Welcome to YODA – Open Digital Archive for CSEM
The YODA archive gives access to CSEM's publications, such as its annual reports and brochures. For technical papers such as scientific publications, bibliographic information is provided, along with the full paper where this is possible.
This comprehensive database is part of CSEM's Open Access Publishing policy. For further information refer to the YODA support guide or contact repository@csem.ch.
Recent Submissions
Item Prozessoptimierung mit KI: Problemstellungen und Ansätze(2025-10-10)In der heutigen wettbewerbsintensiven Industrielandschaft ist die Prozessoptimierung keine Option mehr, sondern Notwendigkeit. Unternehmen in Fertigung, Logistik, Robotik und Chemieingenieurwesen stehen unter zunehmendem Druck, ihre Effizienz zu steigern, Abfall zu reduzieren und die Qualität ihrer Produkte zu sichern. Dafür müssen sie oft komplexe Systeme mit Hunderten voneinander abhängiger Variablen anpassen. Traditionelle Ansätze sind oft zu starr, um mit dynamischen und vielschichtigen Abläufen Schritt zu halten.Item EXPLORING LLM AGENTS FOR CLEANING TABULAR MACHINE LEARNING DATASETS(2025-03-01)High-quality, error-free datasets are a key ingredient in building reliable, accurate, and unbiased machine learning (ML) models. However, real world datasets often suffer from errors due to sensor malfunctions, data entry mistakes, or improper data integration across multiple sources that can severely degrade model performance. Detecting and correcting these issues typically require tailor-made solutions and demand extensive domain expertise. Consequently, automation is challenging, rendering the process labor-intensive and tedious. In this study, we investigate whether Large Language Models (LLMs) can help alleviate the burden of manual data cleaning. We set up an experiment in which an LLM, paired with Python, is tasked with cleaning the training dataset to improve the performance of a learning algorithm without having the ability to modify the training pipeline or perform any feature engineering. We run this experiment on multiple Kaggle datasets that have been intentionally corrupted with errors. Our results show that LLMs can identify and correct erroneous entries—such as illogical values or outliers—by leveraging contextual information from other features within the same row, as well as feedback from previous iterations. However, they struggle to detect more complex errors that require understanding data distribution across multiple rows, such as trends and biasesItem Next-gen diagnostics: CRISPR-chip for monoclonal antibodies quantification at the point-of-care(2025-09-03)Monoclonal antibodies (mAbs) are used to treat numerous cancers, offering specific therapy that improves patient outcomes. Measurements of mAbs in the bloodstream during cancer treatment are essential for monitoring the therapeutic antibody dose, assuring treatment efficacy, reducing side effects, and tracking drug resistance. Enzyme-Linked Immunosorbent Assay (ELISA) and High-Pressure Liquid Chromatography tandem Mass Spectrometry (HPLC-MS) are the gold standard for quantitative mAbs analysis in clinical studies. Both methods involve expensive, bulky, and complicated instrumentation, making them unsuitable for point-of-care (POC) use. To address this need, we introduce a POC platform for quantitative mAb detection, which leverages a recently developed quantitative mAb detection method, based on the use of a DNA circuit for the recognition of the mouse anti-hemagglutinin (antiHA) from influenza, and the activation of the Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) and CRISPR-associated protein (Cas). CRISPR systems' adaptability, low cost, high sensitivity and repeatability make our approach a great candidate for POC applications. Though promising, the conventional protocols run in test tubes, require four incubation steps and consecutive reagent additions. Thus, our study focuses on developing a simplified on-chip CRISPR-based assay without scarifying on the performance. For this purpose, a two-step protocol is developed, including sample dilution and injection, requiring only 10 µL of volume, and providing results within a 60 minutes sample-to-answer timeframe. The assay is integrated on a cost-effective chip, less than 1 € for chemicals, and additionally it is verified that the dispensed reagents can be stored at -20°C for at least 1 month without degradation. In conclusion, we demonstrate a proof-of-concept CRISPR-chip, which has the potential to impact modern diagnostics and the clinical market by introducing a new tool for protein quantification at the POC.Item Automated rigid bodies synthesis for AM compliant mechanisms(2025-12-31)Compliant mechanisms achieve motion through elastic deformation rather than traditional rigid-body joints, eliminating wear, backlash, friction, and the need for lubrication. These advantages make them ideal for high-precision applications and harsh environments. While the design of compliant joints is well-studied, the design of the rigid bodies – connecting the joints and transmitting forces/motions – is often overlooked. Existing approaches such as manual modelling, parametric design, and topology optimisation are inadequate for automation due to their fragmented workflows, limited flexibility, and lack of real-time responsiveness. This paper introduces a computational framework for the design of rigid bodies in compliant mechanisms, considering both functional, non-functional objectives and additive manufacturing constraints. Building on guiding curve-based design approaches, the method enables seamless integration of the rigid bodies’ synthesis into a fully automated compliant mechanism design pipeline. The process involves: (1) initialising a curve network to connect interfaces while minimising mass, (2) optimising the network to avoid mechanical interferences, maximise non-functional criteria, and satisfy AM constraints, (3) synthesising 3D tubes with locally tuned cross-sections to eliminate critical overhangs, and (4) generating smooth geometries with integrated non-sacrificial supports to reduce post-processing. The proposed methodology ensures manufacturable, reliable, and high-performance designs, advancing the automation of functional AM-enabled compliant mechanisms.
Communities in YODA
Select a community to browse its collections.