Position determination of a rail vehicle

Abstract

A method for determining the position of a rail vehicle (202) is provided, comprising: capturing at least one first image (213) of an external environment using an image acquisition system (201) arranged on the rail vehicle (202); analyzing the first image (213) to identify in the first image an image of at least one first object (O1) with a known first object position (rO1); determining a geometric first relative position (rR1) of the first object (O1) relative to the image acquisition system (201) based on the identified image of the first object; determining an estimated rail vehicle position (rSg) based on the first object position (rO1) and the first relative position (rR1); and determining the rail vehicle position (632) based on at least the first estimated rail vehicle position.

Description

AREA OF INVENTION

[0001] The present invention relates to a method and a system for determining the position of a rail vehicle. Furthermore, the present invention relates to a rail vehicle which includes the position determination system. STATE OF THE ART

[0002] Modern train control systems require that the positions of trains on a track system be determined and known to the train control system. Furthermore, the train positions must be made known to the signaling system so that it can control the separation of trains from one another, the train sequence, and individual train movements.

[0003] Traditionally, absolute train tracking is achieved using balises. These must be embedded in the track bed and represent a significant cost factor. Furthermore, a balise antenna must be installed on or inside the train or rail vehicle. Such an antenna requires a large, metal-free space, which complicates installation and consumes considerable space.

[0004] Traditionally, speed measurements are also carried out on trains, but this requires additional sensors. These include, among others, OPG (odometer-pulse generator), radar, accelerometers, and GPS.

[0005] US 10,710,621 B2 discloses a method, system, and apparatus for controlling a vehicle moving along a rail of a transportation network. An image is captured by a video unit on board the vehicle. The image represents a portion of the track surrounding the vehicle and information displayed on an onboard display. The displayed information is processed based on the image to allow the vehicle to be controlled based on the parameter and / or the image. The image, which depicts a portion of the track, can be used to calculate the speed and / or position of the train. Mapping data is generated by three-dimensional photogrammetry, which is periodically provided by a service train equipped with a high-resolution dometer and a high-resolution camera. The mapping data is compared with the captured image.However, the procedure described in this document is very complex and time-consuming.

[0006] However, it has been observed that conventional methods for determining train positions are highly complex, require a lot of space, and necessitate the installation of significant additional track system components. Furthermore, an improvement in the accuracy of position determination is desirable. SUMMARY OF THE INVENTION

[0007] Therefore, it is an object of the present invention to provide a method for determining the position of a rail vehicle and a corresponding system, whereby the complexity, space requirements, installation requirements, or need for elaborate equipment are reduced. Furthermore, it is an object of the present invention to provide a position determination method and system by means of which reliable and sufficiently accurate position determination is enabled.

[0008] According to one embodiment of the present invention, a method for determining the position of a (moving or stationary) rail vehicle is provided, wherein the method comprises: capturing, in particular at a first time, at least a first image of an external environment by means of a device arranged at (e.g. in or on) the rail vehicle (e.g.installed) image acquisition system; Analyzing the first image to identify in the first image an image of at least one first (pre-known) object with a pre-known (geodetic) first object position; Determining a geometric first relative position of the first object relative to the image acquisition system based on the identified image of the first object; Determining a (first) estimated railway vehicle position based on the (geodetic) first object position and the first relative position; Determining the railway vehicle position based on at least the first estimated railway vehicle position.

[0009] The method can be implemented in hardware and / or software. The method can be performed or controlled by a positioning system according to embodiments of the present invention. The method can be performed while the rail vehicle is moving, i.e., while its position is continuously changing, and / or during periods when the rail vehicle is stationary, i.e., while its position is not changing. The method can be configured for absolute position determination, i.e., geodetic position determination of the rail vehicle.

[0010] The rail vehicle in question could be a local train, a long-distance train, a tram, a subway, a light rail system, or similar. The specific position of the rail vehicle could refer, for example, to a reference point within the vehicle, such as its center point, or to its front or rear end.

[0011] The first image (as well as, for example, the second image) can be formed by a two-dimensional field of intensity values, corresponding to light intensities captured by the image acquisition system. The image acquisition system can contain one or more camera chips, such as CCD or CMOS chips, which have an array of light-sensitive sensors arranged in a two-dimensional field. The image acquisition system can, for example, have between one and ten million light-sensitive cells. The image acquisition system can be located, and in particular installed, inside or outside the rail vehicle. For example, the image acquisition system can be installed inside a driver's cab at the front and / or rear of the rail vehicle, allowing it to capture image information from the external environment through a cab window.

[0012] The first object can be a previously known object with a previously known geodetic first object position. The first object can be identified, for example, by its shape and / or by an identifier within a marker of the object. The object can, for example, include a machine-readable marker containing an identifier and / or location information.

[0013] The first object position can therefore be obtained, for example, by analyzing the first image, in particular by analyzing the representation of the first object.

[0014] In other embodiments, an identifier can be derived from the first image, which is then assigned to the first object. The first object's position can then be retrieved, for example, from a database that can provide the object's position in association with the identifier.

[0015] The image acquisition system can be arranged and designed to capture a field of view extending beyond the rail vehicle, covering, for example, a front area in the direction of travel and / or a rear area in the opposite direction. The first object can, for example, be located to the side of a track traversed by the rail vehicle.

[0016] The first relative position can be represented, for example, by a displacement vector that, at the time the first image is captured, is directed from a reference position (e.g., the center) of the image acquisition system to a reference position (e.g., the center) of the first object. The first relative position thus contains information about how far the first object is from the image acquisition system and in which direction at the time of capture. If the first object's position is known, the position of the image acquisition system at the time of capture can be determined using vector operations. Since the image acquisition system also has a specific relative position to a reference point on the rail vehicle, the rail vehicle's position can also be estimated from the first relative position of the first object and the known position of the first object.

[0017] According to embodiments of the present invention, one or more estimated railway vehicle positions (also referred to as railway vehicle position hypotheses) can be determined. Based on one or more of these railway vehicle position hypotheses (and optionally using speed information from the railway vehicle), the railway vehicle position (or at least a preliminary railway vehicle position) can ultimately be determined, for example, by giving less reliable estimated railway vehicle positions a lower weight than relatively more reliable estimated railway vehicle positions. This can be achieved by filtering, for example, using a particle filter. Several different hypotheses, including weightings (called particles), about the state of the system are formed based on the measured values ​​obtained so far.The weighting of the individual hypotheses is adjusted using newly available measurement data.

[0018] For the new hypotheses about the state, in this case the positions of the rail vehicles, of the system, both the new measured values ​​and the newly weighted previous hypotheses are used.

[0019] Weighting different hypotheses can lead to an estimate of a preliminary train position. This preliminary train position can then be further refined / confirmed / substantiated or modified in the evaluation system based on additional information, such as speed data, to obtain the final train position.

[0020] Installing or providing an image acquisition system on (for example, in or on) a rail vehicle does not require a large amount of space and, in particular, does not require a metal-free environment as is necessary for conventional balise antennas. This allows the system's complexity to be kept low. Furthermore, conventional image acquisition systems are relatively inexpensive, and various sensitivity ranges are available. The method can therefore be used both during the day and at night, provided, for example, that the image acquisition system is sensitive to infrared and / or visible radiation. The first object can be illuminated or emit detectable radiation on its own to be imaged. Thus, a reliable and simple method for determining position is provided.

[0021] According to one embodiment of the present invention, analyzing the first image comprises: identifying in the first image several images of several first (previously known) objects with several previously known (geodetic) first object positions; determining several geometric first relative positions of the several first objects relative to the image acquisition system based on the identified images of the several first objects; determining several (first) estimated railway vehicle positions based on the several (geodetic) first object positions and the several first relative positions; determining the railway vehicle position based on the several estimated railway vehicle positions, in particular using a particle filter.

[0022] If several first objects with known first object positions are depicted within the first image, the accuracy of determining the train position can be improved. This is because multiple relative first positions and associated first object positions are used, resulting in multiple estimated train positions. Depending on the reliability of these various estimated train positions, the train position can then ultimately be determined with higher accuracy.

[0023] The process can be carried out continuously over time, for example, while the rail vehicle is in motion. In the first sections of the journey, the images captured by the image acquisition system depict objects (e.g., first objects) whose geodetic positions are known. However, in the second sections, the image acquisition system can only detect other objects (e.g., second objects) whose absolute geodetic positions are unknown. Within these second sections, the rail vehicle's position cannot be derived (solely) by imaging and analyzing first objects with known positions. While in these sections, some of the following embodiments can be implemented.

[0024] According to one embodiment of the present invention, the method further comprises: determining an estimated rail vehicle speed at the first and / or at a second time later than the first time, in particular by means of image analysis; determining the rail vehicle position at the second time later than the first time, based at least on the determined rail vehicle position at an earlier time than the second time, in particular at the first time, and at least on the estimated rail vehicle speed at an earlier time than the second time and / or at the second time.

[0025] According to such embodiments, the speed of the rail vehicle is determined, and together with previously determined positions of the rail vehicle, the positions at a desired time are ascertained. The position of the rail vehicle at the second time can, for example, be determined from the previously determined position of the rail vehicle at the first time and a speed of the rail vehicle at the first time, at the second time, or at a time between the first and second times. A time between the first and second times can, for example, correspond to a sampling interval of the image acquisition system or to a multiple of the sampling interval of the image acquisition system. A time span between the first and second times can, for example, be between 0.1 s and 10 s. Other values ​​are possible.In an image taken at the second time point, there can (need to) be no first object depicted, that is, no object whose geodetic position is known.

[0026] Estimating the speed of a rail vehicle does not necessarily require image analysis of images captured by the image acquisition system. It can also be achieved using conventional methods, such as determining the wheel rotation speed. However, if the speed is determined using image analysis of images captured by the image acquisition system, the complexity can be reduced and costs lowered.

[0027] According to one embodiment of the present invention, determining the estimated speed of the rail vehicle at the first and / or second time point comprises: capturing, at the second time point, at least a second image in order to provide several images captured at different times (together with or without the first image); analyzing the several images in order to determine the estimated speed of the rail vehicle at one or more times point in time, in particular at the first and / or at a second time point.

[0028] This embodiment may require capturing a second image, at least in conjunction with the first image. The multiple images, captured at different times, may include the first image, the second image, and additional images. In other embodiments, the multiple images need not include the first image but may simply include multiple second images. In any case, multiple images are necessary to derive an estimated speed of the rail vehicle from at least two of the multiple images. This requires determining the displacement of a depicted object within the two images and the respective differences in the capture times. The second object may, but need not, also be depicted in the first image and is, in any case, depicted in the second image.The displacement of the second object in the image, captured in successive images, can be determined to calculate the distance traveled by the train between the first and second time points. This distance can then be divided by the difference between the image capture times to estimate the train's speed. Estimating both the train's position and speed utilizes the same image acquisition system, thus minimizing the complexity of the system and the procedure.

[0029] The second object(s) can (must) be stationary, even if their location is unknown. Examples would be axle counters, switch frogs, or signal masts. If a different train is used as the (second) object, the measurement would be distorted by the speed of the other train.

[0030] According to one embodiment of the present invention, the analysis of the multiple images comprises: identifying images of at least a second (location unknown) object in the multiple images; determining the respective geometric second relative positions of the second object relative to the image acquisition system in the multiple images based on the identified images of the second object; deriving at least the estimated speed of the rail vehicle at one or more time points from the second relative positions associated with multiple images, in particular differences therein, and the various acquisition times, in particular differences therein.

[0031] The geodetic location of the second object may be known, but it need not be. According to embodiments of the present invention, images of first objects whose position is known could also be used to estimate the speed of a rail vehicle. However, the accuracy of the position determination of the rail vehicle can be increased if first objects whose position is known are primarily used for direct position determination. The weighting of the determination of the rail vehicle position from analyses of images of first objects (with known positions) and second objects (without known positions) can be adjusted depending on the specific application.

[0032] According to one embodiment of the invention, the analysis of the multiple images comprises: identifying images of several second (position unknown) objects in the multiple images; determining the respective geometric second relative positions for the several second objects relative to the image acquisition system in the multiple images based on the identified images of the several second objects; deriving at least the estimated speed of the rail vehicle at one or more time points from the second relative positions associated with the multiple images and several second objects, in particular differences therein, and the various acquisition times, in particular differences therein.

[0033] If multiple second objects are identified in the multiple images—that is, if multiple second objects are depicted in each of the multiple images—the accuracy of the rail speed determination can be improved. Due to the higher accuracy of the rail vehicle speed determination, the accuracy of the rail vehicle position determination can also be increased.

[0034] According to one embodiment of the present invention, the method is designed such that the position of the rail vehicle is determined based on one or more estimated speeds, if available, and / or one or more estimated positions, if available, in particular using a Kalman filter. The Kalman filter enables the estimation of a system parameter based on measurements from different sources (sensors) and the consideration of the uncertainties of each source. The estimation can also be continued if the source(s) fail completely or only temporarily.

[0035] The one or more estimated train speeds and one or more estimated train positions, as well as the previously determined train positions, can be combined or merged to improve the accuracy of determining the train position. The reliability of the respective train position or train speed can be derived from the consistency or deviation of the various estimated positions or speeds.

[0036] According to one embodiment of the present invention, the method is designed such that the rail vehicle speed at a specific time is derived by weighting the several estimated rail vehicle speeds at that specific time, in particular by means of a particle filter, and / or the rail vehicle position at a specific time is derived by weighting the several estimated rail vehicle positions at that specific time, in particular by means of a particle filter, and by weighting the several estimated rail vehicle speeds at that specific time, in particular by means of a particle filter and / or Kalman filter.

[0037] For example, multiple estimated train speeds can be fed into a particle filter, which then determines the most probable or reliable value from the majority of the estimated values. Similarly, multiple estimated train positions can be fed into a particle filter, which then determines the most reliable value from these inputs. The most probable or reliable train speed and position can then be fed into a Kalman filter, which can determine the final train position.

[0038] According to one embodiment of the present invention, the method is designed such that the respective relative position of the first or second object relative to the image acquisition system is derived from a distance of the respective imaged object from the image acquisition system and the image position of the image in the respective image, in particular using the field of view of the image acquisition system, wherein in particular the distance is determined from an extent of the image and an actual (previously known) extent of the object.

[0039] The camera system can include calibrated cameras with calibrated magnification and distortion. Images captured by the camera system can be rectified using calibration data, allowing precise magnification characteristics to be determined. This enables the determination of distances between the captured objects and the camera system by analyzing the images. Furthermore, the object's orientation relative to the camera system can be determined through image analysis. Therefore, the object's relative position to the image acquisition system can be reliably determined.

[0040] According to one embodiment of the present invention, the method is designed such that the method is carried out continuously over time, wherein the one or more estimated railway vehicle positions and / or the one or more estimated railway vehicle speeds are stored with a timestamp (e.g., image capture time), and wherein the railway vehicle position at a specific time is determined based on one or more estimated railway vehicle positions with timestamps up to the specific time and / or one or more estimated railway vehicle speeds with timestamps up to the specific time, in particular using a Kalman filter.

[0041] The respective measured values ​​or derived values ​​can be stored in an electronic memory. The process can be carried out with the aid of a processor. The determined train position and / or speed can be fed into a train traffic monitoring or control system and / or displayed to a train driver. Thus, reliable train position determination can be carried out during essentially the entire journey of the train and also when stationary.

[0042] According to one embodiment of the present invention, the method is designed such that the image acquisition system comprises one or more cameras; and / or wherein the image acquisition system is arranged in a driver's cab at one and / or both ends of the train; and / or wherein the first and / or second image is composed of partial images captured by several cameras and / or comprises several partial images; and / or wherein a field of view of the image acquisition system comprises an external environment area with an opening angle between 90° and 180°, lying in the direction of travel and / or in the opposite direction of travel of the rail vehicle; and / or wherein a field of view of the image acquisition system comprises a track area and / or a track edge and / or an area of ​​signal signs and / or masts on both sides of the track; and / or wherein the image acquisition system is sensitive to visible light and / or infrared and / or ultraviolet radiation.

[0043] This provides great flexibility and allows the process to be adapted to the respective requirements or applications.

[0044] According to one embodiment of the present invention, the method is designed such that the image acquisition system, in particular a calibrated one, comprises a first camera at a first end of the rail vehicle and a second camera at a second end of the rail vehicle, wherein the method can be carried out according to at least one of the following operating modes: in a first operating mode, the images recorded by the first or the second camera are evaluated to determine the position of the rail vehicle; in a second operating mode, the images recorded by the first and the second camera are evaluated to determine the position of the rail vehicle; in a third operating mode, the images recorded by the first camera are evaluated first to determine a first position of the rail vehicle, and the images recorded by the second camera are evaluated second to determine a second position of the rail vehicle;In a fourth operating mode, a first evaluation of the images recorded by the first camera is carried out to determine a first position of the rail vehicle; a second evaluation of the images recorded by the second camera is carried out to determine a second position of the rail vehicle, whereby the first or the second evaluation is validated by means of the other evaluation in order to determine the position of the rail vehicle.

[0045] This provides various levels of redundancy and / or allows for the compensation or consideration of system failures. Furthermore, it offers a high degree of flexibility. The method can support additional operating modes.

[0046] According to one embodiment of the present invention, the method is designed such that the first and / or the second object comprises at least one of the following: a (locally known) human-readable marker board; a (locally known) machine-readable marker board, in particular having an ID and / or QR code; a signal mast; a switch; a railway sleeper; a switch (e.g., including switch drive); a platform; an axle counter; a marker on a fixed object, in particular a safety point; etc. The above list is not to be understood as exhaustive; other types of objects may be supported.

[0047] It should be understood that features which have been described, mentioned, applied or used individually or in combination in connection with a method for determining the position of a railway vehicle may also be applied or used, individually or in any combination, to a system for determining the position of a railway vehicle according to embodiments of the present invention and vice versa.

[0048] According to one embodiment of the present invention, a system for determining the position of a (moving or stationary) rail vehicle is provided, wherein the system comprises: an image acquisition system arranged on the rail vehicle, configured to acquire at least one first image of an external environment; an image analysis system, configured to evaluate the first image in order to identify in the first image an image of at least one first object with a previously known (first) object position; a determination system, configured to: determine a geometric first relative position of the first object relative to the image acquisition system based on the identified image of the first object; determine an estimated rail vehicle position based on the first object position and the first relative position;to determine the rail vehicle position based on at least the first estimated rail vehicle position.

[0049] The system can be integrated or distributed. For example, while the image acquisition system is located or installed on the rail vehicle, the evaluation of the captured images can be performed by an integrated system located within the rail vehicle or by an external, e.g., distributed, system that is partially located outside the rail vehicle and can be connected to the image acquisition system via a communication network such as the internet or a bus system. The image analysis system and / or the positioning system can be controlled by software, which may contain instructions to execute or control a process for determining the position of the rail vehicle.

[0050] According to one embodiment of the present invention, a rail vehicle is provided comprising: a driver's cab at at least one end of the rail vehicle; a system according to the preceding embodiment, wherein the image acquisition system is installed in particular inside the driver's cab. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Embodiments of the present invention will now be explained with reference to the accompanying drawings. The invention is not limited to the illustrated or described embodiments. Fig. Figure 1 schematically illustrates a system for determining the position of a rail vehicle according to an embodiment of the present invention; Fig. 2 and Fig. Figure 3 illustrates features of a method for determining the position of a rail vehicle according to an embodiment of the present invention together with a rail vehicle according to an embodiment of the present invention; Fig. 4 and Fig. Figure 5 illustrates features of a method for determining a rail vehicle speed and / or position according to embodiments of the present invention; Fig. Figure 6 schematically illustrates a determination system which may be included in a system for position determination according to an embodiment of the present invention; and Fig. Figure 7 illustrates accuracies of certain rail vehicle positions that can be achieved according to embodiments of the present invention. DETAILED DESCRIPTION OF EXECUTION FORMS

[0052] It should be noted that features or components of different embodiments, which are identical or at least functionally equivalent to the corresponding features or components of the embodiment, are provided with the same reference numerals or with other reference numerals which differ only in the first digit and / or by an appended letter from the reference numeral of a corresponding component.

[0053] According to embodiments of the present invention, visual sensors are used for speed measurement and / or the absolute positioning of rail vehicles, particularly trains. These sensors are arranged or installed on the rail vehicle. The visual sensors (in particular, image acquisition systems) can be used in combination with visual markers on the track or in an external area surrounding the rail vehicle. The visual sensors can be mounted at each end of the train, e.g., behind the windshield of the driver's cab. The sensors or the image acquisition system can be sensitive to visible light and / or to infrared light (possibly in conjunction with infrared illuminators).

[0054] The captured sensor image can be processed in an evaluation unit using two different channels. The first channel can determine the current speed of the rail vehicle (velocity channel), and the second channel can determine its absolute position (tracking channel). Both channels can be combined in a sensor data fusion to verify their current plausibility and to provide support in case of temporary channel failure. Tracking Channel

[0055] The tracking channel can search the received sensor information for known visual markers. The distance of the sensor to a detected visual marker can be determined. Since the position of each marker is known, the train's position can be determined. Velocity Channel

[0056] To determine speed, objects (signals, switches, sleepers) are identified in the received sensor information, and their movement over time is observed. If the object is close to the train's path, the train's speed can be calculated. For reliable speed determination, several objects are used simultaneously. If other sensors are available for speed measurement (e.g., OPGs from RST), these can be used to verify the calculated speed value.

[0057] According to embodiments of the present invention, many different operating modes are implemented, or at least one of the following, which satisfy different signal requirements: 1. Only one evaluation unit is active and uses only the sensor information from the assigned driver's position. 2. Only one evaluation unit is active and uses the sensor information from both driver positions. 3. Both evaluation units are active and use the sensor information from their respective driver positions. Both evaluation units transmit their results to a user system. The user system thus receives values ​​from two independent sources. 4. Both evaluation units are active and use the sensor information from their respective driver stations. The evaluation unit of the leading driver station uses the results from the other evaluation unit to validate its own results. Only the results from the evaluation unit of the leading driver station are transmitted to the user system.

[0058] The presented solution significantly reduces the conventionally required effort (balise antennas and corresponding spaces within the train), as only one marker board (or many at various locations along the track) is needed. This can be detected over a wide area, enabling precise localization. Thus, not only are the hardware costs for the balises saved, but also the engineering effort (telegram generation, programming, and calibration).

[0059] Fig. Figure 1 schematically illustrates a system 100 for determining the position of a moving or stationary rail vehicle according to an embodiment of the present invention. A camera system 101 is arranged on a rail vehicle 102. The image acquisition system 101 provides image data 103 to an image processing system 104. The image processing system 104 calculates image analysis data 140. A determination system 141 receives the image analysis data 140 and calculates localization data 106 using external data, such as position data of previously known objects 105. If independent position or speed data of the rail vehicle are available, for example, OPG information 107, these are fused with the localization data 106 in a fusion block 108 to finally determine a rail vehicle position 109, which can then be used by a user system 110.

[0060] The image acquisition system 102 can comprise one or more calibrated cameras, which can be installed, for example, in a driver's cab behind a windshield. One or more cameras can be provided per cab. The cameras can be designed as night vision cameras or can detect objects illuminated by sunlight or rail vehicle headlights.

[0061] Fig. Figure 2 illustrates features of a method for determining the position of a rail vehicle according to an embodiment of the present invention. An image acquisition system 201 is arranged, in particular installed, on the rail vehicle 202 (with reference position 202a). The image acquisition system 201 can register image information from a field of view 211 (with opening angle α). At a first time point, the image acquisition system 201 takes a first image. The first image is processed (for example, by the image evaluation unit 104, as illustrated in Figure 2). Fig. 1) The first image is analyzed to identify an image of a first object O1 (212) with a previously known object position rO1. The first object position rO1 is known with respect to a reference point 0, for example, the Earth's center. According to the position determination procedure, a first relative position rR1 of the first object O1 (212) relative to the image acquisition system 201 is determined based on the identified image of the first object O1 (212). Furthermore, an estimated railway vehicle position rSg is determined based on the first object position rO1 and the first relative position rR1 (and a known relative position rRK of the camera system 201 to the reference position 202a), in particular with the aid of vector operations. From the estimated railway vehicle position rSg, the railway vehicle position is then determined, possibly with the aid of further position and / or velocity information.

[0062] Fig. Figure 3 again illustrates the rail vehicle 202, in which the image acquisition system 201 is arranged and is designed to capture images from a viewing area 211. For example, from Fig. As shown in section 3 above, the field of view 211 includes, for example, the track bed, trackside installations such as signals or platforms.

[0063] In the illustrated embodiment, the first object O1 (212) is a panel on which a QR code is depicted, more generally a machine-readable image or code. The upper part of the Fig. Figure 3 illustrates the situation in a top view, while the lower part illustrates image 213 taken by the image acquisition system 201. Image 213, in particular the first image 213, illustrates part of the interior of the driver's cab with display panels 214, as well as an exterior 215 of the rail vehicle showing a track 216, and an image A_O1 of the first object O1 (212). The absolute position of the first object O1 (212) is known.

[0064] In a first image (for example, 213 in Fig. 3) More than one first object can be depicted, and the position information of all such depicted first objects can be known. The position of the rail vehicle can then be determined using all relative positions of all first objects, as with reference to Fig. 2 was described in detail. This allows a plurality of estimated rail vehicle positions to be determined (also called hypotheses of rail vehicle positions), which can be used to determine the final rail vehicle position.

[0065] Fig. Figure 4 shows in a schematic top view features of a method for determining speed and position according to an embodiment of the present invention. Fig. 4 shows a situation at a certain time interval after the one in Fig. 1 illustrated situation. Meanwhile, rail vehicle 202 has moved further (into Fig. 2 and Fig. 4 to the right) so that the first object O1 (212) is no longer within the field of view or viewing area 211 of the camera system 201. Instead, a second object O2 is within the field of view of the camera system 201. The camera system 201 then takes a second picture at a second time in order to provide (together with or without the first picture) several pictures taken at different times. In the Fig. In the illustrated embodiment 4, the camera system 201 takes place at the second time point (after the one described in Fig. In the illustrated situation (2), a second image is captured, and furthermore, at a third time point, at which the rail vehicle and the camera system are illustrated in a dashed line, a third image is captured. The relative position rR2 between the camera system 201 and the second object O2 at the second time point, as well as the relative position rR3 between the camera system 201 and the second object O2 at the third time point, are determined by the evaluation system. From the several relative positions rR2, rR3 (in particular the difference rD) and the differences of the recording times (i.e., T3-T2, i.e., the third recording time minus the second recording time), the rail vehicle speed is then calculated.

[0066] Fig. Figure 5 illustrates a top view in the upper section and a second image 215 in the lower section, which, for example, was taken at the second time. In the second image 215, railway sleepers 218 are visible, each of which could function as a second object O2. Alternatively, for example, the pillars 217, which are also visible in the second image 215, could serve as second objects, for which, in particular, no absolute knowledge of the absolute position is required. Second objects only need to be stationary in order to be used for speed determination.

[0067] For the transmission of image information, for example, image information transmission 103 illustrated in Fig. 1. For example, multicasting can be used to transmit or transfer image data to a front and / or a rear evaluation unit within the rail vehicle. Image processing

[0068] The purpose of the image processing block (for example, 104 in Fig. 1) Its purpose is to identify known objects within the camera images. The image processing block can extract information from the camera images and forward this information to the velocity channel and / or the tracking channel. Thus, the images only need to be processed once, and subsequent processing steps use the same information extracted from the images. This increases overall performance because the computationally intensive image processing is performed at an early stage of the position determination process. The image processing block (for example, 104 in Fig. 1) can, for example, identify the following known objects: • Marker signs • Track sleepers • Signal masts • Points • Platforms • Curves

[0069] The type of information extracted can depend on the identified object. If the image processing block detects a marker sign, a unique identifier can be identified, along with the information that the marker sign has been detected. This allows the geometric transformation between the marker sign and the camera center to be calculated and passed to the velocity and tracking channel.

[0070] When the image processing block detects track sleepers, signal masts, points, platforms, or curves, the image processing block calculates the geometric transformation between the detected objects and the camera center and forwards this transformation, along with the information that an object has been identified, to the speed and tracking channel. Speed ​​Channel Concept: Objectives and Purpose

[0071] The objective and purpose of the speed channel is to determine the speed of a rail vehicle. The working assumption is that stationary objects along the track are identified in two camera images. The time elapsed between capturing the two images is known. The stationary objects at the edge of the track change their position in the captured images due to the movement of the rail vehicle. Information about how the object position has changed, combined with the time elapsed between the images, allows the rail vehicle speed to be calculated. Speed ​​channel: Inputs and outputs and operating principle

[0072] The input for this channel is a list of objects identified in a camera image and the corresponding timestamp for that image. The channel extracts the positional information of each object based on the transformation from the camera center to the object and stores this information. Only relative object position information is used; the absolute object position is unknown. As soon as the next image (frame) is available, the positions of identified objects are extracted and stored again. The positions of all objects that were in the current and the previous image are also used to generate several speed hypotheses. The channel outputs a list of speed hypotheses (also referred to as estimated train speeds). The list of speed hypotheses is sorted to correspond to the list of identified objects that is the input.Once an object has been identified for the first time, it is not possible to estimate its speed. In this case, the speed hypothesis for that object is marked as invalid. The speed channel also outputs the timestamp of the input camera image or frame. Used track objects

[0073] The following known and fixed track objects can be identified in a camera image: sleepers or known objects with a known and constant spatial distance, signals, points (including point drives), platforms, axle counters, etc.

[0074] Feature points can also be used in known track objects. Feature points are special visual points such as colored points, steep gradients, or other optical anomalies. Using feature points allows for more speed hypotheses to be generated, especially when no known track objects are visible. However, it must be ensured that there is a high probability that a feature point is a stationary point and not some moving object. Otherwise, the estimated speed would be the difference between the feature point's speed and the train's speed. Tracking channel concept: Goal and purpose

[0075] The purpose of the tracking channel is to determine the absolute position of the rail vehicle. The working assumption is that stationary track objects are identified in the camera image and that the position of the track object is fixed and known. Tracking channel: Input and output

[0076] The tracking channel input is a list of objects that were identified at a specific timestamp. For each object, the list includes the transformation from the camera center to the object and information about the object type. For marker signs, an identifier for the marker image must be provided with the input.

[0077] The tracking channel outputs a list of absolute position hypotheses of the rail vehicle as well as the timestamp of the input information. Track objects used and working principle

[0078] The tracking channel uses the following track object types and has a specific operating principle for each: Marker signs: Each marker sign has a unique ID or identifier. This identifier can be encoded as a QR code, for example. The image processing system can detect and decode the identifier and forward it to the tracking channel. Each identifier, along with the marker sign's fixed and absolute position (including its orientation relative to the tracks), should be known to the system. The tracking channel calculates a railway vehicle position hypothesis based on the known absolute position of the marker sign and the transformation between the object and the camera center, which is provided as input to the tracking channel.

[0079] Light signals: The position(s) of light signals should be known. If there is only one light signal in the field of view and the system knows that there is only one signal pole that is a possible candidate based on the last known absolute position of the rail vehicle and the distance traveled since that position, the tracking channel can and should calculate a rail vehicle position hypothesis based on the known absolute position of the signal pole and the transformation between the object and the camera center, which is provided as input to the tracking channel. Thus, the light signal can be used to determine the absolute position even if the light signal does not have an identifier that uniquely identifies it.

[0080] Track curves: The curvature of the track in curves can also be used for absolute positioning. If, based on the last known absolute position of the rail vehicle and the distance traveled since that position, there is only one curve on the track that is plausible at that point, and a matching curvature is detected in the camera images, this information can be incorporated into the calculation of the absolute position. Key points for detecting track curves are the beginning and end points of the curve.

[0081] Switches: If, based on the last known absolute position of the rail vehicle and the distance traveled since that position, there is only one switch that is plausible at that location, and a switch with corresponding characteristics is detected in the camera images, this information can be included in the calculation of the absolute position. Key points for detecting switches can be frogs, limit markers, and / or switch drives.

[0082] Axle counters: The position of axle counters should be known. If only one axle counter is within the camera system's field of view, and the system knows that there is only one axle counter that is a possible candidate based on the last known absolute position of the rail vehicle and the distance traveled since that position, the tracking channel will or should calculate a rail vehicle position hypothesis based on the known absolute position of the axle counter and the transformation between the object and the camera center, which is provided as input to the tracking channel. Sensor Fusion

[0083] Fig. Figure 6 schematically illustrates part of an evaluation system 620, which may be included, for example, in a system for determining the position of a rail vehicle. The evaluation system 620 comprises a speed channel 621 and a tracking channel or position channel 622. The speed channel 621 receives several estimated rail vehicle speeds 623_1, ... 623_n as inputs. The tracking channel 622 (or position channel) receives a plurality of estimated rail vehicle positions 624_1, ..., 624_m. These estimated speeds or positions are fed into a particle filter 625 in the speed channel, which evaluates the many hypotheses in order to deal with unknown probability density functions and to estimate the highly realistic rail vehicle speed 626.Similarly, the many position hypotheses 624 are fed to a particle filter 627, which calculates the most realistic rail vehicle position 628 from them.

[0084] The system behavior is assumed to be non-linear. Therefore, an unscented Kalman filter (UKF) 630 is used, to which the most realistic values ​​of the speed 626 and position 628 of the rail vehicle are fed. The Kalman filter 630 estimates the rail vehicle position based on outputs from the particle filters 625, 627, and also the last output of the Kalman filter 630 via element 631. Thus, the system is robust against temporary failures of one channel (e.g., the position or tracking channel, or the speed channel, or both). The final result of the rail vehicle position is in Fig. 6 is designated by reference numeral 632.

[0085] Fig. Figure 7 illustrates, in a coordinate system with an abscissa representing the positional coordinate of the rail vehicle and an ordinate representing the positional uncertainty, a curve 633 which represents the uncertainty of the rail vehicle's position 632. It is assumed that a rail vehicle has cameras arranged or installed in both the front and rear driver's cabs, both of which are used for position determination.

[0086] At position X1, a first object O1, specifically a marker sign with a QR code, is positioned along the track. This sign is used for position determination by both the front and rear cameras. The front camera detects object O1 at position X1 within a spatial area 634, and the rear camera detects object O1 within a spatial area 635.

[0087] From curve 633 into Fig.Figure 7 shows that the positional inaccuracy (minimum at 638) is very low in the combination of spatial regions 634 and 635, i.e., around the first object O1, and increases continuously outside these spatial regions 634 and 635. In these outer regions 636 and 637, the position of the rail vehicle is additionally determined based on estimated rail vehicle speeds according to embodiments of the present invention, which is why the uncertainty is higher here.

[0088] According to embodiments of the present invention, position determination is supported by various sources. The positional inaccuracy or uncertainty can be adapted to operational requirements: high values ​​can be provided, for example, for journeys from one station to another, and low values ​​can be provided when the rail vehicle is in a station, on a platform, or in sidings. An adjustable positional uncertainty can reduce engineering effort and track installation costs. High accuracy can be achieved as long as the known objects are within the line of sight. Obstacle detection

[0089] An additional feature in the embodiments of the present invention is that the system can be used for obstacle detection. Software can be adapted to analyze the image captured by the image acquisition system and raise an alarm if anything unexpected is detected on the track or if the expected track pattern is disrupted. The advantage compared to existing or conventional obstacle detection systems is that an obstacle can be detected before physical contact between the obstacle detector and the object actually occurs.

[0090] In driverless train systems, scenarios can occur that temporarily prevent driverless operation. In these cases, the train must be remotely controlled from the control center to a location where driverless operation is possible again. This remote control operation can be improved if the ATS operator is aware of the situation in front of the train. A software modification of the presented odometry system can allow the camera image to be transmitted to the ATS HMI. This enables the ATS operator to monitor the RSRM train's journey and take action if necessary. For example, people, animals, or other stationary objects may be recognizable and detectable in the camera image, which can be used to enhance train safety. QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] US 10,710,621 B2

[0005]

Claims

[1] Method of determining the position of a rail vehicle (202), the method comprising: Recording, in particular at a first point in time, at least one first image (213) of an external environmental area by means of an image recording system (201) arranged on the rail vehicle (202); analyzing the first image (213) to identify in the first image an image of at least one first object (O1) with a previously known first object position (rO1); Determining a geometric first relative position (rR1) of the first object (O1) relative to the image recording system (201) based on the identified image of the first object; Determining an estimated rail vehicle position (rSg) based on the first object position (rO1) and the first relative position (rR1); Determining the rail vehicle position (632) based on at least the first estimated rail vehicle position. [2] The method of claim 1, wherein analyzing the first image comprises: Identifying in the first image (213) a plurality of images of a plurality of first objects with a plurality of previously known first object positions; Determining a plurality of geometric first relative positions of the plurality of first objects relative to the image acquisition system based on the identified images of the plurality of first objects; Determining a plurality of estimated rail vehicle positions (624_1,...,624_m) based on the plurality of first object positions and the plurality of first relative positions; Determining the rail vehicle position (632) based on the plurality of estimated rail vehicle positions, in particular using a particle filter. [3] Method according to one of the preceding claims, further comprising: Determining an estimated rail vehicle speed (623_1) at the first time (t1) and / or at a second time (t2), later than the first time, in particular by means of image analysis; Determining the rail vehicle position (632) at the second time, later than the first time, based at least on the determined rail vehicle position at an earlier time than the second time, in particular at the first time, and at least the estimated rail vehicle speed at an earlier time than the second time and / or at the second time. [4] Method according to one of the preceding claims, wherein determining the estimated rail vehicle speed at the first and / or second time comprises: capturing, at the second time, at least one second image (215) to thereby provide a plurality of images (213, 215) captured at different times; Analyzing the plurality of images to determine the estimated rail vehicle speed (623_1) at one or more times, in particular at the first and / or a second time. [5] A method according to any one of the preceding claims, wherein analyzing the plurality of images comprises: Identifying images of at least one second object (O2) in the plurality of images (213, 217); Determining respective geometric second relative positions (rR2, rR3) of the second object (O2) relative to the image recording system (201) in the plurality of images based on the identified images of the second object (O2); Deriving at least the estimated rail vehicle speed (623_1) at one or more times from the second relative positions (rR2, rR3) assigned to a plurality of images, in particular differences thereof, and the various recording times, in particular differences thereof. [6] A method according to any one of the preceding claims, wherein analyzing the plurality of images comprises: identifying images of a plurality of second objects in the plurality of images; Determining respective geometric second relative positions for the plurality of second objects relative to the image acquisition system in the plurality of images based on the identified images of the plurality of second objects; Deriving at least the estimated rail vehicle speed (626) at one or more times from the second relative positions assigned to the plurality of images and the plurality of second objects, in particular differences thereof, and the various recording times, in particular differences thereof. [7] Method according to one of the preceding claims, wherein the rail vehicle position is determined based on one or more estimated rail vehicle speeds (623_1,...,623_n), if present, and / or one or more estimated rail vehicle positions (624_1,...,624_m), if present, in particular using a Kalman filter. [8] Method according to one of the preceding claims, wherein the rail vehicle speed (626) at a specific time is derived by weighting the plurality of estimated rail vehicle speeds (623_1,...,623_n) at the specific time, in particular by means of a particle filter, and / or wherein the rail vehicle position (628, 632) at a specific time is derived by weighting the plurality of estimated rail vehicle positions (624_1,...,624_m) at the specific time, in particular by means of a particle filter, and by weighting the plurality of estimated rail vehicle speeds (623_1,...,623_n) at the specific time, in particular by means of a particle filter and / or Kalman filter. [9] Method according to one of the preceding claims, wherein the respective relative position (rR1, rR2, rR3) of the first (O1) or second object (O2) relative to the image recording system (201) is derived from a distance of the respective imaged object from the image recording system and image position of the image in the respective image, in particular using the field of view (211) of the image recording system, In particular, the distance is determined from an extension of the image and an actual (previously known) extension of the object. [10] Method according to one of the preceding claims, whereby the process is carried out continuously over time, wherein the one or more estimated rail vehicle positions and / or the one or more estimated rail vehicle speeds are stored with a time stamp and wherein the rail vehicle position at a specific time is determined based on one or more estimated rail vehicle positions with a time stamp up to the specific time and / or one or more estimated rail vehicle speeds with a time stamp up to the specific time, in particular using a Kalman filter. [11] Method according to one of the preceding claims, wherein the image recording system (201) comprises one or more cameras; and / or wherein the image recording system (201) is arranged in a driver's cab at one and / or both ends of the train; and / or wherein the first and / or second image is composed of partial images taken by several cameras and / or comprises several partial images; and / or wherein a field of view (211) of the image recording system (201) comprises an external environmental area lying in the direction of travel and / or in the opposite direction of travel of the rail vehicle with an opening angle between 90° and 180°; and / or wherein a field of view (211) of the image recording system (201) comprises a track area and / or a track edge and / or an area of ​​signal signs and / or masts on both sides of the track; and / or wherein the image acquisition system is sensitive to visible light and / or infrared and / or ultraviolet radiation. [12] Method according to one of the preceding claims, wherein the, in particular calibrated, image recording system (201) has a first camera at a first rail vehicle end and a second camera at a second rail vehicle end, wherein the method can be carried out according to at least one of the following operating modes: In a first operating mode, the images taken by the first or second camera are evaluated to determine the rail vehicle position; In a second operating mode, the images taken by the first and second cameras are evaluated to determine the rail vehicle position; In a third operating mode, a first evaluation of the images taken by the first camera is carried out in order to determine a first rail vehicle position, and a second evaluation of the images taken by the second camera is carried out in order to determine a second rail vehicle position; In a fourth operating mode, a first evaluation of the images recorded by the first camera is carried out in order to determine a first rail vehicle position, a second evaluation of the images recorded by the second camera is carried out in order to determine a second rail vehicle position, wherein the first or the second evaluation is checked for plausibility by means of the respective other evaluation in order to determine the rail vehicle position. [13] Method according to one of the preceding claims, wherein the first and / or the second object (O1, O2) comprises at least one of the following: a (locally known) human-readable marking board; a (locally known) machine-readable marking panel, in particular having an ID and / or QR code; a signal mast; a switch; a railway sleeper; a switch; a platform; an axle counter; a marking on a stationary object, especially a protection point. [14] System (100) for determining the position of a rail vehicle (202), the system comprising: an image recording system (201) arranged on the rail vehicle, which is designed to record at least a first image (213) of an external environmental area; an image analysis system (104) configured to evaluate the first image to identify in the first image an image of at least one first object with a previously known first object position (rO1); a determination system (106, 105, 108) which is designed: to determine a geometric first relative position (rR1) of the first object (O1) relative to the image recording system (201) based on the identified image of the first object (O1); to determine an estimated rail vehicle position (rSg) based on the first object position (rO1) and the first relative position (rR1); to determine the rail vehicle position (632) based on at least the first estimated rail vehicle position. [15] Rail vehicle (202), comprising: a driver's cab at at least one end of the rail vehicle; a system (100) according to the preceding claim, wherein the image recording system is installed in particular within the driver's cab.

Images