Alberto Ruiz

This work introduces an advanced computer vision system designed for real-time auto-identification (Auto-ID) of loads in industrial and logistics environments, optimized through a fog computing architecture. The system monitors multiple docks in logistics warehouses, identifying pallet loads in real-time and providing feedback for efficient truck loading to prevent errors. It enhances traditional Auto-ID by incorporating pallet class classification, detection, and size estimation, utilizing foundational models, such as DINOv2, MobileNetV3, SAM, Depth Anything V2, and Depth Pro, alongside few-shot learning to generate efficient training datasets with minimal labeling effort. We propose three key innovations: (1) an embedding analysis approach for precise load classification; (2) DINOv2-based visual feature detection solutions for bounding box estimation, and (3) a depth-guided segmentation strategy for improved load isolation and measurement accuracy. Experimental results on a curated industrial dataset compiled by us demonstrate high mean average precision, with optimized trade-offs between latency and accuracy for fog computing constraints. This solutions reduces cloud dependency, supports rapid online updates, and ensures robust performance in dynamic logistics settings, making it a scalable solution for automated image labeling and load identification in industrial applications.

This study investigates the potential of transient potentiometric signals generated by an array of ion-selective electrodes (ISEs) as the basis for a potentiometric electronic tongue capable of identifying and quantifying a range of inorganic and organic cations. Six distinct polymeric membrane ISEs were fabricated, differing in plasticizer type (either NPOE or DEHS), and in the presence or absence of a lipophilic ion exchanger (KTClPB) and/or an ionophore (DB18C6). Transient potential responses were recorded following the exposure of the electrode array to various cations at different concentrations. A total of 810 transient signals were analyzed through visual inspection and principal component analysis (PCA), revealing characteristic dynamic patterns influenced by membrane composition, ion type, and ion concentration. PCA was conducted both on the transient signals from each individual electrode and on the aggregated dataset comprising signals from the full six-electrode array (electronic tongue). The electronic tongue exhibited a markedly enhanced capacity for discriminating and quantifying ion concentrations in comparison to any single electrode.

In the race for economic growth, many production activities have incorporated automated devices into their processes. This is also true for the agriculture sector, where different sensors and Internet of Things (IoT) architectures have been proposed to perform automatic data gathering and data analyzing in later stages. However, most architectures only consider data from one source, ignoring valuable information from other sources like images. Furthermore, the few solutions that consider information from images employ an approach where the cameras are fixed. In this paper, we propose an IoT architecture called MeloDI that gathers data from traditional IoT sensors and images taken by drones and applies Machine Learning (ML) techniques to evaluate melon quality. The images taken by the drone are both RGB and multispectral, which allow MeloDI to analyze critical growth information. MeloDI is powered by the Cloud and Edge Computing paradigm, and it is divided into three layers: physical, edge, and cloud. The physical layer comprises devices that get information from melons, including drones. The edge layer is responsible for sending the data from sensors to the cloud layer, receiving the corrective actions from the cloud layer, and enforcing them. Finally, the cloud layer is responsible for analyzing the data using ML techniques to assess the quality of the melons. Additionally, we deployed MeloDI in a melon plantation in Southeast Spain. MeloDI gathered data from sensors and drones, extracting new features and indicators to determine melon quality. For comparison purposes with other solutions that only accept one data source, we tested three different configurations of MeloDI: using only data from traditional sensors, using only data from images taken by drones, and using both data sources. We conclude that the configuration using both data sources outperforms the other configurations. In particular, the best-performing model was Random Forest that achieved a Mean Square Error of 1.6111 and a corresponding error rate of 7.9944% on the test set.

Low-cost electrochemical methodologies for the determination of L-ascorbic acid (vitamin C) and the analysis of juices are developed based on its electro-oxidation on carbon screen-printed electrodes. A novel chronoamperometric methodology is developed for the quantification of L-ascorbic acid in fruit juices. The proposed method stands out for its simplicity and rapidity, demonstrating its efficacy in determining L-ascorbic acid content in various fruit juices. Notably, the results obtained with this chronoamperometric approach are compared with those yielded by chromatography, with no significant differences between the two methods being found. Additionally, an electronic tongue is developed for the differentiation of juices based on the square wave voltammetric signals.

A potentiometric electronic tongue (ET) for the analysis of well and ditch irrigation water samples is herein proposed. The sensors’ array is composed of six ion-selective electrodes based on plasticized polymeric membranes with low selectivity profiles, i.e., the membranes do not contain any selective receptor. The sensors differ between them in the type of ion-exchanger (sensors for cations or anions) and the plasticizer used in the membrane composition, while the polymeric matrix and the preparation protocol were maintained. The potentiometric response of each sensor towards the main cations (Na⁺, K⁺, Ca²⁺, Mg²⁺) and anions (HCO₃⁻, Cl⁻, SO₄²⁻, NO₃⁻) expected in irrigation water samples was characterized, revealing a fast response time (2+, Na⁺, and Cl⁻ concentrations and acceptable results for the rest of ions. Overall, the ET is a promising analytical tool for irrigation water quality, exceeding traditional characterization approaches (conductivity, salinity, pH, cations, anions, etc.) in terms of overhead costs, versatility, simplicity, and total time for data provision.

The present paper addresses the development and use of a new potentiometric electronic tongue for both qualitative and quantitative characterization of natural mineral waters. The electronic tongue is particularly related to the conductivity and ion content of/in the water sample. The analytical system is based on six ion-selective electrodes whose membranes are formulated to provide either cationic or anionic response and considering plasticizers with different dielectric constants (bis(2-ethylhexyl) sebacate, 2-nitrophenyl octyl ether or tricresylphosphate), while keeping the polymeric matrix, i.e., poly(vinyl chloride). Notably, the absence of any ionophore in the membrane provides a general response profile, i.e., no selectivity toward any special ion, which is convenient for the realization of an effective electronic tongue. The dynamic response of the tongue toward water samples of different chemical compositions and geographical locations has been obtained. At the optimized experimental conditions, the tongue presents acceptable repeatability and reproducibility (absence of hysteresis). The principal component analysis of the final potential values observed with the six electrodes allows for the differentiation and classification of the samples according to their conductivity, which is somehow related to the mineralization. Moreover, quantitative determination of the six main ions in the water samples (i.e., chloride, nitrate, hydrogen carbonate, sulfate, sodium, calcium, and magnesium) is possible by means of a simple linear calibration (and cross-validation) model.

The race for automation has reached farms and agricultural fields. Many of these facilities use the Internet of Things technologies to automate processes and increase productivity. Besides, Machine Learning and Deep Learning allow performing continuous decision making based on data analysis. In this work, we fill a gap in the literature and present a novel architecture based on IoT and Machine Learning / Deep Learning technologies for the continuous assessment of agricultural crop quality. This architecture is divided into three layers that work together to gather, process, and analyze data from different sources to evaluate crop quality. In the experiments, the proposed approach based on data aggregation from different sources reaches a lower percentage error than considering only one source. In particular, the percentage error achieved by our approach in the test dataset was 6.59, while the percentage error achieved exclusively using data from sensors was 6.71.

A study of the transient potential signals obtained with a cation-selective electrode based on an ion-exchanger was carried out for solutions of the following individual cations at different concentrations: H⁺, Li⁺, Na⁺, K⁺, Rb⁺, Mg²⁺, Ca²⁺, choline (Ch⁺), acetylcholine (AcCh⁺), and procaine (Pr⁺). Three different general types of transient signals were distinguished depending on the value of the selectivity coefficient of the corresponding ion. A principal component analysis (PCA) was performed on the signals, finding that the qualitative identification of the corresponding ion from the scores of two principal components is possible. The study was extended to the transient signals of solutions containing an analyte in the presence of an interfering ion. The PCA of the corresponding signal allows for the detection of the presence of interfering ions, thus avoiding biased results in the determination of the analyte. Moreover, the two principal components of the transient signals obtained for each of the ions at different concentrations allow for the construction of calibration graphs for the quantitative determination of the corresponding ion. All the transient signals obtained experimentally in this work can be reconstructed accurately from principal components and their corresponding scores.