AlbertoPetrucci

In safety-critical software development, changes often span technical and organisational boundaries, e.g., repositories, interfaces, libraries, and teams. Under certification and traceability requirements, these changes create a difficult regression-testing problem: organisations must determine what must be revalidated and provide an auditable justification for those decisions. In practice, the resulting uncertainty often drives overly conservative strategies, where large portions of the system are retested after changes or updates, even when a narrower scope would suffice. In this paper, we present CoCoS, an automated regression-testing approach for safety-critical systems written in C and developed under traceability and certification constraints. Developed end-to-end with Thales Alenia Space Italia, a large international aerospace company, and grounded in avionics practice, CoCoS addresses real certification-driven needs rather than an abstract setting. It combines semantic differencing, abstract syntax tree analysis, and function-level call-graph propagation to identify semantically relevant changes across repository and component boundaries, and to derive conservative regression scopes with human-readable justifications. CoCoS prioritizes conservativeness, explainability, and auditability, thereby supporting defensible regression-scoping decisions in regulated workflows. We evaluate CoCoS in an industrial setting with Thales Alenia Space Italia, using a realistic multi-repository aerospace software environment involving application code and external libraries. The results show that CoCoS supports the conservative bounding of regression scopes while preserving compliance-oriented reasoning and providing practical evidence for test planning in safety-critical environments.

This paper investigates Kolmogorov-Arnold Networks (KANs) as parametrisations of learned nonlinear state-space models in control-relevant settings. We ask whether replacing standard MLP maps with KAN-based ones yields a useful trade-off between predictive accuracy and deployability-oriented properties such as compactness, smoothness, and post hoc inspectability, with particular focus on underwater robotics. To study this question, we use a common autoencoder-based identification backbone and instantiate two KAN-based identification realisations, comparing their performance with their MLP-parametrised counterparts under the same training pipeline. The evaluation covers maritime robotics benchmarks of increasing difficulty, including autonomous underwater vehicles and a disturbance-driven underwater glider, as well as standard nonlinear system-identification datasets. Across these case studies, KAN-based models remain competitive in predictive performance while often using fewer parameters and, in several settings, supporting symbolic extraction of learned nonlinear features.

Safety-critical systems in domains such as aerospace, railways, and automotive require anomaly detection methods that are not only accurate but also transparent, predictable, and certifiable under strict operational constraints. Traditional deep neural networks fall short in these contexts because their black-box nature prevents formal reasoning, and their outputs cannot be restricted within well-defined operational limits. In this paper, we present SafeKAN, developed in collaboration with Thales Alenia Space, to address anomaly detection in satellite telemetry. SafeKAN builds on Kolmogorov-Arnold Networks (KANs) and introduces domain-specific adaptations, including codomain bounds derived from noise statistics, the exploitation of periodicity to learn from minimal datasets, and the estimation of signal period without prior knowledge. The results show that SafeKAN can achieve reliable anomaly detection while operating within safety bounds and providing the transparency required in certification processes.

The integration of renewable energy sources and smart technologies in domestic heating systems has become a critical focus in recent years. However, the complex landscape of competing technologies and need-to-comply normative requirements presents significant challenges for practitioners and end-users. In this paper, we first review the principal state-of-the-art technologies commonly employed in the industry. We then analyze the key normative requirements for implementing such systems in real-world scenarios. Based on this comprehensive review, we design a smart energy coordination system tailored for domestic buildings \, according to state-of-the-art practices, which will serve as a baseline for evaluating and comparing other novel, competing solutions. This testbed provides a common ground for assessing the effectiveness and compliance of various energy management strategies, ensuring they meet both technical and regulatory standards. To validate our approach, we implemented the proposed solution on a real-world residential testbed. Our work contributes to the development of reliable, efficient, and adaptable smart energy systems, supporting the transition towards renewable energy solutions and sustainable home automation.

Spacecrafts increasingly rely on software to fly and talk to Earth. Using Artificial Intelligence and Machine Learning for safety-critical software in space brings advantages and new challenges. We analyze the current way of producing safety-critical software and the reference safety and assurance standards in the domain, namely, ECSS-Q-ST-80C and ECSS-E-ST-40. Then, we explore the readiness of correct practices to ensure that ML- or AI-enabled systems are safe and reliable, and we discuss new practices and methods that should be introduced. The analysis refers to the different criticality classes safety-critical software can have.

Model-based development is a development methodology by which more and more suppliers and manufacturers are responding to increased and fast demands on the software development. The model-based model representation of the requirements or the design of a system has a lot of advantages. It is possible to directly derive several useful information from the modeled system automatically. This paper presents an experience of application / tailoring of a SW Quality Model fitting with the peculiarities of the automatic code generation with the support of the existing reference guidelines and a Simulink auto coding didactic experience.

A peritoneal dialysis machine that includes a preparator (60) as well as a cycler (100) to form the peritoneal dialysis system. The system delivers purified water into one or more containers (50) with different powders to create a concentrate and then moves this concentrate to a mixing bag (PDF-GEN) to create the peritoneal dialysis fluid (PDF). The cycler then delivers fresh PDF to the patient (200) and removes waste fluid via the drain outlet. The containers have unique data tags (1, 2, 3) containing container identification indicia and the machine includes integrated reading technology to retrieve the data from each container.

In medical apparatus, for example a dialysis machine, a pump in conjunction with a multichambered reservoir is provided. The volume pumped is determined by counting the number of fills of the reservoir made during a pumping phase. A less expensive pump may be used whilst maintaining an accurate determination of the volume pumped.

A peritoneal dialysis machine that includes a preparator as well as a cycler to form the peritoneal dialysis system. The system delivers purified water into one or more containers with different powders to create a concentrate and then moves this concentrate to a mixing bag to create the peritoneal dialysis fluid (PDF). The cycler then delivers fresh PDF to the patient and removes waste fluid via the drain outlet. A volumetric approach controls the hydraulic flow paths that introduce purified water to the powder concentrates, provide mixing of the concentrate to form the PDF and delivery of the freshly made PDF to the patient. Different configurations of hydraulic flow/pressure generators are provided in the fluid paths to provide optimization of the flow of water through the fluid system to create the correct powder concentrates and subsequent peritoneal dialysis fluid for cycling, for example being provided in a disposable cassette.

A peritoneal dialysis machine that includes a preparator as well as a cycler to form the peritoneal dialysis system. The system delivers purified water into one or more containers with different powders to create a concentrate and then moves this concentrate to a mixing bag to create the peritoneal dialysis fluid (PDF). The cycler then delivers fresh PDF to the patient and removes waste fluid via the drain outlet. A volumetric approach controls the hydraulic flow paths that introduce purified water to the powder concentrates, provide mixing of the concentrate to form the PDF and delivery of the freshly made PDF to the patient. Different configurations of hydraulic flow/pressure generators (80, 82,84) are provided in the fluid paths to provide optimization of the flow of water through the fluid system to create the correct powder concentrates and subsequent peritoneal dialysis fluid for cycling, for example being provided in a disposable cassette.

The disclosure relates to systems and methods for generating specific peritoneal dialysates from concentrate containers. The systems and methods control an order of adding concentrates from one or more concentrate containers or pouches to generate a peritoneal dialysis fluid. The systems and methods can minimize or avoid formation of a precipitate during the formation of the peritoneal dialysis fluid.