In the paper, the problem of the effects of people motion between two separated population under epidemic spread is considered, referring in particular the COVID-19 case. Model, analysis and design of an optimal switching control emulating the different steps of containment intervention strategies are presented. In particular, the case of an asymmetric people flow is studied, considering one small population receiving a people flow from a second one, more numerous. While the analysis is carried on in the general case, the hypothesis that the small outgoing flux of individuals from the largest population does not change its steady state conditions. After equilibrium and stability analysis of the proposed mathematical model, the philosophy of the switching control is discussed, the control is designed and the results of numerical simulations are reported to support the effectiveness of the proposed approach.
We introduce an efficient, scalable, and ready-to-use approach for generating knowledge graphs (KGs) based on any given ontology. Our method leverages large language model (LLM)-based retrieval-augmented generation (RAG) as a source of high-quality text data. It employs two agents: the first extracts entities and triples from the text corpus maintained by the RAG, while the second links similar entities by applying LLM reasoning grounded in pragmatics. This approach requires no fine-tuning or additional AI training, relying solely on off-the-shelf technologies. Furthermore, its use of RAG enables it to handle a text corpus of arbitrary size. We evaluated our method in the context of high-pressure die-casting, focusing on defect classification. In the absence of annotated datasets, a manual evaluation of the resulting KG and confusion matrix revealed a threat score of approximately 85
This paper presents a model-based control system design for a novel tethered unmanned aerial vehicle (UAV)-unmanned ground vehicle (UGV) system developed in collaboration with Aurrigo, aligning with their autonomous baggage tug, the Auto-DollyTug®. The UAV provides overhead surveillance while tracking the ground vehicle, with power and data transmission managed via a custom-designed tether system. A comprehensive dynamic model of the tethered quadcopter is developed, incorporating aerodynamic drag, environmental disturbances, and tether forces. Simulation studies of a dual-loop control architecture design are presented. The simulated scenarios assess system performance under varying UAV speeds, flight radius, and wind conditions, and make use of integral absolute error (IAE) metrics. Results highlight that UAV velocities (≈2.9 m/s) yield satisfactory tracking, while higher wind speeds and larger radius reduce performance. Preliminary studies involving a fuzzy logic-based tether control strategy are also introduced, with surface plots generated to guide adaptive tether length adjustment.
This study investigates the statistical distribution of inter-event times for lava fountains at Mt. Etna (2011-2022), demonstrating a power-law scaling indicative of Self-Organized Criticality (SOC). Furthermore, a modified version of the Bak-Tang-Wiesenfeld (BTW) model, referred to as the Volcano Sandpile (VS) model, tailored to simulate these inter-event times is investigated. While this preliminary model provides initial insights into the dynamics of lava fountain recurrence, further refinements are necessary to improve its accuracy in capturing the observed variability in eruption inter-event times and its applicability to more complex volcanic systems.
Solving the Stockyard Planning Problem (SPP) is a critical task in the daily business of transhipment points such as harbors where bulk material is freighted. In our previous work presented at the 21st International Conference on Informatics in Control, Automation and Robotics (ICINCO) 2024 [5] we presented a method to solve the SPP in a realistic scenario of a dry bulk port using a hybrid approach of constraint programming and greedy search algorithms. This paper proposes an extension to this method with an optimized score function and refined constraints. With this refined approach it is possible to solve more complex scenarios and the improved score function enables a more accurate representation of the reality. Additionally, we extended our test setting to validate the effectiveness of the new model.
This work presents an affordable educational robot designed to make robotics accessible in resource-constrained settings. Using low-cost hardware, the platform costs just 57.63 55–66
Safety is essential in human-robot collaborative assembly (HRCA), as humans and robots share close working spaces. Traditional safety assessment methods struggle to address the complexities of HRCA, often leading to non-implementation. Our model-based safety control system leverages Colored Petri Nets to reflect the needed flexibility, treating assembly steps as hazardous situations to evaluate task-based risk levels. The system features a modular design with separate human and robot subsystems, enabling a comprehensive risk assessment. Simulated evaluations of an industrial use case highlight the system’s capability to map the entire risk space, accommodate individual requirement restrictions, and assess mitigation strategies effectively.
The paper presents a modelling and identification of a piezoelectric actuator followed by the feed-forward controller design. A commercial piezoelectric bender, model PL140, from Physik Instrumente Co. is considered. Its physical model is derived using Euler-Bernoulli beam theory. This model is designed to produce accurate data for experiments without risking harm to the actual actuator. Nevertheless, the acquired model, due to its complex structure, is unsuitable for control design. For this purpose, a Hammerstein model is proposed. Its structure comprises a static non-linear component that characterizes hysteresis and a dynamic linear component represented by a stochastic auto-regressive model with external inputs (ARX model). The non-linear component of the Hammerstein model is represented by a shallow neural network. The parameters of the ARX model are estimated by the Bayesian approach. The feedforward controller is based on the independently inverted part of the developed Hammerstein model. The results are illustrated by simulations using the proposed physical model.
This study explores and evaluates four Adaptive PID control approaches: the standard PID, Dual-Adaptive PID, fractional PID, and fractional dual PID strategies, applied to a nonlinear process with varying parameters. Adaptive PID controllers are designed to dynamically adjust their parameters to maintain optimal performance in response to changes in system behavior. The dual adaptive PID strategy adds an extra layer of adaptation, improving its capacity to manage uncertainties and process variations. Based on fractional calculus, the fractional PID controller provides enhanced tuning flexibility and is particularly effective for complex, nonlinear systems. The findings reveal the strengths and limitations of each strategy, offering insights into their effectiveness in controlling nonlinear processes with variable parameters. Their performance is evaluated using ISE, ISCO, overshoot, and settling time metrics, which are vital for assessing control precision and energy efficiency, respectively. This comparative study contributes to a deeper understanding of adaptive control techniques and their applications in real-world dynamic systems.
Multi-Agent Pathfinding (MAPF) is a fundamental problem in AI, robotics, autonomous logistics and digital entertainment. This study focuses on the MDD-SAT encoding strategy for the MAPF problem. Traditionally, Conflict-Driven Clause Learning (CDCL) SAT solvers have dominated SAT MAPF solutions. This work explores an alternative approach: leveraging Stochastic Local Search (SLS) SAT solvers, specifically variants of ProbSAT, to solve MAPF with edge conflict prevention using MDD-SAT encoding. Our experiments on standardized benchmarks show that CDCL solvers excel in complex problems. In contrast, ProbSAT variants outperform CDCL in simpler scenarios, when given informed initial assignments derived from Multi-Valued Decision Diagrams (MDDs). Additionally, we propose a restart procedure to enhance SLS solver robustness. This study provides critical insights into the strengths and weaknesses of CDCL and SLS solvers and highlights opportunities for hybrid approaches, paving the way for further optimization in MAPF applications.
This paper presents an analysis and comparison of three control strategies designed for systems with inverse response characteristics and variable reference tracking. Dynamic Sliding-Mode Control (DSMC), Sliding-Mode Control (SMC), and Proportional-Integral-Derivative (PID) control. The efficacy of these controllers was assessed using simulations in a nonlinear isothermal Continuous-Stirred Tank Reactor (CSTR) and through tests in a modified Temperature Control Lab (TCLab). The results from both simulations and experiments reveal that DSMC consistently exceeds SMC and PID controllers in tracking performance to manage the inverse response when the reference varies.
Heating, ventilation, and air conditioning (HVAC) systems present significant opportunities for energy optimization and integration with renewable energy sources. Advanced control strategies are essential to manage energy consumption while enhancing system efficiency and overall performance. A key requirement for such control strategies is the availability of a dynamic model. This paper proposes a novel model-based methodology for real-time compressor control in HVAC systems, utilizing a switching event-triggered model predictive control (MPC) framework. The proposed approach enables dynamic mode switching between different operational states while having configurable constraints, facilitating a multivariable control scheme. The controller models for each operating mode are developed using data-driven system identification techniques and validated with experimental data acquired from air-to-water heat pumps in the test field. The effectiveness and performance of the proposed control strategy are evaluated through Model-in-the-Loop (MIL) simulations on various scenarios systematically testing the control within a single operational mode and upon mode switches. The event-triggered model predictive control method is compared with other control strategies, including event-triggered LQR and event-triggered PID, to highlight its superior performance.
This article addresses the circular take-off and landing (CTOL) strategies for fixed-wing tethered aircraft, particularly for Airborne Wind Energy Systems (AWES). This work introduces a novel bridle actuator to control the aircraft’s roll angle and presents a dynamic model integrated into a hierarchical control architecture. Simulations demonstrate the viability of this approach, exploring loiter phases with varying roll angles, including coordinated turns. The study highlights the advantage of single-tether systems over multi-tether configurations for improved power efficiency in AWES. It identifies operational regions suitable for tethered flight or a coordinated turn based on tether length and height. The findings suggest that CTOL with a single tether is a viable solution for AWES automation.
A multifunctional assistive system engineered to support visually impaired individuals through advanced computer vision and navigation technologies is presented in this paper. The functionalities of the framework are divided into outdoor and indoor modules, each addressing specific challenges in accessibility. For outdoor navigation, the system integrates Google Maps API for geolocation and route planning, coupled with a monocular depth estimation pipeline based on convolutional neural networks to enable real-time obstacle detection and avoidance. Indoor exploration capabilities include a suite of deep learning-based recognition models tailored for object classification and with potential applications towards the semantic understanding of the environment. These include CNN-based modules for color recognition, clothes recognition, food classification via transfer learning on curated datasets, thermal imaging for hot object identification, and object detection models (e.g., YOLOv5) for real-time grocery identification in retail environments. The system is designed for deployment on embedded hardware platforms with optimized inference performance, offering a lightweight yet robust solution for situational awareness. The tests performed show the system’s efficacy in diverse real-world scenarios, highlighting its potential to significantly enhance the mobility, safety, and independence of visually impaired users.
Articial Potential Field (APF)-based control strategies are widely used for the safe navigation of multiple unmanned vehicles due, in part, to their relative ease of implementation and reactive nature. However, a common drawback of these methods is the equal treatment of obstacles regardless of their motion, whether they are moving closer or away, and the general assumption of vehicles and obstacles of spherical shape, which can lead to overly conservative maneuvers and trajectories. To overcome these drawbacks, this paper presents a decentralized, cooperative APF-based control strategy for an arbitrarily large number of Unmanned Underwater Vehicles (UUVs) in cluttered environments that takes into account the relative direction of motion between vehicle and obstacle as well as the non-spherical shape and relative orientation of agents, resulting in more efficient transit through narrow spaces and shorter routes. The approach achieves this by formulating the minimum safe distance as a function of the shape and relative orientation of obstacles and by modulating the distance at which a vehicle starts avoiding an obstacle based on the collision threat, increasing the reaction forces when the obstacle is fast approaching and relaxing the forces when it is moving away. The proposed framework has the added convenience of generating closed-form, continuous control input forces and torques without the need for optimization-based techniques. The stability and safety of the overall control framework are rigorously proven through Lyapunov analysis. Simulation results demonstrate that the method enables safe, cooperative navigation with lower control efforts and shorter trajectories.
The latest research directions have provided a theoretical foundation to the experimental evidence that human walking owns time-harmonic motor patterns: the golden ratio is crucially involved to equal the ratio between the durations of two consecutive walking gait sub-phases within a generalized Fibonacci sequence. The corresponding gait index, named ϕ -bonacci gait number—even involving an intriguing experimental conjecture about the position of the foot relative to the tibia during the double support sub-phase—, is able to fully capture the most reliable and objective (quantitative) outcome measures (and their distortions in pathological subjects) of recursivity, asymmetry, consistency, and self-similarity (harmonicity) of the gait cycle. This paper provides experimental results on healthy and pathological gaits related to Benign Paroxysmal Positional Vertigo (BPPV). They totally support the aforementioned theoretical derivations, especially in the field of walking ability rehabilitation through canalith reposition manoeuvres. Furthermore, the newly introduced concept of (Heart-Rate-Variability-emulating) Harmonic Gait Variability (HGV)—used as a quantitative measure of the distance of the subject’s walking from a corresponding harmonic avatar—further, originally, illustrates in a quantitative fashion, rehabilitation effects.
In this paper, a disturbance compensation method for MIMO linear systems with state delays and delays in the control channels is introduced. First, a modified unknown input observer (UIO) is constructed to estimate the state vector of the system. Subsequently, based on this estimated value, an external disturbance observer is designed. Finally, all these estimated values will be utilized to synthesize the closed-loop control law for the system. The external disturbance is assumed to be the output of a linear generator. The rigor of the algorithm is proven mathematically, and simulations are carried out in MATLAB/Simulink to demonstrate the effectiveness of the proposed method.
The Proportional, Integral, and Derivative (PID) is a linear control technique that constitutes more than 90 ± 5% parametric uncertainty with minimal impact on transient speed.
We revisit rule-based algorithms for multi-agent path finding (MAPF) where the task is being solved by applying a fixed set of movement primitives. MAPF is a task of navigating agents from their initial positions to given individual goal positions via non-conflicting paths. The environment in which agents move is modeled as an undirected graph with agents in its vertices. Rule-based algorithms we study in this paper are complete, polynomial-time, and sub-optimal with respect to common cumulative objectives used in MAPF such as makespan or sum-of-costs. Out contribution consists in a new step that pre-processes the input graph by decomposing it into highly connected components via spectral clustering. Agents are moved to their goal cluster first before the rule-based algorithm is applied. The benefit of this approach is twofold: (1) the algorithms are often more efficient on highly connected clusters and (2) we can potentially run the algorithms in parallel on individual clusters.
In this paper a control strategy to perform the desired orientation of the end-effector in the path following task is presented. It is an extension of the traditional path following task definition for a manipulator as both end-effector position and orientation with respect to a path are taken into account. The path following algorithm is based on a parametric approach–a robot is described with respect to the Serret–Frenet frame associated with the desired path. The formalism of unit quaternions is used to define the relative orientation of the robot end-effector. Such an approach introduces additional constraints of the first order to be satisfied during the motion along the path. In the paper a control law for a redundant holonomic manipulator is proposed to perform the desired position and orientation with respect to the given path. The strategy shows how to deal with the redundant information given by the quaternions. Moreover, the definition of the desired state with respect to the path is proposed so that rapid motions of the manipulator are avoided in spite of significant initial errors, e.g. the robot is located outside the path in the initial state. The theoretical results are validated with a numerical analysis and experimental study. The achieved consistency of the results confirms the practical applicability of the proposed solution.
