Method and device for collision monitoring for a rail vehicle

Abstract

A collision monitoring method for a rail vehicle (100) comprises determining current track alignment data representing a track section (112) of a line (102) currently being traveled by the rail vehicle (100), selecting predetermined track alignment data for the line (102) from a plurality of track alignment data stored for different lines using the current track alignment data, updating the current position of the rail vehicle (100) in the predetermined track alignment data, and determining the probability of a collision between the rail vehicle (100) and an object (120, 122) located in a track section upstream of the current position using the predetermined track alignment data.

Description

State of the art

[0001] The invention is based on a device or method according to the class of the independent claims. The present invention also relates to a computer program.

[0002] A rail vehicle's environmental sensors can be used to detect objects in the area in front of the vehicle. Appropriate analysis can determine whether a detected object is relevant to a potential collision. Disclosure of the invention

[0003] Against this background, the approach presented here presents a method, a device using this method, and finally a corresponding computer program according to the main claims. The measures listed in the dependent claims allow advantageous further developments and improvements of the device specified in the independent claim.

[0004] A pre-stored track layout of a route traveled by a rail vehicle can be used to determine whether there is a probability of collision between the rail vehicle and an object detected in the area of ​​the route. Advantageously, the pre-stored track layout can be selected based on only a section of the route currently traveled by the rail vehicle.

[0005] A method for collision monitoring for a rail vehicle or for determining a collision probability of the rail vehicle with an object comprises the following steps: Determining current rail route data representing a section of a route currently traveled by the rail vehicle; selecting rail course data predetermined for the route from a plurality of rail course data stored for different routes using the current rail course data; Tracking a current position of the rail vehicle in the predetermined rail track data; and Determining a collision probability of the rail vehicle with an object located in a track area in front of the current position using a range of the predetermined rail course data assigned to the track area.

[0006] The rail vehicle can, for example, be a traction vehicle of a tram. The route can represent a branch or a line of a route network. By way of example only, the route section can represent an initial section of the route that is traveled at the start of a journey by the rail vehicle. Alternatively, the route section can also represent a later section of the route. While the route section is being traveled, sensor data from a sensor device on the rail vehicle can be continuously recorded and evaluated. From this, the route of the rails in the traveled section of the route can be determined in a known manner and mapped using the rail route data. The rail route data can relate to the entire route section, i.e. map a curvature along the entire route section already traveled.In the selection step, those rail route data can be selected from the stored rail route data that have a range that can be aligned with the current rail route data. Known methods, such as correlation methods, can be used for this purpose. In this way, existing rail route data for the entire route currently being traveled can be determined very easily, even if only the section of the route has been traveled, which may be very short compared to the total length of the route. Once the predetermined rail route data assigned to the currently traveled route has been determined, the current position of the rail vehicle can be continuously updated using the predetermined rail route data.In this way, the predetermined rail course data can be used at any time to obtain information about the course of the rails at the current position of the rail vehicle. For tracking purposes, for example, a comparison can be made between the course of the rails mapped by the predetermined rail course data and a currently determined course of the rails, or distance information contained in the predetermined rail course data can be used, which can be compared with a distance traveled. The object can be a person, for example, or an object that could cause damage to the rail vehicle. The object and a position of the object relative to the rail vehicle can be determined using an environment detection device of the rail vehicle.Advantageously, the predetermined rail path data can be used to determine whether the position of the object is within the rails, for example, within a track bed, or outside an area traveled by the rail vehicle. If the object is within the traveled area, a high collision probability can be assumed, which may lead to an appropriate response, for example, braking of the rail vehicle. If the object is outside the traveled area, a low or no collision probability can be assumed.

[0007] According to one embodiment, in response to the step of determining the risk of collision, in particular in response to a detected risk of collision, a collision signal is provided, which can be embodied as an information signal and / or a control signal. It is conceivable that the collision risk is provided to a control device of the rail vehicle in order to issue a warning to an operator of the rail vehicle and / or initiate automated emergency braking in response to the collision signal.

[0008] In the selection step, a correlation can be performed between the current rail path data and the plurality of stored rail path data. This allows a proven method to be used to select the predetermined rail path data.

[0009] In the selection step, the predetermined rail path data can be selected using position data representing a current position of the rail vehicle. The position data enables, for example, a preselection of rail path data or a verification option for the selected predetermined rail path data.

[0010] In the selection step, the predetermined rail path data can be selected using environmental data representing a constellation of stationary objects in the current environment of the rail vehicle. The environmental data can be acquired, for example, using an environmental detection device of the rail vehicle. The environmental data can be used in addition to or alternatively to position data.

[0011] The method may include a step of deleting the current rail path data if the predetermined rail path data was selected in the selecting step. This allows storage space to be freed up.

[0012] The step of determining current rail route data can be continued while traveling the route until the predetermined rail route data is selected in the selection step. As a result, the current rail route data is increasingly expanded as the route is traveled. Accordingly, the selection step can be performed repeatedly, for example, at predetermined time intervals, so that the most recent version of the current rail route data can always be used to select the predetermined rail route data.

[0013] The method may include a storing step in which the current rail path data can be stored as rail path data stored for the route. The storing step can occur if, prior to the route being completely traveled, the predetermined rail path data have not been selected in the selecting step. This may be the case if no rail path data had yet been stored for the route just traveled. By storing, the plurality of stored rail path data can be supplemented by the current rail path data that has just been determined for the route just traveled. This can be used when the route is traveled again in the future.

[0014] In the determination step, sensor signals can be read via an interface to a sensor device on the rail vehicle. Using the sensor signals, a curvature of the rails along the track section traveled can be determined as the current rail path data. The sensor device can comprise, for example, a camera or a radar sensor. This allows known methods for environmental detection to be used to detect the rails in front of the rail vehicle.

[0015] Advantageously, the sensor signals can be used to correct a longitudinal error during the tracking step. For example, the current position of the rail vehicle can be corrected at the beginning of a detected bend in the rails. A longitudinal error can be understood, in particular, as a deviation of a specific current position from the actual current position of the rail vehicle along the track traveled or to be traveled.

[0016] This method can be implemented, for example, in software or hardware or in a mixed form of software and hardware, for example in a control unit.

[0017] The approach presented here further provides a device designed to perform, control, or implement the steps of a variant of a method presented here in corresponding devices. This embodiment of the invention in the form of a device also allows the problem underlying the invention to be solved quickly and efficiently.

[0018] For this purpose, the device can have at least one computing unit for processing signals or data, at least one memory unit for storing signals or data, at least one interface to a sensor or an actuator for reading sensor signals from the sensor or for outputting data or control signals to the actuator, and / or at least one communication interface for reading or outputting data embedded in a communication protocol. The computing unit can be, for example, a signal processor, a microcontroller, or the like, wherein the memory unit can be a flash memory or a magnetic storage unit.The communication interface can be designed to read in or output data wirelessly and / or wired, wherein a communication interface that can read in or output wired data can read this data, for example, electrically or optically from a corresponding data transmission line or output it to a corresponding data transmission line.

[0019] In this case, a device can be understood as an electrical device that processes sensor signals and outputs control and / or data signals depending on them. The device can have an interface, which can be implemented in hardware and / or software. In a hardware implementation, the interfaces can, for example, be part of a so-called system ASIC, which contains a wide variety of functions of the device. However, it is also possible for the interfaces to be separate integrated circuits or to consist at least partially of discrete components. In a software implementation, the interfaces can be software modules that are present, for example, on a microcontroller alongside other software modules.

[0020] Also advantageous is a computer program product or computer program with program code that can be stored on a machine-readable carrier or storage medium such as a semiconductor memory, a hard disk memory or an optical memory and is used to carry out, implement and / or control the steps of the method according to one of the embodiments described above, in particular when the program product or program is executed on a computer or a device.

[0021] Examples of the approach presented here are shown in the drawings and explained in more detail in the following description. It shows: Fig. 1 a schematic representation of an embodiment of a rail vehicle; Fig. 2 an embodiment of rail route data; Fig. 3 an embodiment of a route; Fig. 4 is a flowchart of a method according to an embodiment; and Fig. 5 a flowchart of a method according to an embodiment.

[0022] In the following description of advantageous embodiments of the present invention, the same or similar reference numerals are used for the elements shown in the various figures and having a similar effect, whereby a repeated description of these elements is omitted.

[0023] Fig. Figure 1 shows a schematic representation of an embodiment of a rail vehicle 100 traveling along a track 102. The track 102 is represented by two rails 104, 106.

[0024] The rail vehicle 100 has a device 110 for collision monitoring. The device 100 is designed to determine current rail course data representing a section 112 of the route 102. The route 102 is, for example, the entire route to be traveled during the current journey, for example, between two train stations. The route section 112 represents only a section of the route 102 that has already been detected by the rail vehicle 100 during the previous duration of the journey. To detect the course of the route 102, the rail vehicle 100 has a sensor device 114. For example, the sensor device 114 is designed to detect an environment in an area in front of the rail vehicle 100 and to provide sensor signals 116 depicting the environment to the device 110.The device 110 is designed to determine a curvature of the rails 104, 106 along the traveled section 112 as the current rail course data using the sensor signals 116.

[0025] The device 110 is configured to use the current rail course data to select, from a plurality of rail course data stored for different routes, those rail course data associated with the currently traveled route 102. These selected rail course data represent predetermined rail course data for the route 102 and include information about the course of the rails 104, 106 over the entire course of the route 102. Optionally, the device 110 comprises a storage device 118 in which the plurality of rail course data stored for different routes are stored.

[0026] If the predetermined rail path data could be selected using the current rail path data, the device 110 is configured to continuously update the current position of the rail vehicle 100 in the predetermined rail path data as it continues to travel along the route 102. As a result, a section of the predetermined rail path data associated with the current position can always be accessed for the respective current position.

[0027] The device 110 is designed to use the area of ​​the predetermined rail course data assigned to the current position to determine a collision probability of the rail vehicle 100, here for example with a first object 120 and a second object 122. If the objects 120, 122 and their positions are detected, for example using the sensor device 114, it may not be possible to determine solely based on image analysis whether the objects 120, 122 are located on the rails 104, 106 or at a sufficiently large distance from the rails 104, 106. In order to be able to reliably determine the collision probability, the device 110 is designed to use the predetermined rail course data to determine whether the objects 120, 122 are located on the track 102 or next to the track 102.For this purpose, the device 110 is designed, for example, to enter the positions of the objects 120, 122 in a suitable coordinate system and to compare them with a course of the rails 104, 106 in the coordinate system resulting from the predetermined rail course data. For example, the device 110 is designed to recognize, using the predetermined rail course data, that the first object 120 is located at a safe distance outside a track bed of the rails 104, 106 and that a risk of collision therefore exists. With regard to the second object 122, however, the device 110 is designed to recognize, using the predetermined rail course data, that the second object 122 is located directly in the area of ​​the rails 104, 106 and that a risk of collision therefore exists. Optionally, the device 110 is designed to provide a collision signal 124 in response to a detected risk of collision.For example, the collision signal 124 is provided to a control device 126 of the rail vehicle 100, which is designed to issue a warning to an operator of the rail vehicle 100 or to initiate an automated emergency braking in response to the collision signal 124.

[0028] According to one exemplary embodiment, the described approach is implemented as an assistance system or automated system for a collision avoidance system. For this purpose, information is acquired as to whether obstacles, such as objects 120, 122, are located within a clearance profile. For this purpose, the obstacles are detected by an environmental detection sensor system, such as sensor device 114, and the position of the rails 104, 106 being traveled on is determined, since the clearance profile can be derived from this, thus preventing a reaction to irrelevant obstacles, such as the first object 120.

[0029] A common approach is to use a camera as part of the sensor device 114 to record the track layout, here track 102. However, this approach has the disadvantage that the tracks are difficult to identify, particularly on secondary lines, due to overgrowth, fallen leaves, and rust. Bushes also cast shadows on curves, making it impossible to identify the track layout. Even at night, during heavy rain, hail, and snowfall, the track layout is difficult to detect or can only be detected at a short distance. Furthermore, the camera's resolution is usually optimized for the objects 120, 122 to be detected, which are often larger than the more delicate track structures.This effect can be compensated for by stronger focusing (lower FoV), but on the one hand one then loses the possibility of following the track in tight curves and on the other hand the line of view of the camera becomes increasingly flatter into the rails 104, 106 the greater the distance to be estimated, which causes the performance to decrease further with increasing distance.

[0030] If a complete or partially complete rail network map has been (preliminarily) drawn, localization can be performed on this map. Together with line / destination information, the track layout can then be read from the map. However, using GNSS receivers (Global Navigation Satellite System receivers), localization approaches with environmental sensors, or balises for this purpose incurs additional costs, resulting in significant computing and storage requirements, and thus often requires more expensive computing hardware, or is simply unavailable or requires expensive infrastructure investments.

[0031] The described approach takes advantage of the fact that traction vehicles, for example, rail vehicle 100, frequently travel the same routes as route 102 on a recurring basis. This applies, for example, to trams. The track path traveled on route 102 is recorded during the first trip and possibly again later by the environmental detection sensor system and stored in a rail buffer, for example, the storage device 118, in a manner that is both computationally and storage-efficient. For all subsequent trips on this route 102, the track path is read from the rail buffer and is then available in the system. This avoids performance problems and additional costs. Thus, according to one embodiment, the described approach is based on a combination of a rail buffer with environmental detection sensors.

[0032] Using vehicle environment sensors, such as sensor device 114, the track layout is recorded in a buffer. Using the vehicle environment sensors, the driver can locate the recorded section of track again and read and use the track layout from the cache, in this case, the storage device 118.

[0033] In particular, in cases of poor line of sight, it is still possible to react cost-effectively to objects 120, 122 detected by a robust sensor principle, such as radar of the sensor device 118, even if the camera of the sensor device 118 cannot currently detect a rail 104, 106, for example, due to environmental conditions. A reaction without a known rail path would otherwise generally not be possible, since an object could also be located outside the vehicle's own clearance profile and thus would not be relevant to a collision.

[0034] Optionally, the described approach allows for the storage of location-based data that results in a specific reaction pattern, such as a location-based speed limit or a precise stop.

[0035] Fig. Figure 2 shows an embodiment of rail course data 230 in a diagram in which a traveled distance is plotted on the abscissa and a curvature of the rails of the traveled distance is plotted on the ordinate. The rail course data 230 correspond, for example, to the Fig. 1 described current rail profile data, which were determined using the sensor device of the rail vehicle.

[0036] According to one embodiment, the rail vehicle is equipped with a sensor device in the form of a vehicle environment sensor system. This sensor system is capable of determining the distance traveled and a change in the yaw angle, for example, using a camera. According to one embodiment, the rail curvature along the track is determined from the yaw angle over time and the distance traveled over time. For this purpose, the known equations for transforming the sensor coordinate systems into the bogie coordinate system of the rail vehicle are generally used.

[0037] The rail vehicle travels a section of track and stores the rail curvature along the traveled distance in the buffer. When the rail vehicle later travels a route, the course is again recorded in a temporary memory area. A correlation method is used to determine whether the currently traveled route corresponds to one of the entries in the rail buffer. Three cases can be distinguished: If this is clearly the case, the temporary memory area is released again and the track path from the cache is used. If this is not yet clear, the process continues for a longer period and the correlation is repeated until it is clear. If there is no match, the route traveled is stored in the buffer at the end of the line.

[0038] The cache generally contains a finite number of entries. According to one embodiment, known methods for managing the cache are used, for example, deleting the least recently used entry in case of a lack of space.

[0039] For lines that share a section of track, a graph structure can be used in the buffer. In this case, it is unclear which track path will be chosen before individual switches. However, this is indicated, so that the buffer is temporarily deactivated immediately before such a switch, which separates two originally shared lines, until the path after the switch is again clear through correlation.

[0040] Fig. 3 shows an embodiment of a rail course 302, such as it is during a journey of a rail vehicle along a route 102, for example the one shown in FIG. Fig. 1. For example, the rail path 302 is stored as a yaw rate from a video sensor device along with a speed / distance from the radar of the sensor device. The sensor device represents a live sensor system. The rail path 302 is shown starting from a departure 340 of the rail vehicle. To the left of the rail path 302, a variance 304 of the longitudinal position is shown, and to the right of the rail path 302, a lateral error 306 is shown.

[0041] The presented method, in this form, works optimally only for noise-free or low-noise detection of the traveled distance 102 and the yaw angle change using the environment sensors. Noise is generally taken into account. Thus, with noisy yaw angle changes or noisy traveled distances, a perfect correlation never results, and threshold values ​​for correlation coefficients or correlation coefficient differences or correlation coefficient ratios are used.

[0042] Segments with constant curvature and segments with affinely variable curvature occur.

[0043] The method takes advantage of the fact that an error in the track 102 (longitudinal error) often has no impact on the resulting lateral error 306, which is relevant for the object-rail relationship and thus the warning or braking decision. The longitudinal error can be reset for each reliably detected change in curvature, since the currently detected curve can be aligned with the curve stored in the buffer. Furthermore, longitudinal errors on tracks with constant curvature do not affect the lateral error 306, since the rail segment slightly forward or slightly backward is identical.

[0044] Out of Fig. 3 shows how the longitudinal error along the illustrated rail path 302 is initially large due to the lack of initialization, starting at the bottom left, indicated by reference numeral 342, until the first curve 344 is reached. There, the longitudinal error then becomes low due to the reset, indicated by reference numeral 346.

[0045] Over the duration of the straight line following curve 344, the longitudinal error increases continuously on average, identified by the reference numerals 348, 350, 342, until the next curve 352, where the longitudinal error is reset again.

[0046] Although the longitudinal variance is high over long distances, after successful initialization, a critical lateral error 306, identified by the reference symbol 354, results only in a few section of the route, so that the lateral error 306, especially after the first curve 344, is predominantly in the non-critical range, identified by the reference symbol 356.

[0047] According to one embodiment, the longitudinal error is optionally corrected using mini-SLAM, i.e., simultaneous positioning and mapping. To limit the longitudinal error and to ensure reliable rail forecasting even in the event of curvature changes after long sections of constant curvature, such a mini-SLAM approach is used. For this purpose, a few measured variables of locally unique, selected, stationary objects, e.g., approximately periodic structures such as the position of overhead line masts, are stored in the rail buffer when the longitudinal error exceeds a certain threshold. When the buffer is read or used, the selected objects are retrieved. The longitudinal difference between the stored and retrieved object is used to correct the distance traveled.

[0048] Fig. 4 shows a flowchart of a method according to an embodiment. The method is carried out, for example, using devices of the Fig. 1 described rail vehicle.

[0049] If the position in the rail buffer is still unknown, there are several ways to perform an initialization 460, for example using correlation, GNSS or Mini-Siam.

[0050] Initialization 460 can be performed using correlation as described above, but this may require a somewhat longer path to become unique.

[0051] The initialization 460 can be carried out additionally or alternatively by means of satellite-based positioning, for example using a simple GNSS sensor which, although not track-accurate, either makes the correlation process directly unambiguous or greatly reduces the number of possible options and thus greatly reduces the distance to be covered.

[0052] Initialization 460 can also be performed using mini-SLAM. For this purpose, an atypical constellation of stationary objects can be searched for immediately after departure or the beginning of a line. It is advisable to avoid (exclusively) approximately periodic structures, such as a place-name sign or a constellation of building reflections and sign / post reflections. This constellation should be as unambiguous as possible for all line beginnings. This also significantly reduces the number of possible options and thus the distance to be covered.

[0053] After a corresponding optional initialization 460, a traveled distance and a yaw angle are determined in a step 462 using sensors of the rail vehicle. In a step 464, a curvature is determined based on the traveled distance and the yaw angle, in a step 466, curve detection is performed based on the curvature, in a step 468, a correlation of buffer entries is performed, and in a step 470, rail information is used.

[0054] Advantageously, this makes it possible for the rail vehicle to react to an object in its own clearance profile even though the track layout is not visible, for example in a curve with low bushes obscuring the view of the tracks, and even though no other type of localisation, such as a GNSS antenna or a previously conducted reference measurement run, is available.

[0055] Fig. 5 shows a flowchart of an embodiment of a method for collision monitoring for a rail vehicle. The method is carried out, for example, using devices of the type described in Fig. 1 described rail vehicle.

[0056] In a step 501, current rail course data are determined for the section of the route currently traveled by the rail vehicle, as can be seen by way of example from Fig. 2. For this purpose, for example, sensor signals are read via an interface to a sensor device of the rail vehicle and used to determine a curvature of the rails along the track section traveled as the current rail course data.

[0057] In a step 503, the current rail course data are used to select predetermined rail course data for the current route from a plurality of rail course data stored for different routes. Optionally, a correlation is carried out between the current rail course data and the plurality of stored rail course data in order to select the predetermined rail course data. Optionally, position data representing a current position of the rail vehicle and, additionally or alternatively, environmental data representing a constellation of stationary objects in a current environment of the rail vehicle are additionally used to select the predetermined rail course data. The constellation of stationary objects can also be used to determine the current position or to make a preselection of stored rail course data.

[0058] In step 505, the current position of the rail vehicle is continuously tracked using the predetermined rail profile data. Optionally, a longitudinal error is corrected using the sensor signals.

[0059] In step 507, the predetermined rail path data are used to determine a collision probability of the rail vehicle with an object detected in front of the rail vehicle. For this purpose, a region of the predetermined rail path data is used that is associated with the track area in which the detected object is located.

[0060] According to one embodiment, the determining step 501 continues to be performed while the route is traveled until the predetermined rail course data has been selected in the selecting step 503. Optionally, the method comprises a step 509 in which the current rail course data is deleted if the predetermined rail course data was selected in the selecting step 503. Optionally, the method comprises a step 511 of storing the current rail course data as rail course data stored for the route if no predetermined rail course data could be selected in the selecting step 503 until the route has been completely traveled. This is the case, for example, if no rail course data has previously been stored for this route, or if significant changes have occurred to the route in the meantime.

Claims

[1] A method for collision monitoring for a rail vehicle (100), the method comprising the following steps: Determining (501) current rail course data (230) representing a section (112) of a route (102) currently traveled by the rail vehicle (100); Selecting (503) rail course data predetermined for the route (102) from a plurality of rail course data stored for different routes using the current rail course data (230); Tracking (505) a current position of the rail vehicle (100) in the predetermined rail course data; and Determining (509) a collision probability of the rail vehicle (100) with an object (120, 122) located in a track area in front of the current position, using a region of the predetermined rail course data assigned to the track area. [2] The method according to claim 1, wherein in the step (503) of selecting, a correlation is performed between the current rail course data (230) and the plurality of stored rail course data in order to select the predetermined rail course data. [3] Method according to one of the preceding claims, in which the step (503) of selecting the predetermined rail course data is selected using position data representing a current position of the rail vehicle (100). [4] Method according to one of the preceding claims, wherein in the step (503) of selecting the predetermined rail course data are selected using environmental data representing a constellation of stationary objects in a current environment of the rail vehicle (100). [5] Method according to one of the preceding claims, comprising a step (509) of deleting the current rail course data (230) if the predetermined rail course data were selected in the step (503) of selecting. [6] Method according to one of the preceding claims, in which the step (501) of determining is continued during travel of the route (102) until the predetermined rail course data are selected in the step (503) of selecting. [7] Method according to one of the preceding claims, comprising a step (511) of storing the current rail course data (230) as rail course data stored for the route (102) if the predetermined rail course data were not selected in the step of selecting until the route (102) has been completely traveled. [8] Method according to one of the preceding claims, wherein in the step (501) of determining, sensor signals (116) are read in via an interface to a sensor device (114) of the rail vehicle (100) and, using the sensor signals (116), a curvature of rails (104, 106) along the traveled section of track (112) is determined as the current rail course data (230). [9] Method according to claim 8, wherein in the tracking step (505) a longitudinal error is corrected using the sensor signals (116). [10] Device (110) which is arranged to carry out and / or control the steps of the method according to one of the preceding claims in corresponding units. [11] Computer program which is designed to carry out and / or control the steps of the method according to one of claims 1 to 9. [12] A machine-readable storage medium on which the computer program according to claim 11 is stored.

Images