Train positioning and speed measuring system and method and electronic equipment

Abstract

The invention relates to the technical field of train positioning and speed measurement, in particular to a train positioning and speed measurement system and method and electronic equipment, and the system comprises an image collection device which is used for collecting image information of a positioning plate located beside a train line; the laser radar is used for collecting point cloud information reflected by the positioning plate in a laser irradiation mode; the control device is used for controlling the image acquisition device and the laser radar to acquire image information and point cloud information of the positioning plate according to a preset working mode, determining a positioning plate bounding box according to the image information of the positioning plate and identifying position information or serial number information in the positioning plate bounding box; and determining the position and / or speed information of the train according to the positioning plate bounding box, the position information or the serial number information in the positioning plate bounding box and the point cloud information. According to the invention, train speed measurement is realized while positioning is carried out, a redundant backup scheme for positioning and speed measurement is provided for the train, and the running safety of the train is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of train positioning and speed measurement, and in particular to a train positioning and speed measurement system, method and electronic equipment. Background Art

[0002] In recent years, with the vigorous development of urban rail transit construction, the Automatic Train Operation (ATO) system has also been widely used. During the automatic operation of the train, it must be able to determine its position on the line; in addition, in order to limit the speed of the train, improve the stability of the train operation and ensure the on-time arrival of the train, the train also needs to be able to obtain its own speed, which creates the need for train positioning and speed measurement.

[0003] At present, trains are mainly positioned through transponders and speed measured through pulse speed measurement modules. When the transponder fails, the positioning function of the train is affected; when the train is sliding and the wheels are idling, the pulse speed measurement module may have speed measurement errors. The patent with application number CN202011248502.8 and patent name "A Train Positioning Method and System" proposes a train positioning solution based on laser point cloud matching, but in order to implement this solution, it is necessary to pre-build a high-precision electronic map; at the same time, when the solution is required to lay high-reflectivity reflective signs in bridges and tunnels, the shape, installation posture and height of the reflective signs are required to be different, and the requirements for the reflective signs are high; and this solution only involves train positioning and does not involve train speed measurement.

[0004] Therefore, it is necessary to propose a new method for train positioning and speed measurement that does not rely on the existing "balise-pulse speed measurement module". Summary of the invention

[0005] In view of the above problems, the present invention provides a train positioning and speed measurement system, method and electronic equipment.

[0006] The train positioning and speed measurement system provided by the present invention comprises:

[0007] An image acquisition device, used to acquire image information of a positioning plate located beside a train line;

[0008] Laser radar, used to collect point cloud information reflected by the positioning plate through laser irradiation;

[0009] The control device is communicatively connected with the image acquisition device and the laser radar, respectively, and is used to control the image acquisition device and the laser radar to respectively acquire the image information and point cloud information of the positioning plate according to the preset working mode, and to determine the positioning plate bounding box according to the image information of the positioning plate and identify the position information or serial number information in the positioning plate bounding box, and to determine the position and / or speed information of the train according to the positioning plate bounding box, the position information or serial number information in the positioning plate bounding box and the point cloud information.

[0010] The present invention also provides a train positioning and speed measurement method, which is implemented based on the above-mentioned train positioning and speed measurement system, and the method includes:

[0011] The control device acquires the image information of the positioning plate from the image acquisition device, and determines the positioning plate bounding box and the position information or serial number information in the positioning plate bounding box according to the image information;

[0012] Acquire point cloud information of the positioning plate from the laser radar, and determine a preferred point cloud cluster that meets preset conditions from the point cloud information;

[0013] The positioning plate point clouds of the current frame point cloud information and the previous frame point cloud information are determined respectively according to the preferred point cloud cluster, the positioning plate bounding box and the position information or serial number information in the positioning plate bounding box, and the position information and / or speed information of the train at the current moment is determined according to the positioning plate point clouds of the current frame point cloud information and the previous frame point cloud information.

[0014] Furthermore, before acquiring the image information of the positioning plate through the image acquisition device, it also includes:

[0015] The control device, the image acquisition device and the laser radar are preprocessed so that the control device is time synchronized with the laser radar or respectively with the laser radar and the image acquisition device, and the internal parameter matrix and distortion coefficient of the image acquisition device are determined, as well as the pose matrix of the laser radar in the coordinate system of the image acquisition device.

[0016] Further, according to the preferred point cloud cluster, the positioning plate bounding box and the position information or serial number information in the positioning plate bounding box, the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are respectively determined, including: determining the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information by a preset point cloud image association strategy;

[0017] The preset point cloud image association strategies include:

[0018] In the case where the preferred point cloud clusters are determined from both the current frame point cloud information and the previous frame point cloud information, a first point cloud-image matching pair is generated according to the previous frame point cloud information and the first image information, and a second point cloud-image matching pair is generated according to the current frame point cloud information and the second image information; wherein the first image information is image information collected at the same time as the previous frame point cloud information or image information collected at a different time from the previous frame point cloud information determined according to a preset time offset screening strategy, and the second image information is image information collected at the same time as the current frame point cloud information or image information collected at a different time from the current frame point cloud information determined according to a preset time offset screening strategy;

[0019] When the position information or serial number information of the positioning plate bounding box of the first image information is the same as the position information or serial number information of the positioning plate bounding box of the second image information, determining the preferred bounding box of each preferred point cloud cluster in the current frame point cloud information and the preferred bounding box of each preferred point cloud cluster in the previous frame point cloud information;

[0020] After determining a first unique preferred bounding box from the preferred bounding boxes of the previous frame of point cloud information according to the positioning plate bounding box in the first image information, taking the preferred point cloud cluster corresponding to the first unique preferred bounding box as the positioning plate point cloud of the previous frame of point cloud information;

[0021] After determining a second unique preferred bounding box from the preferred bounding boxes of the current frame point cloud information according to the positioning plate bounding box in the second image information, the preferred point cloud cluster corresponding to the second unique preferred bounding box is used as the positioning plate point cloud of the current frame point cloud information.

[0022] Further, after respectively determining the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information according to the preferred point cloud cluster, the positioning plate bounding box, and the position information or sequence information in the positioning plate bounding box, the method further includes:

[0023] Determine whether data fusion is successful based on the point cloud centroid of the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information;

[0024] After the data fusion is successful, the positioning plate point cloud of the current frame point cloud information is saved, and the position information and / or speed information of the train at the current moment is determined based on the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information.

[0025] Further, the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are determined respectively according to the preferred point cloud cluster, the positioning plate bounding box and the position information or sequence information in the positioning plate bounding box, further comprising:

[0026] When the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information cannot be determined through the preset point cloud image association strategy, if the last train positioning and speed measurement data fusion is successful and there is a preferred point cloud cluster in the current frame point cloud information, the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information is determined through the preset point cloud target association strategy.

[0027] Further, in the case where the position information and speed information of the train are measured by the last train positioning and speed measurement, before respectively determining the positioning plate point cloud of the current frame point cloud information and the last frame point cloud information according to the preferred point cloud cluster, the positioning plate bounding box, and the position information or sequence number information in the positioning plate bounding box, the method further includes:

[0028] Obtain the three-dimensional speed and three-dimensional position coordinates of the last train from the position information and speed information of the train measured by the last train positioning and speed measurement;

[0029] According to the 3D speed and 3D position coordinates of the last train and the acquisition time difference between the current frame point cloud information and the previous frame point cloud information, the 3D position coordinates of the positioning plate in the current frame point cloud information are predicted.

[0030] Furthermore, the preset point cloud target association strategy includes:

[0031] The positioning plate point cloud of the current frame point cloud information determined during the last train positioning and speed measurement is used as the positioning plate point cloud of the previous frame point cloud information, and the d value of each preferred point cloud cluster in the current frame point cloud information is determined by a preset model according to the predicted three-dimensional position coordinates of the positioning plate in the current frame point cloud information;

[0032] When the minimum d value is not greater than the preset d value threshold, the point cloud corresponding to the preferred point cloud cluster with the minimum d value is used as the positioning plate point cloud of the current frame point cloud information.

[0033] Furthermore, the preset model includes:

[0034]

[0035] in, D ,y D and z D are the three-dimensional coordinates of D; x , y and λ z x D ,y D and z D The corresponding preset constant coefficient; wherein D is determined by the following formula:

[0036]

[0037] Among them, X lkis the three-dimensional position coordinate of the kth optimal point cloud cluster; It is the three-dimensional position coordinates of the positioning plate in the predicted point cloud information of the current frame.

[0038] Further, in the case where the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are determined by the preset point cloud image association strategy, the position information and / or speed information of the train at the current moment is determined according to the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, including:

[0039] Determine and save the position of the positioning plate according to the position information or serial number information in the positioning plate bounding box identified from the second image information;

[0040] Determine the centroid of the positioning plate point cloud of the current frame point cloud information;

[0041] The position of the train at the current moment is determined according to the position of the positioning plate and the centroid of the positioning plate point cloud of the current frame point cloud information, and the centroid coordinates of the positioning plate point cloud of the current frame point cloud information are used as the three-dimensional position coordinates of the train at the current moment.

[0042] Further, in the case where the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are determined by the preset point cloud target association strategy, the position information and / or speed information of the train at the current moment is determined according to the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, including:

[0043] Obtain the position of the positioning plate determined during the last train positioning and speed measurement;

[0044] Determine the centroid of the positioning plate point cloud of the current frame point cloud information;

[0045] The current position of the train is determined based on the position of the positioning plate determined during the last train positioning and speed measurement and the centroid of the positioning plate point cloud of the current frame point cloud information, and the centroid coordinates of the positioning plate point cloud of the current frame point cloud information are used as the three-dimensional position coordinates of the train at the current moment.

[0046] Further, determining the position information and / or speed information of the train at the current moment according to the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information also includes:

[0047] In the case that this is the first time to calculate the train speed information, the speed and three-dimensional speed of the train at the current moment are determined according to the centroid of the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, and the acquisition time difference between the current frame point cloud information and the previous frame point cloud information;

[0048] When this is not the first time that the train speed information is calculated, the current speed and three-dimensional speed of the train are determined based on the centroid of the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, the acquisition time difference between the current frame point cloud information and the previous frame point cloud information, and the most recently measured train speed.

[0049] The present invention also provides a computer-readable storage medium, wherein the computer program stored in the computer-readable storage medium implements the steps of the above method when executed by one or more processors.

[0050] The present invention also provides an electronic device, comprising a memory and one or more processors, wherein a computer program is stored in the memory, and the memory and the one or more processors are communicatively connected to each other, and when the computer program is executed by the one or more processors, the steps of the above method are executed.

[0051] The train positioning and speed measurement system, method and electronic equipment provided by the present invention have at least the following beneficial effects:

[0052] (1) When performing train positioning and speed measurement, it does not rely on the existing "balise-pulse speed measurement module" method, providing a redundant backup solution for train positioning and speed measurement, thereby improving the safety of train operation.

[0053] (2) When performing train positioning and speed measurement, there is no need to pre-build a high-precision electronic map, the requirements for the positioning board are low, and the method is easier to implement.

[0054] (3) It can measure train speed while performing positioning, meeting the current demand for train operation data detection.

[0055] (4) When locating and measuring train speed, point cloud information and image information are integrated to make the detection results more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0057] Figure 1 This is a schematic diagram of the structure of a train positioning and speed measurement system in Embodiment 1 of the present invention;

[0058] Figure 2 This is a flow chart of the steps of the train positioning and speed measurement method in the second embodiment of the present invention;

[0059] Figure 3This is a flowchart of the method for determining the point cloud of a positioning plate in Embodiment 5 of the present invention;

[0060] Figure 4 This is a flowchart of the steps of a method for determining the position information and / or speed information of a train at the current moment in Embodiment 6 of the present invention;

[0061] Figure 5 This is a flowchart of the method for determining the point cloud of a positioning plate in Embodiment 7 of the present invention;

[0062] Figure 6 This is a flow chart of the steps of a method for determining the position information and / or speed information of a train at the current moment in Embodiment 8 of the present invention;

[0063] Figure 7 This is a flowchart of the steps of a method for determining the position information and / or speed information of a train at the current moment in Embodiment 9 of the present invention;

[0064] Figure 8 This is a schematic diagram of the structure of an electronic device in Embodiment 13 of the present invention;

[0065] Reference numerals:

[0066] Figure 1 In: 101-image acquisition device, 102-laser radar, 103-control device;

[0067] Figure 8 In it: 800 - electronic device, 801 - processor, 802 - communication bus, 803 - user interface, 804 - communication interface, 805 - memory. DETAILED DESCRIPTION

[0068] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the protection scope of the present invention.

[0069] Embodiment 1

[0070] In this embodiment, a train positioning and speed measurement system is provided. Figure 1 As shown, the system includes:

[0071] The image acquisition device 101 is used to acquire image information of a positioning plate located beside a train line.

[0072] In one implementation, the image acquisition device 101 is a camera, and the image acquisition device 101 is installed on the front of the train facing forward.

[0073] The laser radar 102 is used to collect point cloud information reflected by the positioning plate by means of laser irradiation.

[0074] In one implementation, the laser radar 102 is mounted forward on the front of the train.

[0075] The control device 103 is respectively connected to the image acquisition device and the laser radar for communication, and is used to control the image acquisition device and the laser radar to respectively acquire the image information and point cloud information of the positioning plate according to the preset working mode, and to determine the positioning plate bounding box according to the image information of the positioning plate and identify the position information or serial number information in the positioning plate bounding box, and to determine the position and / or speed information of the train according to the positioning plate bounding box, the position information or serial number information in the positioning plate bounding box and the point cloud information.

[0076] Specifically, when the image acquisition device 101 supports triggered acquisition, the preset working mode is that the laser radar and the image acquisition device perform acquisition synchronously. That is, when the laser radar collects the first point of a frame of point cloud information according to its preset acquisition frequency, the image acquisition device is triggered to collect image information once.

[0077] When the image acquisition device 101 does not support trigger acquisition, the preset working mode is that the laser radar and the image acquisition device respectively acquire data according to their own preset acquisition frequencies. The present invention does not limit the preset acquisition frequencies of the laser radar and the image acquisition device. The preset acquisition frequencies of the laser radar and the image acquisition device are set by those skilled in the art according to actual needs, and the preset acquisition frequency of the image acquisition device is set to be higher than the preset acquisition frequency of the laser radar.

[0078] More specifically, along the train line, a plurality of positioning plates are arranged at intervals beside the train line, and each positioning plate is provided with position information or serial number information for identifying the position of the positioning plate in the train line. The position information can be directly represented by numbers, and the numbers on the positioning plate are the positions of the positioning plates. The serial number information can be represented by numbers or letters. When the serial number information is set on the positioning plate, it is necessary to pre-store the corresponding relationship between the serial number of the positioning plate and the position of the positioning plate in the train line in the control device 103. In addition, if the positioning plates are arranged at equal intervals, the position of the positioning plate in the line can also be calculated based on the serial number of the positioning plate and the spacing between the positioning plates.

[0079] Optionally, the position information or serial number information on the positioning plate may be displayed in the form of numbers or letters or a combination of numbers and letters, and in some possible situations may also be displayed in the form of other symbols (eg, a QR code).

[0080] In this embodiment, the existing 100-meter mark beside the train line can be used as the positioning plate (for example, the 100-meter mark on the subway train line). If there is no 100-meter mark on the train line, the positioning plate can be pre-installed beside the train line. The present invention does not limit the material and background color of the positioning plate, and those skilled in the art can set them according to their needs. Preferably, the positioning plate is made of a metal material with high reflectivity, and the background color is a pure color with high brightness, so as to facilitate the identification of the positioning plate and the position information or serial number information on the positioning plate.

[0081] The train positioning and speed measurement system provided in this embodiment does not rely on the existing "transponder-pulse speed measurement module", but instead collects point cloud information and image information of the positioning plate located next to the train line by arranging a laser radar and an image acquisition device on the train, and obtains the position and / or speed information of the train by integrating the point cloud information and image information, thereby providing a redundant backup solution for positioning and speed measurement for the train, improving the safety of train operation, and making the detection results more accurate. When performing train positioning and speed measurement, there is no need to pre-build a high-precision electronic map, the requirements for the positioning plate are relatively low, and the method is easier to implement. In addition, the train speed can be measured while positioning, meeting the current needs for train operation data detection.

[0082] Embodiment 2

[0083] In this embodiment, a train positioning and speed measurement method is provided, and the method is implemented based on the train positioning and speed measurement system provided in the previous embodiment.

[0084] like Figure 2 As shown, the train positioning and speed measurement method provided in this embodiment specifically includes the following steps:

[0085] Step S201: acquiring image information of the positioning plate from the image acquisition device through the control device, and determining the positioning plate bounding box and the position information or serial number information in the positioning plate bounding box according to the image information.

[0086] In actual detection, a trained deep neural network can be used to detect the positioning plate in each frame of image information to obtain a positioning plate bounding box, and output at most one positioning plate bounding box. Subsequently, optical character recognition (OCR) technology is used to identify the position information or serial number information in the positioning plate bounding box. Among them, in order to realize the "trained" deep neural network, a large number of positioning plate images need to be collected in advance for model training. Of course, in actual operation, those skilled in the art may also choose other methods to identify the positioning plate in the image information, obtain the positioning plate bounding box and the position information or serial number information in the positioning plate bounding box, and the present invention is not limited to this.

[0087] In this step, each frame of image information collected is identified and detected. When the positioning plate is detected in the frame of image information, the positioning plate bounding box of the detected positioning plate and the position information or serial number information in the positioning plate bounding box are determined.

[0088] Step S202: acquiring point cloud information of the positioning plate from the laser radar, and determining a preferred point cloud cluster that meets preset conditions from the point cloud information.

[0089] Specifically, a frame of point cloud information includes the three-dimensional position coordinates and reflection intensity of each laser point in the frame of point cloud. It should be understood that the three-dimensional position coordinates of each laser point are the coordinates of the laser point in the laser coordinate system. In one implementation, the laser coordinate system is a three-dimensional coordinate system constructed with the laser radar as the origin and the x, y, and z axes pointing to the front, left, and top directions of the laser radar respectively.

[0090] In one implementation, for a frame of point cloud information, determining a preferred point cloud cluster that meets a preset condition from the point cloud information includes:

[0091] The reflection intensity of each laser point in the frame point cloud information is compared with the preset intensity threshold, and the laser points whose reflection intensity is not less than the preset intensity threshold are selected. According to the preset point number range and clustering radius, the selected laser points are clustered to obtain point cloud clusters. For each point cloud cluster, the centroid, length, width and height are calculated, and the point cloud cluster whose length is within the preset length range, height is within the preset height range, and width is within the preset width range is selected as the preferred point cloud cluster that meets the preset conditions.

[0092] The preset intensity threshold can be set by those skilled in the art according to actual needs, and the present invention does not impose any specific restrictions on this. Specifically, the intensity threshold is related to the laser radar, and the reflection intensity reflects the reflectivity. For example, the laser radar uses 0 to represent almost no reflection and 255 to represent the highest reflectivity (maximum intensity). At this time, the intensity threshold can be set to 200.

[0093] The preset point number range, clustering radius, preset length range, preset height range and preset width range can also be preset by those skilled in the art according to the actual size of the positioning plate, and the present invention does not impose any specific limitation on this.

[0094] It should be understood that there may be multiple preferred point cloud clusters. The preferred point cloud cluster identified in this step is the initially identified positioning plate point cloud.

[0095] Step S203: Determine the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information according to the preferred point cloud cluster, the positioning plate bounding box, and the position information or serial number information in the positioning plate bounding box, and determine the position information and / or speed information of the train at the current moment according to the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information.

[0096] The current frame point cloud information mentioned in the present invention refers to the point cloud information collected at the current moment. In this embodiment, each time a frame of point cloud information is collected, a train positioning and speed measurement is triggered.

[0097] Furthermore, in one implementation method, if the current frame point cloud information is the first frame point cloud information collected after the train positioning and speed measurement function is started, then after the step of determining the optimal point cloud cluster is performed, the current train positioning and speed measurement is directly terminated, and the next train positioning and speed measurement is triggered after the second frame point cloud information is collected.

[0098] The train positioning and speed measurement method provided in this embodiment does not rely on the existing "transponder-pulse speed measurement module" method. It collects point cloud information and image information of the positioning plate located next to the train line by arranging a laser radar and an image acquisition device on the train, and obtains the position and / or speed information of the train by integrating the point cloud information and image information. It provides a redundant backup solution for positioning and speed measurement for the train, improves the safety of train operation, and makes the detection results more accurate. When performing train positioning and speed measurement, there is no need to pre-build a high-precision electronic map, the requirements for the positioning plate are low, and the method is easier to implement. It can also realize train speed measurement while positioning, meeting the current needs for train operation data detection.

[0099] Embodiment 3

[0100] In this embodiment, based on the second embodiment, before acquiring the image information of the positioning plate through the image acquisition device, the method further includes:

[0101] The control device, the image acquisition device and the laser radar are preprocessed so that the control device is time synchronized with the laser radar, or respectively time synchronized with the laser radar and the image acquisition device, and the intrinsic parameter matrix and distortion coefficient of the image acquisition device are determined, and the pose matrix of the laser radar in the coordinate system of the image acquisition device is determined (the pose matrix of the laser radar in the coordinate system of the image acquisition device, that is, the extrinsic parameter matrix of the laser radar to the image acquisition device).

[0102] The image acquisition device includes a camera, and the coordinate system of the image acquisition device includes a camera coordinate system.

[0103] Specifically, if the image acquisition device supports trigger acquisition, then when performing time synchronization, it is only necessary to synchronize the laser radar with the control device. If the image acquisition device does not support trigger acquisition, it is necessary to synchronize the control device with the laser radar and the image acquisition device at the same time. More specifically, the intrinsic parameter matrix and distortion coefficient of the image acquisition device are obtained by calibrating the intrinsic parameters of the image acquisition device; the pose matrix of the laser radar in the coordinate system of the image acquisition device is obtained by jointly calibrating the image acquisition device and the laser radar.

[0104] When the laser radar collects point cloud information, it also records the collection time of each frame of point cloud information. For each frame of point cloud information, the collection timestamp of the first laser point of the frame of point cloud information is taken as the timestamp of the frame of point cloud information, that is, the collection time of the frame of point cloud information. If the image acquisition device supports trigger acquisition, the collection time of each frame of point cloud information is also the collection time of the image information corresponding to the frame of point cloud information. If the image acquisition device does not support trigger acquisition, the image acquisition device will also record the collection time of the image information when collecting image information.

[0105] Embodiment 4

[0106] In this embodiment, based on the second embodiment, when the position information and speed information of the train are measured in the last train positioning and speed measurement, before respectively determining the positioning plate point cloud of the current frame point cloud information and the last frame point cloud information according to the preferred point cloud cluster, the positioning plate bounding box, and the position information or sequence number information in the positioning plate bounding box, the method further includes:

[0107] The three-dimensional speed and three-dimensional position coordinates of the last train are obtained from the position information and speed information of the train obtained by the last train positioning and speed measurement; the three-dimensional position coordinates of the positioning plate in the current frame point cloud information are predicted based on the three-dimensional speed, three-dimensional position coordinates of the last train and the acquisition time difference between the current frame point cloud information and the previous frame point cloud information.

[0108] Specifically, taking the current frame point cloud information as the lth frame point cloud information as an example, the three-dimensional position coordinates of the positioning plate in the current frame point cloud information are predicted by the following formula:

[0109]

[0110] Among them, v *’ is the three-dimensional speed of the train measured during the last train positioning and speed measurement, X l-1 The three-dimensional position coordinates of the train measured during the last train positioning and speed measurement. is the three-dimensional position coordinates of the positioning plate in the predicted point cloud information of the current frame, Δt l is the acquisition time difference between the current frame point cloud information and the previous frame point cloud information.l-1 、v *’ Contains three directions: x, y, and z *’ >0. The train is moving forward; v *’ When <0, the train is moving backwards.

[0111] Furthermore, in one implementation, when the train positioning and speed measurement function is started, the positioning and speed measurement result status can be set to an initialization status.

[0112] Furthermore, after step S202 and before step S203, the method also includes first determining whether the current frame laser point cloud is the first frame laser point cloud received after the train positioning and speed measurement function is activated. If so, the current train positioning and speed measurement is terminated and initialization information is sent. Otherwise, step S203 is continued, and the train positioning and speed measurement result status is updated according to the result of the current train positioning and speed measurement. If the result of the current train positioning and speed measurement is that the train position information and speed information are not detected, the positioning and speed measurement result status is updated to an invalid state after the current train positioning and speed measurement is completed. If the result of the current train positioning and speed measurement is that the train position information and speed information are detected, the positioning and speed measurement result status is updated to a successful state after the current train positioning and speed measurement is completed. When performing the train positioning and speed measurement at the current moment (this train positioning and speed measurement), it can be determined whether the train position information and speed information were measured in the previous train positioning and speed measurement according to the positioning and speed measurement result status.

[0113] Furthermore, based on this embodiment, in another implementation, for a frame of point cloud information in the second embodiment, determining a preferred point cloud cluster that meets a preset condition from the point cloud information includes:

[0114] The reflection intensity of each laser point in the frame point cloud information is compared with the preset intensity threshold, and the laser points whose reflection intensity is not less than the preset intensity threshold are selected. According to the preset point number range and clustering radius, the selected laser points are clustered to obtain point cloud clusters. For each point cloud cluster, the centroid, length, width and height are calculated, and the point cloud cluster whose length is within the preset length range, height is within the preset height range, and width is within the preset width range is selected as the initial point cloud cluster, and the initial point cloud cluster is further screened to obtain the preferred point cloud cluster that meets the preset conditions.

[0115] Specifically, there may be multiple initial point cloud clusters, and screening is performed on each initial point cloud cluster.

[0116] Taking the point cloud information of the current frame as the point cloud information of the lth frame (l>1) as an example, for the kth initial point cloud cluster, the screening method includes:

[0117] The three-dimensional coordinates of the centroid of the kth initial point cloud cluster are marked as X lk , Xlk The three-dimensional position coordinates of the positioning plate in the predicted point cloud information of the current frame The difference is denoted as D x ,in,

[0118] D X The x, y, and z coordinates of D ,y D 、z D , if |x D | is within the preset x value range, and |y D | in the preset y value range, and | z D |Within the preset z value range, the kth initial point cloud cluster is determined as the preferred point cloud cluster.

[0119] The preset x value range, the preset y value range and the preset z value range are set by the technician according to the requirements, and the present invention does not limit this. For example, the x value range can be set to no more than 1.5 meters; the y value range can be set to no more than 1.0 meters; and the z value range can be set to no more than 0.5 meters.

[0120] Embodiment 5

[0121] Based on the second embodiment, in another embodiment of the present invention, the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are respectively determined according to the preferred point cloud cluster, the positioning plate bounding box, and the position information or serial number information in the positioning plate bounding box, including determining the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information through a preset point cloud image association strategy.

[0122] like Figure 3 As shown, the preset point cloud image association strategy includes the following steps:

[0123] Step S301: When the preferred point cloud cluster is determined from both the current frame point cloud information and the previous frame point cloud information, a first point cloud-image matching pair is generated according to the previous frame point cloud information and the first image information, and a second point cloud-image matching pair is generated according to the current frame point cloud information and the second image information.

[0124] It should be understood that the positioning plates are arranged at intervals beside the train line, so when collecting point cloud information, the positioning plates cannot be identified in every frame of point cloud information, that is, the preferred point cloud cluster cannot be identified in every frame of point cloud information.

[0125] Among them, the first image information is image information collected at the same time as the previous frame point cloud information, or image information collected at a different time from the previous frame point cloud information determined according to a preset time offset filtering strategy, and the second image information is image information collected at the same time as the current frame point cloud information, or image information collected at a different time from the current frame point cloud information determined according to a preset time offset filtering strategy.

[0126] Specifically, when the image acquisition device supports triggered acquisition, the first image information is image information acquired at the same time as the previous frame point cloud information, and the second image information is image information acquired at the same time as the current frame point cloud information; when the image acquisition device does not support triggered acquisition, the first image information is image information acquired at a different time than the previous frame point cloud information determined according to a preset time offset filtering strategy, and the second image information is image information acquired at a different time than the current frame point cloud information determined according to a preset time offset filtering strategy.

[0127] More specifically, in this embodiment, the preset time offset screening strategy includes:

[0128] The time difference between the acquisition time and the P1 acquisition time is less than or equal to the preset time difference threshold T s The image information of P2 is selected from which the image information that can detect the positioning plate bounding box and can identify the position information or serial number information in the positioning plate bounding box is formed into the image sequence I1. The time difference between the acquisition time and the acquisition time of P2 is less than or equal to the preset time difference threshold T s The image information of the positioning plate bounding box is selected from which the image information in which the positioning plate bounding box can be detected and the position information or serial number information in the positioning plate bounding box can be identified to form the image sequence I2.

[0129] Among them, P1 is the point cloud information of the previous frame, P2 is the point cloud information of the current frame, and the time difference is defined as the absolute value of the difference between the acquisition time of the image information and the point cloud information.

[0130] Preset time difference threshold T s Less than half of the time interval between two adjacent frames of point cloud information collection, T s The specific value of can be set by those skilled in the art according to actual needs, and the present invention does not limit this. Since the laser radar collects point cloud information according to a fixed preset collection frequency, the collection time interval of two adjacent frames of point cloud information can be determined according to the fixed preset collection frequency of the laser radar, and then the preset time difference threshold T is determined. s size.

[0131] to I k If the position information or serial number of the positioning plate is j, then the image information of this frame is compared with P kThe time difference of the acquisition time is recorded as Δt k (i, j), where k = 1, 2;

[0132] For the identified positioning plate position information or serial number information j, if it does not appear in both I1 and I2, its total time offset is defined as e t (-1,-1,j)=3T s ; If it appears in both the i1th frame image in I1 and the i2th frame image in I2, then its total time offset is defined as e t (i1,i2,j)=Δt1(i1,j)+Δt2(i2,j).

[0133] Traverse the positioning plate position information or serial number information in I1 and I2 in turn, and obtain the total time offset of each positioning plate position information or serial number information respectively, and record the minimum total time offset as like Then the first The frame image information is used as the first image information, and the first The frame image information is used as the second image information.

[0134] Step S302: When the position information or serial number information of the positioning plate bounding box of the first image information is the same as the position information or serial number information of the positioning plate bounding box of the second image information, determine the preferred bounding box of each preferred point cloud cluster in the current frame point cloud information and the preferred bounding box of each preferred point cloud cluster in the previous frame point cloud information.

[0135] Similarly, the positioning plates are arranged at intervals beside the train line, so when collecting image information, the positioning plates cannot be identified in every frame of image information, that is, the positioning plate bounding box and / or the position information or serial number information in the positioning plate bounding box cannot be identified in every frame of image information.

[0136] Specifically, in one implementation, for a frame of point cloud information, determining a preferred bounding box for each preferred point cloud cluster in the frame of point cloud information includes:

[0137] The three-dimensional position coordinates (x, y, z) of each laser point of each preferred point cloud cluster in the frame point cloud information are converted into pixel coordinates (u, v). After the conversion, each preferred point cloud cluster occupies a number of pixels on the image, and its bounding box is calculated, which is called the preferred bounding box.

[0138] It should be understood that the three-dimensional position coordinates of each laser point are coordinates in the laser coordinate system, and this step is to convert the coordinates of each laser point in the laser coordinate system into coordinates in the coordinate system of the image acquisition device. The specific conversion is performed based on the pose matrix determined in Example 3, the intrinsic parameter matrix of the image acquisition device, and the distortion coefficient.

[0139] Furthermore, in one implementation, when the image acquisition device does not support trigger acquisition, the laser radar also records the acquisition time of each laser point in each frame of point cloud information when acquiring point cloud information. When the train position information and speed information are measured by the last train positioning and speed measurement and the image acquisition device does not support trigger acquisition, before converting the three-dimensional position coordinates (x, y, z) of each laser point of each preferred point cloud cluster in the frame of point cloud information into pixel coordinates (u, v), the three-dimensional position coordinates of each laser point of each preferred point cloud cluster are corrected, and then the coordinate conversion and bounding box calculation are performed after correction.

[0140] Among them, for each laser point of each preferred point cloud cluster, its three-dimensional position coordinates are corrected according to the preset coordinate correction strategy. The preset coordinate correction strategy specifically includes:

[0141] Get the three-dimensional position coordinate X of the laser point to be corrected A And the acquisition time t A , the acquisition time of the image information in the point cloud-image matching pair where the laser point to be corrected is t CAMERA , according to the formula:

[0142] Get the corrected three-dimensional position coordinates Where λ is the preset constant coefficient, 0≤λ≤1, v *’ It is the three-dimensional speed of the train measured during the last train positioning and speed measurement. The specific value of λ is set by those skilled in the art according to actual needs, and the present invention does not limit this.

[0143] Step S303: after determining a first unique preferred bounding box from the preferred bounding boxes of the previous frame of point cloud information according to the positioning plate bounding box in the first image information, taking the preferred point cloud cluster corresponding to the first unique preferred bounding box as the positioning plate point cloud of the previous frame of point cloud information;

[0144] Step S304: after determining a second unique preferred bounding box from the preferred bounding boxes of the current frame point cloud information according to the positioning plate bounding box in the second image information, the preferred point cloud cluster corresponding to the second unique preferred bounding box is used as the positioning plate point cloud of the current frame point cloud information.

[0145] Specifically, for a frame of point cloud information, determining a unique preferred bounding box of the frame of point cloud information according to the positioning plate bounding box of the image information in the same point cloud-image matching pair as the frame of point cloud information includes the following steps:

[0146] The intersection over union (IoU) of each preferred bounding box and the positioning plate bounding box is calculated respectively. Where IoU>0 means that the corresponding preferred bounding box intersects with the positioning plate bounding box. IoU=0 means that the corresponding preferred bounding box does not intersect with the positioning plate bounding box.

[0147] If the maximum intersection-over-union ratio is greater than 0, the preferred bounding box corresponding to the maximum intersection-over-union ratio is used as the only preferred bounding box for the point cloud information of the frame.

[0148] If the maximum intersection-over-union ratio = 0, it is necessary to further select a unique preferred bounding box from the preferred bounding boxes.

[0149] More specifically, selecting a unique preferred bounding box from the preferred bounding boxes specifically includes the following steps:

[0150] Calculate the Z value and W value corresponding to each preferred bounding box respectively, and record the Z value and W value of the preferred bounding box with the largest Z value as Z max , W max .

[0151] If Z max ≥T z And W max ≥T W , Z max The corresponding preferred bounding box is used as the only preferred bounding box of the point cloud information of this frame, where T Z is the preset Z value threshold, T W is the preset W value threshold.

[0152] Among them, T Z With T W It is up to the technicians in this field to set it according to actual needs, and the present invention does not limit it. For example, T Z The value can be set to 5, T W The value can be set to 0.5.

[0153] For example, the kth preferred bounding box is taken as an example to calculate its Z value and W value.

[0154] Suppose the projection pixel coordinates of the kth optimal bounding box on the image are (u ki , v ki ), where i = 1, 2, 3, ..., N; N is the number of laser points in the preferred point cloud cluster for generating the kth preferred bounding box.

[0155] The Z value (Z) of the kth optimal bounding box is calculated by the following formula: k ) and W value (W k ):

[0156]

[0157]

[0158]

[0159] Among them, (u0, v0) is the centroid of the positioning plate bounding box (the positioning plate bounding box of the image information in the same point cloud-image matching pair as the point cloud information of this frame), u max is the maximum value of the u coordinate of the pixel point in the positioning plate bounding box (the positioning plate bounding box of the image information in the same point cloud-image matching pair as the point cloud information of this frame), u min v is the minimum u coordinate of the pixel point in the positioning plate bounding box (the positioning plate bounding box of the image information in the same point cloud-image matching pair as the point cloud information of this frame); max is the maximum value of the v coordinate of the pixel point in the positioning plate bounding box (the positioning plate bounding box of the image information in the same point cloud-image matching pair as the point cloud information of this frame), v min is the minimum v coordinate of the pixel point in the positioning plate bounding box (the positioning plate bounding box of the image information in the same point cloud-image matching pair as the point cloud information of this frame), T l is the preset distance threshold. l It is up to the technicians in this field to set it according to actual needs, and the present invention does not limit it. For example, T l The value can be set to

[0160] Furthermore, in another implementation of the present embodiment, after step S304, the method further includes: determining whether the data fusion is successful based on the point cloud centroid of the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information.

[0161] Furthermore, after the data fusion is successful, the positioning plate point cloud of the current frame point cloud information is saved, and the position information and / or speed information of the train at the current moment is determined based on the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information.

[0162] That is, in this implementation method, after determining the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, it is judged whether the data fusion is successful. Only if the data fusion is successful, the position information and / or speed information of the train at the current moment is determined based on the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information.

[0163] Among them, determining whether data fusion is successful based on the point cloud centroid of the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information specifically includes:

[0164] The centroid coordinates of the positioning plate point cloud of the previous frame of point cloud information are marked as P1 (x1, y1, z1);

[0165] The centroid coordinates of the positioning plate point cloud of the current frame point cloud information are marked as P2 (x2, y2, z2);

[0166] According to the formula Δy=|y2-y1|, Δz=|z2-z1|, calculate the increments Δy and Δz of the y and z coordinates:

[0167] If Δy <T y And Δz <T z , the fusion is determined to be successful, otherwise, the fusion is determined to be unsuccessful.

[0168] Among them, T y and T z For the preset increment threshold, those skilled in the art can set it according to actual needs, and the present invention does not limit this. For example, T y The value can be set to 0.5m, T z The value can be set to 0.4m.

[0169] Embodiment 6

[0170] On the basis of the fifth embodiment, when the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are determined by the preset point cloud image association strategy, as follows Figure 4 As shown, the current train position information and / or speed information is determined based on the positioning plate point cloud information of the current frame point cloud information and the previous frame point cloud information, including:

[0171] Step S401: Determine and save the position of the positioning plate according to the position information or serial number information in the positioning plate bounding box identified from the second image information.

[0172] Step S402: Determine the centroid of the positioning plate point cloud of the current frame point cloud information.

[0173] Step S403: Determine the current position of the train based on the position of the positioning plate and the centroid of the positioning plate point cloud of the current frame point cloud information, and use the centroid coordinates of the positioning plate point cloud of the current frame point cloud information as the three-dimensional position coordinates of the train at the current moment.

[0174] Among them, the centroid coordinates of the positioning plate point cloud of the current frame point cloud information are denoted as P2(x2, y2, z2), the position of the train is a scalar, and the position of the train = the position of the positioning plate - x2.

[0175] Among them, the position of the positioning plate is a scalar, and the train position is also a scalar. It should be understood that the centroid coordinates of the positioning plate point cloud of the current frame point cloud information are the coordinates in the laser coordinate system. The laser coordinate system is a three-dimensional coordinate system constructed with the laser radar as the origin, and the x, y, and z axes point to the front, left, and top directions of the laser radar respectively. The laser radar is set at the front of the train, so x2 is equivalent to the distance between the positioning plate and the front of the train in the x direction. Therefore, according to the position of the positioning plate and the distance between the positioning plate and the front of the train in the x direction, the position of the train can be determined.

[0176] Embodiment 7

[0177] On the basis of the fourth and fifth embodiments, in this embodiment, the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are respectively determined according to the preferred point cloud cluster, the positioning plate bounding box and the position information or sequence number information in the positioning plate bounding box, and further includes:

[0178] When the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information cannot be determined through the preset point cloud image association strategy, if the last train positioning and speed measurement data fusion is successful and there is a preferred point cloud cluster in the current frame point cloud information, the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information is determined through the preset point cloud target association strategy.

[0179] like Figure 5 As shown in the figure, the preset point cloud target association strategies include:

[0180] Step S501: The positioning plate point cloud of the current frame point cloud information determined during the last train positioning and speed measurement is used as the positioning plate point cloud of the previous frame point cloud information, and the d value of each preferred point cloud cluster in the current frame point cloud information is determined through a preset model based on the predicted three-dimensional position coordinates of the positioning plate in the current frame point cloud information.

[0181] Specifically, the predicted three-dimensional position coordinates of the positioning plate in the point cloud information of the current frame are predicted by the prediction method described in Example 4.

[0182] Step S502: when the minimum d value is not greater than the preset d value threshold, the point cloud corresponding to the preferred point cloud cluster having the minimum d value is used as the positioning plate point cloud of the current frame point cloud information.

[0183] Since each time a frame of point cloud information is collected, a train positioning and speed measurement is started. Therefore, the previous frame of point cloud information in this train positioning and speed measurement is actually the current frame of point cloud information in the previous train positioning and speed measurement. Therefore, in this embodiment, when the position information and speed information of the train are measured in the previous train positioning and speed measurement and the preferred point cloud cluster is determined from the current frame of point cloud information, the positioning plate point cloud of the current frame of point cloud information determined in the previous train positioning and speed measurement can be directly used as the positioning plate point cloud of the previous frame of point cloud information.

[0184] Specifically, the preset model in step S501 includes:

[0185]

[0186] Among them, x D ,y D and z D are the three-dimensional coordinates of D; x , y and λ z x D ,y D and z D The corresponding preset constant coefficient; for example, λ x =λ y =λ z =1.

[0187] Where D is determined by the following formula:

[0188]

[0189] Among them, X lk is the three-dimensional position coordinate of the centroid of the kth preferred point cloud cluster; It is the three-dimensional position coordinates of the positioning plate in the predicted point cloud information of the current frame.

[0190] Embodiment 8

[0191] On the basis of Embodiment 7, in this embodiment, if Figure 6 As shown, when the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are determined by the preset point cloud target association strategy, the position information and / or speed information of the train at the current moment is determined according to the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, including:

[0192] Step S601: Obtain the position of the positioning plate determined during the last train positioning and speed measurement.

[0193] Step S602: Determine the centroid of the positioning plate point cloud of the current frame point cloud information.

[0194] Step S603: Determine the current position of the train based on the position of the positioning plate determined during the last train positioning and speed measurement and the centroid of the positioning plate point cloud of the current frame point cloud information, and use the centroid coordinates of the positioning plate point cloud of the current frame point cloud information as the three-dimensional position coordinates of the train at the current moment.

[0195] Among them, the three-dimensional position coordinates of the centroid of the positioning plate point cloud of the current frame point cloud information are marked as P2 (x2, y2, z2), the position of the train is a scalar, and the position of the train = the position of the positioning plate determined during the last train positioning and speed measurement - x2.

[0196] Embodiment 9

[0197] In this embodiment, based on the sixth or eighth embodiment, as Figure 7 As shown, the position information and / or speed information of the train at the current moment is determined according to the positioning plate point cloud information of the current frame point cloud information and the previous frame point cloud information, and also includes:

[0198] Step S701: Determine whether this is the first time to calculate the train speed information. If so, execute step S702; otherwise, execute step S703.

[0199] Step S702: Determine the speed and three-dimensional speed of the train at the current moment according to the centroid of the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, and the acquisition time difference between the current frame point cloud information and the previous frame point cloud information.

[0200] In one implementation, the three-dimensional position coordinates of the centroid of the positioning plate point cloud of the current frame point cloud information are marked as P2 (x2, y2, z2), and the three-dimensional position coordinates of the centroid of the positioning plate point cloud of the previous frame point cloud information are marked as P1 (x1, y1, z1). Step S702 specifically determines the speed v2 of the train at the current moment according to the following formula:

[0201]

[0202] Where t1 is the acquisition time of the previous frame of point cloud information, and t2 is the acquisition time of the current frame of point cloud information.

[0203] Step S702 specifically determines the three-dimensional speed v of the train at the current moment according to the following formula: * :

[0204]

[0205] Step S703: Determine the current speed and three-dimensional speed of the train according to the centroid of the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, the acquisition time difference between the current frame point cloud information and the previous frame point cloud information, and the most recently measured train speed.

[0206] In one implementation, the three-dimensional position coordinates of the centroid of the positioning plate point cloud of the current frame point cloud information are marked as P2 (x2, y2, z2), and the three-dimensional position coordinates of the centroid of the positioning plate point cloud of the previous frame point cloud information are marked as P1 (x1, y1, z1). Step S703 specifically determines the speed v2 of the train at the current time according to the following formula:

[0207]

[0208] Where v1 is the most recently measured train speed, k is the preset constant coefficient, t1 is the acquisition time of the previous frame of point cloud information, and t2 is the acquisition time of the current frame of point cloud information.

[0209] Step S703 specifically determines the three-dimensional speed v of the train at the current moment according to the following formula: * :

[0210]

[0211] Further, if the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are determined according to the method in Example five, then k=1; if the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are determined according to the method in Example seven, then 0.8≤k≤1.

[0212] The following example illustrates the meaning of "the latest measured train speed". Assume that 5 train positioning and speed measurements have been completed so far, and this is the 6th train positioning and speed measurement. Among them, only the 2nd and 4th train positioning and speed measurements successfully calculated the train position information and train speed information, and the rest failed. Then when the 6th train positioning and speed measurement calculates the train speed v2, the train speed calculated by the 4th train positioning and speed measurement is the latest measured train speed.

[0213] Embodiment 10

[0214] In summary of the above embodiments, in this embodiment, another method for train positioning and speed measurement is provided. In this embodiment, the image acquisition device is a camera, and the train is a subway train. Specifically, the method includes the following steps:

[0215] Step S1101: Perform preliminary work.

[0216] The preliminary work includes installation of the positioning board, calibration of camera internal parameters, joint calibration of camera and lidar, and time synchronization.

[0217] Among them, if the existing 100-meter mark of the subway line is used as the positioning plate, there is no need to install the positioning plate. Otherwise, the positioning plate is arranged next to the train line. The positioning plate is a flat plate made of metal material with high reflectivity. The base color of the positioning plate is a pure color with high brightness, and numbers are marked on it. The numbers can represent the position information of the positioning plate in the line or the serial number information of the positioning plate in the line. When the number is the serial number of the positioning plate in the line, it is also necessary to store the correspondence between the serial numbers of each positioning plate and the position of the positioning plate in the line. In one implementation, if the positioning plates are arranged at equal intervals, the position of the positioning plate in the line can be calculated based on the serial number of the positioning plate and the spacing between the positioning plates. At this time, it is only necessary to store the correspondence between the serial number of a positioning plate and the position of the positioning plate in the line.

[0218] The laser radar is installed on the front of the train, and the x, y, and z axes of the laser radar coordinate system point to the front, left, and top directions respectively. The camera is installed on the front of the train, and faces forward.

[0219] The camera’s intrinsic parameter matrix and distortion coefficients are obtained through camera intrinsic parameter calibration.

[0220] The camera-lidar joint calibration is used to obtain the pose matrix of the lidar in the camera coordinate system, that is, the external parameter matrix of the lidar to the camera.

[0221] During time synchronization, the laser radar and the control device need to be synchronized. At the same time, if the camera does not support trigger acquisition, the camera and the control device also need to be synchronized. On the contrary, if the camera supports trigger acquisition, only the laser radar and the control device need to be synchronized, and the camera and the control device do not need to be synchronized.

[0222] Step S1102: Collecting laser point cloud and camera image data.

[0223] Camera: If the camera supports triggered acquisition, when the LiDAR acquires the first point of a frame of laser point cloud, the camera is triggered to acquire an image. If the camera does not support triggered acquisition, the camera acquires images at a fixed preset acquisition frequency (generally higher than the preset acquisition frequency of the LiDAR).

[0224] LiDAR: Collect point clouds at a fixed preset acquisition frequency. For each frame of laser point cloud, take the timestamp of the first point as the timestamp (acquisition time) of the laser point cloud of that frame. If the camera does not support trigger acquisition, the timestamp (acquisition time) of each laser point in the laser point cloud of that frame must be recorded at the same time.

[0225] Step S1103: Determine the train location information and speed information.

[0226] This step specifically includes the following sub-steps:

[0227] Step S110301: When the train positioning and speed measurement functions are started, the result status of the train positioning and speed measurement is set to the initialization status.

[0228] Among them, the result status of train positioning and speed measurement is divided into three states: success state, failure state and initialization state.

[0229] Once the train positioning and speed measurement functions are started, after initialization, continue to execute step S110302.

[0230] Step S110302: Laser detection.

[0231] Every time the control device receives a frame of laser point cloud, it performs a train positioning speed to determine the train position information and speed information.

[0232] For the convenience of description, it is assumed that the laser point cloud of the current frame is the laser point cloud of the lth frame.

[0233] Specifically, after receiving the laser point cloud of the current frame, if the result status of the train positioning and speed measurement is an initialization state or an invalid state, point cloud clustering is performed on the laser point cloud of the current frame.

[0234] Among them, for the laser point cloud of the current frame, the laser points whose intensity reaches or exceeds the preset intensity threshold are selected, and the selected laser points are clustered according to the preset point number range and clustering radius; for each point cloud cluster obtained after clustering, the centroid, length, width and height are calculated, and multiple point cloud clusters with a length within the preset length range, a height within the preset height range, and a width within the preset width range are selected as the initial point cloud cluster output.

[0235] If the result status of the train positioning and speed measurement is a success status (that is, the last train positioning and speed measurement successfully measured the train position information and speed information), first predict the three-dimensional position coordinates of the positioning plate point cloud in the current frame laser point cloud, and then perform point cloud clustering.

[0236] Specifically, predicting the three-dimensional position coordinates of the positioning plate point cloud in the current frame laser point cloud includes: obtaining the three-dimensional position coordinates x of the train measured during the last train positioning and speed measurement l-1 and the three-dimensional velocity v * '. According to the acquisition time difference Δt between the current frame laser point cloud and the previous frame laser point cloud l , predict the 3D position coordinates of the positioning plate point cloud in the current frame laser point cloud in,

[0237]

[0238] After the point cloud clustering is performed and the initial point cloud cluster is output, the preferred point cloud cluster is determined. If the result status of the train positioning and speed measurement is an invalid state or an initialization state, all the initial point cloud clusters output in the point cloud clustering are selected as the preferred point cloud clusters; if the result status of the train positioning and speed measurement is a successful state, each initial point cloud cluster is further screened to determine the preferred point cloud cluster.

[0239] Specifically, taking the kth initial point cloud cluster as an example, screening the kth initial point cloud cluster includes: marking the three-dimensional position coordinates of the centroid of the kth initial point cloud cluster as X lk , X lk The three-dimensional position coordinates of the positioning plate point cloud in the current frame laser point cloud are predicted The difference is denoted as D X ,in,

[0240] D X The x, y, and z coordinates of D ,y D 、z D , if |x D | is within the preset x value range, and |y D | in the preset y value range, and | z D |Within the preset z value range, the kth initial point cloud cluster is determined as the preferred point cloud cluster.

[0241] Step S110303: visual inspection.

[0242] For the received image, the trained deep neural network is used to detect the positioning plate in the image, and the bounding box of the positioning plate is obtained. At most one bounding box is output. Subsequently, the optical character recognition (OCR) technology is used to identify the number in the bounding box, that is, the number on the positioning plate. In order to obtain a "trained" deep neural network, a large number of positioning plate images need to be collected in advance for model training.

[0243] Step S110304: First frame laser point cloud judgment: if the current frame laser point cloud is the first frame laser point cloud received after initialization, execute step S110314; otherwise, execute step S110305.

[0244] Step S110305: Determine whether the laser detection of the current frame laser point cloud and the previous frame laser point cloud have both determined the preferred point cloud cluster. If so, execute step S110306; otherwise, execute step S110309.

[0245] Step S110306: constructing point cloud image matching pairs, constructing a first point cloud image matching pair and a second point cloud image matching pair.

[0246] Specifically, if the camera supports trigger acquisition, the laser point cloud and camera image data are collected synchronously, so the image at the corresponding time can be obtained for any frame of laser point cloud, forming a "point cloud-image" matching pair. The first point cloud-image matching pair is formed according to the previous frame of laser point cloud and the image collected at the same time as the previous frame of laser point cloud, and the second point cloud-image matching pair is formed according to the current frame of laser point cloud and the image collected at the same time as the current frame of laser point cloud. If the two frames of images in the first point cloud-image matching pair and the second point cloud-image matching pair can detect the positioning plate (both identify the positioning plate bounding box), and the numbers on the positioning plates identified twice are the same, it is considered that the visual detection is successful, and step S110307 is continued; on the contrary, if the two frames of images in the first point cloud-image matching pair and the second point cloud-image matching pair fail to detect the positioning plate at the same time, or fail to identify the same positioning plate number at the same time, it is considered that the visual detection fails, and then it is directly determined that the "point cloud-image" matching pair construction fails, and step S110309 is subsequently executed.

[0247] If the camera does not support trigger acquisition, define the time difference as the absolute value of the difference between the image and laser point cloud acquisition time. Set the time difference threshold T s , where T s Less than half of the time interval between two adjacent frames of laser point cloud acquisition.

[0248] For any frame of laser point cloud, obtain all the points whose time difference is no more than T s image frames, and select the image frames in which the positioning plate bounding box can be detected and the numbers can be recognized to form an image sequence, denoted as I.

[0249] The image sequence corresponding to the previous frame laser point cloud P1 is recorded as I1, and the image sequence corresponding to the current frame laser point cloud P2 is recorded as I2. k If the number of the positioning plate identified is j, then the image frame is compared with P k The time difference of the acquisition time is recorded as Δt k (i,j), where k=1,2.

[0250] Traverse the location plate numbers in I1 and I2 in turn. For any location plate number j, if it does not appear in I1 and I2 at the same time, its total time offset is defined as e t (-1,-1,j)=3T s , if it appears in both the i1th frame image in I1 and the i2th frame image in I2, then its total time offset is defined as e t(i1,i2,j)=Δt1(i1,j)+Δt2(i2,j).

[0251] Get the total time offset of each positioning plate number in I1 and I2 respectively, and record the minimum total time offset as like Then the first The frame image and the previous frame laser point cloud form the first point cloud-image matching pair. The frame image and the current frame laser point cloud form a second point cloud-image matching pair, and step S110307 is continued.

[0252] like It is considered that the “point cloud-image” matching pair construction has failed, and the subsequent step S110309 is executed.

[0253] Step S110307: For each point cloud-image matching pair, perform point cloud-image target association. If both point cloud-image matching pairs are successfully associated, proceed to step S110308; otherwise, proceed to step S110309.

[0254] If the result status of the train positioning and speed measurement is invalid or initialized, or the camera supports trigger acquisition, for any point cloud-image matching pair, the point cloud-image target association needs to perform the following operations:

[0255] (1) Coordinate conversion. Based on the external parameter matrix, camera intrinsic parameter matrix, and distortion coefficient obtained by joint calibration, the three-dimensional position coordinates (x, y, z) of each laser point in the preferred point cloud cluster of the laser point cloud in the point cloud-image matching pair are converted into pixel coordinates (u, v). Each preferred point cloud cluster occupies a number of pixels on the image, and its bounding box is calculated, which is called the preferred bounding box. Each preferred point cloud cluster has a corresponding preferred bounding box.

[0256] (2) Bounding box association. When visual inspection is performed on an image, only one bounding box (hereinafter referred to as the visual inspection bounding box) is output, but there may be multiple preferred bounding boxes. It is necessary to determine a unique preferred bounding box from the preferred bounding boxes.

[0257] Specifically, the intersection over union (IoU) between the visual detection bounding box and each preferred bounding box is calculated respectively.

[0258] If IoU>0, that is, there is a preferred bounding box that intersects with the visual detection bounding box, then the preferred bounding box corresponding to the maximum IoU is taken as the only preferred bounding box; if all IoUs are 0, that is, all preferred bounding boxes do not intersect with the visual detection bounding box, further screening is performed.

[0259] Specifically, further screening includes:

[0260] The centroid of the visual detection bounding box is recorded as (u0, v0), and the maximum value of the u coordinate of the pixel point in the visual detection bounding box is recorded as u max , the minimum value of u coordinate is recorded as u min , the maximum value of v coordinate is recorded as v max , the minimum value of v coordinate is recorded as v min . Calculate the Z value and W value of each preferred bounding box of the laser point cloud in the point cloud-image matching pair. The Z value and W value of the preferred bounding box with the largest Z value are recorded as Z max , W max , if Z max ≥T z And W max ≥T W , Z max The corresponding preferred bounding box is used as the only preferred bounding box, where T z is the preset Z value threshold, T W is the preset W value threshold.

[0261] Taking the kth preferred bounding box as an example, let the projection pixel coordinates of the kth preferred bounding box on the image be (u ki , v ki ), where i = 1, 2, 3, ..., N; N is the number of laser points in the preferred point cloud cluster for generating the kth preferred bounding box.

[0262] The Z value (Z) of the kth optimal bounding box is calculated by the following formula: k ) and W value (W k ):

[0263]

[0264]

[0265]

[0266] Among them, T l is the preset distance threshold.

[0267] If the result status of the train positioning and speed measurement is a success status and the camera does not support trigger acquisition, before the coordinate conversion, the operations that need to be performed for the above-mentioned point cloud-image target association also include correcting the three-dimensional position coordinates of each laser point of each preferred point cloud cluster according to the preset coordinate correction strategy.

[0268] The preset coordinate correction strategy specifically includes:

[0269] Get the three-dimensional position coordinate X of the laser point to be corrected A And the acquisition time t A, the acquisition time of the image information in the point cloud-image matching pair where the laser point to be corrected is t CAMERA , according to the formula:

[0270] Get the corrected three-dimensional position coordinates Where λ is the preset constant coefficient, 0≤λ≤1, v *’ It is the three-dimensional speed of the train measured during the last train positioning and speed measurement.

[0271] For each point cloud-image matching pair, if the point cloud-image matching pair can determine a unique preferred bounding box, the point cloud-image target association of the point cloud-image matching pair is considered successful.

[0272] Step S110308: Smoothness calculation.

[0273] According to the point cloud-image target association result of the point cloud-image matching pair, the optimal point cloud cluster corresponding to the unique optimal bounding box is obtained, that is, the positioning plate point cloud. According to the first point cloud-image matching pair, the positioning plate point cloud of the previous frame of laser point cloud can be obtained, and according to the second point cloud-image matching pair, the positioning plate point cloud of the current frame of laser point cloud can be obtained.

[0274] The three-dimensional position coordinates of the centroid of the positioning plate point cloud of the previous frame of laser point cloud are marked as P1 (x1, y1, z1);

[0275] The three-dimensional position coordinates of the centroid of the positioning plate point cloud of the current frame laser point cloud are marked as P2 (x2, y2, z2);

[0276] According to the formula Δy=|y2-y1|, Δz=|z2-z1|, calculate the increments Δy and Δz of the y and z coordinates:

[0277] If Δy <T y And Δz <T z , it is considered smooth, the fusion is determined to be successful, and steps S110311-S110312 are executed; otherwise, the fusion is determined to be unsuccessful, and step S110309 is executed.

[0278] Step S110309: Determine whether point cloud target association is required. If the data fusion was successful during the last train positioning and speed measurement, and there is a preferred point cloud cluster in the current frame laser point cloud, it is considered that point cloud target association needs to be performed, and step S110310 is executed. Otherwise, step S110313 is executed.

[0279] Step S110310: For the laser point cloud of the current frame, traverse each preferred point cloud cluster in turn and perform point cloud target association operation. If the point cloud target association is successful, execute steps S110311-S110312, otherwise execute step S110313.

[0280] Specifically, the point cloud target association operation includes:

[0281] Calculate the d value of each preferred point cloud cluster in turn, select the preferred point cloud cluster with the minimum d value, if the minimum d value is not greater than the preset d value threshold, it is considered that the point cloud target association is successful, and the preferred point cloud cluster corresponding to the minimum d value is used as the positioning plate point cloud of the current frame laser point cloud. Otherwise, the point cloud target association fails.

[0282] More specifically, taking the kth preferred point cloud cluster as an example, the d value is calculated by the following formula:

[0283]

[0284] Among them, x D ,y D and z D are the three-dimensional coordinates of D; x , y and λ z x D ,y D and z D The corresponding preset constant coefficient; wherein D is determined by the following formula:

[0285]

[0286] Among them, X lK is the three-dimensional position coordinate of the centroid of the kth preferred point cloud cluster; is the predicted 3D position coordinates of the positioning plate in the laser point cloud of the current frame.

[0287] Step S110311: Calculate the position information and speed information, and execute step S110312 after calculation.

[0288] (1) Calculate location information.

[0289] The position information includes the three-dimensional position information of the train and the position of the train.

[0290] Specifically, if the data fusion is successful, the position of the positioning plate (scalar) is determined according to the number recognized from the image of the second point cloud image matching pair, the three-dimensional position coordinates of the center of mass of the positioning plate point cloud of the current frame laser point cloud are marked as P2 (x2, y2, z2), and the position of the train (scalar) is calculated according to the three-dimensional position coordinates of the center of mass of the positioning plate point cloud of the current frame laser point cloud. Among them, the position of the train = the position of the positioning plate - X2.

[0291] If the point cloud target is successfully associated, the train position is calculated using the positioning plate position determined when the last train positioning and speed measurement data was successfully fused and the three-dimensional position coordinates of the positioning plate point cloud centroid of the current frame laser point cloud. Among them, the train position = the positioning plate position determined when the last train positioning and speed measurement data was successfully fused - x2.

[0292] At the same time, the three-dimensional position coordinates P2 (x2, y2, z2) of the centroid of the positioning plate point cloud of the current frame laser point cloud are used as the three-dimensional position coordinates of the train.

[0293] (2) Calculation speed information

[0294] According to the centroid coordinates P2 (x2, y2, z2) of the positioning plate point cloud of the current frame laser point cloud and the centroid coordinates P1 (x1, y1, z1) of the positioning plate point cloud of the previous frame laser point cloud, the acquisition time t2 of the current frame laser point cloud and the acquisition time t1 of the previous frame laser point cloud, the speed of the train (scalar) is calculated:

[0295] If this is the first time to calculate the speed, the speed v2 of the train is:

[0296]

[0297] If this is not the first time the speed is calculated, the speed v2 of the train is:

[0298]

[0299] Where v1 is the most recently measured train speed, k is the preset constant coefficient, if the data fusion is successful, k = 1; if the point cloud target association is successful, 0.8 ≤ k ≤ 1.

[0300] At the same time, the three-dimensional speed of the train is calculated according to the following formula:

[0301]

[0302] Step S110312: Update the status of the train's positioning and speed measurement results to a successful status, and send a positioning and speed measurement success message, wherein the success message includes the train's position and speed.

[0303] Step S110313: Update the status of the train's positioning and speed measurement results to an invalid state, and send positioning and speed measurement failure information.

[0304] Step S110314: Send positioning and speed measurement initialization information.

[0305] Embodiment 11

[0306] A computer program product is provided in this embodiment. The computer program product includes a computer program or instructions. When the computer program or instructions are executed by a processor, all or part of the steps of the method in the aforementioned method embodiment are implemented, and this embodiment will not be repeated here.

[0307] Further, the computer program product may include one or more computer executable components configured to execute the embodiment when the program is run; the computer program product may also include a computer program tangibly contained on a readable medium, the computer program including program code for executing any method in the embodiments of the present disclosure. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium.

[0308] Embodiment 12

[0309] This embodiment provides a computer-readable storage medium. The computer program stored in the computer-readable storage medium, when executed by one or more processors, implements all or part of the steps of the method in the above method embodiment, which will not be repeated in this embodiment.

[0310] Embodiment 13

[0311] This embodiment provides an electronic device, Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application is shown in FIG. Figure 8 As shown, the electronic device 800 includes: at least one processor 801, at least one communication bus 802, a user interface 803, at least one external communication interface 804, and a memory 805. The communication bus 802 is configured to realize the connection and communication between these components. The user interface 803 may include a display screen, and the external communication interface 804 may include a standard wired interface and a wireless interface. A computer program is stored in the memory 805, and the memory 805 and one or more processors 801 are connected to each other in communication. When the computer program is executed by one or more processors, the processor 801 is configured to execute the computer program stored in the memory to implement all or part of the steps of the method in the above-mentioned embodiment 1, and this embodiment will not be repeated here.

[0312] The train positioning and speed measurement system, method and electronic equipment provided by the present invention do not rely on the existing "transponder-pulse speed measurement module", but instead collect point cloud information and image information of a positioning plate located next to the train line by arranging a laser radar and an image acquisition device on the train, and obtain the position and / or speed information of the train by integrating the point cloud information and image information, thereby providing a redundant backup solution for positioning and speed measurement for the train, improving the safety of train operation, and making the detection results more accurate. When performing train positioning and speed measurement, there is no need to pre-build a high-precision electronic map, the requirements for the positioning plate are relatively low, and the method is easier to implement. In addition, the train speed can be measured while positioning, meeting the current demand for train operation data detection.

[0313] The terms and expressions used in the present specification are for illustration only and are not intended to be limiting. It should be understood by those skilled in the art that various changes may be made to the details of the above embodiments without departing from the basic principles of the disclosed embodiments. Therefore, the scope of the present invention is determined solely by the claims, in which, unless otherwise indicated, all terms should be understood in the broadest reasonable sense.

Claims

1. A train positioning and speed measurement system, characterized in that: The system comprises: An image acquisition device, used to acquire image information of a positioning plate located beside a train line; A laser radar, used for collecting point cloud information reflected by the positioning plate by means of laser irradiation; A control device is communicatively connected to the image acquisition device and the laser radar, respectively, and is used to control the image acquisition device and the laser radar to respectively acquire the image information and point cloud information of the positioning plate according to a preset working mode, and to determine the positioning plate bounding box according to the image information of the positioning plate and identify the position information or serial number information in the positioning plate bounding box, and to determine the position and / or speed information of the train according to the positioning plate bounding box, the position information or serial number information in the positioning plate bounding box and the point cloud information.

2. A train positioning and speed measurement method, characterized in that: Based on the train positioning and speed measurement system according to claim 1, the method comprises: Acquire image information of the positioning plate from the image acquisition device through the control device, and determine the positioning plate bounding box and the position information or serial number information in the positioning plate bounding box according to the image information; Acquire point cloud information of the positioning plate from a laser radar, and determine a preferred point cloud cluster that meets preset conditions from the point cloud information; The positioning plate point clouds of the current frame point cloud information and the previous frame point cloud information are respectively determined according to the preferred point cloud cluster, the positioning plate bounding box and the position information or serial number information in the positioning plate bounding box, and the position information and / or speed information of the train at the current moment is determined according to the positioning plate point clouds of the current frame point cloud information and the previous frame point cloud information.

3. The train positioning and speed measurement method according to claim 2, characterized in that: Before acquiring the image information of the positioning plate by the image acquisition device, the method further includes: The control device, the image acquisition device and the laser radar are preprocessed so that the control device is time synchronized with the laser radar or respectively with the laser radar and the image acquisition device, and the intrinsic parameter matrix and distortion coefficient of the image acquisition device are determined, and the pose matrix of the laser radar in the coordinate system of the image acquisition device is determined.

4. The train positioning and speed measurement method according to claim 2, characterized in that: The method of determining the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information respectively according to the preferred point cloud cluster, the positioning plate bounding box and the position information or serial number information in the positioning plate bounding box comprises: determining the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information by a preset point cloud image association strategy; The preset point cloud image association strategy includes: In the case where preferred point cloud clusters are determined from both the current frame point cloud information and the previous frame point cloud information, a first point cloud-image matching pair is generated according to the previous frame point cloud information and the first image information, and a second point cloud-image matching pair is generated according to the current frame point cloud information and the second image information; wherein the first image information is image information collected at the same time as the previous frame point cloud information or image information collected at a different time from the previous frame point cloud information according to a preset time offset screening strategy, and the second image information is image information collected at the same time as the current frame point cloud information or image information collected at a different time from the current frame point cloud information according to the preset time offset screening strategy; When the position information or serial number information of the positioning plate bounding box of the first image information is the same as the position information or serial number information of the positioning plate bounding box of the second image information, determining the preferred bounding box of each preferred point cloud cluster in the current frame point cloud information and the preferred bounding box of each preferred point cloud cluster in the previous frame point cloud information; After determining a first unique preferred bounding box from the preferred bounding boxes of the previous frame of point cloud information according to the positioning plate bounding box in the first image information, taking the preferred point cloud cluster corresponding to the first unique preferred bounding box as the positioning plate point cloud of the previous frame of point cloud information; After determining a second unique preferred bounding box from the preferred bounding boxes of the current frame point cloud information according to the positioning plate bounding box in the second image information, the preferred point cloud cluster corresponding to the second unique preferred bounding box is used as the positioning plate point cloud of the current frame point cloud information.

5. The train positioning and speed measurement method according to claim 4, characterized in that: After respectively determining the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information according to the preferred point cloud cluster, the positioning plate bounding box, and the position information or sequence information in the positioning plate bounding box, the method further includes: Determine whether data fusion is successful according to the point cloud centroid of the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information; After data fusion is successful, the positioning plate point cloud of the current frame point cloud information is saved, and the position information and / or speed information of the train at the current moment is determined based on the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information.

6. The train positioning and speed measurement method according to claim 5, characterized in that: The method further comprises: determining the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information respectively according to the preferred point cloud cluster, the positioning plate bounding box and the position information or sequence number information in the positioning plate bounding box; When the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information cannot be determined by the preset point cloud image association strategy, if the last train positioning and speed measurement data fusion is successful and there is a preferred point cloud cluster in the current frame point cloud information, the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information is determined by the preset point cloud target association strategy.

7. The train positioning and speed measurement method according to claim 6, characterized in that: In the case where the position information and speed information of the train are measured in the last train positioning and speed measurement, before respectively determining the positioning plate point cloud of the current frame point cloud information and the last frame point cloud information according to the preferred point cloud cluster, the positioning plate bounding box, and the position information or sequence number information in the positioning plate bounding box, the method further includes: Obtain the three-dimensional speed and three-dimensional position coordinates of the last train from the position information and speed information of the train measured by the last train positioning and speed measurement; The three-dimensional position coordinates of the positioning plate in the current frame point cloud information are predicted based on the three-dimensional speed and three-dimensional position coordinates of the previous train and the acquisition time difference between the current frame point cloud information and the previous frame point cloud information.

8. The train positioning and speed measurement method according to claim 7, characterized in that: The preset point cloud target association strategy includes: The positioning plate point cloud of the current frame point cloud information determined during the last train positioning and speed measurement is used as the positioning plate point cloud of the previous frame point cloud information, and the d value of each preferred point cloud cluster in the current frame point cloud information is determined by a preset model according to the predicted three-dimensional position coordinates of the positioning plate in the current frame point cloud information; When the minimum d value is not greater than the preset d value threshold, the point cloud corresponding to the preferred point cloud cluster with the minimum d value is used as the positioning plate point cloud of the current frame point cloud information.

9. The train positioning and speed measurement method according to claim 8, characterized in that: The preset model includes: in, D ,y D and z D are the three-dimensional coordinates of D; x , y and λ z x D ,y D and z D The corresponding preset constant coefficient; wherein D is determined by the following formula: Among them, X lk is the three-dimensional position coordinate of the kth optimal point cloud cluster; The three-dimensional position coordinates of the positioning plate in the point cloud information of the current frame are predicted.

10. The train positioning and speed measurement method according to claim 4, characterized in that: In the case where the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are determined by the preset point cloud image association strategy, the position information and / or speed information of the train at the current moment is determined based on the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, including: Determine and save the position of the positioning plate according to the position information or serial number information in the positioning plate bounding box identified from the second image information; Determine the centroid of the positioning plate point cloud of the current frame point cloud information; The position of the train at the current moment is determined according to the position of the positioning plate and the centroid of the positioning plate point cloud of the current frame point cloud information, and the centroid coordinates of the positioning plate point cloud of the current frame point cloud information are used as the three-dimensional position coordinates of the train at the current moment.

11. The train positioning and speed measurement method according to claim 8, characterized in that: In the case where the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information are determined by the preset point cloud target association strategy, the position information and / or speed information of the train at the current moment is determined based on the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, including: Obtain the position of the positioning plate determined during the last train positioning and speed measurement; Determine the centroid of the positioning plate point cloud of the current frame point cloud information; The position of the train at the current moment is determined based on the position of the positioning plate determined during the last train positioning and speed measurement and the centroid of the positioning plate point cloud of the current frame point cloud information, and the centroid coordinates of the positioning plate point cloud of the current frame point cloud information are used as the three-dimensional position coordinates of the train at the current moment.

12. The train positioning and speed measurement method according to claim 10 or 11, characterized in that: The determining of the position information and / or speed information of the train at the current moment according to the positioning plate point cloud information of the current frame point cloud information and the previous frame point cloud information also includes: In the case that this is the first time to calculate the train speed information, the speed and three-dimensional speed of the train at the current moment are determined according to the centroid of the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, and the acquisition time difference between the current frame point cloud information and the previous frame point cloud information; When this is not the first time that the train speed information is calculated, the speed and three-dimensional speed of the train at the current moment are determined based on the centroid of the positioning plate point cloud of the current frame point cloud information and the previous frame point cloud information, the acquisition time difference between the current frame point cloud information and the previous frame point cloud information, and the most recently measured train speed.

13. A computer-readable storage medium, characterized in that: The computer program stored in the computer-readable storage medium, when executed by one or more processors, implements the steps of the method according to any one of claims 2 to 12.

14. An electronic device, characterized in that: It comprises a memory and one or more processors, wherein the memory stores a computer program, and the memory and the one or more processors are communicatively connected with each other, and when the computer program is executed by the one or more processors, the steps of the method as claimed in any one of claims 2 to 12 are performed.

Images