RU2719499C1 - Method, device and railway vehicle, in particular, rail vehicle, for recognition of obstacles in railway connection, in particular in rail connection - Google Patents

Abstract

FIELD: railway automatics.

SUBSTANCE: invention relates to railway automation for automatic detection of obstacles on the track. In the engineering solution on multi-image base (BI_FSB) located in front of the railway vehicle (BFZ, SFZ) of the haul section (FSB) in the image area (BIB) marked on each of the images, which shows track (FS, GL) used by railway vehicle (BFZ, SFZ), by means of images analysis is recognized by graphically positioned by marking the track and matched with stored in memory known graphical meta information (BMI) or with stored in memory known graphical metainformation (BMI) and additional information (ZI), and on a fragment (BIB_AS) of the marked area of the image, the object recognition method is used to identify whether an object (OBJ) is on the track (FS, GL), wherein in the image area (BIB), preferably in the segment (BIB_AS) of image region an obstacle is marked when object recognition (OBJ) is detected by object recognition method.

EFFECT: higher reliability of automatic recognition of obstacles on the route.

20 cl, 2 dwg

Description

The invention relates to a method for recognizing obstacles in railway communication, in particular in rail communication, according to the restrictive part of claim 1, a device for recognizing obstacles in railway communication, in particular in rail communication, according to the restrictive part of claim 10 and a railway vehicle for detecting obstacles in a railway message, in particular a rail vehicle for recognizing obstacles in a railway message , according to the restrictive part of paragraph 24 of the claims.

Railway vehicles, as an integral part of modern transport infrastructure, are gauge vehicles and vehicles that move, for example, rolling along or under one or two guide rails (rail tracks), hovering over or under a magnetic field or hanging on steel cables. Of the above-mentioned gauge means of communication and transport, the most widespread are rail vehicles based on a wheel-rail system that either (have) their own chassis drive (motor car), or are pulled or moved by a locomotive, and which mainly have steel wheels having a flange bandages are sent along two steel rails or, respectively, rail tracks.

The object of the invention is to propose a method, device and railway vehicle, in particular a rail vehicle, for automatically recognizing obstacles in a railway message, in particular in a railway message, by which obstacles in a railway message are automatically recognized when railway vehicles are on the way on railway lines in the railway network, or, accordingly, obstacles in the rail communicating when rail vehicles are en route on rail hauls in a rail network.

Automatic recognition of obstacles in railway traffic, in particular in rail traffic, which is the subject of this international patent application (application No. PCT / ...; publication No. WO ...) and substantiating its priority for patent application DE (application No. 102016224344.6), taking into account the future automated (autonomous) or with the participation of the driver of the railway vehicles in railway traffic or, accordingly, rail vehicles in rail traffic is an absolute necessity.

So, for automated or with the participation of the driver of the railway / rail vehicles, it is necessary to recognize moving or stationary objects and people in the area of the track / rail track. At the same time, it is necessary to distinguish between acceptable objects and people (e.g. track emphasis on a track / rail track, a maintenance worker next to a track / rail track) from unacceptable objects and people (e.g. a tree torn from the root or children playing).

The problem of automated, or with the participation of a driving operator, has so far been realized with the help of costly additional investments in the haul infrastructure, such as induction loops, computers along the haul and the communication system between the train and the haul components. In addition, special protective fences are used to avoid access to the rail track (e.g. known by airports).

But not only the aspect of automatic recognition of obstacles is important for the future of automated (autonomous) or with the participation of the driver, but also the subsequent technical aspects, all of which are more or less in the technical context with this patent application and therefore are given in In this regard, their contents should be taken into account and, if necessary, even included in it.

These are the following aspects:

1) Automated signal recognition in railway / rail traffic in accordance with the international patent application (application No.PCT / EP2016 / 057804; publication No. WO 2017/174155 A1) and the technical theory disclosed therein.

2) Automatic recognition of dangerous situations in railway / rail traffic in accordance with DE patent application (application No. 102016224358.6) and the international patent application (application No. PCT / ...; publication No. WO ...) and the technical theory disclosed in each of them.

3) Automatic recognition of gauges / rail tracks in railway / rail traffic in accordance with DE patent application (application No. 102016224335.7) and the international patent application (application No. PCT / ...; publication No. WO ...) and the technical theory disclosed in each of them .

4) An alternative definition of provisions in rail communication when the traditional definition of provisions with satellite support does not work or is insufficient, in accordance with DE patent application (application No. 102016224355.1) and the international patent application (application No. PCT / ...; publication No. WO ... ) and the technical theory disclosed in each of them.

5) Performing an analysis of images based on a track / rail track in a rail / rail service in accordance with DE patent application (application No. 102016224331.4) and the international patent application (application No. PCT / ...; publication No. WO ...) and disclosed in each of them technical theory.

The above contextual problem is solved on the basis of the method for recognizing obstacles defined in the restrictive part of claim 1 of the claims using the features specified in the distinctive part of claim 1 of the claims.

In addition, the aforementioned contextual task is solved based on the device for recognizing obstacles defined in the limiting part of clause 8 of the claims of the invention using the features indicated in the characterizing part of clause 8 of the claims.

Further, the above-mentioned contextual problem is solved on the basis of a railway vehicle, in particular a rail vehicle, defined in the restrictive part of clause 20 of the claims, using the features specified in the distinctive part of claim 20 of the claims.

The idea underlying the invention according to independent claims 1, 8 and 20 of the claims is that on the basis of several images of a certain driving area located in front of a railway vehicle in a certain image area marked on each of the images, which essentially shows the track used by the railway vehicle, by analyzing the images, recognize the track graphically positioned by marking and coordinate it with the stored track using meta-information or with known graphic meta-information and additional information stored in memory, and using the object recognition method, recognize on some fragment of the marked area of the image whether there is any object on the track, such as, for example, a person, an animal, a fallen tree, etc. while an obstacle is marked on the image area, preferably in a fragment of the image area, when an object is recognized by the object recognition method. At the same time, graphic meta-information literally contains data of signs and properties of registered images of the haul section.

In this case, the basic principle of the invention is to use metadata about the haul, for example, the haul, in combination with sensors in a railway vehicle, as well as calculation and analytical estimation algorithms to improve recognition of objects and people and to distinguish between acceptable and invalid objects and people.

At the same time, the goal is to make it possible to contribute to fully automated management without additional investment in the haul infrastructure.

To recognize objects and people on a track / rail track or in a certain critical area near a track / rail track, pattern recognition algorithms or pattern comparison algorithms (the so-called pattern matching algorithms) are applied loudly for the driving area, the area “in front of the vehicle” (in direction).

1. For the recognition of objects or people on the track at a great distance, these algorithms work inefficiently, because only a small part of the image is relevant.

2. These algorithms are not able to distinguish valid objects and people from invalid objects and people.

In the automotive field with a focus on roads in accordance with US 6,405,128 B1, an analytical assessment of the so-called “electronic horizon” is known.

Automated recognition of objects and tracks / rail tracks, as well as the distinction between reliable and unreliable objects / people, can preferably be at least partially achieved using the following steps:

A. In the first step, in a railway vehicle, several devices for recording images (eg sensors) of various kinds (eg, video camera, laser sensors, infrared camera, thermal cameras for taking images, radar devices, other devices for collecting images and etc.) to create images or other information about the railway / rail haul in front of the railway / rail vehicle.

So, for example, a radar for detecting metal objects, even in bad weather, can be combined with video cameras and image collection devices, such as thermal cameras for capturing images for recognizing people.

B. In the second step, the current rail track is marked on each image by analyzing the images using external meta-information.

Option 1

In a video image or a similar video image, this can be done using edge recognition algorithms based on the rails immediately in front of the vehicle. When applying additional information, such as rail plans, cartographic material, etc., this recognition can be performed more reliably. In this regard, we refer to the analysis of images based on the track / rail track in the railway / rail service in accordance with the DE patent application (application No. 102016224331.4) and the international patent application (application No. PCT / ...; publication No. WO ...) and disclosed in each of them a technical theory.

Option 2

In the image based on the radar, this can be carried out on the basis of knowledge of the driving distance (the track / rail track relative to the geographical position is known).

C. In the third step, when focusing on the track / rail track for each image recording device used (for example, an image pickup device), object recognition methods recognize whether an object or person is on the track / rail. This means that only a fragment of the image containing the track line / track track and the critical area to the left and right of it are considered. In this case, depending on the device for collecting images, one or both of the following methods of matching patterns are used.

And here the quality of recognition is improved by integrating external information or additional information.

C1. Positive agreement

It is checked whether there are patterns in the relevant image fragment that (relate) to people or objects, such as fallen trees or overtaking railway vehicles, for example, rail vehicles, trains. If yes, an obstacle or potential obstacle is marked.

C2. Negative agreement

It is checked whether the expected pattern, such as, for example, from a solid rail track or from regular rail beams, is recognized in the image or, respectively, images. If this is not the case, then with the help of the image data bank it is checked whether this violation of the law was expected (this information can, for example, be prepared by the driver of a motor car when traveling for the purpose of initialization). If a violation of the law was not expected, a potential obstacle is marked.

D. The fourth step combines the result of marking obstacles from various imaging devices. And here, for example, when applying probabilistic methods of image processing, such as Hiden-Markov models, various sources of information are combined to minimize false recognition and eliminate “false negatives”, that is, false assumption that there is no object on the track / rail, although it does exist.

Using the foregoing analysis of the images of the stage in front of the railway vehicle, it can be achieved that:

- objects and people in the relevant track / rail area are recognized more efficiently than before;

- it is possible to distinguish permissible objects and people in the area in front of the railway vehicle / rail vehicle (but it is outside the track or, accordingly, the track and some critical area to the left and to the right of it) from unacceptable objects and people on the track on a rail track or in a critical area to the left and right of it;

- objects and people under adverse conditions of visibility can be recognized more reliably than by engine drivers;

- Motor car drivers are no longer needed for obstacle recognition, so the track track / track track can be recognized regardless of their presence.

During one of the preferred refinements of the invention, the following additional components a) to c) for an image recording apparatus (e.g., an image pickup apparatus) can also be applied to the obstacle recognition apparatus of claim 8 as follows:

a. The corrective component according to claim 15, which attaches weather and light data for analytical evaluation of graphic material. In this case, for example, in heavy fog, the analytical assessment of video images may be limited to the first 50 meters in front of a railway vehicle or, accordingly, a rail vehicle, and the vehicle speed may be throttled accordingly.

b. The focal length change component according to claim 16, which, depending on the environment (e.g., station, urban area, countryside, etc.) and speed, selects the correct shooting angle in order to optimally support the analytical assessment of the image. For example, then shooting situations on a free stretch (require images from a great distance so that you can react in a timely manner based on speed) and shooting situations in the station area (require images having a large width) can be properly serviced. Additionally, when merging graphic data and haul data, you can focus on particularly interesting areas, such as, for example, a railway crossing.

c. The lighting component according to claim 17, for example, a spotlight that works within or outside the human-visible area, which improves the quality of the graphic material taken by the image recording device or, accordingly, the device for collecting images at night or in bad weather conditions .

d. A roadside analytical assessment station, which is linked via mobile communication and receives images from the device for storing images for which an analytical assessment is possible only with a high uncertainty coefficient. Then these images can be analytically evaluated by a human expert, and then this information can again be returned to the device for storing images, which can be located (in) a device for recognizing obstacles in a railway / rail vehicle (option "A") or outside the device for recognition of obstacles should be intended for him, for example, in the form of a data bank in a railway / rail vehicle or, for example, in the form of a data cloud.

1. With sufficient communication bandwidth and the availability of expert people, this can be done even in real time so that the result of the analytical assessment can be used to drive a rail / rail vehicle.

2. In addition, with the help of a roadside analytical assessment station, graphic material of rail vehicles of one representative sample or several representative samples can be coordinated and distributed.

e. A mobile device of a train driver or a comparable railway employee who, for the purpose of arranging passengers, somehow rides a rail vehicle and, similarly to point d), evaluates images having a high uncertainty coefficient.

In addition, it is possible for an obstacle recognition device to be implemented and function as a virtual machine as “Software Defined Signal Recognition of Rail Traffic Systems” (software-defined recognition of signals from railway systems).

Other advantages of the invention arise from the following description of one embodiment of the invention using FIGS. 1 and 2. They show:

figure 1: recognition on the basis of a railway vehicle obstacles in the form of a tree that fell on a railway stage;

figure 2: the basic design of the device for recognizing obstacles for recognizing obstacles on the basis of the railway transport in the form of a tree that fell on a railway line in accordance with figure 1 obstacles.

Figure 1 shows the recognition on the basis of the railway vehicle of the obstacle in the railway communication BVK, when the railway vehicle BFZ approaches along the track FS of the railway track BST to the object OBJ located in the form of an obstacle on the track FS on the track BST of the railway network BNE, shown in separate sections; in the case shown, a tree falling on the FS rut.

In the present exemplary embodiment, the tracked railway line BST of the railway network BNE is a rail track SST of the rail network SNE on which, in the rail message SVK for detecting obstacles on the basis of the rail vehicle, the rail vehicle SFZ is on the track on the track GL and approaches to an object OBJ, which is in the form of an obstacle on the track GL, in the illustrated case, a tree falling on the track FS. Instead of the illustrated SVK rail service with the SFZ rail vehicle traveling along the SST rail network of the rail network SST, as already discussed, any other, any near or long-distance rail system is also possible and imaginable. So, one could also talk, for example, about a magnetic suspension system for communication of a suspension road (keyword: Transrapid, Maglev, etc.), which has a correspondingly comparable infrastructure consisting of a railway network, a railway stage and a railway vehicle.

In the rail communication system of FIG. 1, in a motor vehicle TRW of an SFZ rail vehicle having a motor vehicle driver TFS console TFS and an integrated display device AZE in which a workstation of a vehicle operator FZF is located, a HEV device is located for detecting an obstacle on the basis of the rail vehicle recognition of obstacles. To this end, the HEV device for detecting obstacles includes preferably a sensor-based BAZG device for recording images, which is made, for example, in the form of a conventional video camera, a laser sensor, a thermal camera for capturing images, a radar device, an infrared camera, etc., and in connection with image collection, it is also called an image collection device.

With the BAZG image recorder, when the SFZ rail vehicle traveling along the GL track approaches an obj object located on the GL track as an obstacle, in the case shown a fallen tree, from an SFZ rail vehicle, for example, from the perspective of the FZF driver a motor car on a TFS console of a motor vehicle driver TRW and / or from a fixed vehicle providing visibility of a position track in or on an SFZ vehicle from an SFZ located in front of the rail vehicle, preferably As the speed of the rail vehicle SFZ is followed by a haul area FSB, a plurality of BI _FSB images representing this haul area FSB can be recorded.

In the _FSB BI images of the haul FSB region, there is an image BIB region having a fragment BIB _{AS of} the image region that represents the GL track used, as well as some critical region for the SVK rail message by which the critical critical region for the SVK rail message is indicated substantially left and to the right of the track GL in the portion of the driving distance _FSB region shown by the BI image area BIB of the FSB driving distance region FSB.

Those. the hopping region FSB also includes a region critical for the SVK rail message.

As now, based on the BI _FSB images of the driving region _FSB region having the image region BIB contained therein and the image region fragment BIB _AS , the obstacle recognition is performed, explained below using the description of FIG. 2.

FIG. 2 shows the principal construction of the obstacle recognition device HEV for recognizing, according to FIG. 1, based on an obstacle rail vehicle of an SFZ railway vehicle that is on the track on the track GL and approaches an obstacle on the track GL OBJ object, in the illustrated case, a fallen tree.

In this case, the starting point for the recognition of obstacles in accordance with the reasoning to figure 1, is the device BAZG for recording images, which registers the image BI _FSB area FSB overtaking for recognition of obstacles.

To this end, the BAZG device for recording images is preferably rotatable for orientation on the image object.

Further, it is possible, and under certain circumstances, for technical reasons related to registration, it is advisable that the HEV device for detecting obstacles contains several BAZG devices for recording images of the same design, for example, several video cameras, or devices of various designs, e.g. several video cameras , laser sensors, radar-based sensors based on wireless location and distance measurement, infrared cameras and / or thermal cameras for shooting images that capture images Nii BI _FSB . Such multiple recording of images or, accordingly, collection of images may be relevant, including for the purposes of redundancy.

To further improve the quality of images recorded or collected using the BAZG image recording device, the following components are preferably contained in the BAZG image recording device:

1. The correction component KOK, with the help of which the data on weather and illumination are attached for the analytical assessment of the graphic material. Using this component, it is possible, for example, in heavy fog, to limit the analytical assessment of video images to the first 50 meters in front of the rail vehicle and, accordingly, to speed the speed of the rail vehicle.

2. The BVK component changes the focal length, which, depending on the environment (eg, station, urban area, countryside, etc.) and speed, selects the correct shooting angle in order to optimally support the analytical assessment of the image. Due to this, then both shooting situations on a free stretch (require images from a great distance so that you can react in a timely manner based on speed) and shooting situations in the station area (require images having a large width) can then be properly serviced. Additionally, when merging graphic data and haul data, you can focus on particularly interesting areas, such as, for example, a railway crossing.

3. The BLK component of lighting, made, for example, in the form of a searchlight, which works within or outside the human visible area, and with the help of which the quality of the graphic material taken by the device for recording images or, accordingly, the BAZG device for collecting images at night or at bad weather conditions.

Images taken in this way are saved by the BAZG device for recording images in the BSPE device for storing images. This BSPE device for storing images is either, in accordance with option “A”, as a component of the HEV device for recognizing obstacles respectively connected to the BAZG device for recording images, or, in accordance with option “B”, outside the HEV device for recognizing obstacles, for example ., in the form of a data memory bank, in a motor car or in a data cloud, is intended for the BAZG device for recording images or, accordingly, can be connected to it.

For the analytical evaluation of recorded or, respectively, collected images for recognition of objects that constitute obstacles to the rail communication by rail, for example, a tree falling on the rail in accordance with Fig. 1, the BAZG device for recording images is connected to the computational / analyzing BAWE, which is also one of the components of the HEV obstacle recognition device. For this purpose, the BAWE computing / analyzing device, like the BAZG device for recording images, is either connected to the BSPE device for storing images in accordance with option “A”, or, in accordance with option “B”, is intended for the BSPE device for storing images or, accordingly, can be connected to it. Thus, a block of individual functions arises from the BAWE computing / analyzing device, the BAZG device for recording images, and the BSPE device for storing images in which the named components of the HEV obstacle recognition device interact with separate functions for the computational / analytical evaluation of obstacle recognition.

In order to form a complete functional block for the recognition of obstacles supported by calculations / analytical evaluation, in which the individual blocks that are its parts interact functionally, the named block of separate functions is expanded by another separate block, the IDB data bank. In this case, the information data bank IDB can, for example, be integrated with the BSPE device for storing images as a structural unit in one common storage device. This memory device, not shown explicitly in FIG. 2, can, in turn, in the same way as the BSPE device for storing images, or, in accordance with option “A”, as a component of the obstacle recognition device HEV, be respectively connected to the BAZG device for recording images and a BAWE computing / analyzing device, or, in accordance with option “B”, outside the HEV device for detecting obstacles, in a motor vehicle or in the data cloud, be intended for the BAZG device for recording images and a computing / BAWE analyzing device or, accordingly, can be connected to it. In this regard, we refer to a device for storing information in the DE patent application (application No. 102016224355.1) and the corresponding international patent application (application No. PCT / ...; publication No. WO ...) on an alternative definition of a position in a rail message, when the traditional definition of provisions with satellite support does not work or is insufficient.

In the IDB data bank, along with the graphic BMI meta-information, which literally contains the characteristics and properties of the FSB driving region recorded on BI _FSB images, additional ZI information is stored, such as, for example, rail plans or cartographic material, etc. In accordance with the image in FIG. 2, the IDB data bank is designed for the HEV device for detecting obstacles or, accordingly, can be connected to it in such a way that the computing / analyzing device The BAWE facility for computational / analytic assessment of obstacle recognition accesses the BMI graphic meta information and additional ZI information stored in the IDB information bank. For this, the IDB data bank is preferably located outside the HEV device for detecting obstacles, for example, in the form of a data bank, in a motor car or in the form of a data cloud.

For computation / analytic evaluation of obstacle recognition, the BAWE computing / analyzing device preferably has a non-volatile, readable memory SP, in which the processor-readable control program commands of the obstacle recognition control program module PGM are stored, and the processor PZ, which executes the control program commands of the PGM program module for the supported computations / analytical assessment of obstacle recognition. For this, the PZ processor additionally, along with access to the graphic meta-information BMI and additional information ZI in the IDB data bank for control and reading data, gets access to the BAZG device for recording images and the BSPE device for storing images.

The BAWE computing / analyzing device or, respectively, the PGM software module having a PZ processor executing the control program instructions of the PGM software module for computationally / analytically assessed obstacle recognition, is now applied to computationally / analytically evaluated obstacle recognition so that each from the BI _FSB images, the image BIB area is marked, which shows the GL track used by the SFZ rail vehicle, in this case, by analyzing the images, the graphically positioned rail track GL of the SFZ rail vehicle is recognized and matched with either the known BMI graphic meta-information stored in memory or the known MMI graphic meta-information (BMI?) stored in memory and additional information ZI.

Image analysis and marking is preferably carried out using edge recognition algorithms, and on the basis of the track GL track recorded in the FSB region in the image BIB region, the course of the GL track used by the SFZ rail vehicle is detected by the fraction changing on the recorded image images of the GL rail track with respect to the entire recorded image.

In addition, it is preferable when BI _FSB images are taken using radar-based sensors based on wireless location and distance measurement, image analysis and, at the same time, marking is performed on the basis of knowledge of the GL track used, because the course of the GL track used is relative to the geographical provisions known.

When, by analyzing images, the graphically positioned rail track GL of the SFZ rail vehicle is recognized and matched with either the BMI known graphic meta-information stored in the memory or the BMI known graphic meta-information stored in the memory and additional information ZI, then for the fragment BIB _{AS of the} marked image BIB region , which represents the GL track used, as well as the region critical for the SVK rail message, is recognized by the object recognition method, Is there any OBJ object on the track GL, such as, for example, a person, an animal, a fallen tree, etc., while, when an OBJ object is recognized by the object recognition method, an obstacle is marked in the image BIB, for example when it is in a fragment of the BIB _{AS of} the image area and / or when it is a potential obstacle.

Using the pattern recognition method, a comparison of samples is carried out, based on a positive comparison and / or a negative comparison, in the case of a positive comparison, it is checked whether the object-specific samples are contained in the BIB _AS fragment of the image area, and if the comparison is negative, it is checked whether the BIB fragment contains _{AS of} the image area is any expected pattern, for example, a solid, used by a rail vehicle SFZ rail track GL or any pattern that forms FS track gauges or track rails between parallel tracks GL.

If, in a negative comparison, the check ends with the result “NO”, then the detected violation of the pattern is consistent with respect to its expectation, used as reference information and taken earlier during the runs in order to initialize the images of the haul on the stage, while when the violation of the pattern was not expected, it is marked an obstacle in the BIB region of the image, for example, in a fragment of the BIB _{AS of} the image region and / or as a potential obstacle.

Obstacle markings made for all BI _FSB images in each BIB image area or, respectively, in the BIB _AS fragment of the image area are preferably combined for combining various image processing image sources using image processing methods, such as, for example, Hiden-Markov models . Thereby, for example, minimization of the probability of false recognition and prevention of the occurrence of “false negatives” can be achieved, i.e. false assumptions that there is no object on the track / rail track, although it does exist.

In addition, for an obstacle recognition device HEV having an integrated or dedicated BSPE device for storing images, for such images for which an analytical assessment is possible only with a high uncertainty factor, an analytical assessment roadside station AWS is provided which is coupled to the image storage device via mobile communication and receives from it the images stored there for a modified analytical evaluation. Then these images can be analytically evaluated by a human expert, and then this information can again be returned to the BSPE device for storing images.

1. With sufficient communication bandwidth and the availability of expert people, this can be done even in real time so that the result of the analytical assessment can be used to drive a rail / rail vehicle.

2. In addition, the graphical material of rail vehicles of one representative sample or several representative samples can be coordinated and distributed using the AWS roadside analytical station.

Alternative to the AWS analytical assessment station for modified analytical assessment of images for which analytical assessment is possible only with a high uncertainty factor, it is also possible for a train driver or a comparable railway employee who, in order to arrange passengers, somehow rides a rail vehicle with Using a mobile device, I evaluated images with a high coefficient of uncertainty, as a human expert on images in AWS stations does. analytical assessment.

With the help of the HVE device for detecting obstacles described above, it is possible to assist or, accordingly, even implement an automated (autonomous) or with the participation of the driver of the railway vehicle BFZ or, accordingly, the railway vehicle SFZ without additional infrastructure for a certain route. This is the case, in particular, when the HVE device for detecting obstacles is implemented as a virtual machine, which is designed and functions as a “Software Defined Signal Recognition of Rail Traffic Systems”.

Claims (56)

1. The method of recognition of obstacles in railway traffic (BVK), in particular in rail traffic (SVK), characterized in that a) from a railway vehicle (BFZ), in particular a rail vehicle (SFZ), in particular from the perspective of the driver (FUF, TFS, TRW) of a motor car and / or from a stationary vehicle providing a visible track gauge (BFZ, SFZ) or on it, a plurality of images (BI _FSB ) are recorded in front of the railway vehicle (BFZ, SFZ), in particular the track-oriented area (FSB) of the track following area representing the track driving area (FSB) ania; b) on each of the images (BI _FSB ) mark the region (BIB) of the image, which shows the track (FS) used by the railway vehicle (BFZ, SFZ), in particular the track (GL), while graphically positioned markings are recognized by image analysis the track (FS, GL) of a railway vehicle (BFZ, SFZ) and coordinate either with the known graphic meta-information (BMI) stored in memory or with the known graphic meta-information (BMI) stored in memory and additional information (ZI), such as p spans plans or maps and charts; c) for the fragment (BIB _AS ) of the marked region (BIB) of the image, which represents the track used (FS, GL), as well as the region critical for railway traffic (BVK, SVK), the object recognition method recognizes whether it is on the track (FS, GL) any object (OBJ), such as, for example, a person, an animal, a fallen tree, etc., while when an object (OBJ) is recognized by the object recognition method, an obstacle in the image area (BIB) is marked, preferably in a fragment ( BIB _AS ) image areas, and / or as a potential obstacle . 2. The method according to claim 1, characterized in that images (BI _FSB ) are taken with several instruments (BAZG) for recording images of various designs, for example, video cameras, laser sensors, radar-based sensors based on wireless location and distance measurement, an infrared camera and / or thermal cameras for taking images. 3. The method according to claim 1 or 2, characterized in that image analysis is performed using edge recognition algorithms, and, based on the track gauge (FS, GL) recorded in the area (FSB), the track (FS, GL) used by the railway vehicle (BFZ, SFZ) is recognized in the image area (BIB) ) according to the proportion of the track image (FS, GL) that varies on the registered image with respect to the entire registered image. 4. The method according to claim 2, characterized in that when images (BI _FSB ) are taken using radar-based sensors based on wireless location and distance measurement, image analysis is performed based on the knowledge of the track used (FS, GL), because the track used (FS, GL) relative to the geographical position is known . 5. The method according to one of claims 1 to 4, characterized in that using the object recognition method, samples are compared based on a positive comparison and / or negative comparison, while a) in a positive comparison, check whether the object-specific samples are contained in the fragment (BIB _AS ) of the image area, and b) in a negative comparison b1) check whether in the fragment (BIB _AS ) of the image area there is any expected pattern, preferably a solid track used by a railway vehicle (BFZ, SFZ) (FS, GL) or any pattern that is formed by track beams (FS ) or, respectively, the rails of the track between parallel passing tracks (GL); b2) if this check ends with the result “NO”, then the detected violation of the laws shall be coordinated in relation to its expectation with those used as reference information and taken earlier during the runs with the aim of initializing the images of the haul on the stage; b3) if a violation of the pattern was not expected, an obstacle is marked in the image area (BIB), for example, in a fragment (BIB _AS ) of the image area and / or as a potential obstacle. 6. The method according to one of claims 1 to 5, characterized in that obstacle markings made for all images (BI _FSB ) in each area (BIB) of the image or, accordingly, in the fragment (BIB _AS ) of the image area are preferably combined for combining various image sources of image processing using image processing methods, such as, for example Hidden-Markov models in order to minimize false recognition and eliminate “false negatives” (false negatives), i.e. false assumptions that there is no object (OBJ) on the track or, accordingly, the rail track, although it really is. 7. The method according to one of claims 1 to 6, characterized in that In this way, they assist the automated (autonomous) or with the participation of the driver driving a railway vehicle (BFZ, SFZ) without additional infrastructure for a certain route. 8. A device (HEV) for detecting obstacles in railway traffic (BVK), in particular in rail traffic (SVK), characterized in that it contains a) at least one device (BAZG) for recording images, with which from a railway vehicle (BFZ), in particular a rail vehicle (SFZ), in particular from the perspective of the driver (FZF, TFS, TRW) of a motor vehicle and / or from a fixed position providing a track gauge in a vehicle (BFZ, SFZ) or on it, a plurality of images (BI _FSB ) of a certain front of a railway vehicle (BFZ, SFZ) can be recorded, in particular oriented to the speed of a railway vehicle means (BFZ, SFZ) of the region (FSB) of the following, representing this area (FSB) of the following, and can be stored in the device (BSPE) for storing images; b) a computing / analyzing device (BAWE), which, when connected and operatively interacting with an image recording device (BAZG), an image storage device (BSPE) and an information data bank (IDB), preferably both, a device (BSPE) for storing images and an information bank (IDB) of data, in the form of a structural unit integrated into one common memory device, is made in this way, in particular has a non-volatile, readable memory (SP) in which readable processors are stored Team manager obstruction recognition program module of the control program (PGM), and a processor (PZ), which executes instructions of software module of the control program (PGM) for the supported calculations / analysis obstacle detection estimate that each of the images (BI _FSB) marked area (BIB ) an image that shows the track (FS) used by a railway vehicle (BFZ, SFZ), in particular a track (GL), while a graphically positioned m by marking the track (FS, GL) of the railway vehicle (BFZ, SFZ) and is consistent either with the known graphic meta-information (BMI) stored in memory or with the known graphic meta-information (BMI) stored in memory and additional information (ZI), such as for example, haul plans or cartographic material; moreover c) the computing / analyzing device (BAWE) is designed in such a way that for the fragment (BIB _AS ) of the marked area (BIB) of the image, which represents the track used (FS, GL), as well as the critical region for railway traffic (BVK, SVK), by the method of recognition of objects it is recognized whether any object (OBJ) is located on the track (FS, GL), such as, for example, a person, an animal, a fallen tree, etc., while when the method of recognition of objects is recognized by an object (OBJ), it is marked image obstruction (BIB), preferred in a fragment of (BIB _AS) area of the image, and / or as a potential obstacle. 9. The device (HEV) according to claim 8, characterized in that contains several devices (BAZG) for recording images of various designs, for example, several video cameras, laser sensors, radar-based sensors based on wireless location and distance measurement, an infrared camera and / or thermal cameras for shooting images that capture images (BI _FSB ) 10. The device (HEV) according to claim 8 or 9, characterized in that the computing / analyzing device (BAWE) is designed in such a way that the image analysis is performed using edge recognition algorithms, and, based on the track following (FS, GL) track spacing recorded in the region (FSB), the path used by the rail is recognized vehicle (BFZ, SFZ) gauge (FS, GL) according to the proportion of the gauge (FS, GL) changing on the registered image in relation to the entire recorded image. 11. The device (HEV) according to claim 9, characterized in that The computing / analyzing device (BAWE) is designed in such a way that when the images (BI _FSB ) are captured using radar-based sensors based on wireless location and distance measurement, image analysis is based on the knowledge of the track used (FS, GL), therefore that the track used (FS, GL) relative to the geographical position is known. 12. The device (HEV) according to one of claims 8 to 11, characterized in that Computing / analyzing device (BAWE) is designed in such a way that the method of object recognition is used to compare samples based on a positive comparison and / or negative comparison, while a) in a positive comparison, it is checked whether the object-specific samples are contained in the fragment (BIB _AS ) of the image area, and b) in a negative comparison b1) it is checked whether in the fragment (BIB _AS ) of the image area there is any expected sample, preferably a solid track used by a railway vehicle (BFZ, SFZ) (FS, GL) or any pattern that is formed by track beams (FS ) or, respectively, the rails of the track between parallel passing tracks (GL); b2) if this check ends with the result “NO”, then the detected violation of the laws shall be coordinated in relation to its expectation with those used as reference information and taken earlier during the runs with the aim of initializing the images of the haul on the stage; b3) if a violation of the pattern was not expected, an obstacle is marked in the image area (BIB), for example, in a fragment (BIB _AS ) of the image area and / or as a potential obstacle. 13. The device (HEV) according to one of claims 8 to 12, characterized in that The computing / analyzing device (BAWE) is designed such that obstacle markings made for all images (BI _FSB ) in each area (BIB) of the image or, respectively, in the fragment (BIB _AS ) of the image area are preferably combined to combine different image sources image processing using image processing methods, such as, for example, the Hidden-Markov model, in order to minimize false recognition and eliminate “false negatives”, i.e. false assumptions that there is no object (OBJ) on the track or, accordingly, the rail track, although it does exist. 14. The device (HEV) according to one of claims 8 to 13, characterized in that the device (BAZG) for recording images is made with the possibility of rotation. 15. The device (HEV) according to one of claims 8 to 14, characterized in that the instrument (BAZG) for recording images has a correction component (KOK), which combines weather and light data for analytical evaluation of graphic material. 16. The device (HEV) according to one of claims 8 to 15, characterized in that the image recording apparatus (BAZG) has a focal length component (BVK) which, depending on the distance to the track (FS, GL), selects the correct shooting angle in order to optimally support multiple analytical evaluations. 17. The device (HEV) according to one of claims 8 to 16, characterized in that an image recording apparatus (BAZG) has an illumination component (BLK), in particular a searchlight, which operates within or outside the human visible region. 18. The device (HEV) according to one of paragraphs.8-17, characterized in that it contains a virtual machine that is implemented and functions as a "Software Defined Signal Recognition of Rail Traffic Systems" (software-defined signal recognition of railway systems). 19. The device (HEV) according to one of claims 8 to 13, characterized in that with the help of the device (HEV), it is possible to assist the automated (autonomous) or with the participation of the driver driving a railway vehicle (BFZ, SFZ) without additional infrastructure for a certain route. 20. Railway vehicle for detecting obstacles in railway traffic (BVK), in particular a rail vehicle (SFZ) for detecting obstacles in rail traffic (SVK), characterized in that The device (HEV) for detecting obstacles according to one of claims 8-19 is integrated into a railway vehicle (BFZ, SFZ).

Images