Enhance 3D Point Cloud Accuracy Through Supervised Machine Learning for Automated Rolling Stock Maintenance: A Railway Sector Case Study

This paper presents findings of a case study conducted to introduce industrial robots into automatic train coupler inspection of Siemens Class 380 rolling-stock. The targets are localized by coalescing RGB and time of flight (ToF) sensor data. The study examines several supervised machine learning techniques to improve the overall accuracy of 3D point clouds. A cost factor which reflects root mean square, mean absolute error and coefficient of determination is defined to evaluate the performance of the learning algorithms. The best-suited models are further validated using simulation data and selected to include in overall robotic sensing system.

[1]  Adolfo Crespo Mrquez.  The Maintenance Management Framework: Models and Methods for Complex Systems Maintenance , 2007 .

[2]  Mustafa Suphi Erden,et al.  Formulation of a Control and Path Planning Approach for a Cab front Cleaning Robot , 2017 .

[3]  Whoi-Yul Kim,et al.  Automated thickness measuring system for brake shoe of rolling stock , 2009, 2009 Workshop on Applications of Computer Vision (WACV).

[4]  Cordelia Schmid,et al.  Combining efficient object localization and image classification , 2009, 2009 IEEE 12th International Conference on Computer Vision.

[5]  Zhengyou Zhang,et al.  A Flexible New Technique for Camera Calibration , 2000, IEEE Trans. Pattern Anal. Mach. Intell..

[6]  Carsten Rother,et al.  Weakly supervised discriminative localization and classification: a joint learning process , 2009, 2009 IEEE 12th International Conference on Computer Vision.

[7]  Yasushi Yagi,et al.  Recovering Transparent Shape from Time-of-Flight Distortion , 2016, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[8]  W. Wiercienski,et al.  Feasibility of robotic cleaning of the undersides of Toronto subway cars , 1989, IEEE 39th Vehicular Technology Conference.

[9]  Tetsuo Tomiyama,et al.  Capturing, classification and concept generation for automated maintenance tasks , 2014 .

[10]  Tetsuo Tomiyama,et al.  Systems and Conceptual Design of a Train Cab Front Cleaning Robot , 2017 .

[11]  P H Bentley.  Development and Application of Auto-Coupler Systems for a Steelworks 1988–92 , 1993 .

[12]  Erik B. Sudderth.  Introduction to statistical machine learning , 2016 .

[13]  Kym Fraser,et al.  Facilities management: the strategic selection of a maintenance system , 2014 .

[14]  Radu Horaud,et al.  Time-of-Flight Cameras , 2012, SpringerBriefs in Computer Science.

[15]  Radu Horaud,et al.  Time-of-Flight Cameras: Principles, Methods and Applications , 2012 .

[16]  Hiroshi Yaguchi.  Robot introduction to cleaning work in the East Japan Railway Company , 1995, Adv. Robotics.

[17]  Anjali K. M. DeSilva,et al.  A Study on Automating Rolling-stock Maintenance in the Rail Industry using Robotics , 2017, ICINCO.

[18]  Milan Sonka,et al.  Object localization and border detection criteria design in edge-based image segmentation: automated learning from examples , 2000, IEEE Transactions on Medical Imaging.

[19]  G. Hayward,et al.  A Noncontact Ultrasonic Platform for Structural Inspection , 2011, IEEE Sensors Journal.

[20]  Jean-Yves Bouguet,et al.  Camera calibration toolbox for matlab , 2001 .