Abstract
We present CIDER, a novel framework for the collaborative editing of 3D augmented scenes. The framework allows multiple users to manipulate the virtual elements added to the real environment independently and without unexpected changes, comparing the different editing proposals and finalising a collaborative result. CIDER leverages the use of 'layers' encapsulating the state of the environment. Private layers can be edited independently by the different subjects, and a global one can be collaboratively updated with 'commit' operations. In this paper, we describe in detail the system architecture and the implementation as a prototype for the HoloLens 2 headsets, as well as the motivations behind the interaction design. The system has been validated with a user study on a realistic interior design task. The study not only evaluated the general usability but also compared two different approaches for the management of the atomic commit: forced (single-phase) and voting (requiring consensus), analyzing the effects of this choice on collaborative behaviour. According to the users' comments, we performed improvements to the interface and further tested their effectiveness.

Abstract
Explainable AI (xAI) emerged as one of the ways of addressing the interpretability issues of the so-called black-box models. Most of the xAI artifacts proposed so far were designed, as expected, for human users. In this work, we posit that such artifacts can also be used by computer systems. Specifically, we propose a set of metrics derived from LIME explanations, that can eventually be used to ascertain the quality of each output of an underlying image classification model. We validate these metrics against quantitative human feedback, and identify 4 potentially interesting metrics for this purpose. This research is particularly useful in concept drift scenarios, in which models are deployed into production and there is no new labelled data to continuously evaluate them, becoming impossible to know the current performance of the model. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Abstract

Abstract
The use of Convolutional Neural Network (CNN) models for image classification tasks has gained significant popularity. However, the lack of interpretability in CNN models poses challenges for debugging and validation. To address this issue, various explanation methods have been developed to provide insights into CNN models. This paper focuses on the validity of these explanation methods for ordinal regression tasks, where the classes have a predefined order relationship. Different modifications are proposed for two explanation methods to exploit the ordinal relationships between classes: Grad-CAM based on Ordinal Binary Decomposition (GradOBDCAM) and Ordinal Information Bottleneck Analysis (OIBA). The performance of these modified methods is compared to existing popular alternatives. Experimental results demonstrate that GradOBD-CAM outperforms other methods in terms of interpretability for three out of four datasets, while OIBA achieves superior performance compared to IBA.

Abstract
Municipal solid waste (MSW) management has become a critical issue today, posing substantial economic, environmental, and social challenges. Identifying and analyzing dominant themes in this field is essential for advancing research and policies towards sustainable MSW management practices. This study aims to explore the key issues related to MSW management that have been addressed by both the scientific community and policymakers through funded projects. By doing so, the study seeks to guide the scientific community as a knowledge producer and the EU as a key funder. Two Latent Dirichlet Allocation (LDA) models were applied to analyze the themes from two corpora: one representing scientific literature and another focusing on EU-funded projects. Additionally, this analysis was complemented by a quantitative estimation of the similarity between the two corpora, providing a measure of alignment between the scientific community and policymakers. The results generally indicate that the two spheres are aligned and highlight the diversity of topics explored by the scientific community. Nevertheless, it is concluded that there are opportunities for further research on specific topics, such as leaching and the extraction of heavy metals. Additionally, the popularity of topics identified in European Union-funded projects has fluctuated considerably over time, focusing primarily on waste management rather than its prevention. In light of these findings, waste prevention emerges as a promising avenue for future EU-funded research initiatives.

Abstract
This paper presents a method for position correction in collaborative robots, applied to a case study in an industrial environment. The case study is aligned with the GreenAuto project and aims to optimize industrial processes through the integration of various hardware elements. The case study focuses on tightening a specific number of nuts onto bolts located on a partition plate, referred to as “Cloison”, which is mounted on commercial vans produced by Stellantis, to secure the plate. The main challenge lies in deviations that may occur in the plate during its assembly process, leading to uncertainties in its fastening to the vehicles. To address this and optimize the process, a collaborative robot was integrated with a 3D vision system and a screwdriving system. By using the 3D vision system, it is possible to determine the bolts’ positions and adjust them within the robot’s frame of reference, enabling the screwdriving system to tighten the nuts accurately. Thus, the proposed method aims to integrate these different systems to tighten the nuts effectively, regardless of the deviations that may arise in the plate during assembly.