Keywords: Intelligent transportation systems, Unmanned systems, Modelling and simulation
Abstract: Traffic state estimation (TSE) is a challenging task due to the collection of sparse and noisy measurements from fixed points in the traffic network. Unmanned Aerial Vehicles (UAVs) have been gaining popularity as traffic sensors due to their ability to monitor a number of important traffic parameters over space and time. In this work, we develop a novel UAV-based sensing architecture which provides sparse, noisy measurements of traffic densities and transfer flows of the traffic network. Assuming free-flow conditions, we construct a Kalman filter approach that utilises knowledge of regional split ratios along with the UAV-based measurements. To avoid the assumption of known split ratios, we further develop a weighted least-squares optimization approach that minimizes measurement and process errors over a moving horizon window subject to linear traffic dynamics to accurately estimate traffic densities. We compare the UAV-based sensing architecture to an all-measurement method where we assume that measurements for all traffic densities and transfer flows are available at every time-step. Results show that the UAV-based sensing architecture compares favourably to the all-measurement scenario and the proposed optimization based estimator achieves similar results to the Kalman filter, even when regional split ratios are unknown.

Keywords: Predictive control, Multi-agent systems, Unmanned systems
Abstract: This paper presents a nonlinear model predictive control (NMPC) approach for trajectory tracking of a multi-UAV system with a high risk of collision. The proposed solution focuses on minimizing unnecessary complex maneuvers while ensuring collision avoidance without compromising the final position accuracy. Allocating different levels of passing priority to the agents enables fewer alternations of the initially planned path as only the agents with lower passing priority handle collision avoidance when the risk arises. The agent with a higher passing priority tracks its reference trajectory along the planned path, without considering collision avoidance. This strategy aims to perform fewer alternations and aggressive maneuvers resulting in increased safety of the multi-UAV mission. All the agents solve an unconstrained optimal control problem in a distributed manner, as collision avoidance is also defined as a cost function term that penalizes the proximity between the agents. Finally, the performance of the proposed approach is studied in simulation in the case of a two-quadcopter mission, highlighting its efficiency and robustness against external disturbances and model uncertainties.

Keywords: Unmanned systems
Abstract: Real-time wildfire perimeter tracking provides situational awareness and enhances decision-making during firefighting. This paper proposes a UAV-based system that integrates real-time data collection (using onboard sensors) into a fire propagation model to provide accurate state information on the wildfire perimeter and improve fire prediction. Firstly, a data fusion scheme is devised to employ available historical data in combination with real-time measurements to provide updated inputs to the fire propagation model. Then the model is used to predict the future fire perimeter and uses these predictions to guide the UAV to track the fire perimeter better. The proposed system is evaluated in extensive simulation experiments, demonstrating its effectiveness for real-time wildfire perimeter propagation tracking.

Keywords: Unmanned systems, Autonomous systems, Optimisation
Abstract: In this paper, a route selection algorithm is proposed for aerial and ground vehicle cooperative control. The main goal is to determine a feasible trajectory for a drone, which satisfies several limitations. Moreover, during the route selection, the flight time is also minimized to increase the efficiency of the entire system. The route selection is performed by a graph-based method, which is evaluated for different initial conditions. Then, the proposed algorithm determines the trajectories for the drone, and the predefined limitations are also considered. The method is validated in a MATLAB-based simulation environment, in which the whole algorithm is implemented.

Keywords: Unmanned systems, Nonlinear systems, Optimisation
Abstract: This work presents a novel multirotor UAV that incorporates a spatial configuration and inclined rotors with the aim of overcoming the under-actuation issue. The modeling of the multirotor is based on the Newton-Euler formalism, taking into consideration the asymmetry of the structure and the center of gravity, which may not coincide with the origin of the body frame. In addition, a Proportional Integral Derivative (PID) is used to control the considered drone. In order to select the best control gains for the applied control method, a meta-heuristic algorithm based on Particle Swarm Optimization (PSO) is employed. The simulation results for attitude change with hovering and helical trajectory tracking without changing orientations demonstrate the efficiency of the proposed design and controller. At last, the paper concludes with a discussion on the potential for further improvements in terms of mechanical design and control.

Keywords: Prognostics and diagnostics
Abstract: In this paper, an assessment of two methods of uncertainty quantification in prognostics is under-taken. Two methods, the Inverse First Order Reliability Method (IFORM) and set-based reachability analysis for prognostics are considered. By quantifying the uncertainties using the IFORM, an assessment of the quality of Remaining Useful Life (RUL) prediction. The IFORM approach permits the generation of confidence bounds that allows for the calculation of RUL values corresponding to the specified user-based probability levels. On the other hand, uncertainty quantification can be achieved by means of set-based reachability analysis. A Zonotopic Kalman filter (ZKF) is proposed to take into account a damage-model such that at each propagation time, with the estimated state (degradation) and its uncertainty, a propagation of zonotopic sets can be produced. Coming from two different schools of thought, the statistical and set-based theory, both schemes are explored and tested on a case study in simulation.

Keywords: Fault diagnosis, Linear systems, Power systems and smart grid
Abstract: This paper proposes a model-based fault diagnosis scheme for incipient faults in the DC voltage and AC current sensors of the grid-tied converter, considering the coupling phenomena between the two sides of the converter that might cause fault propagation. First, a linear descriptor estimator is designed, based on a linear augmented descriptor model of a nonlinear grid side converter (GSC) system, while considering the effect of sensor noise and modeling uncertainty. Second, a time varying estimator gain matrix for the design of the descriptor estimator is obtained by minimizing the sum of the variances of the estimation error. Lastly, the diagnosis of incipient sensor faults is designed using the derived analytical redundancy relations, based on the estimations provided by linear descriptor estimator. The results of this work are useful for the early diagnosis of incipient sensor faults, which can improve the safety and reliability of GSCs.

Keywords: Fault diagnosis, Modelling and simulation, Nonlinear systems
Abstract: In this paper, a new distributed adaptive observer-based fault detection and isolation (FDI) approach is developed for an intensified interconnected heat exchanger/reactor. In the distributed FDI architecture, an FDI component is designed for each subsystem in the interconnected system. For each FDI component, multiple adaptive observers, which correspond to the possible faulty parameters of the corresponding subsystem, are designed to monitor the states and derive a local diagnostic decision. Thanks to the parameter estimations provided by the adaptive observers, both local and global faults could be isolated and identified. Simulation results show that this method can quickly and accurately diagnose different types of faults occurring in the interconnected heat exchanger/reactor system.

Keywords: Fault diagnosis, Nonlinear systems, Disturbance rejection
Abstract: This paper investigates the problem of state and actuator/sensor fault (ASF) estimation for nonlinear systems described by Takagi-Sugeno (T-S) fuzzy models subject to external disturbances. A robust adaptive observer (RAO) is designed to estimate the system state, sensor faults and actuator faults conjointly. For the convergence analysis of all estimation errors, a fuzzy Lyapunov functional candidate combined by free weighting matrices have been constructed to obtain more relaxed results. The design conditions, taking into account the H_infty performance, are formulated in terms of Linear Matrix Inequalities (LMIs). Finally, a comparative study is presented to prove the superiority of the proposed method.

Keywords: Modelling and simulation, Prognostics and diagnostics
Abstract: Rolling Element Bearings (REB) are one of the key components of rotating machinery. Their correct functioning and failure have been the object of many studies and today many models are available that can reproduce their vibration response. Most of them are applied for diagnosis purposes and simulate the bearing behaviour in steady state considering fixed surface defect. Such vibration signals are useful to perform bearing diagnosis but they lack the necessary information for predictive algorithms conceived for prognosis applications. The objective of the work presented here is using an already existing dynamic model to simulate vibration signals under unsteady degradation conditions. Different degradation profiles have been proposed to simulate the evolution of local surface defects on the bearing components to form a synthetic database for future prognosis applications. The obtained signals can be very useful for data-drive prognosis algorithm training. As proof, they were used for RUL (Remaining Useful Life) estimation with a simple approach and proved to be effective.

Keywords: Neural networks, Cyber-physical systems, Fault diagnosis
Abstract: With the increasing interconnectivity of cyber-physical systems (CPSs) in various fields, such as manufacturing plants, power plants, and smart networked systems, large amounts of multivariate data are generated through sensors and actuators, also other data sources such as measurements and images. This paper focuses on the anomaly detection (AD) problem, also known as fault detection or outlier detection, depending on the type of dataset, which involves identifying anomalous values in datasets using analytical methods. However, datasets often contain missing values, which can lead to incorrect outcomes and affect the availability of anomalous samples that are fewer in amount, making incomplete datasets. Therefore, a generalized AD method is proposed for incomplete datasets, which involves two steps: data imputation (DI) to obtain complete datasets using GAN and later AD for the complete datasets. While statistical-based imputation methods are commonly used, they do not consider data distribution for datasets with anomalous samples. The capabilities of GAN-based DI are tested under different hyperparameter settings and percentages of missing values. The AD problem is then addressed using seven unsupervised anomaly detection methods on six different datasets, including a real dataset from a steel manufacturing plant in Italy. Each dataset is analyzed to determine which DI and AD method combination performs the best. The results show that GAN-imputed data provides the best DI performance, while the reweighted minimum covariance determinant (RMCD) method offers the overall best AD results combined with GAN.

Keywords: Nonlinear systems
Abstract: In this paper, we develop an energy-based dynamical system model driven by a Markov input process to present a unified framework for stochastic thermodynamics predicated on a stochastic dynamical systems formalism. Specifically, using a stochastic dissipativity, losslessness, and accumulativity theory, we develop a nonlinear stochastic port-Hamiltonian system model characterized by energy conservation and entropy nonconservation laws that are consistent with statistical thermodynamic principles. In particular, we show that the difference between the stored system energy and the supplied system energy for our stochastic thermodynamic model is a martingale with respect to the system filtration, whereas the system entropy is a submartingale with respect to the system filtration.

Keywords: Modelling and simulation, Computing and communications
Abstract: The power of modern computing equipment, from small devices such as laptops through a variety of cases up to entire data centres, makes cooling vital. Especially in large-scale systems, delivering the right cooling to the right place at the right time is crucial for both computing performance and energy efficiency. As such, modern cooling systems require a lot of controls. Given the many cases to face, designing and assessing such controls requires tools to rapidly and modularly build and manage computationally efficient simulation models, sometimes concentrating on the thermal policies aboard on a chip, sometimes on the cooling of a rack, sometimes on an entire date centre with its fluid conditioning and transport machinery, and so forth. Though technology exist to address many such cases individually, a holistic approach to embrace them all within a unified modelling methodology and workflow is still the subject of research. In this paper we distil our experience over the last years, and discuss how a solution based on joining purpose-specific chip modelling (using the 3D-ICE simulator) and Equation-Based Object-Oriented Modelling (employing the Modelica language) can help the joint design of a computing system and its cooling.

Keywords: Modelling and simulation, System identification
Abstract: The problem of simultaneous online co-estimation of the battery state of charge (SOC) and the parameters of the open circuit voltage (OCV) vs. SOC characteristic is investigated. It is shown that any co-estimation technique requires at least one known point in the function that approximates the OCV vs. SOC map. A co-estimation strategy based on the equivalent circuit model of the battery is then proposed and its well-posedness is analyzed. The technique is validated on real data coming from an automotive application.

Keywords: Energy efficient systems, Intelligent control systems, Cyber-physical systems
Abstract: Modern factories collect and process a large volume of different types of industrial process data. These data are used to develop metrics and Key Performance Indicators to monitor and improve productivity and the efficiency of a factory. This work develops an ontology-based framework that semantically describes an industrial process, and in particular it describes the elements of physical connectivity, industrial behaviour, and KPIs. Using a notion of sub-process hierarchy, a Decision Support System %based on the proposed framework explores and suggests options for reconfiguring the elements of the industrial process, to improve efficiency. A proof-of-concept use-case from the KIOS Water System Testbed is presented. The pumping station (connectivity, behaviour and energy efficiency KPIs) of the Testbed is semantically modelled, whereas the DSS suggests reconfiguration options for improving its overall energy efficiency.

Keywords: Energy efficient systems, Renewable energy and sustainability, Wireless sensor networks
Abstract: Greenhouse systems offer a promising solution for meeting the growing demand for food production, especially during off-season crops, without compromising the quality and quantity of the products. However, the significant amount of energy consumption in greenhouses, including heating, cooling, and lighting, should not be ignored, as it contributes up to half of the production cost. Given the adverse impact of fossil fuels on climate change and the wider political, social, and economic context of the agricultural sector, it is essential to prioritize the use of renewable energy sources like solar energy. Unfortunately, the need for more land remains a major issue, causing spatial and economic challenges. Thus, a promising solution is to install semi-transparent photovoltaics on the roof of greenhouse units, creating dual use of land for both food and energy production. This paper aims to present the Greenhouse Integrated Photovoltaic System (GIPV) using the above elements for complete energy autonomy, focusing on Aloe Vera production.

Keywords: Predictive control, Real-time control, Renewable energy and sustainability
Abstract: Fuel cell systems are a viable alternative for stationary and mobile applications. Advanced control algorithms are the main levers to ensure safe operation in transients and increase the applicability of fuel cell systems in research and industry. This paper focuses on the control of the fuel cell air path and the net power output for a small-scale fuel cell system. For safe operation and durability even in transients, tight bounds on stoichiometry and compressor operation must be ensured at all times. To tackle this challenge, a data-based nonlinear model predictive controller is implemented and experimentally validated on a cathode path test bench with a real-time fuel cell stack simulation. Our results show accurate tracking, safe operation, and a reduction in settling time to new power reference set points of approximately 50% compared to a reference controller.

Keywords: Autonomous systems, Cyber-physical systems, Robotics
Abstract: This paper presents a bibliometric analysis of the research literature on potential applications of digital twins in autonomous vehicles, aiming to identify its main features, the current research trends and their evolution and potential gaps for future studies. The set of publications under study is collected through the most popular scientific databases by performing targeted queries and after removing erroneous entries. Different types of analysis (trend analysis, co-occurrence analysis and citation analysis) are performed and the results obtained are presented through graphs and tables, discussed to extract useful conclusions and widened to propose future extensions and suggestions for the involved stakeholders.

Keywords: Neural networks, Image processing, Autonomous systems
Abstract: Autonomous Vehicles have long leveraged Artificial Intelligence to be capable of self-driving without the need for a human supervisor. To achieve self-driving autonomy, various sensors are installed onboard the vehicle in order to be able to perceive information from its surroundings. However, since autonomous vehicles’ capabilities rely heavily on sensor readings, various challenges arise in terms of security and privacy. Thus, it is of the essence to design methodologies able to detect anomalies caused by malicious threat actors or sensor malfunctions. This paper proposes an anomaly detection algorithm for autonomous vehicle camera sensors. By utilizing Recurrent Neural Networks in combination with Convolution operations, it is possible to obtain a sequence of images and reconstruct the next frame in real-time. By leveraging image similarity techniques such as Mean Squared Error and Structural Similarity Index, it is possible to compare the ground truth with the predicted image and draw conclusions about whether an anomaly is present. The experiments in real datasets captured from autonomous vehicles within the European-funded nIoVe project highlighted that the proposed framework is able to detect anomalies and malfunctions with high accuracy, clearly indicating the necessity of such algorithms to enhance the security of autonomous vehicles.

Keywords: Autonomous systems, Intelligent transportation systems, Modelling and simulation
Abstract: Simultaneous Localization and Mapping (SLAM), which is characterized as a core problem in autonomous vehicles, involves the estimation of the vehicle’s position and the concurrent building of the map of the environment. The use of deep learning-based feature extractors has gain increasing popularity since they possess the ability to extract reliable and repeatable features from raw sensor data. However, the performance of deep learning-based approaches varies depending on the application, environmental conditions, and the type of implemented technology. In this paper, we evaluate the performance of several deep learning-based feature extractors integrated into a SLAM system, using real and synthetic data as input, which implement common odometry problems. To our knowledge, this is the first work that benchmarks the accuracy of deep-learning based algorithms in estimating the vehicle’s trajectory in specific odometry corner cases.

Keywords: Intelligent transportation systems, Automotive control, Autonomous systems
Abstract: Connected vehicles have introduced a vast array of possibilities to improve vehicle fleet performance. Vehicle-to-Vehicle (V2V) and vehicle-to-network (V2N) interactions allow collecting and communicating data on the vehicle's environment, traffic and safety. Sharing information on the road profile can ensure a safer mobility. In such framework, the goal of the paper is to explore the already existing sensors of the DS7 Crossback suspension system in order to estimate the road profile through a structure of Luenberger and unknown inputs observers. The unmeasured variables are estimated by resolving a linear matrix inequality (LMI) satisfying a H-Infinity criterion to reject external disturbances. The unknown input observer is provided by the sensor measurements enhanced by the estimated variables obtained from the Luenberger observer. Linear matrix inequality (LMI) tool is used for design observer gain and thus estimate the unknown road profile. Simulations are performed to show that the proposed structure successfully estimates the unknown road profile even in the presence of disturbances.

Keywords: Intelligent transportation systems, Autonomous systems, Automotive control
Abstract: The topic of in-vehicle and V2X communication in autonomous vehicles consists of a variety of different communication protocols, mechanisms, and devices. The implementation and cooperation between these entities and protocols in such a complex system is a rigorous and complicated process that should not only be efficient, robust, flexible, and scalable, but also secure. The security of critical systems such as autonomous vehicles requires a deep understanding of all the individual and distinct components that compose the system. This paper presents a cybersecurity architecture having as purpose to shield the communication security in the autonomous vehicles. For this reason, several well-established cybersecurity tools (e.g. Keycloak, Cloudflare) and communication mechanisms (e.g. MQTT, Kafka) have been combined in this architecture along with a novel statistical-based Intrusion Detection System. All the aforementioned cybersecurity defense mechanisms were selected to protect the entire system pipeline and meet the requirements for Confidentiality, Integrity, and Availability regarding vehicle communication. To test the performance of the proposed architecture abnormal data have been injected to the system and the results from the experiments conducted highlighted that the proposed solution can achieve its purpose of increased cybersecurity.

Keywords: Image processing, Autonomous systems, Robotics
Abstract: In this paper, we propose an algorithm that can measure obstacle distances in real time independent of brightness. The proposed method acquires camera information from a robot equipped with a cross-line laser and a monocular camera, binarizes only the cross-line laser light data using YUV values as threshold values, and calculates the ratio of color and luminance, thereby improving the method to measure the linear distance to obstacles in real time even in bright environments. Experimental results confirm that the proposed method can accurately detect the distance to obstacles. It was also able to accurately calculate the distance to obstacles even when the measurement environment was bright (717 Lux).

Keywords: Image processing, Robotics, Neural networks
Abstract: Several factors may compromise the effectiveness of algorithms for relatively localizing specific objects in outdoor unstructured environments using robotic platforms, such as the complexity of the environment, and changes in lighting conditions. Consequently, methods that rely solely on instantaneous detection may not be reliable in such application scenarios. In this work, we propose an architecture that utilizes an RGB-D camera mounted on a mobile robot and combines a state-of-the-art detection system with a purposely designed tracking algorithm. Specifically, we employ the latest You Only Look Once (YOLO) version to detect and segment the target in the image. We extract relevant relative information of the robot with respect to the object, i.e., its position and the relative orientation, by exploiting the depth map. Finally, we design an Extended Kalman Filter to track this relative information while taking into account the robot kinematic model. We implement this architecture in the ROS middleware and validate it within a precision agriculture setting for trap monitoring in a pest detection system.

Keywords: Image processing, Unmanned systems, Neural networks
Abstract: This paper proposes a system architecture that uses deep learning image processing techniques to automatically identify forest fires in real-time using neural network models for small UAV applications. Considering the strict power and payload constraints of small UAVs, the proposed model runs on a compact, lightweight Raspberry Pi4B (RPi4B) and its performance is comparable to the state-of-the-art metrics (accuracy and real-time response) while achieving significant reduction in CPU usage and power consumption. The proposed YOLOv5 optimization approach used in this paper includes: 1) Replacing the backbone network to ShuffleNetV2, 2) Pruning the Head and Neck network following the backbone baseline, 3) Sparse training to implement the model-pruning method, 4) Fine-tuning of the pruned network to recover the detection accuracy and 5) Hardware acceleration by overclocking the RPi4B to improve the inference speed of the algorithm. Experimental results of the proposed forest fire detection system show that the proposed algorithm compared to the state-of-the-art that run on RPi single board computer, achieves 50% higher inference speed (9 FPS), reduction in CPU usage and temperature by 35% and 25% respectively and 10% reduced power consumption while the accuracy (92.5%) is only compromised by 2%. Finally, it is worth noting that the accuracy of the proposed algorithm is not affected by deviations in the bird-eye view angle.

Keywords: Neural networks, Image processing, Autonomous systems
Abstract: The application of robotic platforms for precision agriculture is gaining traction in modern research. However, the demand for a complete fruit dataset is still not satisfied. In this paper, we present fruity, a multi-modal fruit dataset with a variety of use cases such as 6D-pose estimation, fruit detection, fruit picking applications, etc. To the best of our knowledge, this dataset is the first-ever multi-modal fruit dataset tailored specifically for fruit 6D pose estimation in precision agriculture. The dataset is collected over a range of multiple sensors consisting of an RGB-D camera, thermal camera and an indoor tracking camera for ground truth poses. Fruity features RGB images, stereo depth images, thermal images, camera 6D- poses, fruit 6D-poses and relative 6D-poses between the cameras and fruits. The classes of the dataset are commonly harvested fruits which include: apples, oranges, bananas, avocados and lemons. It is also enriched with a clustered class to account for occlusion scenario. The dataset is recorded over multiple trajectories implemented with multiple platforms encompassing a robotic manipulator and an Unmanned Aerial Vehicle (UAV). The dataset alongside the documentation and utility tools is publicly available at: https://github.com/MahmoudYidi/Fruity.git.

Keywords: Predictive control, Image processing, Automotive control
Abstract: With cameras being one of the most vital sensors for perception and planning it is more intuitive to design controllers that are able to operate directly on the camera data. In this work we present two approaches for designing a model predictive controller (MPC) that is able to directly operate in the perspective coordinates and show their equivalence to the standard MPC formulation. Consequently, it eliminates the need for dedicated modules for converting the output of the planner and estimating the state of system in the 3D coordinates, thereby enabling a lean design of the system architecture. We apply this method for the task of automated lane following and lane changing, a common use case arising in autonomous driving, and demonstrate its effectiveness.

Keywords: Predictive control, Unmanned systems, Autonomous systems
Abstract: In this work, a nonlinear Model Predictive Control (NMPC) strategy is presented for stabilizing an Unmanned Aerial Vehicle (UAV) with a cable-suspended liquid collection device during water sampling from aquatic environments. Building upon our previous work, an NMPC scheme is developed which incorporates the disturbances acting on the multirotor and attains the accurate hovering of the vehicle while simultaneously state and input constraints are satisfied. Once the UAV is stabilized above the water surface, a custom electromechanical mechanism is activated to collect water samples. The performance of the proposed controller and the reliability of the sampling device are demonstrated through real-world experiments in a river with high water flow.

Keywords: Unmanned systems, Multi-agent systems, Autonomous systems
Abstract: Cooperative control of multi-UAV systems has attracted substantial research attention due to its significance in various application sectors such as emergency response, search and rescue missions, and critical infrastructure inspection. This paper proposes a distributed control algorithm to generate collision-free trajectories that drive the multi-UAV system to completely inspect a set of 3D points on the surface of an object of interest. The objective of the UAVs is to cooperatively inspect the object of interest in the minimum amount of time. Extensive numerical simulations for a team of quadrotor UAVs inspecting a real 3D structure illustrate the validity and effectiveness of the proposed approach.

Keywords: Autonomous systems, Intelligent control systems, Robotics
Abstract: We present a multisensor fusion framework for the onboard real-time navigation of a quadrotor in an indoor environment. The framework integrates sensor readings from an Inertial Measurement Unit (IMU), a camera-based object detection algorithm, and an Ultra-WideBand (UWB) localisation system. Often the sensor readings are not always readily available, leading to inaccurate pose estimation and hence poor navigation performance. To effectively handle and fuse sensor readings, and accurately estimate the pose of the quadrotor for tracking a predefined trajectory, we design a Maximum Correntropy Criterion Kalman Filter (MCC-KF) that can manage intermittent observations. The MCC-KF is designed to improve the performance of the estimation process when is done with a Kalman Filter (KF), since KFs are likely to degrade dramatically in practical scenarios in which noise is non-Gaussian (especially when the noise is heavy-tailed). To evaluate the performance of the MCC-KF, we compare it with a previously designed Kalman filter by the authors. Through this comparison, we aim to demonstrate the effectiveness of the MCC-KF in handling indoor navigation missions. The simulation results show that our presented framework offers low positioning errors, while effectively handling intermittent sensor measurements.

Keywords: Autonomous systems, Optimisation, Unmanned systems
Abstract: This work considers the problem of passively monitoring multiple moving targets with a single unmanned aerial vehicle (UAV) agent equipped with a direction-finding radar. This is in general a challenging problem due to the unobservability of the target states, and the highly non-linear measurement process. In addition to these challenges, in this work we also consider: a) environments with multiple obstacles where the targets need to be tracked as they manoeuvre through the obstacles, and b) multiple false-alarm measurements caused by the cluttered environment. To address these challenges we first design a model predictive guidance controller which is used to plan hypothetical target trajectories over a rolling finite planning horizon. We then formulate a joint estimation and control problem where the trajectory of the UAV agent is optimized to achieve optimal multi-target monitoring.

Keywords: Marine control, Sampled-data systems, Autonomous systems
Abstract: This paper proposes a low-cost solution for localizing a remotely operated vehicle (ROV) inside a fish net pen. The solution consists of a kinematic Kalman Filter capable of estimating the absolute ROV position and orientation in a fish net pen using primarily the onboard compass, laser-camera triangulation, and a model of the cylindrical net pen. The solution is demonstrated in a real fish net pen, under realistic operating conditions, and the performance is comparable to that of specialized positioning sensor systems such as ultra short baseline systems and Doppler velocity loggers.

Keywords: Autonomous systems, Unmanned systems, Nonlinear control
Abstract: This paper presents a comparison of three different control approaches for UAS flight controllers. A cascaded PID structure from Pixhawk, an energy-based controller and an incremental nonlinear dynamic inversion approach are implemented in a simulation environment based on MATLAB. The precision in position and heading angle, the required flight and calculation time and the effort to implement the algorithms are taken into account. Further, the results are evaluated for the specific UAS use cases of an infrastructure inspection and the drop-off of a sensor.

Keywords: Aerospace control, Fault tolerant control, Optimisation
Abstract: We propose the design of an optimal control problem (OCP) for generating recovery trajectories to a safe set in the state space in case of an actuator failure. The OCP is developed for recovery of hybrid quad-plane unmanned aerial vehicles (UAV) with a complete rotor loss in the hover flight regime. The UAV is recovered to the safe fixed-wing flight while minimizing the loss of altitude. For solving the optimization problem sequential convex programming is used in order to turn infeasible initial guesses into feasible trajectories with respect to the dynamics and constraints. The applicability of the proposed method is shown with numerical results where recovery from various initial attitudes is demonstrated.

Keywords: Fault diagnosis, Fault tolerant control, Integrated control and diagnostics
Abstract: This paper proposes an output-feedback fault-tolerant control approach for multiple faults. The proposed approach is able to deal with both sensors and actuator faults. Moreover, the disturbances are assumed to be bounded within an ellipsoidal sets. The proposed strategy boils down to solving a set of LMIs along with an auxiliary parameter, which determines the convergence rate of the approach. Finally, the proposed strategy is illustrated with two-rotor aerodynamical system.

Keywords: Fault tolerant control, Aerospace control
Abstract: This paper presents a fault-tolerant control scheme for a scale model of a High-Altitude Long Endurance UAV. The aircraft considered in this paper is a scale model glider that has a similar configuration to typical HALE platforms. The proposed control system was designed using sliding mode and control allocation to handle actuator faults. To evaluate the performance of the system, simulations were conducted using a nonlinear fixed-aerodynamic model. The results are promising, since the control system was able to handle multiple actuator failure cases, including a total actuator failure in the left surfaces.

Keywords: Fault tolerant control, Aerospace control, Fault diagnosis
Abstract: This paper deals with fault tolerant control for space missions. A new integral--based control allocation algorithm is developed, for optimal spacecraft actuator selection. The algorithm is developed in a general manner so that the allocation can be done under l1,l2,linfinity optimisation criteria. Stability and convergence properties of the algorithm are formally proved, using the small gain theory and the scaled bounded real lemma. The proposed solution is evaluated through intensive simulations from a functional engineering simulator that accurately simulates an in-orbit autonomous rendezvous, on a circular orbit. The obtained results demonstrate the efficiency of the proposed fault-tolerant control allocation algorithm.

Keywords: Fault tolerant control, Unmanned systems
Abstract: This paper proposes a new approach for fault tolerant control of hexarotor UAVs against motor failure subject to maximum motor speed constraint. The proposed approach consists of two parts: sliding mode control and dynamic control allocation. The sliding mode control is designed to generate the desired forces and torques that achieve and maintain flight stability and performance. As the control design is independent of plant model parameters, it remains the same across the cases of fault-free and motor failure. The dynamic control allocation is then applied to redistribute the desired forces and moments among the remaining healthy motors subject to maximum motor speed constraint. In the proposed approach, the control allocation problem is formulated in terms of solving a nonlinear optimization problem. Then the optimization problem is transformed to a stability problem where the convergence is established by using the Lyapunov stability theory. The simulation on a hexarotor UAV demonstrates the effectiveness of the proposed approach and the advantages over existing approach in terms of motor speed limits.

Keywords: Unmanned systems, Fault tolerant control, Formation control
Abstract: To overcome input saturation and loss of thrust effectiveness problems for multiple fixed-wing unmanned aerial vehicles (UAVs), a fault-tolerant cooperative control (FTCC) based on adaptive backstepping sliding mode control (BSMC) method is developed. An auxiliary dynamic system is constructed to solve the input saturation problem. Furthermore, adaptive laws are proposed to estimate the thrust effectiveness and lumped unknown term. Stability of the system and finite-time convergence of the error signals are proved by Lyapunov analysis. Finally, the effectiveness of the proposed control scheme is verified by the simulations.

Keywords: Power systems and smart grid, Nonlinear control, Networked systems
Abstract: In this paper, a novel distributed controller for multimachine power systems is proposed to guarantee grid frequency restoration and accurate real and reactive power sharing among the generator units, while maintaining the generator inputs (mechanical torque and field excitation voltage) within given bounds. The boundedness of the controller outputs (generator inputs) is rigorously proven using vector field theory. It is additionally shown that even if one generator input reaches its upper/lower limit, the remaining units can still accomplish the desired control tasks without modifying the controller structure or dynamics; hence introducing enhanced system resilience using the proposed approach. This has been accomplished for the first time in a unified control structure while using neighbour-to-neighbour communication, thus maintaining the distributed nature of the controller. An example of a 10-bus, 4-machine power system is simulated to verify the proposed controller performance under sudden changes of the load demand.

Keywords: Power systems and smart grid, Nonlinear systems, Distributed systems
Abstract: This paper proposes a robust control scheme for isolated AC Microgrids, where each node is connected locally to a constant power load (CPL). Contrary to many approaches in the literature, we consider the explicit model of the inverter dynamics and separate the overall system into two parts; a nominal subsystem parametrized by a nominal load and an error subsystem describing the difference between the true and the nominal voltage, resulting from perturbations of the load demand. In the presented analysis, we investigate the non-linear structure of the CPL in order to analytically describe its geometric effect on the network dynamics. We exploit this information to propose mild conditions on the tuning parameters such that a positive invariant set for the error dynamics exists and the distance between the true and the nominal voltage trajectories is bounded at all times. We demonstrate the properties of the proposed control scheme in a simulated scenario.

Keywords: Nonlinear control, Mechatronic systems, Adaptive control
Abstract: In this paper, we consider the problem of flux, position and speed observer design for permanent magnet synchronous motors (PMSMs) with uncertain parameters. It is assumed that the only measured signals are stator currents and control voltages. The key feature of the proposed approach is that it requires the knowledge of only one structural parameter of PMSM model - the number of pole pairs. Thus, all electrical and mechanical parameters, namely, the stator resistance and inductance, constant flux from permanent magnets, motor inertia and viscous friction coefficient are assumed to be unknown. A new nonlinear parameterization of motor model is proposed that is resulted in the regression model of eleven unknown parameters including the stator resistance and inductance as well as two parameters involved in the state observer design. The dynamic regressor extension and mixing (DREM) estimator is used to provide good performance and fast estimation of unknown parameters which is more efficient than the standard gradient approach in the case of high-dimensional regression models. Simulation results carried out for a typical scenario of motor operation illustrate good performance of the designed observer and parameter estimators.

Keywords: Renewable energy and sustainability, Power systems and smart grid, Nonlinear control
Abstract: The most common techniques employed for the control of doubly-fed induction generator (DFIG) wind turbine systems are restricted to either the well-known field-orientation control (FOC) or the direct-power control (DPC), with each one of them, however, suffering in one way or another from distinctive drawbacks. Instead of these standard methods, in this paper, a novel and nonlinear model-based control approach is adopted, which is developed in view of the entire system structure and characteristics. The key novelties introduced by the proposed design are due to an innovative technique, defined as 3s-FOC, which is formulated to enable the implementation of a simple cascade-mode PI-based control scheme that i) achieves stator field orientation without the need for estimating the actual flux, ii) guarantees system stability while simultaneously provides a relaxation on the transient response, iii) improves the closed-loop system dynamic behavior by employing extra damping terms in the inner-loop current regulators. The stability and state convergence properties of the complete system is firmly ensured as it is verified by a rigorous analysis based on advanced Lyapunov-based methods and input-to-state stability (ISS) techniques. Finally, a thorough simulation is conducted, which firmly verifies the theoretical results and the superior controlled system dynamic performance.

Keywords: Distributed systems, Networked systems, Switching systems
Abstract: Latest standards for DC fast charging (DCFC) stations enable up to 900 kW of charging power per Electric Vehicle (EV). Unfortunately, for a limited-capacity isolated Microgrid (MG), this high amount of power may lead to MG instability. In this research, a fully distributed event-triggered Dynamic Consensus (DNC) technique is proposed for controlling DCFC fast charging process. The proposed control determines a proper charging rate considering the estimated real-time sparse capacity of the MG. Meanwhile, a safety factor is used to account for the system losses, demand growth, and sparse capacity estimation error. Moreover, the admissible Communication (COM) delay effect is evaluated. The control scheme is validated using a MATLAB/Simulink model. Different case studies with different conditions were simulated. The simulated cases show success of the proposed control strategy to have successful fast vehicle charging under different operating conditions without loss of stability.

Keywords: Predictive control, Adaptive control, Intelligent control systems
Abstract: We propose a novel Model Predictive Control (MPC) scheme based on online-learning (OL) for microgrid energy management, where the control optimisation is embedded as the last layer of the neural network. The proposed MPC scheme deals with uncertainty on the load and renewable generation power profiles and on electricity prices, by employing the predictions provided by an online trained neural network in the optimisation problem. In order to adapt to possible changes in the environment the neural network is online trained based on continuously received data. The network hyperparameters are selected by performing a hyperparameter optimisation before the deployment of the controller, using a pretraining dataset. We show the effectiveness of the proposed method for microgrid energy management through extensive experiments on real microgrid datasets. Moreover, we show that the proposed algorithm has good transfer learning (TL) capabilities among different microgrids.

Keywords: Assistive technology, Robotics, Autonomous systems
Abstract: This article presents MSL3D, a novel framework for pointcloud-based muck pile Segmentation and Localization in unknown Sub-Terranean (Sub-T) environments. The proposed framework is capable of progressively segmenting the muck piles and extracting their location in a global constructed point cloud map. MSL3D is structured in a two layer novel architecture that relies on the geometric properties of muck piles in underground tunnels, where the first layer extracts a Volume Of Interest (VOI) proposal area out of the registered point cloud and the second layer is refining the muck pile extraction of each VOI proposal in the global optimized point cloud map. This action is performed by using a first instance of VOI that is then refined by utilizing a progressive RANSAC in order to extract the ceilings, walls, and ground of the scene. Once the refined VOI is extracted, it is transmitted to the second layer, where it is converted to the world frame coordinates. In the sequel, a progressive morphological filter is applied, in order to segment ground and nonground points, followed by RANSAC once again to extract the remaining points corresponding to the right and left walls. In this approach, euclidean clustering is utilized to keep the cluster with the majority of points, which is assumed to belong to the muck pile. The efficacy of the proposed novel scheme was successfully and experimentally validated in real and large scale SubT environments by utilizing a custom-made UAV.

Keywords: Formation control, Multi-agent systems, Swarms
Abstract: This paper proposes a swarm-based approach for coordinating a multi-agent system (MAS) in a 3D environment to encircle a target for monitoring tasks in precision agriculture. Specifically, we are motivated by the objective of encircling large tree canopies in order to collaboratively gather information on tree health status. This goal is achieved by enhancing classical potential-based swarm design with a novel topology switching policy allowing the desired encirclement behavior to emerge. The resulting interaction protocol requires agents to utilize only local information, ensuring collision-free trajectories without restrictive assumptions on the undirected time-varying graph encoding the network topology. Numerical results are presented to demonstrate the effectiveness of the proposed approach.

Keywords: Networked systems, Modelling and simulation, Prognostics and diagnostics
Abstract: Along with the evolution of communication technologies, cybersecurity has evolved, and so have its new directions and demands. There is a wide range of tools to detect, analyse, or protect systems from malicious activity. Yet, as new technologies are emerging and maturing, the need for particular domain solutions arises. This paper proposes a methodology for a honeypot network organisation mimicking vital autonomous vehicle sensors inside the Internet of Vehicles (IoV) infrastructure, along with attack propagation patterns analysis based on the logs collected from the honeypots. The discovery of sequential patterns is based on Markov Chain models applied in the honeyfarm data. Further, these trained models are applied with graph-based algorithms to discover the interaction patterns between honeypots targeting the discovery of segments that were attacked in series. The intelligence produced from the analysis is used to rank and estimate the relative importance of the honeypots in their framework. The results of our study allowed us to identify common attacks on the IoV system, detect the geolocation of each attacker, and specify the usage of each honeypot node from the attacker’s perspective.

Keywords: Nonlinear systems, Modelling and simulation
Abstract: In this paper, we study the performance of a class of cruise-controllers for automated vehicles on a macroscopic scale. We first show that the solution of the corresponding second-order macroscopic model can be approximated by the solution of a nonlinear heat-type equation and introduce an appropriate notion of a weak solution that requires certain entropy-like conditions. To study the behavior induced on the macroscopic model by the first-order approximation, we derive a conservative finite-difference scheme that respects the corresponding entropy conditions. Certain links between the weak solution of the nonlinear heat equation and the solution produced by the proposed numerical scheme are also provided. Finally, a traffic simulation scenario and a comparison with the Lighthill-Witham-Richards (LWR) model are given, illustrating the benefits of the use of cruise-controlled vehicles.

Keywords: Nonlinear systems, System identification
Abstract: Modern transportation research has given a lot of attention to autonomous vehicles. For the control of these vehicles, the longitudinal states of vehicle dynamics play a critical role. The objective of the article is to develop an LMI-based nonlinear observer which estimates the states of vehicles under the presence of disturbance. To achieve this, the observer proposed in~cite{Mohite_LMI} is extended by using the mathcal{H}_infty criterion, and a new LMI condition is derived. Further, the proposed observer is implemented and validated on a third-order ``Position-Velocity-Acceleration" nonlinear autonomous vehicle model.

Keywords: Predictive control, Nonlinear control, Modelling and simulation
Abstract: Skid-steered mobile robots are widely used in several applications due to their simple mechanical structure and their high maneuverability. This paper proposes a hierarchical control system for such robots that, at a high level, uses Nonlinear Model Predictive Control (NMPC) with a simplified prediction model, optimizing the longitudinal forces between the wheels and the ground to follow the desired trajectory. Using Pacekja's formula, the reference slip for each wheel is obtained by interpolation, allowing the computation of the reference angular speeds of the wheels. Then, a proportional control is employed to find the required wheel torques, which are applied to a complete model of the skid-steer robot that takes into account the longitudinal slippage on the wheels. The proposed hierarchical controller is compared with a purely NMPC approach using the full model of the robot and a classical proportional control. Double-lane change and a circular trajectory tracking was performed. The resulting torque variation is compatible with values obtained in physical systems and the wheel skidding remained within the allowed limit.

Keywords: Image processing, Neural networks
Abstract: Deep learning (DL)-based solutions have been extensively researched in the medical domain in recent years, enhancing the efficacy of diagnosis, planning, and treatment. Since the usage of health-related data is strictly regulated, processing medical records outside the hospital environment for developing and using DL models demands robust data protection measures. At the same time, it can be challenging to guarantee that a DL solution delivers a minimum level of performance when being trained on secured data, without being specifically designed for the given task. Our approach uses singular value decomposition (SVD) and principal component analysis (PCA) to obfuscate the medical images before employing them in the DL analysis. The capability of DL algorithms to extract relevant information from secured data is assessed on a task of angiographic view classification based on obfuscated frames. The security level is probed by simulated artificial intelligence (AI)-based reconstruction attacks, considering two threat actors with different prior knowledge of the targeted data. The degree of privacy is quantitatively measured using similarity indices. Although a trade-off between privacy and accuracy should be considered, the proposed technique allows for training the angiographic view classifier exclusively on secured data with satisfactory performance and with no computational overhead, model adaptation, or hyperparameter tuning. While the obfuscated medical image content is well protected against human perception, the hypothetical reconstruction attack proved that it is also difficult to recover the complete information of the original frames.

Keywords: Image processing, Neural networks, Unmanned systems
Abstract: Human detection in critical missions with unmanned aerial vehicle (UAV) support gains nowadays more and more important in the actual context when tension at borders builds up for an increasing number of countries. Although convolutional neural networks are continuously evolving, the required computational resources pose a great problem when implemented on portable embedded systems such as UAVs, with limited processing power and autonomy. This demand becomes even more drastic when running real-time human detection. The paper proposes an improved implementation of the YOLO v7 network, trained on a custom dataset, for real-time human detection and tracking with confidence scores above 80% on the NVIDIA Jetson TX2 neural processing unit equipped on DJI Matrice 100 UAV. The authors created a YOLO v7 model running independently on an embedded system for real-time human detection and tracking.

Keywords: Neural networks, Image processing
Abstract: Pest detection and identification in a timely manner is a crucial step for precision agriculture. Halyomorpha Halys is a common pest whose negative effects are known in agricultural areas and on various crops. The present work implemented and studied four performant neural networks, VGG19_BN, EfficientNetB7, DenseNet161, and ResNet152 for the detection of these insects. Although the detection of these insects in the natural environment through automated means, excluding traps, is a challenge, the results obtained are promising.

Keywords: Neural networks, Image processing, Computational intelligence
Abstract: Since drone technology has recently advanced, human detection and tracking techniques have increased, and these technologies have a variety of uses, particularly close to borders. In this research, we examined methods to enhance people detection performance in diverse outdoor scenarios. A broad variety of light and color changes, as well as different target distances, angles, and postures, were all considered in the dataset's design. The experimental data were based on images taken under various environmental situations, like changing the drone's flight height, capturing pictures in low light, and so on. To calculate the metrics and useful performance indicators, this study proposes an enhanced version of the generic YOLOv5 model. This is accomplished by applying the data gathered to each model, including the baseline YOLOv5 model and the enhanced custom model. The main metrics are the improved YOLOv5 model loss functions, recall, accuracy, and mAP50. An evaluation was reached after contrasting the outcomes of the standard YOLOv5 model and the modified YOLOv5 model against the same testing set.

Keywords: Nonlinear control, Complex systems, Intelligent control systems
Abstract: Data-driven control (DDC) algorithms have been developed in the last decades, whose design is based only on the data collected from the controlled plant, without using a process model. These techniques that do not use an explicit model of the system have become very attractive for the control of complex processes with high nonlinearities. This paper presents two DDC algorithms, one model-free adaptive control (MFAC), and the other model-free intelligent P (iP), whose performances are experimentally evaluated using the AERO 2 platform, a highly nonlinear aerospace system made by Quanser. The similarities and differences between the two DDC are succinctly presented and based on the results obtained through real-time experiments, the performances are compared.

Keywords: Switching systems, Modelling and simulation, Networked systems
Abstract: Multiprotocol Label Switching (MPLS)combines multiple advanced systems for embedding techniques like ATM and IP with Quality of Service (QoS). This paper investigates and provides solutions for implementing a local Label Switched Path (LSP) recovery mechanism. This mechanism allows for an alternative path to be set up when an accidental drop in the topology affects an LSP that carries a privileged data flow. The new path has equivalent properties to its predecessor, thus privileged traffic can be redirected. The analysis was performed on the traffic of medical image data.

Keywords: Aerospace control, Navigation
Abstract: Nowadays, along with the trend of developing highly autonomous spacecrafts, there is a strong motivation to improve real-time Precise Orbit Determination (POD), in particular for Low Earth Orbit (LEO) satellites. The development of Global Navigation Satellite System (GNSS) sensors allows to obtain low-noise measurements and provide a spacecraft with autonomous continuous tracking onboard. Following the deactivation of Selective Availability, a representative real-time positioning accuracy of 10 m is presently achieved by means of Global Positioning System (GPS) receivers on LEO satellites. The introduction of dynamical filtering methods has opened a new way to improve this accuracy by making use of measurements such as pseudorange or carrier-phase. This paper presents a Kalman filtering approach using pseudorange and pseudorange-rate measurements instead of pseudorange and carrier-phase ones, with advantages in terms of storage and processing requirements. An error of around 0.2 m and 1e-3 m/s for position and velocity is obtained, which is in line if not better w.r.t. other approaches.

Keywords: Autonomous systems, Navigation, Automotive control
Abstract: In this study we propose a model-based dynamic path planning algorithm that is designed to navigate Autonomous Vehicles through complex and dynamic environments. To achieve that, a novel spline-based approach is utilized for the production of several candidate paths along a predetermined route and a Gaussian-based function is utilized for their evaluation. Our algorithm takes into account various factors, such as static and dynamic objects, to make the appropriate decisions for the vehicle's path, making it a promising solution for such objects during an autonomous vehicle navigation. The algorithm was tested in high-fidelity scenarios using CARLA Simulator, which is a powerful tool for simulating autonomous vehicle scenarios. The results indicate that the proposed algorithm is capable of generating efficient and safe paths for the vehicle to follow.

Keywords: Autonomous systems, Navigation, Robotics
Abstract: This paper proposes methods to enable autonomous operation, specifically for localization and motion planning, of net grooming robots in aquaculture net pens and validates the proposed methods in both simulations and experimental fieldwork. Moreover, this paper suggests enabling uninterrupted operation by investigating the use of data from an inertial measurements unit that is a common sensor in underwater vehicles, rather than investing and upgrading to costly sensory systems that often require additional installation and calibration. In particular, the presented work consists of a localization method capable of estimating a robotic system's cylindrical position in an aquaculture net pen, a 3~DOF cylindrical robotic model, a method for path planning and collision avoidance, and a heading guidance and control system. The simulations demonstrate successful localization of the robotic system, while simultaneously planning and following collision-free trajectories in an environment obstructed by obstacles. Furthermore, the field trials successfully demonstrate that the system, when applied to net crawling robots, is capable of localization, path planning, and collision avoidance in an aquaculture setting. As follows, the presented work contributes to establishing net grooming robots as competitive candidates for biofouling management.

Keywords: Navigation, Formation control, Robotics
Abstract: In this article, a novel combined aerial cooperative tethered carrying and path planning framework is introduced with a special focus on applications in confined environments. The proposed work is aiming towards solving the path planning problem for the formation of two quadrotors, while having a rope hanging below them and passing through or around obstacles. A novel composition mechanism is proposed, which simplifies the degrees of freedom of the combined aerial system and expresses the corresponding states in a compact form. Given the state of the composition, a dynamic body is generated that encapsulates the quadrotors-rope system and makes the procedure of collision checking between the system and the environment more efficient. By utilizing the above two abstractions, an RRT path planning scheme is implemented and a collision-free path for the formation is generated. This path is decomposed back to the quadrotors’ desired positions that are fed to the Model Predictive Controller (MPC) for each one. The efficiency of the proposed framework is experimentally evaluated.

Keywords: Neural networks, Autonomous systems, Navigation
Abstract: LiDAR point-cloud segmentation is a crucial issue for autonomous cars applications. The standard method for segmenting large-scale point clouds is to project 3D point cloud onto a 2D LiDAR image and apply convolutions to it. In this paper, we also follow this method and we want to detect and classify occurrences of road-objects, namely cars, cyclists, and pedestrians. To achieve this goal, we adapted the SqueezeSeg deep neural network.To address the challenge of obtaining labeled data for training autonomous driving systems, we used the CARLA autonomous driving simulator to generate a synthetic dataset in a simulation environment. The proposed network is initially trained on real-world LiDAR point-cloud data acquired from the KITTI dataset. Then, we created a synthetic dataset using the CARLA autonomous driving simulator in order to obtain more data and determine its impact on the validation accuracy of real-world data. To compare our current work to earlier work, we employ the same method. Our synthetic dataset has additional classes, such as cyclists and pedestrians, and when combined with real-world data, it significantly improves validation accuracy for each class, surpassing previous work. This demonstrates the effectiveness of our approach in detecting and classifying road-objects using LiDAR point-clouds, which is essential for the safe operation of autonomous vehicles. Index Terms— Semantic Segmentation, LiDAR Point Cloud Segmentation, Spherical Projection, CARLA Simulator, CNN, Conditional Random Field

Keywords: Robotics, Optimisation, Navigation
Abstract: Control Barrier Functions (CBFs) when paired with Quadratic Programming offer an efficient way to generate safety-critical controllers. In this paper, we utilize CBFs for guiding multiple robots to their goals while avoiding collisions with the environment and among themselves. However, in more complex scenarios, with many robots and non-convex obstacles, these approaches often fail to guide the robots towards their desired goals because there can be other stable and undesirable equilibrium points in the system other than the desired one (reaching the goal). The proposed approach in this paper mitigates this issue by including constraints in the formulation that force the robots to circulate the boundary of the obstacles as well as each other when in close proximity. This ensures that the system does not get stuck in an undesirable equilibrium. Simulation studies show the efficacy of the proposed approach for a multi-agent problem.

Keywords: Cyber-physical systems, Decentralized control, Distributed systems
Abstract: Large systems are typically partitioned in many subsystems to reduce computational load. For this reason, the Interlaced Extended Kalman Filter (IEKF) was created, in which each subsystem estimates only its own state while utilizing information from other subsystems. The information shared is normally the a-priori and a-posteriori state, as well as the a-priori and a-posteriori covariance matrix.

Subsystems, however, cannot, for technological reasons, always operate at the same rate. To address this issue, we propose a multirate distributed filter, in which the subsystems operate independently and only share information when a novel measurement activates each subsubsystem. The only information exchanged is the a-posteriori state and covariance matrix. In the paper, we demonstrate that the proposed filtering technique is accurate and effective by examining the convergence property. A water tank case study is detailed, and two subsubsystem with different but fixed rates are discussed, illustrating the efficiency of the proposed solution. The same approach can be modified to take into account numerous instances of subsubsystem as well as missing data due to an unreliable communication route.

Keywords: Cyber-physical systems, Switching systems, Disturbance rejection
Abstract: In this paper, first, a robust covert attack is designed for switched cyber-physical systems with synchronous switching from the attacker-viewpoint in which the attacker makes the system follow their specified reference signal while it remains stealthy in the monitoring system. This attack is defined in the form of H∞ control problem such that the objectives of the attacker will be achieved. Next, as a defender, a novel detection method will be presented that can detect covert attacks in the switched system. In the proposed method, we do not need any secure channel and even in the case that the attacker can find the model of the auxiliary system and injects another signal on the corresponding communicated information, the cyber-attack can be detected. The only protected information in the proposed method is the considered delay in the mode information of the auxiliary system that needs to be exactly estimated by the attacker to have a completely stealthy attack. Simulation results demonstrate and illustrate the significant performance and capabilities of the proposed method.

Keywords: Process control, Event based systems, Optimisation
Abstract: A good cyber risk assessment is nowadays a matter of paramount importance for industrial systems and critical infrastructures. In a radical change and continuous development scenario such as that represented by Industry 4.0 plants, it is no longer sufficient to consider only static risks relating to the analysis of past data, but there is a need for a risk assessment that takes into account risks arising from emergent threats. In this paper, we propose a novel methodology for dynamic risk assessment that takes into account both the known values related to the static components of the system and the risks related to the emergence of new threats that have not yet been verified but are plausible according to experts. To achieve this, as part of the national "DRIVERS" project, an analysis of the most significant cyber-security factors was conducted to classify them in terms of relevance, considering both risk acceleration and risk mitigation aspects. This assessment is carried out by means of the multi-criteria decision support technique Analytic Hierarchy Process (AHP), performed by dividing the threat into a hierarchical structure.

Keywords: Multi-agent systems, Fault tolerant control, Event based systems
Abstract: The consensus control problem for the multi-agent system (MAS) against periodic Denial-of-Service (DoS) attacks is focused on in this paper. The characteristic of periodic DoS attacks is that the attack duration is fixed, and the attack start time of each attack is periodic. Considering arbitrary periodic DoS attacks under a communication topology, a resilient controller with a switching mechanism is proposed by using the present and delayed neighbor information. Besides, an event-triggered mechanism (ETM) is designed to adjust the controller triggering frequency according to the periodic DoS attacks. The consensus and convergency of MAS against periodic DoS attacks are theoretically analyzed via Lyapunov stability criteria. Numerical simulations are provided to demonstrate the effectiveness of the proposed method.

Keywords: Signal processing, Optimisation, Intelligent transportation systems
Abstract: The new era of Industry 4.0 and its key-enabling Internet of Things technologies promises fundamental advances during data collection, processing and analysis from a variety of agents and sensors, for the collective benefit of society. In this regard, connected and autonomous vehicles equipped with integrated perception sensors and communication abilities formulate a cluster or swarm of intelligent nodes capable to transform the transportation sector into a new smart mobility system. However, its feasible operation may be potentially threatened by hijackers whose goal is to cause malfunctioning to critical vehicular sensors, harnessing the perception system of vehicle. Therefore, in this paper we discuss the impact of cyberattacks such as GPS spoofing on autonomous vehicles, and design efficient detection and mitigation centralized schemes which provide location awareness and security monitoring over the whole cluster of vehicles. More specifically, we exploit the cooperation among the interacting vehicles, and develop robust sparse coding solutions based on graph signal processing and Alternating Direction Method of Multipliers. Cooperative based approach is further benefited by a in-vehicle module which provides spoofing detection alerts at the level of individual vehicle. Experimental analysis using the renowned CARLA simulator indicates highly efficient mitigation performance for different rates of compromised vehicles, as well as spoofing detection metrics greater than 94%.

Keywords: Wireless sensor networks, Intelligent control systems, Real-time control
Abstract: The smart grid era relies on the large deployment of advanced monitoring, automation and communication infrastructures. Features such as flexibility and scalability of digital services to sustain smart grid operation functions are core requirements for smart grids architectures. This work elaborates on the use of 5th generation wireless communication as key technology for enabling secure, scalable and flexible digital services for electric power and energy stakeholders. Specifically, this work introduces the Network Application concept defined by the Smart5Grid Project for rolling out such type of smart grid digital services. A concrete example in terms of development, implementation and testing of Wide Area Monitoring services as a 5G-enabled Network Application in a Hardware-In-the Loop testbed is also provided. The results of this integration and testing process demonstrate the viability of the proposed solution.

Keywords: Automotive control, Computational intelligence, Neural networks
Abstract: Autonomous vehicles (AVs) will be part of everyday life in the near future. In order to accelerate this process, many subsystems need to be optimised and validated. One of the most important subsystem of AVs is the steering controller. It’s task is to keep the vehicle on track, which is the reason, why many steering controllers have been designed for a large variety of applications. However, the validation of such controllers is a labour-intensive task, which is why in this paper, an Artificial Intelligence (AI) is trained to find an edge case path that brings the steering controller to its limits. This path is a sufficient substitute for a large set of paths and enables fast validation of steering controllers. This contribution describes the development of a reinforcement learning (RL) based path planner using the PPO-Algorithm to train a so called agent. Comparing the resulting key feature maps shows that the agent adapts to each controllers characteristics during the learning process. The result is demonstrated for three different state of the art path tracking controllers. For each controller the agent finds a path that leads to the controllers failure within seconds.

Keywords: Automotive control, Energy efficient systems, Nonlinear control
Abstract: In order to enhance the energy economy of a four independent wheel motor drive electric vehicle (4-IWMDEV), this paper proposes an optimal based energy-saving torque distribution. The proposed algorithm can adapt to different driving conditions while ensuring vehicle stability control. The controller consists of a hierarchical structure, a reference model which generates the desired vehicle dynamics parameters, and an upper-level control that computes the integrated traction force and yaw moment. The lower-level control employs a multi- objective optimization that considers energy efficiency and steer- ing stability to calculate the optimal torque distribution for each motor. The yaw moment control of 4-IWMDEV, integrated in the latest version of Carsim, with the classical tire workload control, were chosen to compare and evaluate the proposed controller. It has been shown from simulation studies, that vehicle steering stability and energy efficiency can be effectively improved.

Keywords: Automotive control, Integrated control and diagnostics, Modelling and simulation
Abstract: This paper is concerned with the VGT (Variable Geometry Turbocharger) and EGR (Exhaust Gas Recirculation) control design for the air path of a Diesel engine. The purpose is to validate a Diesel engine model by choosing the right operating conditions based on a series of open-loop tests. For the control design, an observer-based control with integrator is developed for the estimation of the compressor power and a good tracking of reference pressure signals which correspond to emission standards. The efficiency of this method is illustrated through simulation results on the Amesim software.

Keywords: Automotive control, Linear systems, Optimisation
Abstract: A functional interval estimation method is proposed for Linear Parameter-Varying (LPV) multivariable systems with unknown but bounded disturbances and measurement noises. Based on the interval analysis, the proposed estimator provides the upper and lower bounds of the linear functional state. An L1 formalism is used to improve stability and estimation accuracy. The design conditions are given in terms of Linear Matrix Inequalities (LMIs). Finally, a numerical example is applied to check the effectiveness of the proposed methodology.

Keywords: Automotive control, Neural networks, Nonlinear systems
Abstract: Tightened diesel pollutants emissions regulations rendered the performance of steady-state map controls, which are commonly used in Internal Combustion Engine (ICE) management, unsatisfactory. To overcome these performance constraints, control systems must deal with engine transient operation, subsystem coupling and the trade-off between different requirements to efficiently manage the engine. The research demonstrates the utility of a reference generator for coordinating the air path and combustion control systems of a turbocharged diesel engine for heavy-duty applications. The control system coordinator is based on neural networks and allows following different engine-out Nitrogen Oxides (NOx) targets while satisfying the load request. The main idea is to generate air path targets, intake O2 concentration and Intake MAnifold Pressure (IMAP), in accordance with the ones of the combustion control system, engine load, in the form of Brake Mean Effective Pressure (BMEP), and NOx. As a result, the air path control system provides the global conditions for the engine proper operation, while the combustion control responds to rapid changes in the engine operating state and compensates for any remaining deviations from load and NOx targets. The reference generator, as well as the two controllers, are suitable for real-time implementation on rapid-prototyping hardware. The performance was overall good, achieving average deviations of 0.1 bar for the BMEP and 150 ppm for the NOx.

Keywords: Nonlinear systems, Automotive control, Complex systems
Abstract: This paper proposes an LMI-based nonlinear unknown input observer to estimate the rollover index of the autonomous vehicle, whether the rollover is tripped or untripped. To tackle the rollover prevention-related model, some transformations are introduced with the goal to satisfy specific necessary rank conditions and to simplify the structure of the nonlinear model. In addition, to propose a more general method, the mathcal{L}_2-optimality criterion is considered in order to get an estimation with mathcal{H}_infty performance with respect to system disturbances and measurement noises. After presenting the generic numerical design procedure, simulations using Matlab/Simulink and CarSim are provided. The results obtained from CarSim simulations show that the developed nonlinear observer reliably estimates the vehicle states, the unknown normal tire forces, and the rollover index to predict tripped and untripped rollovers.

Keywords: Adaptive control, Distributed systems, Intelligent transportation systems
Abstract: In this paper, the effectiveness of network-wide parallel application of PC and MP strategies embedded in a two-layer control framework is assessed in a link-level dynamic traffic simulation environment. With the aim of reducing MP implementation cost without significant performance loss, partial MP deployment to node subsets is examined, based on a node importance assessment method that is proposed. Different configurations of the two-layer framework for a real large-scale network are tested in moderate and highly congested conditions. Results show that: (i) combined control of MP and PC outperforms individual MP/PC application in both demand scenarios tested; (ii) MP control in critical node sets leads to similar or even better performance compared to full-network implementation, thus allowing for significant cost reduction.

Keywords: Intelligent transportation systems, Game theory, Decentralized control
Abstract: A popular remedy for the morning commute bottleneck congestion is to split the highway capacity into a managed lane that is kept in free-flow and a general purpose lane that is subject to congestion. A classical theoretical result is that the more capacity is allocated to the managed lane the less the resulting congestion. However, existing approaches to restrict access to the managed lane are primarily monetary, e.g., tolls, which severely limits the public willingness to accept them due to equity concerns. Following up on recent work which introduces karma as a completely non-monetary credit used to control access to a so-called Karma Priority (KP) lane, we first review the strategic problem of the commuters which is modeled as a dynamic population game. We then numerically investigate the effect of varying the KP lane capacity. The karma scheme is equitable with respect to different income classes irrespective of the capacity split, meanwhile achieving near-optimal traffic reduction. Thus, managing a larger fraction of the bottleneck could be more socially feasible under a karma scheme than a monetary scheme.

Keywords: Intelligent transportation systems, Optimisation
Abstract: This paper presents a Quadratic Programming (QP) problem formulation, employing a modified multi-lane version of the well-known macroscopic Cell Transmission Model (CTM), to determine integrated optimal control actions for motorway networks. These include mainstream traffic flow control (MTFC), lane change control (LCC), ramp metering control (RM) and dynamic traffic assignment (DTA) actions to be applied by Connected and Automated Vehicles (CAVs). An algorithm based on the barrier method provides a fast solution of the convex QP problem. A case study, for a hypothetical motorway network with multiple destinations and routes, demonstrates the efficiency of the open-loop solutions.

Keywords: Intelligent transportation systems, Optimisation, Intelligent control systems
Abstract: This paper deals with electric automated buses that have to follow a given route in inter-urban roads including stops, with a given timetable. Some stops are provided with a charging infrastructure allowing to charge the batteries while others are not. In order to control these buses, it is necessary to account for the traffic conditions along the road and to minimize two objectives, respectively related to the minimization of the deviations from the timetable and the minimization of the energy lack, at the end of the bus route, with respect to a desired final energy level. To address this problem and to investigate the conflicting nature of these objectives, two multi-objective methods based on the epsilon-constraint approach are applied in this paper, allowing to find different sets of efficient solutions for the problem. The results obtained in a real case study show that the two objective are in conflict, and compromise solutions can be found using the methods proposed in this paper.

Keywords: Modelling and simulation, System identification
Abstract: This paper studies the capacity drops phenomena on a macroscopic, first-order model for freeway traffic. In particular, we focus on the effect that a service station (ST) has on the mainstream traffic evolution. We propose a modified formulation of the Cell Transmission Model with service station (CTM-s) that considers this important phenomenon, and use a micro-simulation based on Aimsun Next to identify the model parameters via a structured identification procedure. Finally, we validate the ability of the CTM-s with capacity drops to better describe the traffic evolution with respect to the classical formulation.

Keywords: Intelligent transportation systems, Modelling and simulation, Optimisation
Abstract: This work proposes a novel formulation for the On-Time Arrival (OTA) problem based on macroscopic traffic dynamics. The OTA problem is formulated as a multi-objective optimization problem considering two objective criteria. The first criterion aims to avoid the emergence of congestion by minimizing the travel time of all vehicles in the network. The second criterion seeks to reduce the discrepancy between the actual and desired arrival time. The proposed formulation results in a nonlinear multi-objective program solved efficiently through an approximated convex solution. Finally, simulation results show that the proposed methodology can significantly reduce congestion while ensuring that most vehicles arrive at their desired time.

Keywords: Education and training, Neural networks, Adaptive control
Abstract: Behavioural Cloning is a Machine Learning method concerning how a machine attempts to autonomously mimic the actions of a human, or in general a complex controller, performing a given task. This work innovatively exploits Behavioural Cloning in support of Pediatric Neurorehabilitation. In particular, an Artificial Neural Network Classifier has been implemented to autonomously adapt the difficulty, through a set of tunable parameters, of a Serious Game that was specifically developed to stimulate some relevant cognitive capabilities of the patient. Data augmentation via Behavioural Cloning allows such autonomous difficulty adaptation system to improve its classification performances and, thus, to enforce a control logic that, in turn, improves the effectiveness of the cognitive training. The system is validated through an experimental assessment on a Serious Game that trains motor coordination: experimental results of children gameplay are analyzed and discussed.

Keywords: Intelligent control systems, Aerospace control, Neural networks
Abstract: Among the most common issues to be faced in optical satellite communications, weather conditions play a fundamental role for a correct transmission of the information. In case of heavy rain, hailstorm or snow, but even dense clouds, a communication channel between the satellite and a optical ground station (OGS) may suffer significant interference, causing errors in delivering information. Since satellite transmissions cover in general very spread areas, it usually happens that different zones are characterized by different weather conditions. This property is exploited by the site diversity technique, that tries to limit bad weather effects on the overall availability of the communication channel. When implementing such a site diversity technique, the satellite should be able to switch between the OGSs, by evaluating the rain events probability either through direct measurement campaigns or exploiting statistical models. The setup studied in this work is composed by a geostationary satellite equipped with two laser communication terminals (LCTs). In order to dynamically decide the OGSs to be pointed by those LCTs, a Deep-Learning based proactive control algorithm for site diversity, performing weather forecasting and consequent preventive LCT switching on the basis of current and past weather conditions has been developed. Simulative results show the ability of our proposed algorithm in achieving the maximum possible link availability, which is bounded by the weather conditions of all the OGSs.

Keywords: Intelligent control systems, Distributed systems, Neural networks
Abstract: Having access to a reliable and accurate prediction of the short-term power demand is a fundamental step for the widespread adoption of Electric Vehicles (EVs), as their charges may have a significant impact on the power system balancing. In this direction, we propose a short-term load demand predictor, based on distributed Long Short-Term Memory Networks, that employs consensus and fully-decentralized Federated Learning (FL) algorithms to seek cooperation among multiple points of charge without the requirement of sharing any user-related data.

Keywords: Neural networks, Biomedical engineering, Image processing
Abstract: Deep image inpainting is a computer vision task that uses Deep Neural Networks to generate plausible content to complete an image, for example for the restoration of a damaged image or the removal of unwanted elements captured in the picture. This paper uses deep image inpainting to restore endoscopic images that are affected by various types of artifacts. To this end, we developed a transfer learning-based procedure that uses the CSA inpainting model, which was originally proposed for unrelated tasks including the restoration of images from the Paris StreetView Dataset. The opposed system is trained and validated on the EndoCV2020 dataset, consisting of images from real endoscopies, highlighting how deep image inpainting may be a promising technology for frame restoration during medical procedures.

Keywords: Neural networks, Image processing, Intelligent control systems
Abstract: Landslides are critical natural hazards whose frequency and severity are increasing due to climate change and human activities. The consequences of landslides are severe and can lead to the destruction of homes, infrastructures and the contamination of water supplies, with severe impact also on the local ecosystems and the disruption of natural habitats. This article examines the application of an ad-hoc neural network-based intelligent system to evaluate the landslide susceptibility of the terrain on the basis of satellite data. The proposed system is validated on data from Lombardia and Abruzzo, two Italian regions that have been particularly subject to the landslide phenomenon. Results indicate that the CNN model is able to correctly identify landslide occurrences with high accuracy, demonstrating that CNNs are capable of providing accurate susceptibility mapping at a local scale and surpassing the performance of existing solutions available in the literature.

Keywords: Neural networks, Intelligent control systems, Game theory
Abstract: In recent years, stock price forecasting has become a challenging task commonly used to evaluate the performance of various machine learning solutions. This work explores a Federated Learning (FL) framework within a competitive collaboration scenario with the aim of training a centralised model advised by non-recoverable decentralised strategies so that no exchange of private data is required. The proposed Vertically-Advised Federated Learning (VAFL) framework combines elements from both horizontal and vertical FL, as each client trains two independent models. Furthermore, a novel forecasting architecture, based on a stochastic variant of an Attention-based Long Short Term Memory (LSTM) network, is proposed and validated on a simulated scenario based on real data from the stock market.

Keywords: Neural networks, Intelligent control systems, Robotics
Abstract: The perception of the environment is essential in mobile robotics applications as it enables the proper planning and execution of efficient navigation strategies. Optical sensors offer many advantages, ranging from precision to understandability, but they can be significantly impacted by lighting conditions and the composition of the surroundings. In contrast, millimeter wave (mmWave) radar sensors are not influenced by such adverse conditions. and are capable of detecting partially or fully obstructed obstacles, resulting in more informative point clouds. However, such point clouds are often sparse and noisy. This work presents a cross-modal contrastive learning approach based on Conditional Generative Adversarial Networks (cGANs) to transform sparse point clouds from mmWave sensors into depth images, preserving the distance information while producing a more comprehensible representation. An extensive data collection phase was conducted to create a rich multimodal dataset with each information associated with a timestamp and a pose. The experimental results demonstrate that the approach is able to produce accurate depth images, even in challenging environmental conditions.