Fraunhofer-Institut für Produktionsanlagen und Konstruktionstechnik IPK

Filters

2
2
1 1
1 1
2
Reset filters

Settings
Now showing 1 - 2 of 2
  • Publication
    Inertial Measurement Unit based Human Action Recognition Dataset for Cyclic Overhead Car Assembly and Disassembly
    ( 2022)
    Kuschan, Jan
    ;
    Filaretov, Hristo
    ;
    Krüger, Jörg
    Motion datasets in industrial environments are essential for the research on human-robot interaction and new exoskeleton control. Currently, a lot of Activities of Daily Living (ADL) datasets are available for researchers, but only a few target an industrial context. This paper presents a dataset for a semi-industrial Overhead Car Assembly (OCA) task consisting of synchronized video and 9-Degrees of Freedom (DOF) Inertial Measurement Unit (IMU) data. The dataset was recorded with a soft-robotic exoskeleton equipped with 4 IMUs covering the upper body. It has a minimum sampling rate of 20 Hz, lasts approximately 360 minutes and comprises of 282 cycles of a realistic industrial assembly task. The annotations consist of 6 mid-level actions and an additional Null class. Five different test subjects performed the task without specific instructions on how to assemble the used car shielding. In this paper, we describe the dataset, set guidelines for using the data in supervised learning approaches, and analyze the labeling error caused by the labeler onto the dataset. We also compare different state-of-the-art neural networks to set the first benchmark and achieve a weighted F1 score of 0.717.
  • Publication
    Inertial Measurement Unit based Human Action Recognition for Soft-Robotic Exoskeleton
    ( 2021)
    Kuschan, Jan
    ;
    Burgdorff, Moritz
    ;
    Filaretov, Hristo
    ;
    Krüger, Jörg
    Absence from work caused by overloading the musculoskeletal system lowers the life quality of the worker and gains unnecessary costs for both the employer and the health system. Exoskeletons can present a solution. Typically, such systems struggle with stiffness and discomfort and primarily a lack of battery lifetime. Soft-robotic exoskeletons offer a possibility to overcome these problems by increasing the system flexibility, not limiting the supported DoF and being actuator and joint together. Since soft-robotic exoskeletons can be designed only using power when supporting the wearer, it is possible to increase the battery lifetime by only acting on those actions for which the wearer needs support. Dealing with controls for soft-robotic exoskeleton one major difficulty is to find a compromise between saving energy and supporting the wearer. Having an action-depending control can reduce the supported actions to cover only relevant ones and increase the lifetime of the battery. The system conditions are to detect the user actions in real-time and distinguish between actions which require support and those which do not. We contribute an analysis and modification of human action recognition(HAR) benchmark algorithms from activities of the daily living, transferred them onto industrial use cases containing short and mid-term action and reduce the models to be compatible using embedded computers for real-time recognition on soft exoskeletons. We identified the most common challenges for inertial measurement units based HAR and compare the best-performing algorithms using a newly recorded data set overhead car assembly for industrial relevance. As a benchmark data set we focused on the "Opportunity" data set. By introducing orientation estimation, we were able to increase the F1 scores by up to 0.04. With an overall F1 score without a Null-class of up to 0.883, we were able to lay the foundation to use HAR for action dependent force support.