Collaborative robotics and computer vision systems are increasingly important in automating complex industrial tasks with greater safety and productivity. This work presents an integrated vision system powered by a trained neural network and coupled with a collaborative robot for real-time sorting and quality inspection in a food product conveyor process. Multiple object detection models were trained on custom datasets using advanced augmentation techniques to optimize performance. The proposed system achieved a detection and classification accuracy of 98%, successfully processing more than 600 items with high efficiency and low computational cost. Unlike conventional solutions that rely on ROS (Robot Operating System), this implementation used a Windows-based Python framework for greater accessibility and industrial compatibility. The results demonstrated the reliability and industrial applicability of the solution, offering a scalable and accurate methodology that can be adapted to various industrial applications.

Non-destructive testing (NDT) is one of the best alternatives to perform inspections and maintenance operations in aerospace and aeronautics industries. In Lock-in Thermal Tests (LTT) the stimulation is modulated in a sinusoidal wave using mechanical loads, ultrasounds, microwaves or, as in this work, visible light through halogen lamps. This work assesses the influence of the parameters of LTT, such as defect geometry, cycle period, and number of cycles, interpolation method, and the type of image to identify the sensitivity of the LTT (parameter c). Several samples were manufactured with precise notches to simulate defects (slots). And performed several LTT in a controlled environment and with a custom jig to secure the samples. The performed tests permitted the analysis of various results for numerous types of controlled situations and defects, such as the slot width, depth, and cycle period. This work compared the number of cycles used during the test (1–15), the interpolation method (Harmonic or DFT) and the type of analysis (phase or amplitude). The cycle period indirectly defines the amount of energy applied during the test; therefore, it was expected to have a great impact in the results. Shorter cycles produced lower thermal differences, while longer cycles resulted in blurred images. The type of image was also found to be one of the most important setting, with the phase delay analysis presenting a higher differentiation of defects and its boundaries. The results from the variation of the number of cycles revealed these should be kept between three and nine. Additionally, the optical stimulation may also be a decisive setting, depending the defect geometry. As a major conclusion, the current LTT can detect defects with a width to depth ratio of 1.25, far less than 2.0, as is stated by the current literature.

Among the Non-Destructive Testing (NDT) techniques available today, Active Infrared Thermal Testing (AIRTT) is certainly one of the most flexible and promising. The goal of this work was to compare the results obtained with a common Lock-in Thermal Test (LTT) and the same test using a true sinusoidal stimulation obtained through a closed-loop controller. The results showed a poor dynamic response of the common system and a lack of proportionality between the reference signal and the generated optical stimulation. To improve its response, it was implemented a PID controller using a light sensor to close the feedback loop. The amplitude images obtained with this controller showed a significant improvement in the results. Defects invisible with the common LTT were now identifiable. The phase images obtained using the controller with feedback revealed higher sensitivity with lower noise. Despite only one system was tested, the results show that the optical stimulation used in LTT is not very accurate and can/should be improved and, that a sensitivity 2.5 times higher than the common LTT was achieved with a real sinusoidal stimulation. Abbreviation: NDT: Non-Destructive Tests; AIRTT: Active Infrared Thermal Testing; LTT: Lock-in Thermal Test; cLTT: common LTT; PID: Proportional, Integral and Derivative; CFRP: Carbon Fibre Reinforced Polymers; TTT: Transient Thermal Tests; LDR: light-dependent resistor; PMMA: Poly(methyl methacrylate)

Several studies demonstrate that the structure of the brain increases in hierarchical complexity throughout development. We tested if the structure of artificial neural networks also increases in hierarchical complexity while learning a developing task, called the balance beam problem. Previous simulations of this developmental task do not reflect a necessary premise underlying development: a more complex structure can be built out of less complex ones, while ensuring that the more complex structure does not replace the less complex one. In order to address this necessity, we segregated the input set by subsets of increasing Orders of Hierarchical Complexity. This is a complexity measure that has been extensively shown to underlie the complexity behavior and hypothesized to underlie the complexity of the neural structure of the brain. After segregating the input set, minimal neural network models were trained separately for each input subset, and adjacent complexity models were analyzed sequentially to observe whether there was a structural progression. Results show that three different network structural progressions were found, performing with similar accuracy, pointing towards self-organization. Also, more complex structures could be built out of less complex ones without substituting them, successfully addressing catastrophic forgetting and leveraging performance of previous models in the literature. Furthermore, the model structures trained on the two highest complexity subsets performed better than simulations of the balance beam present in the literature. As a major contribution, this work was successful in addressing hierarchical complexity structural growth in neural networks, and is the first that segregates inputs by Order of Hierarchical Complexity. Since this measure can be applied to all domains of data, the present method can be applied to future simulations, systematizing the simulation of developmental and evolutionary structural growth in neural networks.

Pelvic organ prolapse (POP) is a common condition among women, characterized by the descent of pelvic organs through the vaginal canal. Although traditional synthetic meshes are widely utilized, they are associated with complications such as erosion, infection, and tissue rejection. This study explores the design and fabrication of biodegradable auxetic implants using polycaprolactone and melt electrowriting technology, with the goal of developing implants that closely replicate the mechanical behavior of vaginal tissue while minimizing implant-related complications. Four distinct auxetic mesh geometries—re-entrant Evans, Lozenge grid, square grid, and three-star honeycomb—were fabricated with a 160 $\mathrm{\mu }$m diameter and mechanically evaluated through uniaxial tensile testing. The results indicate that the square grid and three-star honeycomb geometries exhibit hyperelastic-like behavior, closely mimicking the stress–strain response of vaginal tissue. The re-entrant Evans geometry has been observed to exhibit excessive stiffness for applications related to POP, primarily due to material overlap. This geometry demonstrates stiffness that is approximately five times greater than that of the square grid or the three-star honeycomb configurations, which contributes to an increase in local rigidity. The unique auxetic properties of these structures prevent the bundling effect observed in synthetic meshes, promoting improved load distribution and minimizing the risk of tissue compression. Additionally, increasing the extrusion diameter has been identified as a promising strategy for further refining the biomechanical properties of these meshes. These findings lay a solid foundation for the development of next-generation biodegradable implants.

Thermography is an imaging technique that uses the infrared radiation to create a thermal image. Due to changes induced in the thermal conductivity of materials, the superficial temperature pattern of a defected component can reveal delamination, voids, insertions, moisture, or changes in material continuity when heated. High power light can be used to produce, without contact, a slight increase in the temperature of the component that creates a specific thermal pattern in the vicinity of a defect. The aim of this work is to analyze the influence of thermal NDT parameters on the performance of composite materials boards with simulated defects test using lock-in thermography, and thus find the optimal parameters that maximizes the signal/noise ratio. Four samples were submitted to 60 thermal tests to evaluate the detectability performance c (defect with/defect depth). The results were processed using custom scripts developed in LabVIEW to crop, scale, and extract the thermal information at the sample' slots and sound areas. The results indicated that the parameter with higher influence is the stimulation duration. The best time to conduct a thermal analysis is also very important, which is usually right after the end of the stimulation in order to have the maximum contrast in the thermal images and to minimize the artifacts. In this work, the defect detectability was between 1.25 and 1.67, which shows an higher sensitivity than reported in previous works (c = 2).