Non-iterative object detection methods in electrical tomography for robotic applications

• Stephan Mühlbacher-Karrer ^[1] ; Juliana Padilha Leitzke ^[1] ; Lisa-Marie Faller ^[1] ; Hubert Zangl ^[1]
  1. [1] Alpen-Adria-Universität Klagenfurt
• Localización: Compel: International journal for computation and mathematics in electrical and electronic engineering, ISSN 0332-1649, Vol. 36, Nº 5 (Special Issue: IGTE 2016), 2017, págs. 1411-1420
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • Purpose This paper aims to investigate the usability of the non-iterative monotonicity approach for electrical capacitance tomography (ECT)-based object detection. This is of particular importance with respect to object detection in robotic applications.

    Design/methodology/approach With respect to the detection problem, the authors propose a precomputed threshold value for the exclusion test to speed up the algorithm. Furthermore, they show that the use of an inhomogeneous split-up strategy of the region of interest (ROI) improves the performance of the object detection.

    Findings The proposed split-up strategy enables to use the monotonicity approach for robotic applications, where the spatial placement of the electrodes is constrained to a planar geometry. Additionally, owing to the improvements in the exclusion tests, the selection of subregions in the ROI allows for avoiding self-detection. Furthermore, the computational costs of the algorithm are reduced owing to the use of a predefined threshold, while the detection capabilities are not significantly influenced.

    Originality/value The presented simulation results show that the adapted split-up strategies for the ROI improve significantly the detection performance in comparison to the traditional ROI split-up strategy. Thus, the monotonicity approach becomes applicable for ECT-based object detection for applications, where only a reduced number of electrodes with constrained spatial placement can be used, such as in robotics.