EP4029758A1 - Safety-critical on-board monitoring of the environment of a rail vehicle - Google Patents

Abstract

A method for on-board monitoring of the surroundings (U) of a rail vehicle (2) is described. In the method, sensor data (SD) from a monitoring area (BU) that is adjacent to a track section (GA-G2) of a track (G2) that is adjacent to a track (G1) on which the rail vehicle (2) is moving, as well as its Environment (U) includes detected. In addition, an object (O, L, 2a) is detected and localized in the monitored area (BU) on the basis of the sensor data (3D-SD). It is then determined whether the detected object (O, L, 2a) could pose a risk to rail traffic. This determination takes place depending on the recorded sensor data (3D-SD) and the position (P_O) and the dimensions (AB, b_2a) of the detected and localized object ( O, L, 2a). An environmental monitoring device (30) is also described. Furthermore, a rail vehicle (2) is described.

Description

The invention relates to a method for onboard monitoring of the surroundings of a rail vehicle. The invention also relates to an environment monitoring device. The invention also relates to a rail vehicle.

Rail vehicle drivers, also referred to as drivers for short, have to fulfill many additional tasks in addition to actually controlling the journey of a rail vehicle. One of the critical tasks of the rail vehicle driver is to observe the surroundings of his rail vehicle and to report any observations that could affect a potential problem for the operation of the rail transport system. This includes, for example, the observation of whether the vegetation growing in the periphery has gotten too close to the tracks or whether other components of the environment have changed and could possibly pose a future hazard. Examples of this are fallen trees, branches that have fallen on the tracks, boulders that have fallen on the tracks and a thick layer of snow in the track area. It should be noted that drivers monitor not only their own track, but also neighboring tracks. Drivers also monitor oncoming or passing trains to identify sources of danger, such as cargo on an oncoming train that has not been properly secured and has therefore shifted. If railway operations are switched to driverless operation, there is no person on board a rail vehicle who could perform the aforementioned tasks. Consequently, the tasks described must be performed by other institutions and/or people.

So far, therefore, a person on board, for example a driver or another person, has always been assigned to take on the task of observing the surroundings. Among the monitoring tasks described above, only vegetation monitoring attempts to accomplish this task through alternative methods. One such alternative method uses a special surveillance vehicle that accurately records the location of vegetation. Alternatively or additionally, the position of the vegetation can also be recorded by satellites or aircraft. Additionally, an onboard video camera may also be pointed at a track to capture video data to determine where vegetation is growing into the track bed or embankment of a track.

However, these methods have the following problems: the use of a vehicle specially designed for surveillance is very expensive, and since few of these special surveillance vehicles are usually available, it is impossible to cover the entire rail network with sufficient frequency , in order to identify problems and dangers promptly and in good time. Image acquisition with satellites is limited in terms of resolution and image acquisition of the rails from above cannot be used between overhanging vegetation, which has no influence on railway operations, and the vegetation, which is located in the height range of the rail vehicles and therefore a danger for this can mean to distinguish. Onboard surveillance of vegetation by onboard cameras is performed with dedicated cameras aimed at the track bed to identify small growing plants. However, such a camera cannot be used to identify any vegetation that grows next to the track bed and may grow into the driving area of the rail vehicles in the future.

The object is therefore to provide a method and a device for monitoring the surroundings, which can at least partially relieve a driver of a rail vehicle of having to monitor the surroundings of the rail vehicle and are at least as effective as a human person.

This object is achieved by a method for on-board monitoring of the surroundings of a rail vehicle according to patent claim 1, an environment monitoring device according to patent claim 12 and a rail vehicle according to patent claim 13.

In the method according to the invention for on-board monitoring of the surroundings of a rail vehicle using a sensor unit of the rail vehicle, sensor data are recorded from a monitoring area which includes a track section of an adjacent track and its surroundings. The sensor data is then checked and evaluated to determine whether there is an object in the monitored area. In this case, an object is therefore detected and localized in the monitored area on the basis of the sensor data recorded. Sensor units are preferably used as the sensor unit, which are also used for detecting potential collision obstacles on the track on which the rail vehicle itself travels. Advantageously, therefore, no additional sensors for monitoring the adjacent track need to be installed on the rail vehicle.

It is then determined whether the detected object could pose a risk to rail traffic, that is to say in particular to a rail vehicle traveling on the adjacent track or to a rail vehicle traveling on its own track. This determination step takes place depending on the sensor data and the position and the dimensions of the detected and localized object.

The position of the track section of the adjacent track is preferably also determined. As an adjacent track is to be understood in this context, a track that is in the monitoring area of the sensors of the rail vehicle and therefore of the rail vehicle while driving or can also be monitored when stationary. In the narrower sense, an adjacent track should be understood to mean a track that is located next to, or more precisely right next to, the track on which the rail vehicle being monitored is driving or standing.

Furthermore, with the method according to the invention for on-board monitoring of the surroundings of a rail vehicle, dimensions of the track section of the track adjacent to the rail vehicle are preferably also determined. The dimensions include in particular the width of the area occupied by the adjacent track. As a rule, track widths are known, but the track body can also be wider than the track width, for example in the area of a switch.

To determine the position of the adjacent track section, a position and orientation of the rail vehicle itself is determined. This position can take place, for example, by self-localization and an independent measurement of the orientation of the rail vehicle. Since the tracks or parts of the track sections can be obscured from the sensors, whether due to a curve or an object lying on the tracks, or poor weather conditions, the tracks or the track on which the rail vehicle is traveling as well as the adjacent track will also be on Localized based on map data.

Determining whether the detected object could pose a risk to rail traffic preferably includes a comparison of the position and dimensions of the detected and localized object with the position and dimensions of the monitored track section of the adjacent track.

This comparison also preferably takes place as a function of the determined position and the dimensions of the monitored track section of the adjacent track. For example, it is determined whether the detected and localized object overlaps or is in the monitored track section protrudes into it or is even located completely in the monitored track section.

Furthermore, a position and corresponding dimensions of a driving channel of a rail vehicle possibly traveling on the adjacent track are also taken into account. In the simplest case, the position and the dimensions of the driving channel of a rail vehicle can be determined using the position and the track width of the adjacent track. The position and the dimensions of a driving channel of the rail vehicle being monitored and/or a rail vehicle moving on the track on which the rail vehicle being monitored is driving can also be taken into account in order to determine whether an object is even protruding into the driving channel on this side. The determination step includes a comparison of the aforementioned variables. If it is determined that the object is in the driving channel of the adjacent track or the track on which the monitoring rail vehicle is moving, a hazard message is generated and sent to a suitable location, for example a central stationary monitoring device, such as a central traffic control center. or passed on to a rail vehicle traveling on the adjacent track.

Advantageously, with the method according to the invention, the rail network can be examined and monitored with regard to disruptive or dangerous objects at a higher frequency than is the case when inspecting with the aid of special monitoring vehicles. In addition, the accuracy of the localization and the resolution of the monitoring achieved are much higher than when using satellite-based monitoring of a rail network, since the sensors installed on a rail vehicle, due to their much greater proximity to the monitored area, can achieve a much higher resolution of the monitored area with comparatively little technical effort can achieve. The monitoring method according to the invention also has advantages over aircraft-based monitoring methods Method has the advantage that a three-dimensional positioning and extent of an obstacle can be determined, while monitoring from a bird's-eye view often only enables two-dimensional data of an object in the track area to be determined. With the method according to the invention, branches or boulders as well as passing rail vehicles can be advantageously monitored in addition to the vegetation located at the edge of the tracks. A further advantage of the method according to the invention is that the on-board sensors already present for detecting obstacles on one's own track can also be used for the inspection of neighboring tracks, so that there is little or no additional hardware outlay. Consequently, the monitoring method according to the invention can be implemented in a particularly resource-saving manner and without much effort.

The environmental monitoring device according to the invention, which is at least predominantly on-board, has a sensor unit for acquiring sensor data from a monitoring area. The monitoring area includes a track section of a track that is adjacent to a track on which a rail vehicle that includes the environmental monitoring device according to the invention is moving, as well as its surroundings. As mentioned, the sensor unit is preferably already present on board the rail vehicle for other monitoring tasks and is only used for monitoring the adjacent track, so that resources are saved and the conversion effort for implementing the environmental monitoring device according to the invention is reduced.

The area monitoring device according to the invention has an object recognition unit for detecting and localizing an object in the area surrounding the adjacent track on the basis of the sensor data. The detection and localization include determining a position and the dimensions of the detected and localized object.

Advantageously, knowledge of the position of the object and of the area occupied by the object, as well as knowledge of the position and the nature of the monitored track section based on the sensor data from the monitoring area, can be used for a subsequent comparison.

For this purpose, the environmental monitoring device according to the invention includes an evaluation unit for determining whether the detected object could pose a risk for rail traffic, ie in particular for a rail vehicle traveling on the adjacent track or for a rail vehicle traveling on its own track. This determination is made by a comparison based on knowledge of the position and dimensions of the detected and localized object and the sensor data.

The surroundings monitoring device according to the invention shares the advantages of the method according to the invention for on-board monitoring of the surroundings of a rail vehicle.

The environmental monitoring device according to the invention preferably comprises a track detection unit for determining a position and dimensions of a track section of a track adjacent to a rail vehicle on the basis of a position and orientation of the rail vehicle and on the basis of map data which depict the track on which the rail vehicle is traveling and the adjacent track. The position of the adjacent track is preferably determined by sensors that are already on board the rail vehicle for other tasks. For example, rail vehicles have sensors for detecting obstacles and identifying such obstacles on their own track in order to prevent a collision with them. Advantageously, the edge area of the sensor field, which includes one or more neighboring tracks, can also be used to carry out an inspection of the neighboring track and its surroundings with regard to interfering objects, in particular obstacles.

In this embodiment, the evaluation unit is set up to determine whether the detected object poses a risk to rail traffic by comparing the position and dimensions of the detected and localized object with the position and dimensions of the monitored track section of the adjacent track. A relative position of the object and the area occupied by the object in relation to the position and extent of the adjacent track section can advantageously be determined and used for a subsequent comparison.

This determination step is therefore carried out by a comparison based on knowledge of the position and dimensions of the adjacent track and/or, if necessary, based on the position and dimensions of a driving channel of a rail vehicle traveling on the adjacent track and based on knowledge of the position and dimensions of the detected object. The position and the dimensions of a driving channel of the rail vehicle being monitored can also be taken into account in order to determine whether an object protrudes into the driving channel on this side. If it is determined that the area occupied by the object and the track section overlap or an object protruding from a railway vehicle traveling on the adjacent track section protrudes from said travel channel and possibly into the travel channel of an adjacent track, so generates a danger message and forwards it to a suitable point, such as a central monitoring device or a traffic control center or a rail vehicle traveling on one of the two tracks for warning.

The rail vehicle according to the invention has the environment monitoring device according to the invention and a data transmission device for transmitting a hazard message to a stationary device, such as a traffic control center, or to another rail vehicle. That rail vehicle according to the invention shares the advantages of the environmental monitoring device according to the invention.

Some components of the environmental monitoring device according to the invention can, if necessary after the addition of hardware systems, such as a sensor unit, for the most part be designed in the form of software components. This relates in particular to parts of the track detection unit, the object recognition unit and the evaluation unit.

In principle, however, these components can also be partially implemented in the form of software-supported hardware, for example FPGAs or the like, particularly when particularly fast calculations are involved. Likewise, the interfaces required, for example when it is only a matter of taking over data from other software components, can be designed as software interfaces. However, they can also be in the form of hardware interfaces that are controlled by suitable software.

A largely software-based implementation has the advantage that computer systems already present in a rail vehicle can be easily retrofitted with a software update after a possible supplementation with additional hardware elements, such as additional sensor units, in order to work in the manner according to the invention . In this respect, the object is also achieved by a corresponding computer program product with a computer program, which can be loaded directly into a memory device of such a computer system, with program sections in order to execute the steps of the method according to the invention that can be implemented by software when the computer program is executed in the computer system.

Such a computer program product can, in addition to the computer program, optionally contain additional components, such as e.g. B. a documentation and/or additional components, too Hardware components, such as hardware keys (dongles, etc.) for using the software.

A computer-readable medium, for example a memory stick, a hard disk or another transportable or permanently installed data medium, on which the program sections of the computer program that can be read and executed by a computer unit are stored, can be used for transport to the storage device of the computer system and/or for storage on the computer system. The computer unit can, for. B. this have one or more cooperating microprocessors or the like.

The dependent claims and the following description each contain particularly advantageous configurations and developments of the invention. In particular, the claims of a claim category can also be developed analogously to the dependent claims of another claim category and their descriptive parts. In addition, within the scope of the invention, the various features of different exemplary embodiments and claims can also be combined to form new exemplary embodiments.

In one embodiment of the method according to the invention for on-board monitoring of the area surrounding a rail vehicle, the position and orientation of the rail vehicle is based on a GNSS-based position and orientation estimate and/or on sensor data from the area surrounding the rail vehicle and on a recognition of features in the area position and orientation estimate based on the rail vehicle. The abbreviation "GNSS" stands for satellite navigation or in English "global navigation satellite system". The features of the environment or the position and orientation of the rail vehicle determined based on the sensor data can be compared with a map in order to further improve the accuracy of the position determination and orientation determination. Because the position of a rail vehicle restrictive conditions by being tied to a rail route, it can easily be cross-checked with map data to improve position estimation, e.g. by satellite navigation.

• Landmarken,
• den Gleiskörper,
• markante Objekte in der Umgebung des überwachenden Schienenfahrzeugs, die sich beispielsweise auch im für die Selbstlokalisierung bzw. Ermittlung der Orientierung des Schienenfahrzeugs zur Verfügung stehenden Kartenmaterial wiederfinden lassen.

In one embodiment of the method according to the invention for on-board monitoring of the area surrounding a rail vehicle, the features include at least one object of the following object types:

• landmarks,
• the track body,
• prominent objects in the vicinity of the rail vehicle being monitored, which can also be found, for example, in the map material available for the self-localization or determination of the orientation of the rail vehicle.

Advantageously, these marks and objects and their positions, which were detected and determined by on-board sensors, can be compared with map data in which the objects mentioned are recorded, and so a calibration or adjustment of the position determination of objects in the vicinity of the rail vehicle as well as enables calibration or adjustment of the self-localization of the rail vehicle. Since the position of the rail vehicle influences the determination of the position of the adjacent track, the position of the monitored area or a monitored adjacent track section can also be determined more precisely in this way. In this way, the sensor-based determination of the position of objects in the area surrounding the rail vehicle and the evaluation with regard to a possible source of danger in the area surrounding the rail vehicle are made more precise.

In one embodiment of the method according to the invention for on-board monitoring of the surroundings of a rail vehicle, 3D data are recorded and/or generated from the monitoring area. The 3D data from the surroundings of the rail vehicle advantageously allow an exact assessment of a positioning and whether an object detected in the area poses a risk to rail traffic.

In one embodiment of the method according to the invention for on-board monitoring of the area surrounding a rail vehicle, depth sensor data are recorded as 3D data from the area surrounding the adjacent track. Such a depth sensor allows three-dimensional scanning of an area to be monitored, as a result of which a position, an orientation and the extent of an object in three-dimensional space can be determined more precisely.

The 3D data can be captured from the monitored environment by a lidar unit or a stereo camera, for example. Lidar units or stereo cameras are also used to detect collision obstacles for the monitoring rail vehicle. Advantageously, these special sensor units can also be used to monitor track sections on adjacent tracks.

The 3D data can preferably be reproduced as a depth image or as a point cloud. Point clouds are particularly suitable for capturing the environment using lidar systems or laser-based systems in general, as they limit the amount of data to be processed.

In one embodiment of the method according to the invention for on-board monitoring of the surroundings of a rail vehicle, the step of determining the risk includes a comparison of the 3D data with the position and the dimensions of the driving channel of a rail vehicle running on the adjacent track. Advantageously, it can be determined whether an object protrudes into the driving channel of the adjacent track and, if necessary, a danger message can be issued.

In one embodiment of the method according to the invention for on-board monitoring of the surroundings of a rail vehicle, the 3D data is based on video data from a mono camera and determined based on optical flow detection of the captured video data. A mono camera should be understood as a camera with only one lens that generates 2D image data. If image data is now captured from different positions and/or viewing angles of the camera, a 3D image of the environment can still be generated with reduced hardware expenditure. The concept of capturing the optical flow is used here. The concept of determining 3D data based on the acquisition of the optical flow can be implemented, for example, using a "structure from motion" algorithm. By using a mono camera with two-dimensional recordings, the structure-from-motion method does not calculate the 3D environment directly from the recordings of two cameras with known relative positions, as is the case when using stereo cameras. Rather, the missing third degree of freedom is determined from the movement of the mono camera between two images, without the use of additional sensors being necessary. The environment is then reconstructed three-dimensionally and in real time in the form of a point cloud. At the same time, the trajectory of the camera and thus the movement of the rail vehicle in the area can be calculated. This results in a simultaneous location determination and environment recognition. In a next step, information on the course of the track, on surfaces and on objects in the vicinity of the rail vehicle is extracted from the point cloud. Advantageously, a complex 3D camera can be dispensed with and three-dimensional information about the surroundings of the rail vehicle can still be generated.

In one embodiment of the method according to the invention for on-board monitoring of the surroundings of a rail vehicle, the detected object is classified. Furthermore, the hazard is determined on the basis of the determined class of the detected object. For example, the detected object is classified as vegetation or as a rock or as a tree. The danger of a rock or tree will then rated as significantly higher than the risk of relatively harmless vegetation.

In one embodiment of the method according to the invention for on-board monitoring of the surroundings of a rail vehicle, in the event that a rail vehicle traveling on the adjacent track is classified as an object, the dimensions and the position of the rail vehicle on the adjacent track are recorded and a comparison is made between the Driving channel of the rail vehicle carried out on the adjacent track and the dimensions and the position of the rail vehicle to determine whether the rail vehicle protrudes at least partially from the driving channel. Alternatively or additionally, the dimensions and the position of the rail vehicle on the adjacent track can also be compared with the position and the dimensions of the driving channel of the observing or monitoring rail vehicle. An object may protrude from the rail vehicle on the adjacent track, endangering rail vehicles on the track on which the detecting rail vehicle is traveling. If a sensor measurement determines that the protruding object is protruding into the traffic channel on this side, a danger message can also be issued in order to warn rail vehicles that are running on the track of the rail vehicle being observed or to prevent a collision yourself.

Analogously, the dimensions and position of the rail vehicle driving on the adjacent track can also be compared with driving channels on other tracks or the infrastructure gauges at the edge of a track in order to avoid the risk of trains on other tracks or the infrastructure colliding with the track to be able to determine and assess the detected rail vehicle. The safety of the monitored rail network is advantageously further improved by these refinements.

• Satellitennavigationsdaten,
• IMU-Daten,
• Geschwindigkeitssensordaten,
• Odometriedaten,
• infrastrukturbasierte Positionsdaten, vorzugsweise von Beacons,
• in der Umgebung des Schienenfahrzeugs detektierte Merkmalsdaten, die mit einer Landkarte abgleichbar sind.

At least one of the following types of sensor data is preferably recorded for the environment detection or position determination:

• satellite navigation data,
• IMU data,
• speed sensor data,
• odometry data,
• infrastructure-based position data, preferably from beacons,
• feature data detected in the vicinity of the rail vehicle, which can be compared with a map.

IMU data is to be understood as data from an inertial measurement unit, which is referred to as an “inertial measurement unit” in English.

A beacon is a so-called radio beacon.

The feature data can include landmarks or the track body, for example.

The sensor data recorded can also be combined in order to be able to determine the position and orientation of the rail vehicle or the position of objects in the vicinity of the rail vehicle more precisely. For example, certain sensors are particularly suited to certain weather conditions. If necessary, this sensor data can be weighted according to the current weather conditions in such a way that a particularly precise measurement result is achieved.

Individual sensors of the rail vehicle can also be calibrated with a map to correct for effects of vibration, drift in an extrinsic calibration of the rail vehicle sensors.

• Verwenden einer geschätzten Position und Orientierung des Schienenfahrzeugs als Startpunkt,
• Verfeinern der Position und Orientierung des Sensors in der Karte durch Abgleich mit Landmarken und anderen Merkmalen, die in der Umgebung des Schienenfahrzeugs detektiert wurden.

Such a calibration can include the following steps:

• using an estimated position and orientation of the rail vehicle as a starting point,
• Refine the position and orientation of the sensor in the map by comparing it with landmarks and other features that have been detected in the vicinity of the rail vehicle.

A calibration and adjustment of the sensor system used for monitoring the surroundings is advantageously achieved, so that a hazard determination becomes more accurate and reliable due to the more precise data basis.

The evaluation of the sensor data recorded by the rail vehicle for object detection or obstacle detection can be carried out by a specialist who recognizes possible dangers or collision objects using image data based on the recorded sensor data. This process can take place in semi-real time, with the sensor data first being transmitted to the person via a radio network, or afterwards, with the data only being uploaded or transmitted to the person when the rail vehicle stops at a station or a depot. where access to a network is possible.

The semi-real-time operation enables a timely response to a detected obstacle. The evaluation afterwards can be carried out much more quickly, since the sensor data can be played back at a much higher data rate, with the obstacles becoming recognizable. However, in contrast to the semi-real-time approach, it may take place with a significant time delay at the time a source of danger occurs.

The sensor data can also be automatically analyzed by a computer. This process can take place in real time when the sensor data is analyzed with an onboard computer, or afterwards when the sensor data is first transferred to the cloud and evaluated there.

A combination of the two procedures is also possible, with only potentially dangerous objects that were automatically detected being examined again by a person in order to determine whether the detected object is actually dangerous. Advantageously, the automated detection does not have to be carried out so precisely, as a result of which computing capacity and data transmission capacity can be saved and robust hazard detection is nevertheless achieved.

• FIG 1 ein Flussdiagramm, welches ein Verfahren zur Überwachung der Umgebung eines Schienenfahrzeugs gemäß einem Ausführungsbeispiel der Erfindung veranschaulicht,
• FIG 2 eine schematische Darstellung eines Szenarios einer Überwachung eines benachbarten Gleises durch ein Schienenfahrzeug gemäß dem in FIG 1 veranschaulichten Verfahren,
• FIG 3 eine schematische Darstellung einer Umgebungsüberwachungseinrichtung gemäß einem Ausführungsbeispiel der Erfindung,
• FIG 4 eine schematische Darstellung eines Szenarios einer Überwachung eines einem ersten Schienenfahrzeug auf einem benachbarten Gleis entgegenkommenden zweiten Schienenfahrzeugs,
• FIG 5 ein Flussdiagramm, welches ein Verfahren zur Überwachung der Umgebung eines Schienenfahrzeugs gemäß einem zweiten Ausführungsbeispiel der Erfindung veranschaulicht.

The invention is explained in more detail below with reference to the attached figures using exemplary embodiments. Show it:

• FIG 1 a flow chart illustrating a method for monitoring the area surrounding a rail vehicle according to an embodiment of the invention,
• FIG 2 a schematic representation of a scenario of monitoring an adjacent track by a rail vehicle according to in FIG 1 illustrated procedures,
• 3 a schematic representation of an environment monitoring device according to an embodiment of the invention,
• FIG 4 a schematic representation of a scenario for monitoring a second rail vehicle approaching a first rail vehicle on an adjacent track,
• 5 a flow chart illustrating a method for monitoring the surroundings of a rail vehicle according to a second embodiment of the invention.

In FIG 1 a flow chart 100 is shown, which illustrates a method for monitoring the area U of a rail vehicle 2 according to an exemplary embodiment of the invention.

In step 1.I, a position P _s and an orientation OR _s of the rail vehicle 2 being monitored are determined. For this purpose, satellite navigation data are recorded using a GNSS receiver and landmarks LM from the area surrounding the rail vehicle 2 using an object sensor, for example a lidar sensor or a radar sensor or a camera. The landmarks LM are identified and a distance d and an orientation of the rail vehicle relative to the landmarks LM are determined using the sensor data. An absolute orientation OR _s of the rail vehicle 2 can be determined on the basis of the determined position P _s of the rail vehicle 2 and the relative orientation OR _s of the rail vehicle 2 to the landmark LM.

Furthermore, in step 1.II, map data KD are used, which reflect the track G1 on which the rail vehicle 2 travels and an adjacent track G2.

The map data KD in combination with the knowledge of the position P _s and the orientation OR _s of the rail vehicle 2 are then used in step 1.III to determine a position P and dimensions of the rail vehicle 2 adjacent track G2 or a track section GA- G2 of the adjacent track G2 to determine.

In step 1.IV, 3D data 3D-SD are recorded from a surveillance area BU. The monitoring area BU includes the monitored track section GA-G2 of the adjacent track G2 and its environment U. For example, the environment U includes a strip to the right and left of the adjacent track G2, in which no disturbing objects such as vegetation or boulders should occur , so that the movement of a rail vehicle, which is located on the track section GA-G2 that is also monitored, is not disturbed by the objects.

In step 1.V, an object O is detected and localized in the monitoring area BU on the basis of the 3D data 3D-SD recorded in step 1.IV. For this step, for example, an AI-based image evaluation with a detection or classification of objects O can be carried out.

In step 1.VI, it is also determined whether the detected object O could pose a risk to rail traffic. This determination is made depending on the position P and the dimensions b of the monitored track section GA-G2 of the adjacent track G2. Furthermore, the position P _F and the dimensions b _F of a driving channel FK of a rail vehicle 2a possibly traveling on the adjacent track G2 are compared with the position P _o and the dimensions of the detected object O when determining the risk. If, based on this comparison, it is determined, for example, that the object O is in the driving channel FK, the object O is classified as dangerous and corresponding danger information is transmitted by radio to a central dispatcher.

In FIG 2 is a schematic representation 20 of a scenario in which a neighboring track G2 is monitored by a rail vehicle 2 according to FIG FIG 1 illustrated method. The rail vehicle 2 travels in the direction of the arrow, i.e. from left to right, on a first track G1, which in figure 2 as the upper track or left track, and detects a wide triangular area B _D in front of the rail vehicle 2, which is in FIG 2 is marked with a dashed line. The sensor is actually used to detect an area of the first track G1 in front of rail vehicle 2, i.e. in FIG 1 to the right of the rail vehicle 2, in order to avoid collisions of the rail vehicle 2 with an object (not shown) located in the driving channel of the rail vehicle 2.

The triangular area B _D includes a rectangular monitoring area BU, which includes a track section GA-G2 of a second track G2 adjacent to the first track G1, in FIG 2 the lower track or right track, includes. This track section GA-G2 is now monitored more or less alongside or in addition to the section of the first track G1 lying in front of the rail vehicle 2 . If an object O is now detected and localized by the environmental monitoring device 30 in the monitoring area BU, its position P _o and its dimensions AB are compared with the position P _F of a driving channel FK and its dimensions b _F on the second track G2. If it is determined in this comparison that the object O is in the driving channel FK, then a hazard message GI is transmitted to a central dispatcher line 22 via a radio unit 21 of the rail vehicle 2 . Rail vehicles 2a traveling on the adjacent track G2 (see FIG 4 ), e.g. by radio, are warned in good time of the detected danger.

In 3 is an environment monitoring device 30 of a rail vehicle 2 (see FIG 2 ) shown schematically according to an embodiment of the invention. The environment monitoring device 30 includes a track detection unit 31 for determining a position P and dimensions b of a track section GA-G2 of a track G2 adjacent to the track G1 on which the rail vehicle 2 is traveling (see FIG FIG 2 ). For this purpose, the track detection unit 31 receives from a GNSS unit 32a position data P _s of the rail vehicle 2 and from additional sensor units (not shown) and on the basis of map data KD information about the orientation OR _s of the rail vehicle 2. Furthermore, the track detection unit 31 receives from a 3D Sensor unit 32 3D sensor data 3D-SD from a monitoring area BU, which includes the track section GA-G2 of the adjacent track G2 and its surroundings U. On the basis of the received sensor data P _s , OR _s , 3D-SD and on the basis of stored The track detection unit 31 uses map data KD to determine the position P and the dimensions b of the adjacent track G2 or of the track section GA-G2 of the adjacent track G2.

The position data P and dimensions b of the adjacent track section GA-G2 are transmitted to an evaluation unit 34, which is also part of the environmental monitoring device 30. Furthermore, the area monitoring device 30 includes an object detection unit that is set up to detect and localize an object O in the area U of the adjacent track G2 on the basis of the sensor data 3D-SD which it receives from the 3D sensor unit 32 . Such a detection can be done, for example, by comparing it with comparative data from the track section from a database or by using an AI-based approach. A localization of the detected object O in the monitoring area BU or a determination of position data P _o and dimensions AB of the detected object O can be realized, for example, by determining the distance and orientation based on the detected 3D sensor data 3D-SD.

As already mentioned, the environmental monitoring device 30 also includes an evaluation unit 34, which is set up to determine, based on the determined position data P _o of the detected object O and the position P and the dimensions b of the adjacent track G2, whether the detected and localized object O could pose a hazard to rail traffic. The object O can be a boulder or a fallen tree, for example. When determining the danger, it can be determined, for example, whether the object O is in the driving channel FK (see FIG 2 ) of a rail vehicle 2a running on the adjacent track G2 or into the driving channel of a rail vehicle running on its own track G1. The hazard is determined on the basis of knowledge of the position P and the dimensions b of the adjacent track G2 and knowledge of the position P _F and the dimensions b _F of a driving channel FK of a rail vehicle 2a traveling on the adjacent track G2 and on the basis of knowledge of the position P _o and the dimensions AB of the detected object O. If it is determined that the object O is located in the driving channel FK, a central driving service line 22 (see FIG 2 ) issued a hazard warning GI so that the central dispatcher 22 could detect a rail vehicle 2a traveling on the adjacent track G2 (see FIG 4 ) can warn in time.

In FIG 4 a scenario 40 of a monitoring of an oncoming rail vehicle 2a on a second track G2 adjacent to a first track G1 is illustrated by a first rail vehicle 2 traveling on the first track G1. The first rail vehicle 2 uses its surroundings monitoring device 30 to monitor a triangular area B _D in the direction of travel, which is marked with an arrow. Part of this triangular area B _D is also a rectangular monitoring area BU, which includes a track section GA-G2 of the adjacent track G2. A second rail vehicle 2a approaching the first rail vehicle 2 is located in the rectangular monitoring area BU. The second rail vehicle 2a has loaded a load L, which, however, is not fixed sufficiently firmly on the second rail vehicle 2a, so that it has already slipped to the right, ie in the direction of the first track G1, and out of the actually intended driving channel FK of the second rail vehicle 2a protrudes to the right.

With the help of the environment monitoring device 30, a position P _o and dimensions AB of the detected object or the detected load L are now determined and there is a comparison of the position P _o and the dimensions AB of the detected load L and the width b _2a of the through the Last L widened second rail vehicle 2 with a position P _F and a maximum allowable width b _F of a driving channel FK on the second track G2. If determined by the comparison If it is found that the load L protrudes from the driving channel FK of the second track G2, then it is concluded that the load L has become detached or is not properly fastened to the second rail vehicle 2a. In this case, it is determined that the second rail vehicle 2a could pose a risk to other rail vehicles.

Additionally or alternatively, the position P _o and the dimensions AB of the detected object or the detected load L or the width b _2a of the second rail vehicle 2 can be compared with the position and a maximum permitted width of a driving channel of a rail vehicle on the first track G1 take place. If it is determined on the basis of this additional comparison that the load L protrudes into the driving channel on the first track G1, a hazard message can also be issued stating that the second rail vehicle 2a poses a risk to rail vehicles traveling on the first track G1 .

In 5 a flowchart 500 is shown, which illustrates a method for monitoring the surroundings of a rail vehicle 2 according to a second exemplary embodiment of the invention. At the in 5 illustrated procedure is based on an in FIG 4 scenario 40 shown responds. First, as in the method according to the first exemplary embodiment, a position P _s and an orientation OR _s of a first rail vehicle 2 are determined in step 5.1. As already mentioned, for this purpose satellite navigation data are detected by sensors with the aid of a GNSS receiver and landmarks LM in the vicinity of the rail vehicle 2 . The landmarks LM are identified and a distance d and an orientation of the first rail vehicle 2 to a landmark LM are determined using the sensor data. An absolute orientation OR _s of the rail vehicle 2 can be determined on the basis of the determined position P _s of the rail vehicle 2 and the relative orientation of the rail vehicle 2 to the landmark LM.

Furthermore, in step 5.II, map data KD are used, which reflect the track G1 on which the rail vehicle 2 travels and an adjacent track G2.

The map data KD in combination with the knowledge of the position P _s and the orientation OR _s of the rail vehicle 2 are then used in step 5.III to determine a position P and dimensions b of the rail vehicle 2 adjacent track G2 or a track section GA -G2 on the adjacent track G2, also referred to as adjacent track section GA-G2 for short.

In step 5.IV, 3D data 3D-SD are recorded from a surveillance area BU. The monitoring area BU includes the track section GA-G2 of the adjacent track G2 and its environment U. For example, the environment U includes a strip to the right and left of the adjacent track G2, in which no disturbing objects such as vegetation or boulders should occur.

In step 5.V, a second rail vehicle 2a is detected and localized in the monitoring area BU on the basis of the 3D data 3D-SD.

In step 5.VI, the dimensions b _2a of the second rail vehicle 2a are then determined on the basis of the 3D data 3D-SD and the width b _2a of the second rail vehicle 2a at the widest point of the rail vehicle 2a is compared with a maximum permissible Channel width b _F for the second track G2. For example, the width b _2a can be increased by a shifted load L hanging out on the second rail vehicle 2a in such a way that the load protrudes from the maximum permitted driving channel FK. It is therefore clear that the load L is not correctly attached to the second rail vehicle 2a and thus represents a safety risk. In the event that such an exceeding of the limits of the driving channel FK was determined by the load L, which is 5 marked with "j" becomes the Step 5.VII passed and a corresponding danger message GI transmitted, for example, to the second rail vehicle 2a and a central dispatcher. If the load L does not exceed the limits of the maximum permissible width b _F of the driving channel FK of the second track G2, which in 5 is marked with "n", no hazard message GI is issued and a return is made to step 5.I and the monitoring process is repeated.

Finally, it is pointed out once again that the methods and devices described above are merely preferred exemplary embodiments of the invention and that the invention can be varied by a person skilled in the art without departing from the scope of the invention insofar as it is specified by the claims. For the sake of completeness, it is also pointed out that the use of the indefinite article "a" or "an" does not rule out the possibility that the relevant characteristics can also be present more than once. Likewise, the term "unit" does not rule out that this consists of several components, which can also be spatially distributed if necessary.

Claims (15)

Method for on-board monitoring of the surroundings (U) of a rail vehicle (2), having the steps: - Acquisition of sensor data (3D-SD) from a monitoring area (BU) which is a track section (GA-G2) of a track (G2) which is adjacent to a track (G1) on which the rail vehicle (2) is moving, and its surroundings (U), - Detecting and locating an object (O, L, 2a) in the surveillance area (BU) based on the sensor data (3D-SD), wherein a position (P _o ) and the dimensions (AB, b _2a ) of the detected and localized object (O, L, 2a) can be determined, - Determine whether the detected object (O, L, 2a) could pose a risk to rail traffic, depending on - the recorded sensor data (3D-SD) and - the position (P _o ) and the dimensions (AB, b _2a ) of the detected and located object (O, L, 2a). A method according to claim 1, comprising the steps of: - determining a position (P) and dimensions (b) of the track section (GA-G2) of the track (G1) adjacent to the track (G1) on which the rail vehicle (2) moves (G2) on the basis - a position (P _s ) and orientation (OR _s ) of the rail vehicle (2) and - On the basis of map data (KD), comprising the track (G1) on which the rail vehicle (2) travels and the adjacent track (G2), wherein when determining whether the detected object (O, L, 2a) could pose a risk to rail traffic, a comparison of the position (P _o ) and the dimensions (AB) of the detected and localized object (OL, 2a) with the Position (P) and the dimensions (b) of the monitored track section (GA-G2) of the adjacent track (G2). Method according to claim 1 or 2, wherein determining whether the detected object (O, L, 2a) could pose a risk to a rail vehicle on the first track (G1) comprises the steps of: - Determining the position (P _F ) and the dimensions (b _F ) of a first driving channel (FK) of a rail vehicle (2a) traveling on the adjacent track (G2) and/or - Determining the position and the dimensions of a second driving channel of a rail vehicle traveling on the track (G1) on which the monitoring rail vehicle (2) travels and - Comparing the determined position (P _F ) and the dimensions (b _F ) of the first and/or second driving channel (FK) with the determined position (P _o ) and the dimensions (AB, b _2a ) of the detected and localized object (O , L, 2a). Method according to one of the preceding claims, wherein the position (P _s ) and orientation (OR _s ) of the monitoring rail vehicle (2) is determined on the basis of - a GNSS-based position and orientation estimation and/or - Sensor data (3D-SD) from the environment of the rail vehicle (2) and - A position and orientation estimation, which is based on features in the environment of the rail vehicle (2) which are comparable to a map (LK). The method of claim 4, wherein the features include: - landmarks, - the track body, - Distinctive objects in the vicinity of the rail vehicle being monitored (2). Method according to one of the preceding claims, wherein 3D data (3D-SD) of the monitored area (BU) are recorded and/or generated as sensor data. Method according to Claim 6, depth sensor data being recorded as 3D data (3D-SD) from the monitoring area (BU) of the adjacent track (G2). Method according to claim 6 or 7, wherein the 3D data (3D-SD) are acquired by one of the following devices: - a lidar unit, - a stereo camera. Method according to one of Claims 6 to 8, the 3D data (3D-SD) being reproduced in one of the following types of representation: - through a depth image, - by a point cloud. Method according to one of the preceding claims, wherein the 3D data (3D-SD) on the basis - from video data from a mono camera and - be determined on the basis of the detection of the optical flow of the recorded video data. Method according to one of the preceding claims, in which the detected object (O) is classified and the risk is determined on the basis of the determined class of the detected object (O). An environmental monitoring device (30) comprising: - a sensor unit (32) for acquiring sensor data (3D-SD) from a monitoring area (BU) which covers the track section (GA-G2) of a track (G2) which corresponds to a track (G1) on which the rail vehicle (2nd ) moves, is adjacent, and includes its surroundings (U), - An object detection unit (33) for detecting and locating an object (O, L, 2a) in the monitoring area (BU) of the adjacent track (G2) on the basis of the sensor data (3D-SD), wherein a position (P _o ) and the dimensions (AB, b _2a ) of the detected and localized object (O, L, 2a) are determined,
and - An evaluation unit (34) for determining whether the detected object (O, L, 2a) could pose a risk to rail traffic, depending on - the sensor data (3D-SD) and - A position (P _o ) and the dimensions (AB, b _2a ) of the detected and located object (O, L, 2a). Rail vehicle (2), having an environment monitoring device (30) according to claim 12. Computer program product with a computer program which can be loaded directly into a memory unit of a control device of a rail vehicle (2), with program sections in order to carry out all the steps of a method according to one of Claims 1 to 11 when the computer program is executed in the control device. Computer-readable medium on which program sections that can be executed by a computer unit are stored in order to carry out all the steps of the method according to one of Claims 1 to 11 when the program sections are executed by the computer unit.

Images