Imbalanced data are a non-trivial problem in deep learning. The high variability in the number of samples composing each category might force learning procedures to become biased towards classes with major cardinality and disregard classes with low instances. To overcome such limitations, common strategies involve data balancing using resampling techniques. The cardinality of overnumbered categories is often lowered by sample deletion, thus reducing the data space where the model can learn from. This paper introduces a new approach based on data balancing without sample deletion, allowing for biasing reduction in instance localization and classification tasks. The method is a multi-stage pipeline starting with data cleaning and data filtering steps and ending with the actual data balancing process, during which overnumbered samples are not deleted but divided into multiple sub-classes. In this way, the model can learn from balanced data distribution in which some classes have a high correlation factor. To evaluate the effectiveness of the method in real-life scenarios, a case study in the field of precision agriculture has been developed, motivated by the fact that the publicly available datasets for pest classification often reflect the real-world imbalanced distribution of pests, making the task challenging. Two models for the localization and recognition of pests belonging to several species are also indicated. The obtained results show the method’s validity as the performance both in the detection and classification tasks outperforms the state-of-the-art methods. The general nature of the conceived balancing technique may make the approach useful in other application fields.

Copper contamination in the environment poses significant risks to both soil and human health, making the need for reliable monitoring methods crucial. In this study, we report the use of the EmStat Pico module as potentiostat to develop a portable electrochemical biosensor for copper detection, utilizing yeast Saccharomyces cerevisiae cells immobilized on a polydopamine (PDA)-coated screen-printed electrode (SPE). By optimizing the sensor design with a horizontal assembly and the volume reduction in the electrolyte solution, we achieved a 10-fold increase in current density with higher range of copper concentrations (0–300 µM CuSO₄) compared to traditional (or previous) vertical dipping setups. Additionally, the use of genetically engineered copper-responsive yeast cells further improved sensor performance, with the recombinant strain showing a 1.7-fold increase in current density over the wild-type strain. The biosensor demonstrated excellent reproducibility (R² > 0.95) and linearity over a broad range of copper concentrations, making it suitable for precise quantitative analysis. To further enhance portability and usability, a Bluetooth-enabled electrochemical platform was integrated with a web application for real-time data analysis, enabling on-site monitoring and providing a reliable, cost-effective tool for copper detection in real world settings. This system offers a promising solution for addressing the growing need for efficient environmental monitoring, especially in agriculture.

We present Swarm Manager, a novel distributed computation framework to augment perception accuracy of computationally-constrained platforms in heterogeneous multi-robot systems. The proposed approach exploits computation offloading and software defined networking allowing robots to offload heavy computation (e.g. lidar odometry, object detection) to other more resourceful peers in the team following the decisions of a globally-aware and dynamically-eligible central orchestrator. We demonstrate in simulation enhancements in the perception accuracy of a computationally-constrained platform belonging to an heterogeneous team of four autonomous robots involved in the exploration of a multi-level, GPS-denied and communication-constrained unfinished power plant, gathering real-world data from the Urban Circuit of the DARPA Subterranean Challenge.

We present Swarm Manager, a novel distributed computation framework to augment perception accuracy of computationallyconstrained platforms in heterogeneous multi-robot systems. The proposed approach exploits computation offloading and software defined networking allowing robots to offload heavy computation (e.g. lidar odometry, object detection) to other more resourceful peers in the team following the decisions of a globally-aware and dynamically-eligible central orchestrator. We demonstrate in simulation enhancements in the perception accuracy of a computationally-constrained platform belonging to an heterogeneous team of four autonomous robots involved in the exploration of a multi-level, GPS-denied and communication-constrained unfinished power plant, gathering real-world data from the Urban Circuit of the DARPA Subterranean Challenge.

Any data structure used to store information can be considered a database. Blockchain technology, at its core, is no more than a ledger to store information about transactions. In recent years, such technology has gained immense popularity in a wide range of applications. However, scalability and throughput are major challenges of the Blockchain technology since applications and APIs have much lower throughput compared to non‐Blockchain ones. Several studies showed that the integration of the NoSQL paradigm along with a Blockchain pipeline enhances the overall throughput and scalability of the system, as well as the possibility of handling both on‐chain and off‐chain data. The goal of this paper is therefore to study how the advantages of both technologies can be capitalized on to enhance the system capabilities and to finally put forward a novel hybrid architecture that mixes NoSQL and Blockchain characteristics using Deep Learning techniques.