Fraunhofer-Institut für Produktionsanlagen und Konstruktionstechnik IPK

Filters
6
Reset filters

Settings
Now showing 1 - 6 of 6
  • Publication
    With synthetic data towards part recognition generalized beyond the training instances
    ( 2024)
    Koch, Paul
    ;
    Schlüter, Marian
    ;
    Krüger, Jörg
    In this work we investigate the effect of using synthetic data, generated in a simulation, in order to pre-train an AI-based image classification for industrial components. After pre-training we use real camera-captured training images to fine-tune the AI with the aim to close the Sim2Real domain gap. We compare our approach to purely using real training images of a single candidate object instance. In an exemplary case study for screw recognition, we found that a given AI classification algorithm dropped its recognition rate from 99.8% to 88.5% when testing the algorithm with known and unknown screw instances of the learned object classes, respectively. Employing our pre-training method on the basis of synthetic data, the drop in recognition rate is decreased from 99% to 96.95%. Thus, our proposed method has only a relative drop of 2.05% when shifting towards a generalized domain (including unknown part instances), while a compared approach on the basis of real camera-captured data showed a drop of 11.3%.
  • Publication
    Green incremental learning - Energy efficient ramp-up for AI-enhanced part recognition in reverse logistics
    ( 2023)
    Schlüter, Marian
    ;
    Schimanek, Robert
    ;
    Koch, Paul
    ;
    Briese, Clemens
    ;
    Chavan, Vivek Prabhakar
    ;
    Bilge, Pinar
    ;
    Dietrich, Franz
    ;
    Krüger, Jörg
    Artificial Intelligence (AI) has made significant progress in supporting circular economy and reverse logistics by learning from diverse data to predict, e.g., routes or to assist workers in sorting. However, it remains an open question how AI can be integrated and trained into such operational processes, where little to no data has been collected previously. Traditionally, AI models would only be rated by their accuracy. This paper aims to introduce the concept of green incremental learning, i.e. rating AI models not only for their accuracy but to evaluate energy efficiency as well. A ramp-up of a data-driven AI system for part recognition is explored under consideration of energy efficiency. Therefore, we combine online and incremental learning, working with growing data sets to simulate a ramp-up phase. We present experiments of incremental learning on business and image data, partially supported by regular joint training steps. We start local CPU-based machine learning and prediction on business data from the first sample. Finally, we compare incremental learning to traditional batch learning and show energy-saving potential of up to 62 % without a significant drop in accuracy.
  • Publication
    Decentralised identification of used exchange parts with a mobile application
    ( 2020)
    Lehr, Jan
    ;
    Schlüter, Marian
    ;
    Krüger, Jörg
    Sustainable product development and use requires an extended life cycle of used and defective mechanical parts. Remanufacturing saves resources and helps the industry to utilise the product more efficiently. Reverse logistics is one of the most important challenges towards efficient remanufacturing. To improve this process, we propose an on-site part identification at the workshops. A fast on-site identification is essential for assisting repair shop personnel and saving time on searching for the right spare parts. Based on images taken by a mobile device our application provides various machine vision services, e.g., visual identification of used parts, already successfully tested in a sorting facility for remanufacturing parts. The mobile application provides a robust visual identification for different environments. We show that enhancing data for machine vision approaches with images from decentral sensors, i.e., mobile devices, leads to an improved identification accuracy.
  • Publication
    Classification of Similar Objects of Different Sizes Using a Reference Object by Means of Convolutional Neural Networks
    ( 2019)
    Lehr, Jan
    ;
    Schlüter, Marian
    ;
    Krüger, Jörg
    Part identification is relevant in many industrial applications, either for direct recognition of components or assemblies, either as a fully automated process or as an assistance system. Convolutional Neural Networks (CNNs) have proven their worth in image processing, especially in classification tasks. It therefore makes sense to use them for industrial applications. There are major problems with parts that look very similar and can only be identified by their size. In this paper we have considered a subset of screws that all conform to the same norm but are of different sizes. The implicit learning of the screw size is only possible if the images are taken in a fixed distance setup and larger screws are shown larger on the images. In this paper we show that CNNs are able to implicitly measure target objects with the help of reference objects and thus to integrate the object size into the learning process.
  • Publication
    Deep learning for part identification based on inherent features
    ( 2019)
    Krüger, Jörg
    ;
    Lehr, Jan
    ;
    Schlüter, Marian
    ;
    Bischoff, Nils
    The identification of parts is essential for the efficient automation of logistic processes such as part supply in assembly and disassembly. This paper describes a new method for the optical identification of parts without explicit codes but based on inherent geometrical features with Deep Learning. The paper focusses on the improvement of training of Deep Learning systems taking into account conflicting factors such as limited training data and high variety of parts. Based on a case study in turbine industry the effects of steadily growing training data on the robustness of part classification are evaluated.
  • Publication
    Vision-based Identification Service for Remanufacturing Sorting
    ( 2018)
    Schlüter, Marian
    ;
    Niebuhr, Carsten
    ;
    Lehr, Jan
    ;
    Krüger, Jörg
    One of the main goals of sustainability is to reduce the ecological footprint. As a result the automotive industry has been encouraged to become more efficient in using existing resources to reach a target value of at least of 85 % of a car's weight for reuse and recycling as of 2015. The trade of used parts is expanding in total amount as well as in diversity of items. In industry practice employees have to decide upon the further use of a product based on experience or a reference list. We introduce a machine vision -based service for the identification of exchange parts. Images and weights of used parts serve as input whereby extracted inherent object features determine the identification of respective parts. First, in two main steps data is pre-filtered by its dimensions and volume out of a low-level 3D-model, created by a Shape-From-Silhouette algorithm. Secondly, a feature-based matching process is performed on the images. Two different feature matching approaches, a classic key point-based as well as a convolutional neural network, are evaluated. First results show the proof of concept recognition rates up to 96 %.