BG113786A - Equipment installed onboard railway trains for detection of objects and their subsequent classification - Google Patents

Abstract

The present invention pertains to a device installed onboard railway trains and used for detecting objects and their subsequent classification. Said device consists of detector sensors connected to a computing unit (8) with a module (7) of a global satellite navigation and positioning system. Furthermore, mounted at the front of the railway vehicle is a console (6) containing sensors with a lidar array (1), a stereo camera (2), a thermo-imaging camera (3) and a high-resolution camera (4), all of them powered by a power supply and control module (5) and connected to a command console (9) inside the rail vehicle with a computing unit (8) and a module (7) of the global satellite navigation and positioning system, wherein said computing unit (8) is responsible for the detection and classification of objects and is connected to a train expert system (11) for decision making, to an automatic driverless train control system (12) and to a monitoring terminal (13) of the dispatcher of driverless trains.

Description

Railway train equipment for object detection and their subsequent classification

Technical field

[0001] The technical solution relates to railway train equipment for the detection of objects and their subsequent classification.

Prior art

[0002] Like the automobile industry, the railway industry is also beginning to develop systems that should gradually make possible the complete replacement of human labor. This trend in the railway world is mainly aimed at replacing the driver's sensory, decision-making and execution skills. The complete replacement of these capabilities corresponds to GoA4 (degree of automation). The fully automated GoA4 metro has been running around the world for many years, but projects equivalent to GoA2 are only now starting to operate on conventional railways. This means that the system can start and stop the train automatically, but the driver must be present in the cab at all times to monitor the train. Compared to the closed environment of the subway, the open environment of the conventional railway line is much more demanding in terms of perception and the number of opportunities to interact with the surrounding environment.

[0003] The entire driverless train control system must be a very complex and sophisticated system of interacting subsystems. These subsystems include, for example, the Automatic Train Operation (ATO) system, known in the Czech Republic by the abbreviation AVV (automatic train control), as well as an expert train decision-making system that replaces the driver's decision-making skills and, last but not least, a number of detectors that replace the driver's sensory perception. These detectors include a detector for

I objects, which is intended to replace the driver's eyes or vision when observing the area in front of and around the railway vehicle.

Nature of the invention

[0004] The subject of the presented technical solution is the design and technical solution of the object detection equipment and their subsequent classification, which is railway train equipment for object detection.

[0005] The essence of the technical solution is that detection sensors (a group of lidars, a stereo camera, a thermal camera and a high-resolution camera), located in an adaptive console at the front of the railway vehicle, continuously monitor what is happening in the area in front of the train and in close proximity to the train. Thanks to the use of different types of sensors, a given scene can be captured in multiple parts of the electromagnetic spectrum both visible and invisible to the human eye. All raw sensor data is analyzed and stored in the computing unit of the object detector. All data processing and object detection calculations are performed in this block in real time and regardless of the speed of the moving vehicle. The data is processed with the help of analytical algorithms, and also with the help of artificial intelligence, which is able to classify not only ordinary objects (people, animals, vehicles), but also special railway objects (signals, their signal signs, etc. .). Two approaches are used to define the free space profile and define the safe zone. The first is to use a digital map created by scanning the track, and the second approach is to detect the track (or rails) to refine the entire route. The result of the calculation block of the object detector in this case is mainly information about the position and nature (for example, the position of the detected object in relation to the front of the train or the type and properties of this object). This information is ready for distribution to downstream train subsystems (eg an expert train decision-making system, a driverless automatic train control system or a driverless train dispatcher monitoring terminal).

Brief description of the drawings

Figure 1

Figure 1 represents an embodiment of the device according to this technical solution.

Description of variants

[0007] The variants of the invention are described in detail in the following example.

An example

[0008] The subject of the technical solution is a railway device 10 for detecting objects and their subsequent classification, including detection sensors connected to a computing unit 8 with a module 7 of the global positioning system (hereinafter referred to as GNSS).

[0009] The essence of the technical solution consists in the fact that in the console 6, located in the front of the rail vehicle, there are sensors including a group of lidars 1, a stereo camera 2, a thermal camera 3 and a high-resolution camera 4, together powered by power supply and control module 5, which are connected to a control panel 9 inside the rail vehicle with a computing unit 8 and module 7 of the global navigation satellite system GNSS. The computing unit 8 for detecting and classifying objects can be connected to the expert train system 11 for making decisions, to the system 12 for automatic train control without a driver and to the monitoring terminal 13 of the train dispatcher without a driver. The computing unit 8 is adapted to provide position and nature information, including the distance of the detected object from the front of the train and the type of the detected object.

[0010] The device 10 consists of the following parts:

• detection sensors in a certain configuration, which are a group of lidars 1, a stereo camera 2, a thermal camera 3 and a high-resolution camera 4;

• protective and support console 6 with adaptive attachment to the front of the rail vehicle, • power supply and control module 5, • GNSS module 7 and communication module, • object detector computing unit 8, • control panel 9 inside the rail vehicle .

[0011] Each sensor, due to its technological nature, usually perceives the environment at different distances, at different viewing angles and in different parts of the electromagnetic spectrum. The aforementioned sensor configuration can also capture some of the information beyond the current vision of the human eye or the driver's eye.

[0012] The principles of individual detection are as follows:

[0013] For lidar 1, this involves accurately measuring the distance to objects and creating a 3D scene using the reflection of laser beams. For stereo camera 2, it is about calculating the disparity map and rendering the 3D scene. In the case of the thermal camera 3, this is an image of emitted infrared radiation, and in the case of a high-resolution camera 4, it is an image in the visible spectrum of electromagnetic radiation. All raw sensor data ie. lidars 1, stereo cameras 2, thermal cameras 3 and camera 4, are analyzed and stored in the computing unit 8 of the object detector. The individual sensors are powered by a power supply and control module 5.

[0014] The sensors, ie. lidars 1, stereo cameras 2, thermal cameras 3 and camera 4, including the power and control module 5, are placed in a special console 6, which is located in the front part of the rail vehicle.

This console 6 has not only a carrying but also a protective function. One of the key features of this bracket 6 is that it can be mounted on any standard rail vehicle thanks to its adaptable mounting. It consists of two parts: a universal fixed support and protection box for lidars 1, stereo cameras 2, thermal cameras 3 and camera 4, and a power module 5 and adapter (structure mounted on the vehicle) corresponding to the design of the railway vehicle type.

[0015] Directly in the train there is a railway control panel 9 with a computing unit 8 of the object detector and a GNSS module 7 (and a communication module). The computational unit 8 of the object detector performs all computational operations necessary to detect and classify objects using artificial intelligence. In addition to standard objects (person, animal, vehicles), it can also classify special railway objects (signals, their signs, etc.). In addition, the track (or rail) itself is also taken into account so that it can be determined whether the detected object is on the free space profile line or not. In this case, the product of the calculation unit 8 of the object detector is mainly information about the position and nature (for example, the distance of the detected object from the front of the train or the type of the detected object). All operations are performed in real time, regardless of the speed of the moving vehicle.

[0016] The computing unit 8 of the object detector connects not only to its own subsystems and modules, i.e. lidars 1, stereo cameras 2, thermal cameras 3 and camera 4, power module 5 and module 7, but also with associated surrounding rail subsystems.

[0017] These subsystems include in particular the train decision expert system 11 (and therefore the downstream automatic train control system 12). The computing unit 8 sends status information and detected object information to the expert decision-making system 11 so that it can decide to take the appropriate action. In addition, it is possible for the object detection device 10 to send visual information to downstream systems outside the train, namely to remote control and control of the train (monitoring terminal 13 of the driverless train dispatcher).

Industrial applicability

[0018] In addition to the main use of this device in a train without a driver (GoA4), where the driver is not present on board, the device can also be used when the train is operated at lower levels of automation or in degraded conditions. In these cases, it will serve as an assistance system for the driver or other authorized persons present on the train, or for authorized persons outside the train.

List of reference numbers

[0019] 1 - Group of lidars

- Stereo camera

- Thermal camera

- High resolution camera

- Power supply and control module

- Console

- GNSS module

- Computing block — Command console

- Object detector

- Expert train decision making system

- Driverless automatic train control system

- Driverless train dispatcher monitoring terminal

Claims (2)

1. A railway train device for detecting objects and their subsequent classification, including detection sensors connected to a computing unit (8) with a module (7) of a global navigation satellite positioning system, characterized in that the console (6) , located at the front of the railway vehicle contains sensors including a group of lidars (1), a stereo camera (2), a thermal camera (3) and a high-resolution camera (4), together powered by a power and control module (5), which are connected to a control panel (9) inside the rail vehicle with a computing unit (8) and module (7) of the global navigation satellite system, wherein the computing unit (8) for detecting and classifying objects is connected to the expert train system (11) for decision-making, to the automatic train control system (12) and to the monitoring terminal (13) of the train dispatcher. 2. The device according to claim 1, characterized in that the computing unit (8) is adapted to provide information about the position and nature, including the distance of the detected object from the front of the train and the type of the detected object.