Development of Intelligent Obstacle Detection System on Railway Tracks for Yard Locomotives Using CNN

The paper proposes an approach to the development of obstacle detection systems on railway tracks for yard locomotives. The proposed approach is illustrated by full-stack technology comprises of hardware construction and software implementation. Original video capture device with double cameras making stereoscopic image recording in the realtime mode has been developed. The novel modified edge detection algorithm recognizes railway tracks and obstacles with on-line noise filtering. Pretrained object detection model containing deep convolution neural network able to distinguish and classify obstacles by its type and size has been implemented. Thus, the yard locomotive equipped with a proposed system can be classified as an intelligent vehicle achieving an autonomous safe-operating unit.

[1]  Jie Lin,et al.  Smart Transportation Systems: Architecture, Enabling Technologies, and Open Issues , 2017 .

[2]  Hiroshi Murase,et al.  Moving camera background-subtraction for obstacle detection on railway tracks , 2016, 2016 IEEE International Conference on Image Processing (ICIP).

[3]  Kim Steenstrup Pedersen,et al.  Deep Learning for Detection of Railway Signs and Signals , 2019, CVC.

[4]  Yong Chen,et al.  Research on Environment Perception and Obstacle Detection for Unmanned Vehicle Based on Machine Vision , 2020 .

[5]  José Barata,et al.  Laser-Based Obstacle Detection at Railway Level Crossings , 2016, J. Sensors.

[6]  Ilias K. Savvas,et al.  Detection of Point Anomalies in Railway Intelligent Control System Using Fast Clustering Techniques , 2018, Advances in Intelligent Systems and Computing.

[7]  Alessandro Fantechi,et al.  Connected or Autonomous Trains? , 2019, RSSRail.

[8]  Andrey V. Chernov,et al.  Mobile Smart Objects for Incidents Analysis in Railway Intelligent Control System , 2017 .

[9]  Wang Jie,et al.  TrackNet - A Deep Learning Based Fault Detection for Railway Track Inspection , 2018, 2018 International Conference on Intelligent Rail Transportation (ICIRT).

[10]  Bisheng Yang,et al.  Automated Extraction of 3-D Railway Tracks from Mobile Laser Scanning Point Clouds , 2014, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[11]  Aleksandar Miltenović,et al.  INTELLIGENT MACHINE VISION BASED RAILWAY INFRASTRUCTURE INSPECTION AND MONITORING USING UAV , 2019 .

[12]  Tiago Andrade,et al.  Automatic Detection of Obstacles in Railway Tracks Using Monocular Camera , 2019, ICVS.

[13]  Kenneth Gavin,et al.  A New Methodology for Assessment of Railway Infrastructure Condition , 2016 .

[14]  Xuan Song,et al.  DeepRailway: A Deep Learning System for Forecasting Railway Traffic , 2018, 2018 IEEE Conference on Multimedia Information Processing and Retrieval (MIPR).

[15]  Andrey V. Chernov,et al.  Knowledge Representation Method for Intelligent Situation Awareness System Design , 2018 .

[16]  Hiok Chai Quek,et al.  A novel framework for making dominant point detection methods non-parametric , 2012, Image Vis. Comput..

[17]  Yong Shi,et al.  Efficient railway tracks detection and turnouts recognition method using HOG features , 2012, Neural Computing and Applications.

[18]  Andrey V. Chernov,et al.  An Approach for Distributed Reasoning on Security Incidents in Critical Information Infrastructure with Intelligent Awareness Systems , 2019 .

[19]  V. Mohanavel,et al.  Fundamentals and development of self-driving cars , 2020 .

[20]  M. Heddebaut,et al.  Automatic radar target recognition of objects falling on railway tracks , 2012 .

[21]  I. Nourbakhsh,et al.  On the future of transportation in an era of automated and autonomous vehicles , 2019, Proceedings of the National Academy of Sciences.

[22]  Martin Dobrovolny,et al.  The obstacle detection on the railway crossing based on optical flow and clustering , 2013, 2013 36th International Conference on Telecommunications and Signal Processing (TSP).