Object Detection

Because of the large stopping distances, the presence of large radii vertical and horizontal curves and linear nature of rail lines, detecting track obstructions including vehicles, humans and animals is a continuing challenge. The GoSAFE RAIL project will address this specifically on safety-critical sectors of infrastructure, such as level crossings, bridges and tunnels', and on open-tracks by developing a new train mounted multiple sensor system. The train-mounted system consists of high-resolution cameras used to capture a series of 2D images. Image processing techniques applied in other domains will be transferred to the rail sector by implementing a new, innovative, virtual instrument-based tool supported by neural network pattern recognition toolbox with custom made architecture used to form an algorithm for efficient obstacle tracking. A fixed (off-track) sensor network including geophones to sense vibration and fixed camera systems will be adopted.

In addition GoSAFE RAIL partner NGI will extend ongoing work with the Norwegian rail authorities (JBV) and the Norwegian road authorities (SSV) to develop systems for detection and warning of geohazards. This proposed system will use embedded geophones to detect avalanches through the generated seismic waves. The seismic events are classified using real-time processing. In this project a wireless sensor networks will be deployed on the Norwegian Rail network to demonstrate the efficacy of the technology. A prototype of a just-in-time avalanche warning system that has four nodes equipped with seismic sensors that detect avalanches through ground vibrations is shown in Figure on the right. The advantage of this approach is that a relatively large area could be monitored with a single distributed measurement system, containing only a few easily installable sensors.