Receiver Obstacle Railway Transmitter Road Figure 1: Existing obstacle-detecting system (light-interrupting type). Millimetre-wave-based System In ord

Transcription

1

2 Receiver Obstacle Railway Transmitter Road Figure 1: Existing obstacle-detecting system (light-interrupting type). Millimetre-wave-based System In order to improve the weather tolerance, the millimetre-wave-based system has been developed [1]. The obstacle is detected when the millimetre-wave ( GHz) reflected from the obstacle is received or the refector, which does not need wiring, becomes invisible by using the antenna. Fig. 3 shows the outlook of millimetre-wave antennas and reflectors. The range is from 2 m to 40 m. In order to avoid the interference with other ladars such as anti-collision systems on motorcars, a new modulation (FM-CW) has been designed for the exclusive use of railways. This system meets the international safety guide ICNIRP and its power density is limited below 7.97 W/m2 rms. The interference between the antennas, which are used for obstacle detection at an identical level crossing, is prevented by achiving a sharp electromagnetic field of each Bars antenna. Consequently, the optical Antenna transmitters and the corresponding receivers in the conventional light- Reflector interrupting type in Fig. 1 can be replaced by the millimetre-wave antennas and reflectors as shown Fig. 2. Obstacle The test results have shown that an equivalent obstacle-detecting accuracy is achieved by the millimetre-wavebased system. Because this millimetre-wave-based sysyem does not need the unobstructed optical view, which is essential to the conventional light-interrupting type, it has exhibited a good weather tolerance in the snow scattered by passing trains at level crossings as shown in Fig. 4. Distance is measured Cable Figure 2: Millimetre-wave-based system (millimetre -wave-interrupting type).

3 Figure 3: Millimetre-wave-based antennas (left) and reflectors (right). Figure 4: Test site in the snow scattered by passing trains. Laser Range Scanner with Automatic Pedestrian Tracking Algorithm For detecting small objects video-based solutions exhibit some disadvantages, such as narrow field of view, coverage limitation of setting, and susceptibility to change in light conditions. Therefore, multiple laser scanners are investigated [4]. Single-row laser range scanners with high scanning ratio, wide viewing angle and long-range measurement capability are used. They are set on the ground so that horizontal cross sections of the surroundings, containing moving feet of pedestrians as well as still objects, are obtained in a rectangular coordinate system of real dimension. The distance between the scanner and an object is calculated based on the measured time-of-flight of the pulse laser. Integration of multiple laser scanners, which is performed by using Hermart transformation that deals with shift and rotation, has the advantage of reducing the effect of occlusion. Since an interface for manual setting up is implemented in the software, a number of laser scanners can easily be calibrated. Fig.5 shows a sample laser scan, where laser points are coloured through background subtraction. The green image is the still object and the moving motorcar, bicycles and pedestrians are white. Background images are initially generated by each client computer, and sent to the server computer for two purposes. Firstly, they are registered to find transformation matrixes from each sensors local coordinate system to a pre-defined global one. Secondly, they are shown as well as the data of moving feet that are recorded in each scan for real-time monitoring. Background information 0 Scanner 1 Motorcar Scanner 2 Bicycle Pedestrian Scanner 3 Scanner 4 Figure 5: Detected objects by using one laser range scanner.

4

5

6