Method for magnetic field-based detection of a reference point

Abstract

The invention relates to a method for magnetic field-based detection of a reference point when performing multiple series of measurements to create a magnetic field map for a rail network. The method comprises the following steps: A first series of measurements is performed to determine a magnetic field at multiple reference points within a specified section of the rail network. A second series of measurements is performed to determine the magnetic field at multiple reference points within the specified section of the rail network. At each of the reference points, the magnetic field is detected using a magnetic field sensor and the position of the reference point is detected using a position sensor. A local magnetic field signature is recorded at each reference point and stored in a database. Each local magnetic field signature represents the local magnetic field in the area around the reference point.At least one magnetic field signature recorded during the second measurement series is compared with the magnetic field signatures recorded during the first measurement series. Identity between a reference point recorded within the second measurement series and a reference point already recorded in the first measurement series is detected if the magnetic field signature to the reference point from the second measurement series exhibits a minimum similarity to the magnetic field signature to the reference point from the second measurement series.

Description

The present invention relates to a method for magnetic field-based detection of a reference point when carrying out a plurality of measurement series for creating a magnetic field map for a rail network. The invention further relates to a method for creating a magnetic field map, in which the magnetic field-based detection of a reference point according to the invention is used. In addition, the present invention relates to a method for magnetic field-based position determination for a rail vehicle, in which a magnetic field map produced by the method according to the invention is used.Position determination methods play a particularly important role in various fields of application. For example, the precise determination of the position of a rail vehicle is indispensable if it is intended to be autonomously controlled or if the position data is intended to be used for optimizing the traffic guidance.Various positioning methods are widely known. Frequently, such methods are based on global navigation satellite systems (also referred to as global navigation satellite system or GNSS). However, this approach cannot be implemented, or can be implemented only to a limited extent, in certain scenarios. For example, GNSS methods cannot be used in tunnels. Therefore, in certain scenarios, it may be preferable to rely on magnetic field-based methods.Magnetic field-based methods for position determination are known in various embodiments. Some of these methods use a magnetic field map in which information about a magnetic field is stored at different reference points. The magnetic field may be based on the earth's magnetic field or alternatively may be caused by magnets actively provided in a particular environment to generate a (locally) significantly varying magnetic field. If a precise magnetic field map is present for a specific environment, within which the magnetic field differs at different positions, the current position of an object can be deduced by knowing the magnetic field.However, the precise creation of magnetic field maps presents a particular challenge in practice. This is particularly due to the fact that, for the creation of a magnetic field map, on the one hand, the magnetic field at individual reference points must be known, but on the other hand, the exact position of these reference points must also be known. However, this presupposes that the position of the reference point can be determined with sufficient accuracy without recourse to GNSS data.Various methods are known from the prior art for creating the magnetic field maps mentioned. In particular, graph-based methods for simultaneous position determination and map creation (also known as graph simultaneous localization and mapping or graph SLAM) have proven to be suitable for this purpose (see [1], [2]). In these methods, a route is traversed several times. Each time the distance is covered, a series of measurements is carried out, in which the magnetic field is measured at predetermined reference points (also referred to as nodes) (position and magnetic field of the respective reference point). From the information that was determined within the individual measurement series, the probable position of the individual reference points is then determined using statistical means.In the above method, it is necessary to match the reference points detected in a measurement series with those from previous measurement series in order to identify reference points already recorded previously. If a reference point detected during a measurement series is recognized in a subsequent measurement series, this information can be used to define the position of the reference points within the map data. However, the detection of the reference points presents a non-trivial challenge in practice.It is therefore an object of the present invention to provide a method which allows reliable detection of a reference point when carrying out a plurality of measurement series for creating a magnetic field map for a rail network.To achieve the object mentioned above, the present invention proposes a method for detecting a reference point on the basis of a magnetic field when a plurality of measurement series are carried out in order to generate a magnetic field map for a rail network, wherein the method has the following steps:performing a first series of measurements for determining a magnetic field at a plurality of reference points within a predetermined section of the rail network, wherein at each of the reference points the magnetic field is detected using a magnetic field sensor and the position of the reference point is detected using a hodometer, wherein at each reference point a local magnetic field signature is recorded and stored in a database, and each local magnetic field signature reproduces the local magnetic field in the environment around the reference point, wherein the magnetic field signatures recorded during the first series of measurements have a length I sig1 ;performing a second series of measurements for determining the magnetic field at a plurality of reference points within the predetermined section of the rail network, wherein at each of the reference points the magnetic field is detected using the magnetic field sensor and the position of the reference point is detected using a hodometer, wherein a local magnetic field signature is recorded for each reference point, and each local magnetic field signature reproduces the local magnetic field in the environment around the reference point, wherein the magnetic field signatures recorded during the second series of measurements have a length I sig2 ;matching at least one magnetic field signature recorded during the second measurement series or a section of this magnetic field signature with the magnetic field signatures recorded during the first measurement series; anddetecting the identity between a reference point detected within the second measurement series and a reference point already detected in the first measurement series, provided that the magnetic field signature or the section of the magnetic field signature to the reference point from the second measurement series has a minimum similarity with the magnetic field signature to the reference point from the first measurement series.The method according to the invention allows reliable and precise detection of previously detected reference points within the predefined route section. This can increase the accuracy in the creation of a magnetic field map. In the method according to the invention, it is advantageously not dependent on additional magnets being introduced into the section of the track. Optionally, however, additional magnets may be provided to specifically adjust the magnetic field based on the earth's magnetic field.The magnetic field signatures recorded during the first and second measurement series comprise N measurement values, which can be adjacent to the reference point, for example. N can be 100, 200 or 300, for example, wherein a magnetic field sensor can be recorded at a measurement frequency of 100 Hz, for example. The measurement data can preferably be interpolated, whereby the number of values within a magnetic field signature can be increased. Each measurement value can contain, for example, a vector which describes the magnetic field at the respective measurement point. It can also be provided that the measured values are designed such that the reference point lies within the defined measurement range. For example, the reference point can also be located centrally within the measurement range.A hodometer in the sense of the present invention generally measures a distance travelled by a vehicle and is therefore also referred to as a distance measuring device. In this case, the distance covered relative to a reference point is measured, for which reason the measurement results have an increasing measurement error as the measurement distance increases.The local magnetic field signatures are each recorded using the hodometer and the magnetic field sensor, wherein a local magnetic field signature describes the magnetic field in the immediate vicinity of a reference point.The length of the magnetic field signatures recorded during the first series of measurements are preferably identical (I sig1= I sig2)Even if the method according to the invention has been described above with two series of measurements, it is obvious to the person skilled in the art that in practice it is desirable to carry out more than two series of measurements and contributes to increased precision of the method.In some embodiments of the method according to the invention, it can be provided that the method has the following steps:matching a section of a magnetic field signature recorded during the second measurement series with the magnetic field signatures recorded during the first measurement series, wherein the section of the magnetic field signature recorded during the second measurement series has a length Isection, and wherein Isection<I sig2 applies; whereinduring the matching of the section of the signature recorded during the second measurement series with the magnetic field signatures recorded during the first measurement series, a successive comparison of said section with sections of a magnetic field signature recorded during the first measurement series that are displaced relative to one another is effected.By using the section of the magnetic field signature (instead of the entire magnetic field signature) and the sliding comparison with a magnetic field signature from the first series of measurements, several technical advantages are achieved. In particular, this achieves the effect that a reference point is also recognized when this reference point of the first and in the second measurement series is recognized at positions which are displaced with respect to one another. As a result, the reliability of the reference point detection is significantly increased and the method according to the invention is thus more robust with respect to measurement inaccuracies of the hodometer. The smaller the section opposite I sig2 is selected, the greater the recognition tolerance when matching the section of the magnetic field signature from the second measurement series with the magnetic field signatures from the first measurement series. In addition, the sliding comparison makes it possible not to recognize the one reference point, but also to simultaneously recognize the distance of the recognized reference point between the first and the second measurement series.In some preferred embodiments of the invention, it can be provided that Isection maxima is 75% of I sig2 that Isection maxima is 50% of I sig2 and that Isection maxima is 25% of I sig2 with particular preference. The selection of the length of the section of the magnetic field signature Isection depends on several factors. If the cutout is selected to be too small, the precision of the recognition during the matching decreases. If the section is selected to be too large, the recognition tolerance decreases, so that reference points that are slightly shifted between the first and the second measurement series are not recognized.According to one embodiment of the method according to the invention, it can additionally be provided that IsectionMin is 20% of I sig2 that IsectionMin is preferably 40% of I sig2 and that IsectionMin is particularly preferably 70% of I sig2.According to some embodiments of the method according to the invention, it can be provided that at least one local magnetic field signature within the first measurement series and / or within the second measurement series reproduces the magnetic field in a range from 100 m to 500 m, preferably in a range from 200 to 400 m, wherein the range comprises the associated reference point. The selection of such measurement ranges for recording the magnetic field signatures has proven to be suitable in practice, so that a reliable detection of the reference points can be achieved. If the magnetic field signature represents a smaller area, the detection of the reference points is made more difficult, while the precision of the position determination decreases if the magnetic field signature represents a significantly larger area. The value ranges mentioned therefore represent a compromise which has proven itself in practice.Furthermore, in the method according to the invention, it can be provided that the recording of at least one local magnetic field signature is carried out by detecting the magnetic field in the vicinity of a reference point with a measurement frequency of ≥10 Hz or ≥30 Hz or ≥50 Hz, preferably with a measurement frequency of ≥100 Hz and particularly preferably with a measurement frequency of ≥200 Hz. As a result, a high-resolution detection of a characteristic magnetic field around a reference point can be achieved, as a result of which the detection accuracy of the reference points can be increased. In addition, the high-resolution detection of the magnetic field in the vicinity of a reference point has the effect that altogether fewer reference points have to be observed.Preferably, it can be provided that the matching of at least one magnetic field signature recorded during the second measurement series or a section of this magnetic field signature with the magnetic field signatures recorded during the first measurement series takes place based on a correlation calculation. As a result, the degree of similarity can be determined in a simple manner, so that a reliable decision can be made with regard to the identity of two reference points on the basis of the result of the correlation calculation. If the correlation calculation takes place between a section of a magnetic field signature recorded during the second measurement series and the magnetic field signatures of the first measurement series, it is possible to determine how great the distance between two reference points identified as identical in the first and the second measurement series is. In this case, the magnetic field signatures of the first measurement series are longer than the portion of the magnetic field signature recorded during the second measurement series. Therefore, the correlation coefficients for all positions of the section within the local measurement signature of the first measurement series can be advantageously evaluated by the section "sliding" over the local measurement signature of the first measurement series. In this case, the largest correlation coefficient determined can be determined and stored together with the position at which this maximum was determined. After the comparison for a section of a magnetic field signature of the second measurement series has been matched to all possible local magnetic field signatures from the first measurement series, it can finally also be checked whether the greatest of the correlation coefficients determined exceeds a predefined threshold value. If this is the case, then on the one hand the identity between the reference points can be recognized and at the same time also the displacement between the first and the second measurement series. In this way, the sliding alignment in combination with the correlation calculation allows reliable recognition of a reference point and its distance between two measurement series. If the local maps of the different reference points overlap, the identity to a plurality of reference points and the respective displacement can also be determined. This is the case, for example, when a reference point is provided every 50 m and the local maps have a length of 200 m, for example. A plurality of correct loop closures can then be detected (at least partially) and inserted into the graph. Accordingly, according to some embodiments of the present invention, it may be provided that a magnetic field signature describes the magnetic field along a distance that is longer than the distance between two reference points. It can also be provided that the distance described by a magnetic field signature is at least twice as long, at least three times as long, or at least four times as long as the distance between two reference points.In some preferred embodiments of the invention, it can be provided that in each case a correlation coefficient for determining the similarity of a magnetic field signature recorded during the second measurement series is calculated with a plurality of magnetic field signatures recorded during the first measurement series, wherein the correlation coefficient with the greatest value is compared with a predefined correlation limit value and an identity between a reference point recorded during the second measurement series and a reference point recorded during the first measurement series is detected, provided that the correlation coefficient with the greatest value exceeds the predefined correlation limit value. As a result, the sensitivity of the detection of a reference point can be set, wherein the limit value can be set depending on the field of application and requirements.In the method according to the invention, it can also be provided that the matching of the reference point detected during the second measurement series with the reference points detected during the first measurement series is carried out selectively, so that the matching is carried out only with selected reference points at which there is an increased probability of an identity of the reference points. In other words, the adjustment is thus limited to a predefined range, wherein the limitation can preferably be carried out on the basis of the hodometer measurement data. If reference points have been determined in the first measurement series 100 and it is to be assumed during the second measurement, in the case of a newly detected reference point, on the basis of the error-containing hodometer measurement data that are present, that the newly detected reference point can only have an identity with one of the reference points 83 to 87 of the first measurement series, the alignment of the newly detected reference point (during the second measurement series) can be restricted to the reference points 83 to 87 of the first measurement series. As a result, the efficiency of the method can be increased, so that the correlation calculation is only limited to a selection of the reference points. In addition, the precision of the recognition method is also increased, since the risk of an erroneous assignment of the reference points is reduced.In addition, in the method according to the invention, it can be provided that the values of the recorded local magnetic field signatures are interpolated.Preferably, it can be provided that the hodometer has a rotational speed sensor which is designed to detect the rotational speed of a wheel of a rail vehicle, and / or a Doppler radar sensor which is designed to measure the speed of the rail vehicle. However, further methods known from the prior art for direct or indirect distance measurement can also be implemented. In the indirect distance measurement, for example, speed sensors can be used, the measurement results of which are integrated.In addition, the present invention proposes a method for creating a magnetic field map for a route section within a rail network, the method having the following steps:creating a magnetic field map for a route section using a graph-based method for simultaneous position determination and map creation, SLAM,detecting reference points already detected in past measurement series using one of the above reference point detecting methods.Moreover, it can be provided that an edge within the SLAM graph is added between a reference point from the second measurement series and a reference point from the first measurement series, provided that an identity between the reference points mentioned has been recognized. The added edge describes the displacement between two reference points. The distance is also determined by the magnetic field.Furthermore, the present invention proposes a method for determining the position of a rail vehicle within a rail network, wherein the rail vehicle has a magnetic field sensor and the method has the following steps:creating a magnetic field map for a route section within a rail network according to one of the preceding methods for creating magnetic field maps;detecting a magnetic field using the magnetic field sensor;matching the magnetic field with the magnetic field map; anddetermining the position of the rail vehicle within the rail network as a function of the magnetic field detected beforehand and the magnetic field map produced.Further advantageous embodiments of the present invention are discussed in more detail below.As stated above, the present invention relates, among other things, to a method for creating magnetic field maps along railway rails, even in scenarios in which GNSS localization is not possible. The proposed method can be used to improve existing SLAM methods (in particular in pose graph SLAM). The advantage of pose graph SLAM is that the problem can be formulated as a sparse graph that can be efficiently solved in terms of computational complexity and memory requirements. According to an aspect of the present invention, each node in the pose graph may be associated with a local magnetic field map (also referred to as a local magnetic field signature) generated from measurements of an odometer and magnetometer. These local maps are then used to identify loop conclusions (also referred to as identifying identical reference points or loop-closes) between nodes and to calculate their relative positions. Since loop conclusions can only be detected for nodes or reference points located close to one another, the resulting graph is sparsely populated and the optimization can be carried out efficiently on it, so that the algorithm can also be used for long distances. To evaluate the proposed method, a data record is used which was recorded with the German lane advanced TrainLab which was travelling at speeds of up to 100 km / h. In addition, simulations were performed to evaluate a scenario not covered by the measurements.Motifation MotifationMagnetic field-based train localization has the potential to enable localization even in areas in which no GNSS is available, such as e.g. in tunnels, without special trackside devices for localization such as balises or radio beacons having to be installed. Magnetic localization is a fingerprinting technique and requires a card that associates a particular magnetic fingerprint or pattern with the location at which the pattern is located. Unfortunately, map creation requires a position reference system (e.g., GNSS), which results in a kind of Henne-Ei problem. The magnetic location system requires a map and the creation of the map requires a reference position. This type of problem occurs in many applications, e.g., robotics, which has led to the development of a variety of simultaneous location and mapping (SLAM) algorithms. Initially, SLAM was based on extended Kalman filters, the complexity of which increased quadratically with the number of landmarks observed and which, given highly nonlinear problems, had only a low accuracy. Therefore, particle filters and Rao-Blackwellized particle filters have been introduced alternatively in the form of the known FastLAM algorithm. While the above solutions are based on filters, alternatives have also been proposed that attempt to find a solution to the SLAM problem by optimization. Above all, graph-based methods have been introduced, which are nowadays considered as prior art for SLAM because of their computing efficiency and stability. In the case of SLAM, a distinction is also made between landmark-based and landmark-free approaches. In the present invention, a landmark-free approach may preferably be used, which may be found, for example, in [1]. In addition, the present invention proposes improvements over the method described in [2], wherein a new loop closure detection is introduced and preferably a priori information edges and absolute measurement edges are introduced, which are required on account of the limited freedom in the trajectory of the train when taking measurements.Graph-based SLAMThe fundamentals of the graph-based SLAM will be summarized below. More specifically, the SLAM method with pose graph optimization will be explained, according to which the map is not estimated directly. Instead, only the series of poses is optimized. On the basis of the trajectory obtained, the desired map can then be created.The starting point for the derivation of graph-based SLAM is the definition of the cost function to be optimized. In this particular case, the cost function is the total posterior probability density function (pdf) over the trajectory of a train x where two sets of measurements are and the trajectory is represented by a sequence x of train positions merged into a vector X = x 1, ···, x N] where x i is the ith train position along the path. The set contains measurements describing how two positions x i and x j with i ≠ j in the trajectory are related to each other. Since the position of a train along the route is considered here, the position is one-dimensional and is basically only a measure of the distance between x i and x j. This type of measurements can be obtained from a hodometer or from a distance measuring device which measures the distance travelled by the train between two positions, or, as will be explained in more detail in the next sections, this information can also be obtained from the magnetic field itself. In contrast to the set, the set contains absolute information about a specific position. Therefore, an element y i of only one position x i is connected. Hereinafter, a position x i is also referred to as a node, and measurements are referred to as edges. This will be explained in more detail below in connection with the derivation of the cost function.The goal of the optimization is to find the maximum a posteriori (MAP) estimateTo perform the optimization, it is advantageous to disassemble the back expression into a more suitable form. In a first step, the conditional probability law is applied to the back expression to obtainNote that the right side of the above expression is only proportional to the posterior expression. However, this does not change the position of the maximum. In a second step, the right side is further decomposed in (3)The elements of the set are pairs of node indices for which a relative measurement is included, and the set includes the indices of positions for which an absolute measurement exists. For simplicity, only one relative measurement per node pair and one absolute measurement per node are considered here, but the extension to several measurements is possible without any problem.The last two terms in (4) are the product of the probability densities of the individual measurements. For each of the measurements, the probability density is assumed to be Gaussian. For the relative measurements in, the probability density is given by the mean value being given by the predicted measurement, which is simple for the 1D case considered here.For the absolute measurements in y, the probability depends only on one nodeFor pdf p(x), it is assumed that only information about the first node is present. Therefore, it is set to. Setting Ω 0 high anchors the node to the mean value x≅ 0 of p(x). This mean value can be determined, for example, by GNSS before the trains enter a tunnel. If absolute measurements are not available, the first node may be used to define the coordinate system.For cost functions of the above form, it is convenient to optimize the logarithm of the cost function. The Gaussian pdfs in (5)-(7) then give the following expression, neglecting all constant parts which are not important for the optimization and multiplying the entire function by -2 to simplify the expressions. By multiplying the cost function by -2, the problem is converted to a minimization problem which will be the basis for the optimizationThe cost function c(x) may be visually represented by a graph, wherein each edge represents a term of the cost function and each node is a position to be optimized. An example of a graph as found in the railway environment is shown in FIG. 1. The graph in FIG. 1 has a discontinuity between the poses x 5 and x 6. This is because in the railway environment, a train may pass through a tunnel and then not return for a long time. The SLAM algorithm must therefore be able to combine the data from multiple discontinuous measurement runs on the same track. These discontinuities are the motif to preferably include absolute position measurements in the graphs as they connect the poses of multiple measurement trips on the same track, even if the train always traverses the tunnel from the same direction. In addition, absolute measurements at the beginning and end of a tunnel help optimize to reduce the overall position error in the tunnel when only relative information is present between different poses.Examination of (8) shows that the MAP estimate of the trajectory is a weighted nonlinear least squares problem. In the context of the present invention, the least-square solution can be found using the Gaussian-Newton method. This requires linearization of the error functions for the a priori information e p( x), relative measurements e ij( x) and absolute measurements e i( x). The linearized version of these functions results from their Taylor series developments about the operating point x*J p, J ij and J ¡ are the corresponding Jacobi matrices evaluated at the location x*, i.e. the Jacobi matrix results from. Since the error functions are scalar, the Jacobi matrices are row vectors of dimension N and the product with vector Δx = x-x* is a scalar. For the Gaussian-Newton algorithm, the linearized error functions can be inserted into the cost function in (8)The Gaussian-Newton equations are obtained by setting the gradient of the cost function with the linearized error functions in (11) to zeroThe gradients of the individual parts of the cost function then result in an equation of the shape with the vector and the matrixTo minimize the cost function, it is now possible to solve for Δx and calculate the estimated value for xThis method is performed iteratively by setting x*=x̂ after each iteration until the estimate has converged.Complexity considerationsThe matrix H, which must be inverted during the optimization, has the dimension N x N, where N is the number of nodes and thus the dimension of x. For large vectors x, this leads to a high complexity, in particular because the matrix H has to be recomputed at the position of the new estimate x̂after each iteration step. In the simplest implementation, this would have a complexity of enjoyability it can be shown that the matrix H is typically sparsely populated by its construction.Relative MeasurementsIn the previous section, the general idea of the graph-based SLAM approach has been generally explained. To illustrate some preferred aspects of the invention, the relative observations and their possible acquisition will be described in more detail in this section. 1) Odometer edges: As shown in the example graph in FIG. 1, there is an edge between successive nodes if they were created during the same run or the same measurement series. The measurements for the edges between successive nodes are obtained, for example, from an odometer. In railways, the odometer can typically be determined by integrating the speed of a wheel speed sensor or a Doppler radar. 2) Detection of a loop closure or a previously detected reference point: in addition to the edges on the basis of hodometer measurements, edges between non-consecutive nodes are preferably also taken into account in the present case. An edge between two non-consecutive nodes may be introduced when a loop closure is detected. A loop closure should be detected when a train re-actuates a position or a reference point on the track. The advantage of a loop closing edge is that it connects newly added nodes to older nodes already present in the graph. Roughly speaking, this connection between new and old nodes "limits" the overall trajectory and can therefore, if sufficient loop conclusions are detected, significantly improve the estimated trajectory compared to the pure odometer.As mentioned above, a core aspect of the present invention relates to the manner in which loop inclusions are detected or how reference points detected in previous measurement series are detected in subsequent measurement series. In the present case, it is proposed in particular to identify loop inclusions by creating a local map of the magnetic field (also as a local magnetic field signature) in its environment for each newly added node. How these local maps can be created is explained in more detail in the next section. First, it is sufficient to know that a local map is preferably a function that takes a one-dimensional position as input and returns the corresponding magnetic field vector at this position. Since the map is "local", the input position is preferably relative to the position of the corresponding reference point. With this local map, the loop closure can preferably be detected in three steps when a new node is added• Search for the set of indices of all nodes located in a predefined search radius R LC around the newly created node• Compare the magnetic field recorded on the last d LC meters of the distance around the new node with the local maps of all nodes and find the position at which it best matches each local map• Add an edge between the current node and a node in if the similarity between the current magnetic field and the local map is above a defined threshold T LC. The relative dimensions of the edge are given by the position at which the current magnetic field best matches the local map, which dimensions describe the distance of the points from one another.In the following it will be explained how the similarity can be determined and how the local maps can be constructed. 3) Local maps: To create a local map and magnetic field signature, respectively, the magnetic field measurement results and the speed of the train may be stored in a buffer during these measurements. The train speed is also measured, for example, with a wheel speed sensor. From this buffer, a local map can now be calculated each time a new node is created. To calculate the map, first the measured speed can be integrated back in time starting with the latest available measurement. By integrating the velocity, the relative position in relation to the position of the newly created node is now also assigned for each magnetic field measurement. In a second step, an interpolation of the magnetic data onto an equidistant position grid can be carried out. This interpolation is advantageous because the relative position from the velocity integration depends on the velocity of the train during the measurements and therefore the magnetic field data in the spatial region is not present at equal distances, which can influence the further processing. An example of a magnetic map and the relative position measurement is shown in FIG. 2. The length L map of the card can be chosen freely. However, initial studies have shown that setting the length to several hundred meters is advantageous. If the maps become too long, the error in relative position resulting from the integration of the measured train speed becomes too large. The choice of card length is a compromise between maximizing the number of loop ends between the nodes and the accuracy and reliability of the loop end. 4) Detector: Using the local maps from the previous section, a detector for loop conclusions can now be implemented. The detector, which is preferably proposed in the present case, is basically a correlator. When a new node is created, its local map can be created and stored together with the node. Subsequently, a piece of the card of length L sig may be cut out, which is called a magnetic signature or magnetic field signature. The signature may be cut out, for example, at the beginning of the card. In general, the equation L map > L sig should be satisfied, since the entire signature should be matched to the local maps of the other nodes in order to obtain good and reliable results.To identify a loop closure, the local map of each candidate node included in the index set may be compared to the signature of the newly created node. The comparison is preferably based on a correlation coefficient. If the card is longer than the signature, the correlation coefficient for all possible positions of the signature in the local card can be evaluated by "pushing" the signature over the card. For the detector, only the highest value of the correlation coefficient and the relative position at which it was calculated are stored. After the comparison is performed for all candidate nodes, a loop closure may be detected if the stored maximum coefficient is above the threshold T LC. For each detected loop closure, a relative measurement edge is inserted between the corresponding candidate node and the newly created node. The measurements of the inserted edge correspond to the position in the local map at which the maximum coefficient was observed. Here, attention should be paid to the correct sign, since the sign depends on the direction of the edge and the definition of the relative position. If the local map and the signature were recorded during journeys with different travel directions, the map or the signature must be adapted accordingly.EvaluationIn this section, the proposed approach is to be evaluated with measurement data recorded with the advanced TrainLab of the Deutsche Web. The measurements were recorded on the distance between gottingen and cassel at a speed of up to 100 km / h. The magnetic field was measured with a low-cost magnetometer from KMX and the speed of the train was determined with a wheel encoder from Deuta. For the evaluation, the data of three journeys on a route section of about 3 km length are used. During the three drives, the train always carried the track from the same direction, which is not absolutely necessary for the proposed approach. This means that the data set is discontinuous and GNSS measurements at the beginning and at the end of the route section were used to align the coordinate systems of the different journeys. In this case, GNSS data were used only at the beginning and at the end, since this corresponds to a tunnel scenario.In FIG. 3, the final graph after the three runs can be seen. As can be seen, despite the distance travelled of about 9 km, the graph has only 166 nodes. A node was created only when the train has traveled 50 m compared to the last node according to the odometer. The sparseness of the nodes can be achieved because local maps are used that closely represent the magnetic field between the nodes. The improvement in SLAM algorithm position error versus the clean unsupported odometer is shown in FIG. 4. After the optimization, the error is usually less than 10 m, while the pure odometer (hodometer) has errors of up to 60 m. In the measurement data, the wheel diameter was well calibrated, so that the error of the odometer was very small, which is why during the evaluation the wheel diameter was reduced by one centimeter in order to obtain larger errors in the odometer and to check whether the optimization still provides good results. It should also be mentioned that the odometer appears to function quite well here. In the case of ETCS, the error of the odometer after 3 km may be up to 5 m+5% x 3000 m=155 m. For an accurate mapping of the magnetic field over longer distances, the pure hodometer measurement is therefore rather unsuitable.Additional processing should be done to create the desired magnetic map, since the graph contains one node only every 50 m. For mapping, therefore, interpolation can be performed between the nodes in order to obtain the positions of the individual magnetic field measurements for which no node was created. Due to the low dynamic and acceleration capacity of a train, it can be assumed that a linear interpolation should be sufficient in this case.To complete the evaluation, a train was simulated which moved twice forwards and backwards on the same distance, about 5 km long. The magnetic field map used for the simulation was measured with the BRB between Augsburg and Friedberg. To simulate the measurements, the value of the map at the current position of the train was taken and noise was added. Compared to the measurement data, the simulated data set is continuous (in the sense that the train position does not jump from one trip to the next) and therefore no external information will need to perform SLAM, i.e. no GNSS is needed. Only the first node has been anchored in position to fix the coordinate system, as is common in the SLAM method.In FIGS. 5 and 6, the graph and the resulting position error associated therewith are shown. It can be seen here that only 362 nodes are required for covering the almost 20 km of data. It can also be seen from the graph that the different journeys are now connected to one another via a hodometer edge, since the train decelerates at the end of the route and then starts to travel in the opposite direction. The hodometer in the simulation has a systematic error that results in a position error of several hundred meters. As can be seen in FIG. 6, the error of the odometer can be significantly reduced by using the graph-based SLAM method. This is accomplished by the loop conclusions connecting nodes created at the beginning of the record to nodes created near the end. This connects nodes with a low uncertainty to nodes with a high uncertainty. The introduction of these connections makes it possible to correct the positions of the nodes created with a hodometer position already having a large error.This work has emerged as part of a statement from the impulse and crosslinking fields of the Hermann Helmholtz community of Deutscher Forschungszcentre e.V. (contract number ZT-I-PF-5-49 ("Ubiquitous Spatio-Temporal Learning for Future Mobility (ULearn4Mobility)")).LITERATURE DIRECTORY[1] G. Grisetti, R. Kummerle, C. Stachniss, and W. Burgard, "A Tutorial on Graph-Based SLAM," IEEE Intelligent Transportation Systems Magazine, vol. 2, no. 4, pp. 31-43, 2010. [2] J. Jung, J. Choi, T. Oh, and H. Myung, "Indoor Magnetic Pose Graph SLAM with Robust Back-End," in Robot Intelligence Technology and Applications 5th Springer International Publishing, 2019, pp. 153-163.References included in the specificationThis list of documents cited by the applicant has been produced in an automated manner and is only included for the better information of the reader. The list is not part of the German patent application or utility model application. The DPMA does not take any adhesion for any faults or omissions.Cited Non-Patent LiteratureG. Grisetti, R. Kummerle, C. Stachniss, and W. Burgard, "A Tutorial on Graph-Based SLAM," IEEE Intelligent Transportation Systems Magazine, vol. 2, no. 4, pp. 31-43, 2010

[0060] J. Jung, J. Choi, T. Oh, and H. Myung, "Indoor Magnetic Pose Graph SLAM with Robust Back-End," in Robot Intelligence Technology and Applications 5th Springer International Publishing, 2019, pp. 153-163

[0060]

Claims

Method for magnetic field-based detection of a reference point when carrying out a plurality of measurement series for creating a magnetic field map for a rail network, wherein the method has the following steps: - carrying out a first measurement series for ascertaining a magnetic field at a plurality of reference points within a predefined section of the rail network, wherein at each of the reference points the magnetic field is detected using a magnetic field sensor and the position of the reference point is detected using a hodometer, wherein at each reference point a local magnetic field signature is recorded and stored in a database, and each local magnetic field signature reproduces the local magnetic field in the environment around the reference point, wherein the magnetic field signatures recorded during the first measurement series have a length I sig1 ; performing a second measurement series for determining the magnetic field at a plurality of reference points within the predefined section of the rail network, wherein at each of the reference points the magnetic field is detected using the magnetic field sensor and the position of the reference point is detected using a hodometer, wherein a local magnetic field signature is recorded for each reference point, and each local magnetic field signature reproduces the local magnetic field in the environment around the reference point, wherein the magnetic field signatures recorded during the second measurement series have a length I sig2 ; matching at least one magnetic field signature recorded during the second measurement series or a section of this magnetic field signature with the magnetic field signatures recorded during the first measurement series; and - detecting the identity between a reference point detected within the second measurement series and a reference point already detected in the first measurement series, provided that the magnetic field signature or the section of the magnetic field signature to the reference point from the second measurement series has a minimum similarity with the magnetic field signature to the reference point from the first measurement series.Method according to claim 1, characterised in that the method has the following steps: - matching a section of a magnetic field signature recorded during the second measurement series with the magnetic field signatures recorded during the first measurement series, wherein the section of the magnetic field signature recorded during the second measurement series has a length Isection and wherein Isection<I sig2 applies; wherein - during the matching of the section of the signature recorded during the second measurement series with the magnetic field signatures recorded during the first measurement series, a successive comparison of said section with mutually displaced sections of a magnetic field signature recorded during the first measurement series takes place.Method according to Claim 2, characterized in that Isection maxima is 75% of I sig2 in that Isection maxima is preferably 50% of I sig2 and in that Isection maxima is particularly preferably 25% of I sig2.Method according to one of Claims 1 to 3, characterized in that at least one local magnetic field signature within the first measurement series and / or within the second measurement series reproduces the magnetic field in a range from 100 m to 500 m, preferably in a range from 200 to 400 m, wherein the range comprises the associated reference point.Method according to one of Claims 1 to 4, characterized in that the recording of at least one local magnetic field signature is carried out by detecting the magnetic field in the vicinity of a reference point with a measurement frequency of ≥ 10 Hz or ≥ 30 Hz or ≥ 50 Hz, preferably with a measurement frequency of ≥ 100 Hz and particularly preferably with a measurement frequency of ≥ 200 Hz.Method according to one of Claims 1 to 5, characterized in that the matching of at least one magnetic field signature recorded during the second measurement series with the magnetic field signatures recorded during the first measurement series takes place on the basis of a correlation calculation.Method according to Claim 6, characterized in that in each case a correlation coefficient for determining the similarity of a magnetic field signature recorded during the second measurement series is calculated with a plurality of magnetic field signatures recorded during the first measurement series, wherein the correlation coefficient with the greatest value is compared with a predefined correlation limit value and an identity between a reference point recorded during the second measurement series and a reference point recorded during the first measurement series is detected, provided that the correlation coefficient with the greatest value exceeds the predefined correlation limit value.Method according to one of Claims 1 to 7, characterized in that the comparison of the reference point detected during the second measurement series with the reference points detected during the first measurement series is carried out selectively, with the result that the comparison is carried out only with selected reference points at which there is an increased probability of an identity of the reference points.Method according to one of Claims 1 to 8, characterized in that the values of the recorded local magnetic field signatures are interpolated.Method according to one of Claims 1 to 9, characterized in that the hodometer has a rotational speed sensor which is designed to detect the rotational speed of a wheel of a rail vehicle, and / or a Doppler radar sensor which is designed to measure the speed of the rail vehicle.Method for creating a magnetic field map for a route section within a rail network, the method comprising the following steps: - creating a magnetic field map for a route section using a graph-based method for simultaneous position determination and map creation, SLAM, - detecting reference points which have already been recorded in past measurement series using one of the methods according to claims 1 to 10.Method according to claim 11, characterised in that an edge within the SLAM graph is added between a reference point from the second measurement series and a reference point from the first measurement series, provided that an identity between these reference points has been detected.Method for determining the position of a rail vehicle within a rail network, wherein the rail vehicle has a magnetic field sensor and the method has the following steps: - creating a magnetic field map for a section of track within a rail network according to one of Claims 11 or 12; - detecting a magnetic field using the magnetic field sensor; - matching the magnetic field with the magnetic field map; and - determining the position of the rail vehicle within the rail network as a function of the detected magnetic field and the magnetic field map.

Images