Research

Roberto Marani’s research activity began in 2006 with his Bachelor's Degree thesis with the modeling and design of photonic crystals by means of BPM (Beam Propagation Method) techniques. In 2008 he was engaged in his master thesis activity regarding the development of efficient algorithms for the design of functional devices in photonic crystal technology. The scientific activity of Roberto Marani has then continued in the fields of Electromagnetism, Optoelectronics, and Applied Electronics. Since 2012 he is working in the challenging field of Artificial Intelligence for industrial applications, precision agriculture, robotics, geology, archaeology, reverse engineering, and human monitoring.

His research activity regards:

• • • Automatic processing of thermal images for structural defects detection in composite materials (glass fibers and carbon fibers) through:

      • • Comparison of test components with master (pristine) references;

        • Extraction of features with physical meanings and corresponding classification with machine learning approaches;

        • Development of deep learning networks for complete automatic detection of complex targets.

    • Analysis of indoor environments returned by range sensors for the realization of 3D maps to look for possible changes affecting the scenes. This vast activity includes:

      • • Acquisition of range samples via the use of time-of-flight lasers sensors or triangulation-based sensors;

        • Noise reduction through the definition and application of suitable filters;

        • Extraction of features for scenes registration regardless of the initial point of view;

        • Object segmentation, identification and tracking, and autonomous estimation of scene alterations.

    • Development of complex setups for automatic three-dimensional reconstruction of micro-targets with sizes smaller than one millimeter;

    • Development of novel algorithms for scene flow computation using RGB-D data;

    • Design and prototyping of novel systems for monitoring productive processes and for the quality control of manufactured goods;

    • Modeling and design of reliable techniques for robot localization in structured environments;

    • Image processing for:

      • • Understanding complex environments by the analysis of natural images. This activity is mainly targeted to precision agriculture;

        • Estimation of products available on shelves in smart retail systems through the automated analysis of indoor images;

        • Fast and accurate detection of archeological embedded traces.

    • Video processing for:

      • • Indoor surveillance;

        • Monitoring of people with neurodegenerative diseases targeted to the automated estimation of the development of the disease;

        • Gesture and action recognition for human-robot cooperation.

    • Study of gain dynamics in active media made of dye molecule in epoxy for lasing applications and optical amplification of surface plasmons;

    • Analysis and design of metal systems exhibiting a plasmonic bandgap (Plasmonic Crystals, PlCs) for sensing, photovoltaics and optical interconnections. In more detail:

      • • Modeling and design of Lab-on-a-chip optical sensors in plasmonic technology based on the spectral shift of transmission resonances due to the presence of surface plasmonic modes;

        • Development of techniques for the control of plasmonic propagation aimed to the enhancement of the absorption efficiency in thin-film solar cells;

        • Design of optical nanoantennas for the detection of high-frequency electromagnetic fields and for beaming purposes in intra- and inter-chip connections;

    • Definition of a tridimensional vectorial model for the analysis of guided-wave photonic crystals (PhCs) based on the use of the Green function;

    • Design of PhC accelerating cavities for cancer hadrontherapy;

    • Design and characterization of electronic systems for remote health monitoring through the screening of physiological and biological parameters with non-invasive techniques;

    • Study of the scaling principles in nanoscale MOS devices;

    • Analysis of electronic devices at the hyper-frequencies (MESFET, HEMT and HBT) for the prediction of  internal and external thermal effects;

    • Modeling of field effect transistors on carbon nanotubes (CNTFETs) through the definition of SPICE models for the comprehensive design of analog and digital electronic circuits.

Dr. Marani has developed many numerical codes for:

• • • Advanced data analysis:

      • • Reconstruction of 3D models of environments and corresponding segmentation;

        • Optical flow analysis from 3D and 2D images and videos;

        • Deep learning algorithms for recognition of specific phenomena in challenging contexts

        • Machine learning algorithms for target classification;

        • Algorithms for detection of surface defects by the inspection of 3D data coming from production lines.

    • Electromagnetics throughout the use of: 

      • • Finite-Difference Time-Domain (FDTD) technique;

        • Multiple Scattering Method (MST); 

        • Transfer Matrix Method (TMM).

He has also developed a deep knowledge and mastery of many commercial software packages often used in Computer Science, Image Processing, Electromagnetisms and Electronics:

• • • Mathworks Matlab;

    • MVTec Halcon;

    • Microsoft Visual Studio (coding language C++);

    • Microsoft Visual Studio Code;

    • MEEP;

    • Rsoft Package including BandSolve, FullWave e BeamProp;

    • COMSOL Multiphysics Package;

    • Cadence PSpice.

His research activity in the field of computer vision has led Dr. Marani to improve his skills in the use of active lasers (till class IIIb), both in visible and infrared ranges, and detectors, with a particular focus on high-definition cameras by many manufacturers, such as Mikrotron, Dalsa, IDS, Imperix, Allied Vision, to mention a few. At the same time, he has established comprehensive know-how in the use of laser range finders (AccuRange AR4000), range cameras (Fotonic B- and C-series), and RGB-D sensors (Microsoft Kinect, Microsoft Kinect v2, and Microsoft Azure Kinect with both Microsoft e PrimeSense SDKs; Intel Realsense R200, D415, D435 with Intel Realsense SDK, v2.0). Moreover, Dr. Marani has been involved in software integration for controlling automatic mechanical stages (FESTO axes), robots and autonomous vehicles.