https://ediss.sub.uni-hamburg.de:443/handle/ediss/2 2026-05-20T00:05:03Z 2026-05-20T00:05:03Z Ahlers, Daniel https://ediss.sub.uni-hamburg.de:443/handle/ediss/12390 2026-05-19T23:35:18Z 2026-05-19T11:46:39Z

Title: In Situ Error Correction for 3D Printed Objects with Integrated Electronics on 5-Axis Printers Authors: Ahlers, Daniel Abstract: 3D printing has altered manufacturing by enabling the production of complex objects. The emerging field of printed electronics makes it possible to integrate functional parts with embedded electronics. The full potential of combining 3D printing and printed electronics is unlocked by 5-axis printing. This technique allows circuits to be routed along the surfaces of complex objects. This thesis presents a method to reliably print electronics, correct upcoming errors during printing, and generate a digital twin of the printed object, all achieved on low-cost hardware. 5-axis printing on low-cost printers is challenging due to misalignments in the rotary axes. This challenge is amplified for printed electronics that require a constant distance to the surface and continuous deposition along the path. To address this issue, this work presents a software pipeline that measures the misalignments of the printer's rotary axes and incorporates them into a URDF based model of the printer. An IK solver uses this model to generate compensated toolpaths that account for these deviations. To achieve continuous deposition, surface normals extracted from the underlying object are incrementally adapted along the path. This enables reliable deposition of printed electronic wires onto arbitrary surfaces using imprecise low-cost hardware. Even with a well calibrated system, errors can still occur during printing, and printed electronics are highly sensitive to these faults. A small imperfection in a wire can make the circuit nonfunctional and cause failure of the entire object. This work presents a method for in itu error detection and repair to address this problem. A neural network segments the wires in images captured during the printing process. By comparing the segmented wires with the intended toolpath, defects are identified. From the identified defects, repair toolpaths are generated to fix them. This results in more reliable electronics with known circuit properties. For structural objects, reliability of the printing process is also crucial, especially in critical industries such as medical or aerospace. The first step toward quality control and certification is the creation of a digital twin of the printed object. Each printed layer is reconstructed by segmentation with a neural network using two inputs: an image of the current layer and an image of the previous layer. By stacking these individual layer segmentations, a 3D reconstruction of the printed object is created. The reconstruction archives high precision with a resolution of 12µm per pixel and a mean geometric deviation of 61.5µm. This digital twin is accurate enough to enable future quality inspection, adjust printing parameters, and serve as a basis for certification. In conclusion, the methods proposed in this thesis enable 5-axis printing of objects with embedded electronics on low-cost hardware, ensure the reliability of the printed electronics, and generate accurate reconstructions of structural objects.

2026-05-19T11:46:39Z Jaehnke, Sabrina https://ediss.sub.uni-hamburg.de:443/handle/ediss/12396 2026-05-19T23:35:11Z 2026-05-19T11:02:08Z

Title: Crown Ether-Polymers for Lithium Recovery: From Molecular Binding to Solid-State Architecture Authors: Jaehnke, Sabrina

2026-05-19T11:02:08Z Laxminarayanan Raj Prabhu, Navin https://ediss.sub.uni-hamburg.de:443/handle/ediss/12386 2026-05-18T23:34:47Z 2026-05-18T11:14:21Z

Title: Probabilistic and Generative Modeling for Emotion Recognition and Synthesis Authors: Laxminarayanan Raj Prabhu, Navin Abstract: Artificial Intelligence (AI) technologies have made remarkable progress in recent years, enabling systems that enhance human productivity, creativity, and decision-making across diverse domains. Despite these advances, current AI systems remain largely deficient in social intelligence: the ability to perceive, interpret, and express social signals in ways that support seamless human–agent interaction. This limitation is particularly critical in purposive social interactions, where success depends on interdependence, shared goals, coordinated action, and mutual trust. In such settings, a lack of social awareness can fundamentally compromise interaction quality and outcomes. Scenarios such as meetings, job interviews, collaborative design workshops, and healthcare consultations demand that agents accurately interpret social cues, adapt to interpersonal dynamics, and respond appropriately to sustain effective cooperation. For AI systems to be meaningfully integrated in these contexts, they must demonstrate the ability to understand and generate socially appropriate behavior. This dissertation adopts the framework of social signal processing (SSP), which distinguishes two essential components of social intelligence: recognition of social signals and their synthesis. Among the wide array of social signals—including personality, dominance, rapport, and regulation—affect plays a central role. Experienced as conscious feelings such as pleasure–displeasure or energy–tiredness, affect permeates social interactions and shapes outcomes at individual, group, and organizational levels. Accordingly, affect constitutes the focus of this thesis, with research contributions spanning both its recognition and synthesis. The first line of work focuses on the recognition of affective expressions. A central challenge arises from the inherently subjective and ambiguous nature of affect labels—annotators frequently disagree, and collapsing their judgments into a single “ground truth” discards meaningful variation. To address this, we develop methods that explicitly model label uncertainty for individual-level emotion recognition. By combining Bayesian neural networks (BNN) with label distribution learning (LDL), our approach captures both the central tendencies and the variability in annotations, yielding improvements in predictive accuracy, calibrated uncertainty, and robustness to annotator disagreement. Furthermore, we investigate how uncertainty can be modeled effectively under limited annotation conditions. By replacing Gaussian assumptions with a Student’s t-distribution, we directly connect annotation sparsity to uncertainty, resulting in higher predictive accuracy, faster convergence, and improved cross-corpus robustness. Together, these findings reinforce a central aim of this thesis: rethinking affect modeling by treating uncertainty not as noise to be eliminated, but as an intrinsic property of affective data that provides valuable insight into emotional ambiguity. While individual-level affect recognition has been extensively explored, the collective level of affect—how group-level affect emerges and evolves within groups—remains under-researched. To extend affect modeling from individuals to the group level, we introduce an annotation protocol grounded in psychological theory to capture the dynamic ebb and flow of group affect. Building upon these annotations, we propose a graph-based multimodal framework that models collective affect at scale, capturing both convergence and divergence across group members and revealing systematic patterns of group dynamics. The second line of work addresses the synthesis of affective expressions, with a particular emphasis on speech. Prior approaches largely rely on acted corpora with parallel data, which fail to capture the richness and spontaneity of natural emotional speech. We investigate generative modeling for speech emotion conversion in in-the-wild data using a disentanglement–resynthesis framework: by separating speaker, lexical, and affective embeddings and reconstructing them with a modified HiFiGAN vocoder, we achieve conversion without parallel corpora. This yields improvements in naturalness, controllability, and pitch modulation. We also explore diffusion-based decoders as an alternative to vocoder-based methods, showing that they broaden the range of emotions, improve expressivity, and better capture rare states, albeit sometimes at the expense of perceptual naturalness. Finally, we study the role of prosodic cues such as rhythm and stress, integrating a dedicated duration predictor (DP) that modulates speech rate with arousal and enhances overall quality and naturalness. Together, we frame seven research questions that advance both the perception and response aspects of socially intelligent systems. On the recognition side, this thesis contributes probabilistic methods for modeling label uncertainty, introduces novel group-level annotation protocols, and develops scalable, multimodal graph-based frameworks. On the synthesis side, it proposes unsupervised generative models for in-the-wild emotion conversion, explores diffusion models for controllable affect synthesis, and demonstrates the importance of duration modeling. Beyond technical improvements, these contributions reconceptualize affect recognition and synthesis by treating ground truth as distributions rather than absolutes, group affect as dynamic and emergent, and affective speech as inherently diverse and ecologically grounded. By addressing these challenges, this thesis lays the groundwork for AI systems that can both perceive and express affect appropriately in purposive social interactions. Such systems hold promise for enhancing trust, collaboration, and effectiveness across societal domains, including education, healthcare, and organizational teamwork.

2026-05-18T11:14:21Z Blanke, Christian https://ediss.sub.uni-hamburg.de:443/handle/ediss/12381 2026-05-18T23:35:32Z 2026-05-18T10:08:03Z

Title: Empirische Erfassung der Holzverwendung im Baubereich Authors: Blanke, Christian Abstract: Die Quantifizierung der Holzverwendung im Bausektor hat eine große Bedeutung in der Stoffstromanalyse Holz, da Gebäude eine komplexe und oft langfristige Materialnutzung darstellen. Auch gesellschaftlich wächst die Bedeutung über das Wissen über die Holzverwendung im Bausektor, was verschiedene Holzbauinitiativen und aufkommende Regulatoriken wie „EU Carbon Removals and Carbon Farming“ (CRCF) zeigen. Die Fähigkeit differenzierte Darstellungen, über die Holzverwendung im Bau anstellen zu können und bioökonomische Auswertungen anzuschließen, erfordern ein systematisches Monitoring. Bisherige Erhebungen wie das Projekt „KlimaBau“ haben bereits wertvolle Grundlagen geschaffen. Aus den Herausforderungen des KlimaBau-Projekts entwickelte sich diese Arbeit, die praxistaugliche Lösungen für die Umsetzung eines belast-baren Monitorings zur Erhebung der Holzverwendung im Bauwesen aufzeigt. Für ein kontinuierliches Monitoring bestehen systematische Lücken, zu deren Schließung diese Arbeit Lösungen aufzeigt: Es fehlt eine systematische Methodik für ein umfassendes Monitoring und eine Bewertung der Belastbarkeit verwendeter Datengrundlagen. Die vorliegende Arbeit schließt diese Forschungslücken durch drei zentrale Beiträge: die Entwicklung eines systematischen Bewertungsschemas für Datenquellen, die Analyse und Bewertung von sechs identifizierten Datengrundlagen zur Erhebung der Holzverwendung im Bauwesen, sowie die methodische Entwicklung einer Datenaufbereitungs- und Integrationsmethodik für ein kontinuierliches Monitoring der Holzverwendung im deutschen Bauwesen. Forschungsfrage: Kann die Integration verschiedener Datengrundlagen mit unterschiedlichen In-formationstiefen zu einem systematisierbaren und in Datenqualität bewertbaren Monitoring der Holzverwendung führen? Methodisches Vorgehen: Die Arbeit analysiert fünf bestehende Studien zur Materialverwendung im Baukörper („Holzverwendung im Bauwesen", „IÖR ISBE Materialkennziffern", „KlimaBau", „BAMB2020", „HolzImBauDat") und identifiziert daraus sechs zentrale Datengrundlagen: Zielgruppenbefragungen, BKI-Gebäudedokumentation, Leistungsverzeichnisse, Gebäuderessourcenpässe, Belege sowie BIM-basierte Materiallisten. Zur systematischen Bewertung der Informationsqualität des Monitorings zur Holzverwendung im Bauwesen wird erstmals ein dreischichtiges Bewertungsschema entwickelt. Die Kriterien sind nach Priorität für eine praxisnahe Anwendbarkeit strukturiert: Erreichbarkeit, inhaltliche Eignung und qualitative Eignung. Die Bewertung er-folgt auf einer dreistufigen Skala und ermöglicht die systematische Analyse von Stärken und Schwächen verschiedener Datenquellen. Zentrale Ergebnisse: Die Bewertung zeigt unterschiedliche Stärkenprofile: Leistungsverzeichnisse und BKI-Daten bilden durch ihre systematische Struktur und breite Verfügbarkeit die Hauptdatengrundlage für ein kontinuierliches Monitoring von Neubau und Modernisierung im Nichtwohnbau. Zielgruppenbefragungen sind unverzichtbar für Segmente ohne strukturierte Daten-quellen, insbesondere für die private Modernisierung, sollten jedoch durch hohe Stichprobenumfänge oder Kombination mit Belegen abgesichert werden. BIM-Materiallisten können die höchste Detailtiefe bieten, sind jedoch derzeit in der Anwendung nicht weit verbreitet. Gebäuderessourcenpässe könnten bei gesetzlicher Verpflichtung einen erheblichen Beitrag zur Datenbasis leisten. Die Arbeit entwickelt Standardprozesse zur Datenharmonisierung und -integration und stellt Standardanforderungen zur Einbindung in die Stoffstromanalyse über Umrechnungsfaktoren. Ein zentraler Mehrwert ist die erstmalige Schaffung von Transparenz über die Informationsqualität verwendeter Datengrundlagen: Bioökonomische Auswertungen können mit Metainformation zur Belastbarkeit der zugrundeliegenden Daten ergänzt werden, was die Fundiertheit von Aussagen zu Kohlenstoffspeicherung und Ressourceneinsatz erheblich stärkt. Wissenschaftliche Bedeutung: Die Arbeit erweitert den Forschungsstand durch einen systematischen konzeptionellen Rahmen, der Anforderungen, Datenauswahl, qualitative Bewertung und Integrationsmethodik verbindet. Die entwickelte Bewertungsmethodik ist auf andere Baustoffe übertragbar und schafft die Grundlage für fundierte politische und wirtschaftliche Entscheidungen im Kontext der Bioökonomie und des nachhaltigen Bauens. Die Forschungsfrage wird positiv beantwortet: Unter Gewährleistung von Transparenz der Informationsqualität und konsequenter Datenaufbereitung kann die Integration verschiedener Datenquellen zu einem systematisierbaren und bewertbaren Monitoring führen.

2026-05-18T10:08:03Z