Advanced voxel-based CAD modelling for FSI simulations for automotive structures design

Additive Manufacturing (AM) is nowadays considered an important alternative to traditional manufacturing processes. AM technology shows several advantages in literature as design flexibility, and its use increases in automotive, aerospace and biomedical applications. As a systematic literature review suggests, AM is sometimes coupled with voxelization, mainly for representation and simulation purposes. Voxelization can be defined as a volumetric representation technique based on the model’s discretization with hexahedral elements, as occurs with pixels in the 2D image. Voxels are used to simplify geometric representation, store intricated details of the interior and speed-up geometric and algebraic manipulation. Compared to boundary representation used in common CAD software, voxel’s inherent advantages are magnified in specific applications such as lattice or topologically structures for visualization or simulation purposes. Those structures can only be manufactured with AM employment due to their complex topology. After an accurate review of the existent literature, this project aims to exploit the potential of the voxelization algorithm to develop optimized Design for Additive Manufacturing (DfAM) tools. The final aim is to manipulate and support mechanical simulations of lightweight and optimized structures that should be ready to be manufactured with AM with particular attention to automotive applications. A voxel-based methodology is developed for efficient structural simulation of lattice structures. Moreover, thanks to an optimized smoothing algorithm specific for voxel-based geometries, a topological optimized and voxelized structure can be transformed into a surface triangulated mesh file ready for the AM process. Moreover, a modified panel code is developed for simple CFD simulations using the voxels as a discretization unit to understand the fluid-dynamics performances of industrial components for preliminary aerodynamic performance evaluation. The developed design tools and methodologies perfectly fit the automotive industry’s needs to accelerate and increase the efficiency of the design workflow from the conceptual idea to the final product.

An approach to digitalise activities of combine harvesters through CANBUS

The increase in the efficiency of agricultural machinery is a theme that attracted the attention and investments of the industrial and research community. In addition, in a global market, where the prices of agricultural commodities are so volatile and the prices of the inputs increase, farmers and agricultural contractors struggle to obtain at the end of the agricultural season a consolidated profit. For these reasons, it is important to carefully plan the usage of combine harvesters, to reduce the unproductive time and the input usage such as the fuel, that at the end of the harvesting season could increase costs. This study aims to develop an algorithm able to automatically identify and evaluate the time spent by the combines in each of the identified activities, identify the field boundaries of the harvested fields and perform a performance evaluation. To be able to develop the algorithm, during the harvesting seasons of 2020 and 2022, two combine harvesters operating in real-world conditions in Bologna’s Province were monitored. The data necessary to perform the analysis were acquired as CANBUS data and processed by using the MATLAB ® suite. The results obtained from this analysis show that the monitored combines have spent over 60% of the time performing harvesting activities, 13% of the time idling at the field, 10% performing headland turn, the 3% and 4% of the time respectively in transport on the field and road and 2% of the time in unloading. In addition, the performance of the monitored combines resulted similarly to the performance reported in other studies.

An architecture for network acceleration as a service in the cloud continuum

The pervasive availability of connected devices in any industrial and societal sector is pushing for an evolution of the well-established cloud computing model. The emerging paradigm of the cloud continuum embraces this decentralization trend and envisions virtualized computing resources physically located between traditional datacenters and data sources. By totally or partially executing closer to the network edge, applications can have quicker reactions to events, thus enabling advanced forms of automation and intelligence. However, these applications also induce new data-intensive workloads with low-latency constraints that require the adoption of specialized resources, such as high-performance communication options (e.g., RDMA, DPDK, XDP, etc.). Unfortunately, cloud providers still struggle to integrate these options into their infrastructures. That risks undermining the principle of generality that underlies the cloud computing scale economy by forcing developers to tailor their code to low-level APIs, non-standard programming models, and static execution environments. This thesis proposes a novel system architecture to empower cloud platforms across the whole cloud continuum with Network Acceleration as a Service (NAaaS). To provide commodity yet efficient access to acceleration, this architecture defines a layer of agnostic high-performance I/O APIs, exposed to applications and clearly separated from the heterogeneous protocols, interfaces, and hardware devices that implement it. A novel system component embodies this decoupling by offering a set of agnostic OS features to applications: memory management for zero-copy transfers, asynchronous I/O processing, and efficient packet scheduling. This thesis also explores the design space of the possible implementations of this architecture by proposing two reference middleware systems and by adopting them to support interactive use cases in the cloud continuum: a serverless platform and an Industry 4.0 scenario. A detailed discussion and a thorough performance evaluation demonstrate that the proposed architecture is suitable to enable the easy-to-use, flexible integration of modern network acceleration into next-generation cloud platforms.

Directly training spiking neural networks for cyber-physical systems: from supervised to reinforcement learning

Spiking Neural Networks (SNNs) are bio-inspired Artificial Neural Networks (ANNs) utilizing discrete spiking signals, akin to neuron communication in the brain, making them ideal for real-time and energy-efficient Cyber-Physical Systems (CPSs). This thesis explores their potential in Structural Health Monitoring (SHM), leveraging low-cost MEMS accelerometers for early damage detection in motorway bridges. The study focuses on Long Short-Term SNNs (LSNNs), although their complex learning processes pose challenges. Comparing LSNNs with other ANN models and training algorithms for SHM, findings indicate LSNNs' effectiveness in damage identification, comparable to ANNs trained using traditional methods. Additionally, an optimized embedded LSNN implementation demonstrates a 54% reduction in execution time, but with longer pre-processing due to spike-based encoding. Furthermore, SNNs are applied in UAV obstacle avoidance, trained directly using a Reinforcement Learning (RL) algorithm with event-based input from a Dynamic Vision Sensor (DVS). Performance evaluation against Convolutional Neural Networks (CNNs) highlights SNNs' superior energy efficiency, showing a 6x decrease in energy consumption. The study also investigates embedded SNN implementations' latency and throughput in real-world deployments, emphasizing their potential for energy-efficient monitoring systems. This research contributes to advancing SHM and UAV obstacle avoidance through SNNs' efficient information processing and decision-making capabilities within CPS domains.