced by using a data pruning procedure. The optimization of the SVM-based estimator is performed by using a Particle Swarm Optimization method. The design approach proposed to develop to estimator is validated using a simulation model of the battery and an Estimator Design Tool in MATLAB software which provides a user-friendly interface for the different algorithms that may be used in the estimator design. The approach is applicable to a wide range of applications including automotive systems.

Background: Human-Machine Interaction (HMI) has been an important field of research in recent years, since machines will continue to be embedded in many human actvities in several contexts, such as industry and healthcare. Monitoring in an ecological mannerthe cognitive workload (CW) of users, who interact with machines, is crucial to assess their level of engagement in activities and the required effort, with the goal of preventing stressful circumstances. This study provides a comprehensive analysis of the assessment of CW using wearable sensors in HMI. Methods: this narrative review explores several techniques and procedures for collecting physiological data through wearable sensors with the possibility to integrate these multiple physiological signals, providing a multimodal monitoring of the individuals’CW. Finally, it focuses on the impact of artificial intelligence methods in the physiological signals data analysis to provide models of the CW to be exploited in HMI. Results: the review provided a comprehensive evaluation of the wearables, physiological signals, and methods of data analysis for CW evaluation in HMI. Conclusion: the literature highlighted the feasibility of employing wearable sensors to collect physiological signals for an ecological CW monitoring in HMI scenarios. However, challenges remain in standardizing these measures across different populations and contexts.

This paper introduces a kernel-based method for learning feedforward controllers within an Iterative Learning Control (ILC) framework tailored for nonlinear processes. Unlike traditional ILC algorithm that relies on the knowledge of first principle-based models, this approach leverages a data-driven methodology to develop an iterative control update rule using kernel-based training. We compared this method against a traditional ILC scheme and a baseline neural network-based approach. The effectiveness of the proposed method is demonstrated through a unicycle path-following control problem, evaluated across various simulated test scenarios. Performance metrics include vehicle tracking error and ILC convergence speed, confirming the effectiveness of the proposed data-driven approach.

The ongoing demographic transition, characterized by a projected rise in the old-age dependency ratio from 28% to 50% by the year 2060, indicates that age-related illnesses will provide a significant issue in the future. In this perspective, developing technological solutions able to support older individuals and people with special needs in their autonomous mobility could be crucial. However, when delivering such solutions, it is fundamental to monitor the affective state of the users to observe their acceptance of the technology. The approach proposed in this study integrates a smart wheelchair with a physiological computational module, composed of co-registered RGB and Infrared Cameras, with integrated artificial intelligence algorithms for affective computing, delivering a classification of the stress and engagement condition of the user. This study aims to showcase the technical viability of such an approach, monitoring and comparing the stress and engagement states of individuals during autonomous and manual smart wheelchair navigation. The results did not deliver significant differences in stress and engagement condition between the two driving modalities, demonstrating the acceptability of the proposed framework.

The performances of energy management systems or electric vehicles and hybrid electric vehicles are highly dependent on the forecast of future driver torque/power request sequence that affects vehicle efficiency and economy. Since the behaviour of the driver is challenging to model/predict by first-principles models, modern artificial intelligence algorithms would represent feasible methods for approaching this problem in real-world automotive systems. This work provides a comparative study and analysis of performances of different data-driven torque prediction strategies. The studied and compared torque demand prediction techniques are exponentially varying model, linear regression, shallow and deep neural networks, and least square support vector machine-based approaches. The prediction performance and computational cost of these techniques are evaluated and reported, and the possibility of exploiting these techniques in real-world scenarios is also discussed.

In this paper, we propose an efficient method for handling large datasets in linear parameter-varying (LPV) model identification. The method is based on least-squares support vector machine (LS-SVM) identification in the primal space. To make the identification computationally feasible, even for very large datasets, we propose estimating a finite-dimensional feature map. To achieve this, we propose a two-step method to reduce the computational effort. First, we define the training set as a fixed-size subsample of the entire dataset, considering collision entropy for subset selection. The second step involves approximating the feature map through the eigenvalue decomposition of the kernel matrices. This paper considers both autoregressive with exogenous input (ARX) and state-space (SS) model forms. By comparing the problem formulation in the primal and dual spaces in terms of accuracy and computational complexity, the main advantage of the proposed technique is the reduction in space and time complexity during the training stage, making it preferable for handling very large datasets. To validate our proposed primal approach, we apply it to estimate LPV models using provided inputs, outputs, and scheduling signals for two nonlinear benchmarks: the parallel Wiener-Hammerstein system and the Silverbox system. The performances of our proposed approach are compared with the dual LS-SVM approach and the kernel principal component regression.

This paper presents a Model Predictive Control (MPC) based autopilot for a fixed-wing Unmanned Aircraft Vehicle (UAV) for meteorological data sampling tasks, named Aerosonde. Aerosonde missions are featured by predetermined operating conditions, allowing the design of ad-hoc controllers for each control task by using the future knowledge of the reference signals driving the aircraft during operations. To develop the controller, the nonlinear dynamics of the vehicle has been described by a Linear Parameter-Varying (LPV) model identified from the plant data by using a subspace identification technique. The LPV model is used to design a MPC to drive the UAV. Two different Linear Parameter-Varying MPC (MPCLPV) algorithms have been proposed by introducing the previewing technique in the controller due to the a priori knowledge of full reference signals. In the design of the inner Attitude Controller (AC), a future LPV scheduling parameters estimation policy has been introduced (PF −MPCLPV) for improving the control results of the standard Previewing MPCLPV (P-MPCLPV). Furthermore, an anticipative switching approach (PS −MPCS) has been considered for the altitude External Controller (EC) to improve the control performances of the standard previewing switching MPC (P-MPCS). Both PF −MPCLPV and PS −MPCS algorithms have been compared to the P-MPCLPV and P-MPCS baseline algorithms, showing the effectiveness of proposed methods.