17 results
Now showing1 - 10 of 17
- Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , Embodied AI in industrial operations: A systematic literature reviewEmbodied Artificial Intelligence (EAI) is increasingly seen as a promising route toward flexible automation, also in industrial operations. Yet the term is used inconsistently, and existing surveys often address isolated components rather than the combined role of embodiment, learning, perception, action, data, and industrial deployment constraints. This paper presents a PRISMA-guided systematic literature review of recent EAI research in industrial operations. We first consolidate the terminology of EAI and define four guiding characteristics: embodiment, learning, perception-action coupling, and situated intelligence. Based on this framework, we introduce a manipulation-mobility matrix for classifying physical embodiments and analyze the reviewed literature across industrial applications, research focuses, hardware forms, input and output modalities, learning architectures, datasets, simulation environments, and training paradigms. The results show that current research is dominated by assembly and manufacturing scenarios, stationary robotic arms, vision- and language-based interfaces, and modular foundation model-centered architectures. However, tightly coupled sensorimotor learning, tactile and force feedback, situated intelligence, and validated real-world deployment remain limited. This review clarifies the current state of industrial EAI and identifies the key gaps that must be closed to enable scalable deployment in industrial operations. Conducts a PRISMA-guided review of 42 recent Embodied AI (EAI) industrial studies. Defines formal industrial EAI terminology via four key characteristics. Proposes a manipulation-mobility matrix classifying robotic embodiments. Finds a dominant focus on stationary cobots and vision/language inputs. Highlights key deployment bottlenecks: decoupled foundation models, precision limitsPublicationtype: Journal Article52 21 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , Mobile, multimodal, vision-based data acquisition system for passive monitoring in production and intralogistics(Elsevier, 2026); ; ; ; The digitalization of chaotic intralogistics and production processes, including, e.g., humans and otherwise dynamic or static, non-tracked assets, as in the case of lot size one and large-object production facilities, requires non-invasive sensor solutions. One approach is to equip already movable assets on the shopfloor with multimodal 2D/2.5D/3D optical sensor systems that perceive the surrounding environment – such a solution requires methods for sensor calibration, sensor fusion, localization, and mapping. Besides, to comply with data privacy regulations, data must be de-personalized. This work proposes a mobile, multimodal sensor system that passively monitors the surroundings, localizes itself, outputs depersonalized data online, and can recreate the environment as a geometric digital twin.Publicationtype: Conference Paper78 22 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , Leveraging passive monitoring applications in production and intralogistics(Elsevier, 2026); ; ; ; Today’s rapidly evolving industrial landscape and the progressive digitization of products, processes, and resources require massive amounts of online data on the current production and logistical status. Fixed-positioned sensor systems have a constrained Field of View (FoV), while a mobile system can significantly extend these, enabling further applications due to the larger observed space. In this work, we elaborate on potential applications that utilize incidentally acquired data from various locations from the perspective of a mobile, vision-based sensor system. We exemplarily demonstrate an application within the scope of capturing and triggering the intralogistics chain with data from a vision-based sensor system that can be carried by an Autonomous Guided Vehicle (AGV).Publicationtype: Conference Paper26 27 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , Industry 5.0 in aircraft production and MRO: challenges and opportunitiesGlobally interconnecting machines, processes, and resources driven by exploring and advancing new technologies defined Industry 4.0 (I4.0), resulting in, e.g., Cyber-Physical Production Systems (CPPS). The aircraft industry particularly struggled with transforming production and Maintenance, Repair, and Overhaul (MRO) processes, replacing humans with machines and automating as well as digitalizing significant parts of their value- and non-value-adding activities. However, in the face of current social and environmental challenges, future industries will need to shift from purely technology-driven to value-driven, working sustainably with resources, including human capital. Together, these approaches constitute the idea of Industry 5.0 (I5.0). On the one hand, the aviation industry faces the challenge that even I4.0 concepts and technologies are not yet fully exploited or implemented. On the other hand, due to the specific characteristics of aircraft production and MRO as well as the environmental impact of the product, a tremendous potential arises regarding placing human well-being back into the center of adding value and decreasing environmental footprint while building an industry that is resilient and fortified against disruptions of this era. In line with the I5.0 terminology, in this work, we outline the challenges and opportunities of integrating I5.0 principles into the aircraft production and MRO industries, focusing specifically on the scope of selected use cases.Publicationtype: Journal Article65 43 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , Legal implications of vision-language foundation models (VLFM) in industrial applications in Europe: an inquiry into data protection, copyright, and AI regulationThe rise of Foundation Models (FM) starts to revolutionize different aspects of machine vision in industrial applications. Increased generalizability, robustness, reliability, and the capacity for few-shot learning make vision systems adaptable and flexible to an unprecedented degree. Yet, the development and deployment of fine-tuned models in the industrial domain face unique legal challenges in Europe that arise from stringent regulations, including Copyright Law, the General Data Protection Regulation (GDPR), and the recently enacted AI Act. In this work, we investigate the legal implications of pre-training (1), fine-tuning (2), and utilizing VFMs (3) in industrial contexts in Europe, with a specific focus on manufacturing. With this contribution, we aim to raise awareness on how various regulations will impact the use of VLFMs in industrial settings.Publicationtype: Conference Paper31 26 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , Industrial segment anything - A case study in aircraft manufacturing, intralogistics, maintenance, repair, and overhaul(2023-07-24); ; ; ; Deploying deep learning-based applications in specialized domains like the aircraft production industry typically suffers from the training data availability problem. Only a few datasets represent non-everyday objects, situations, and tasks. Recent advantages in research around Vision Foundation Models (VFM) opened a new area of tasks and models with high generalization capabilities in non-semantic and semantic predictions. As recently demonstrated by the Segment Anything Project, exploiting VFM's zero-shot capabilities is a promising direction in tackling the boundaries spanned by data, context, and sensor variety. Although, investigating its application within specific domains is subject to ongoing research. This paper contributes here by surveying applications of the SAM in aircraft production-specific use cases. We include manufacturing, intralogistics, as well as maintenance, repair, and overhaul processes, also representing a variety of other neighboring industrial domains. Besides presenting the various use cases, we further discuss the injection of domain knowledge.Publicationtype: Preprint147 69 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , A methods-time-measurement based approach to enable action recognition for multi-variant assembly in Human-Robot CollaborationAction Recognition (AR) has become a popular approach to ensure efficient and safe Human-Robot Collaboration. Current research approaches are mostly optimized for specific assembly processes and settings. This paper introduces a novel approach to extend the field of AR to multi-variant assembly processes. The approach is based on generalized action primitives derived from Methods-Time-Measurement (MTM) analysis that are detected by an AR system using skeletal data. Subsequently a search algorithm combines the information from AR and MTM to provide an estimate of the assembly progress. One possible implementation is shown in a proof of concept and results as well as future work are discussed.Publicationtype: Conference PaperScopus© Citations 14 203 408 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , Towards synthetic AI training data for image classification in intralogistic settingsObtaining annotated data for proper training of AI image classifiers remains a challenge for successful deployment in industrial settings. As a promising alternative to handcrafted annotations, synthetic training data generation has grown in popularity. However, in most cases the pipelines used to generate this data are not of universal nature and have to be redesigned for different domain applications. This requires a detailed formulation of the domain through a semantic scene grammar. We aim to present such a grammar that is based on domain knowledge for the production-supplying transport of components in intralogistic settings. We present a use-case analysis for the domain of production supplying logistics and derive a scene grammar, which can be used to formulate similar problem statements in the domain for the purpose of data generation. We demonstrate the use of this grammar to feed a scene generation pipeline and obtain training data for an AI based image classifier.Publicationtype: Conference Paper210 296 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: without files , Visual AI applications on smart delivery units(IEEE, 2021-12); ; ; ; As actors in an IoT production environment, smart delivery units are tasked with identifying loaded components and acquiring shopfloor events such as consumption of material. Conventional identification procedures rely heavily on tags and markers that are applied on components. For processes that require marker-less identification procedures, AI-based object identification can be incorporated. In this paper, we present a novel integration of such visual applications on smart delivery units. We address the main challenges of this approach, namely the need for computational resources and integration with low-cost components. Additionally, we propose a scalable IoT concept for the distribution of the AI functionalities on those delivery units by utilizing containerized applications. We demonstrate the validity of this AI integration with a real-world implementation on delivery units, tested in an application near environment.Publicationtype: Conference Paper89 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Access status: with files , Smart material delivery unit for the production supplying logistics of aircraftDespite recent advantages, internal logistics for aircraft production is mainly performed manually. Missing or wrongfully loaded components can cause costly delays. Transforming delivery units into smart participants of a digitalized logistic chain has the potential to avoid such delays. In the scope of this work, we present a concept of a smart delivery unit for use in intralogistic processes on aircraft production sites. Its main functionality, the detection of loaded components and material, handles a high variety of identification principles that are defined by pre-existing processes. We therefore conceptualize smart sensor boards. Through a novel modularity structure, demand-driven equipment of these boards with different sensor types can be achieved. This includes AI-based, visual detection of components on delivery units. We display the versatility of our concept with a practical implementation based on low-cost sensors and demonstrate how our approach leads to demand-driven delivery units.Publicationtype: Conference PaperScopus© Citations 4 173 330