Rail transit obstacle detection and early warning system

Abstract

The embodiment of the invention relates to a rail transit obstacle detection and early warning system. The system comprises a laser radar, a camera, an inertial navigation module, a data exchange module, a main control module, a sound-light alarm module, a storage module and a start-stop button module, according to the invention, the driving safety guarantee can be improved, and the operation complexity of drivers and passengers / system users can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a rail transit obstacle detection and early warning system. Background Art

[0002] When rail transit vehicles (such as subway trains, intercity railway trains, national network railway trains, etc.) are traveling, if obstacles (such as people, animals, vehicles, etc.) appear in the front track area or the proximal area on both sides of the track, it is easy to cause major traffic accidents. To reduce the probability of such accidents, it is necessary to provide rail transit vehicles with a technical means that can actively detect obstacles, which is also the technical problem that the present invention needs to solve. Summary of the invention

[0003] The purpose of the present invention is to provide a rail transit obstacle detection and early warning system in view of the defects of the prior art, including: a laser radar, a camera, an inertial navigation module, a data exchange module, a main control module, an audible and visual alarm module, a storage module and a start-stop button module; wherein, the data exchange module is a data exchange device of the whole system (similar to a switch), the laser radar, the camera and the inertial navigation module are three types of perception components of the system, the audible and visual alarm module is an active early warning component of the system, the storage module is an early warning log storage component of the system, the main control module is a core component of the system, and the start-stop button module is a one-button start-stop component for controlling the start / stop of the main control module; the main control module is used to actively identify whether there are obstacles in the front driving track area + the proximal area on both sides of the track according to the real-time perception data (scene point cloud, scene video, self-vehicle operation parameters) provided by the three types of perception components, and when it is confirmed that an obstacle has occurred, it is promptly warned through the audible and visual alarm module and the early warning log is stored through the storage module. On the one hand, the present invention can provide rail transit vehicles with a technical means for actively detecting obstacles, thereby improving driving safety and reducing the probability of collision accidents; on the other hand, based on the one-button start-stop function, the operation complexity of drivers and passengers / system users can be reduced.

[0004] To achieve the above-mentioned purpose, an embodiment of the present invention provides a rail transit obstacle detection and warning system, the system comprising: a laser radar, a camera, an inertial navigation module, a data exchange module, a main control module, an audio and visual alarm module, a storage module and a start-stop button module;

[0005] The data exchange module includes at least six Ethernet exchange ports, which are respectively connected to the laser radar, the camera, the inertial navigation module, the main control module, the sound and light alarm module and the storage module through RJ45 interface cables; the main control module is directly connected to the start-stop button module;

[0006] The laser radar is installed on the front of the rail transit vehicle; the laser radar is used to perform radar scanning on the front scene of the rail transit vehicle to obtain a corresponding first scene point cloud; and regularly send the latest first scene point cloud to the main control module according to a preset first upload frequency;

[0007] The camera is installed at the front of the rail transit vehicle; the camera is used to capture a video of a scene in front of the rail transit vehicle to obtain a corresponding first scene video; and regularly sends the latest captured first scene video to the main control module at a preset second upload frequency; the video length of each first scene video is the preset first video length; the second upload frequency is less than the first upload frequency; the first video length is greater than the time period corresponding to the first upload frequency;

[0008] The inertial navigation module is used to measure the acceleration, speed, position and posture of the rail transit vehicle in real time to calculate the corresponding first acceleration, first speed, first position and first posture to form a corresponding first motion parameter group; and regularly send the latest first motion parameter group to the main control module according to the preset third upload frequency; the first position is the track travel distance of the rail transit vehicle after it departs from the departure station; the first posture includes pitch angle, yaw angle and roll angle; the third upload frequency is greater than the first and second upload frequencies;

[0009] The data exchange module is used to forward data to the six Ethernet switch ports according to the Ethernet data exchange protocol;

[0010] The main control module is used to load and run the built-in detection and warning software upon receiving the start signal sent by the start-stop button module; and when the loading and running is successful, the preset system startup prompt audio is used as the corresponding first audio; and the corresponding first mode is set to a non-warning mode; and the first audio and the first mode form the corresponding first warning data; and the first warning data is sent to the sound and light alarm module through the data exchange module; the first mode includes a non-warning mode and a warning mode;

[0011] The main control module is also used to stop the running detection warning software when receiving the stop signal sent by the start-stop button module; and when the stop operation is successful, the preset system has stopped prompting audio is used as the corresponding first audio; and the corresponding first mode is set to a non-warning mode; and the first audio and the first mode form the corresponding first warning data; and the first warning data is sent to the sound and light alarm module through the data exchange module;

[0012] The main control module is also used to create three cache queues as the corresponding first, second and third cache queues when the detection and early warning software is running normally; and store the first scene point cloud received at that time into the corresponding first cache queue each time the first scene point cloud is received; and store the first scene video received at that time into the corresponding second cache queue each time the first scene video is received; and store the first motion parameter group received at that time into the corresponding third cache queue each time the first motion parameter group is received;

[0013] The main control module is also used to extract the latest first scene point cloud from the first cache queue as the corresponding current scene point cloud when the detection and warning software is running normally; and extract the first scene video whose storage time in the second cache queue is closest to the storage time of the current scene point cloud as the corresponding current scene video; and extract the first motion parameter group whose storage time in the third cache queue is closest to the storage time of the current scene point cloud as the corresponding current motion parameter group; and record the video period of the current scene video as the corresponding current video period; and extract the first motion parameter group whose storage time in the third cache queue meets the current video period and sort them in chronological order as the corresponding current period parameter group sequence; and perform obstacle detection based on the current scene point cloud, the current scene video, the current motion parameter group and the current period parameter group sequence to obtain the corresponding first detection result; and when the first detection result is that there is an obstacle, call the sound and light alarm module and the storage module to perform sound and light warning and warning log storage processing; the first detection result includes no obstacle and obstacle;

[0014] The sound and light alarm module includes a voice playback component and an alarm light component; the sound and light alarm module is used to extract the corresponding first audio and the first mode from the first alarm data received at that time when receiving the first alarm data sent by the main control module; and identify the first mode; if the first mode is a non-alarm mode, the first audio is played by the voice playback component and the alarm light component is turned off; if the first mode is an alarm mode, the first audio is played by the voice playback component and the alarm light component is turned on for light alarm;

[0015] The storage module is used to store multiple first warning logs;

[0016] The start-stop button module includes a start button and a stop button; the start-stop button module is used to send the start signal to the main control module when the start button is pressed; and send the stop signal to the main control module when the stop button is pressed.

[0017] Preferably, the main control module is specifically used for obtaining a corresponding first detection result when obstacle detection is performed based on the current scene point cloud, the current scene video, the current motion parameter group and the current time period parameter group sequence.

[0018] According to the first acceleration, the first speed and the first posture of the current motion parameter group, the emergency braking distance of the rail transit vehicle is estimated to obtain the corresponding first distance L; and according to the first position of the current motion parameter group, the vehicle height H and the vehicle width W of the rail transit vehicle and the preset high-precision map of the vehicle route of the rail transit vehicle, the three-dimensional space R that the rail transit vehicle will pass through in the future during the period of traveling the first distance L is confirmed; and according to the preset vehicle single-side lateral safety area width △w, the left and right widths of the three-dimensional space R are laterally expanded; the cross-sectional area of ​​the three-dimensional space R that has completed the lateral expansion is (W+2△w)×H, and the length is L;

[0019] And performing target detection according to the three-dimensional space R and the current scene point cloud to obtain a corresponding first detection target sequence; when the first detection target sequence is not empty, it is composed of one or more first detection targets; the first detection target includes a first target size and a first target ego-vehicle coordinate;

[0020] And according to the three-dimensional space R, the current scene video, the current motion parameter group and the current time period parameter group sequence, target detection is performed to obtain a corresponding second detection target sequence; when the second detection target sequence is not empty, it is composed of one or more second detection targets; the second detection target includes a second target size, a second target ego-vehicle coordinate and a second target predicted trajectory;

[0021] The first and second detection target sequences are identified; if the first and second detection target sequences are both empty, the corresponding first detection result is set to no obstacle; if the first and second detection target sequences are not both empty, the corresponding first detection result is set to have an obstacle.

[0022] Further, the main control module is specifically used for, when the target detection is performed according to the three-dimensional space R and the current scene point cloud to obtain a corresponding first detection target sequence,

[0023] Extracting the point cloud in the three-dimensional space R from the current scene point cloud as the corresponding first space point cloud;

[0024] and performing ground point filtering on the first spatial point cloud to obtain a corresponding second spatial point cloud;

[0025] and performing target detection on the second spatial point cloud to obtain a corresponding first target detection frame set; when the first target detection frame set is not empty, it is composed of one or more first target detection frames; the first target detection frame includes a first detection frame size and a first detection frame center point coordinate;

[0026] and identifying whether the first target detection frame set is empty;

[0027] If the first target detection frame set is empty, setting the corresponding first detection target sequence to empty;

[0028] If the first target detection frame set is not empty, the coordinates of the center points of each first detection frame are converted from the radar coordinate system to the vehicle coordinate system of the rail transit vehicle based on the preset radar internal and external parameters to obtain the corresponding first detection frame self-vehicle coordinates; and the first detection frame size and the first detection frame self-vehicle coordinates corresponding to each first target detection frame are used as the corresponding first target size and the first target self-vehicle coordinates to form a corresponding first detection target; and all the obtained first detection targets are sequentially sorted to form a corresponding first detection target sequence.

[0029] Further, the main control module is specifically used for obtaining a corresponding second detection target sequence by performing target detection according to the three-dimensional space R, the current scene video, the current motion parameter group and the current time period parameter group sequence.

[0030] The current scene video is subjected to frame-by-frame image extraction processing to obtain a corresponding first image sequence; the first image sequence is composed of a plurality of first images P i The first images P are arranged in chronological order, and the index i is an integer greater than 1; each of the first images P i Corresponds to a video time point;

[0031] and storing the parameter group in the parameter group sequence of the current period with each of the first images P i The first position and the first posture of the first motion parameter group closest to the video time point are extracted to form a corresponding first parameter group C i ; and extracting the first position and the first posture of the current motion parameter group to form a corresponding current parameter group C 0 ; and for each of the first parameter groups C i With the current parameter group C 0Perform position and posture differential calculation to obtain the corresponding differential parameter group △c i ;

[0032] And for each of the first images P i Performing target detection processing to obtain a corresponding second target detection frame set; the second target detection frame set includes multiple second target detection frames; the second target detection frame includes a second detection frame size and a second detection frame center point coordinate;

[0033] Based on the preset camera internal and external parameters, the coordinates of the center points of each second detection frame are transformed from the image coordinate system to the vehicle coordinate system of the rail transit vehicle to obtain the corresponding first transformation coordinates; then all the differential parameter groups △c i traverse; and during the traversal, the differential parameter group △c currently traversed i as the corresponding current differential parameter group, and the first image P corresponding to the current differential parameter group i as the corresponding current image, and based on the current differential parameter group, the first transformation coordinates corresponding to the current image are subjected to displacement and rotation processing to obtain the corresponding second transformation coordinates; and the center point coordinates of the second detection frame corresponding to the current image are reset to the corresponding second transformation coordinates; each of the first transformation coordinates is the first parameter group C corresponding to the first transformation coordinates. i The coordinates of the vehicle coordinate system at the corresponding time point; each of the second transformed coordinates and the current parameter group C 0 The coordinates of the vehicle coordinate system at the corresponding time point;

[0034] All the obtained second target detection frame sets are sorted in chronological order to form a corresponding second target detection frame set sequence; target association processing is performed according to the second target detection frame set sequence to obtain a corresponding plurality of first target association sequences; target motion trajectory prediction processing is performed based on each of the first target association sequences to obtain a corresponding first predicted trajectory; each of the first target association sequences is formed by one or more second target detection frames corresponding to the same detection target being sorted in chronological order; the first predicted trajectory includes a plurality of first trajectory points; each of the first trajectory points corresponds to a first point coordinate; the first point coordinate is the coordinate corresponding to the current parameter set C 0 The coordinates of the vehicle coordinate system at the corresponding time point;

[0035] and taking the second detection frame size and the second detection frame center point coordinates of the last second target detection frame in each first target association sequence as the corresponding second target size and second target ego-vehicle coordinates, and taking the first predicted trajectory corresponding to each first target association sequence as the corresponding second target predicted trajectory; and forming a corresponding second detection target by the second target size, the second target ego-vehicle coordinates and the second target predicted trajectory corresponding to each first target association sequence;

[0036] And delete the second detection targets whose second target self-vehicle coordinates or the second target predicted trajectory have no intersection with the three-dimensional space R; and identify whether the total number of the remaining second detection targets is zero, if so, set the corresponding second detection target sequence to empty, otherwise the corresponding second detection target sequence is composed of all the remaining second detection targets.

[0037] Preferably, the main control module is specifically used for, when the sound and light alarm module and the storage module are called to perform sound and light warning and warning log storage processing,

[0038] The preset obstacle-appearing-ahead warning audio is used as the corresponding first audio; the corresponding first mode is set as the warning mode; the first audio and the first mode form the corresponding first warning data; and the first warning data is sent to the sound and light alarm module through the data exchange module;

[0039] And use the current time as the corresponding first end time; and use the time of the first end time minus the preset first duration as the corresponding first start time; and the first start time and the first end time form a corresponding first time period; and extract the first scene point cloud whose storage time in the first cache queue meets the first time period to form a corresponding first backup point cloud set; and extract the first scene video whose storage time in the second cache queue meets the first time period to form a corresponding first backup video set; and extract the first motion parameter group whose storage time in the third cache queue meets the first time period to form a corresponding first backup parameter set; and form a corresponding first warning log by the first start time, the first end time, the first backup point cloud set, the first backup video set and the first backup parameter set; and store the first warning log in the storage module through the data exchange module.

[0040] The embodiment of the present invention provides a rail transit obstacle detection and warning system. As can be seen from the above content, the system includes: a laser radar, a camera, an inertial navigation module, a data exchange module, a main control module, an audio and video alarm module, a storage module and a start-stop button module; wherein the data exchange module is a data exchange device of the whole system (similar to a switch), the laser radar, the camera and the inertial navigation module are three types of perception components of the system, the audio and video alarm module is an active warning component of the system, the storage module is an early warning log storage component of the system, the main control module is a core component of the system, and the start-stop button module is a one-button start-stop component that controls the start / stop of the main control module; the main control module is used to actively identify whether there are obstacles in the front driving track area + the near-end area on both sides of the track according to the real-time perception data (scene point cloud, scene video, self-vehicle operation parameters) provided by the three types of perception components, and when it is confirmed that an obstacle has occurred, it will be warned in time through the audio and video alarm module and the early warning log will be stored through the storage module. On the one hand, the system of the embodiment of the present invention can provide rail transit vehicles with a technical means for actively detecting obstacles, thereby improving driving safety and reducing the probability of major collision accidents; on the other hand, based on the one-button start-stop function, it also reduces the operational complexity of drivers and passengers / system users and improves the convenience of using the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A module structure diagram of a rail transit obstacle detection and warning system provided by an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of a three-dimensional space R provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings. 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 creative work are within the scope of protection of the present invention.

[0044] The embodiment of the present invention provides a rail transit obstacle detection and warning system. Figure 1 As shown in the module structure diagram of a rail transit obstacle detection and warning system provided in an embodiment of the present invention, the system mainly includes: a laser radar 1, a camera 2, an inertial navigation module 3, a data exchange module 4, a main control module 5, an audio and video alarm module 6, a storage module 7 and a start-stop button module 8.

[0045] The connection relationship between the components / modules of the system of the embodiment of the present invention is as follows: the data exchange module 4 includes at least six Ethernet switching ports, which are respectively connected to the laser radar 1, the camera 2, the inertial navigation module 3, the main control module 5, the sound and light alarm module 6 and the storage module 7 through RJ45 interface cables; the main control module 5 is directly connected to the start-stop button module 8.

[0046] (I) Laser radar 1:

[0047] The laser radar 1 of the embodiment of the present invention is installed on the front of a rail transit vehicle.

[0048] The laser radar 1 is used to perform radar scanning on the front scene of the rail transit vehicle to obtain the corresponding first scene point cloud; and regularly send the latest first scene point cloud to the main control module 5 according to a preset first upload frequency.

[0049] Here, the first uploading frequency is a preset time frequency.

[0050] (ii) Camera 2:

[0051] The camera 2 in the embodiment of the present invention is installed on the front of a rail transit vehicle.

[0052] The camera 2 is used to capture the scene in front of the rail transit vehicle to obtain a corresponding first scene video; and to regularly send the latest captured first scene video to the main control module 5 at a preset second upload frequency.

[0053] Here, the video length of each first scene video is the preset first video length; the second upload frequency is a preset time frequency, and the second upload frequency is less than the first upload frequency; the first video length is a preset time length, and the first video length is greater than the time period corresponding to the first upload frequency, and the time period = 1 / first upload frequency.

[0054] (III) Inertial Navigation Module 3:

[0055] The inertial navigation module 3 of the embodiment of the present invention is an inertial measurement unit (I MU) or an inertial navigation system (I NS).

[0056] The inertial navigation module 3 is used to measure and calculate the acceleration, speed, position and posture of the rail transit vehicle in real time to obtain the corresponding first acceleration, first speed, first position and first posture to form a corresponding first motion parameter group; and regularly send the latest first motion parameter group to the main control module 5 according to a preset third upload frequency.

[0057] Here, the first position is the track travel distance of the rail transit vehicle after it departs from the starting station; the first posture includes the pitch angle, yaw angle and roll angle; the third upload frequency is a preset time frequency, and the third upload frequency is greater than the first and second upload frequencies, that is, the third upload frequency>first upload frequency>second upload frequency.

[0058] (IV) Data exchange module 4:

[0059] The data exchange module 4 of the embodiment of the present invention is a software / hardware module, device, equipment, terminal, server or system with switch function. The data exchange module 4 is used to forward data to six Ethernet switch ports according to the Ethernet data exchange protocol.

[0060] (V) Main control module 5:

[0061] The main control module 5 of the embodiment of the present invention is a software / hardware module, device, equipment, terminal, server or system.

[0062] The main control module 5 is used to load and run the built-in detection warning software when receiving the start signal sent by the start-stop button module 8; and when the loading and running is successful, the preset system startup prompt audio is used as the corresponding first audio; and the corresponding first mode is set to a non-warning mode; and the first audio and the first mode form the corresponding first warning data; and the first warning data is sent to the sound and light alarm module 6 through the data exchange module 4;

[0063] Among them, the first mode includes a non-warning mode and a warning mode.

[0064] Here, the system startup prompt audio is a pre-recorded voice audio, which is used to prompt the driver and passenger / system operator that the current system has completed startup.

[0065] The main control module 5 is also used to stop the running detection and warning software when receiving the stop signal sent by the start-stop button module 8; and when the operation is stopped successfully, the preset system stop prompt audio is used as the corresponding first audio; and the corresponding first mode is set to a non-warning mode; and the first audio and the first mode form the corresponding first warning data; and the first warning data is sent to the sound and light alarm module 6 through the data exchange module 4.

[0066] Here, the system stopped prompting audio is a pre-recorded voice audio, which is used to prompt the driver and passengers / system operators that the current system has stopped running.

[0067] The main control module 5 is also used to create three cache queues as corresponding first, second and third cache queues when the detection and early warning software is operating normally; and when each first scene point cloud is received, the first scene point cloud received that time is stored in the corresponding first cache queue; and when each first scene video is received, the first scene video received that time is stored in the corresponding second cache queue; and when each first motion parameter group is received, the first motion parameter group received that time is stored in the corresponding third cache queue.

[0068] The main control module 5 is also used to extract the latest first scene point cloud from the first cache queue as the corresponding current scene point cloud when the detection and warning software is operating normally; and extract the first scene video in the second cache queue whose storage time is closest to the storage time of the current scene point cloud as the corresponding current scene video; and extract the first motion parameter group in the third cache queue whose storage time is closest to the storage time of the current scene point cloud as the corresponding current motion parameter group; and record the video period of the current scene video as the corresponding current video period; and extract the first motion parameter group in the third cache queue whose storage time meets the current video period and sort it in chronological order as the corresponding current period parameter group sequence; and perform obstacle detection based on the current scene point cloud, the current scene video, the current motion parameter group and the current period parameter group sequence to obtain the corresponding first detection result; and when the first detection result is that there is an obstacle, call the sound and light alarm module 6 and the storage module 7 to perform sound and light warning and warning log storage processing;

[0069] The first detection result includes no obstacle and obstacle.

[0070] In a specific implementation of the embodiment of the present invention, the main control module 5 is specifically used to perform obstacle detection based on the current scene point cloud, the current scene video, the current motion parameter group and the current time period parameter group sequence to obtain a corresponding first detection result:

[0071] Step A1, according to the first acceleration, the first speed and the first posture of the current motion parameter group, the emergency braking distance of the rail transit vehicle is estimated to obtain the corresponding first distance L; and according to the first position of the current motion parameter group, the vehicle height H and the vehicle width W of the rail transit vehicle and the preset high-precision map of the vehicle route of the rail transit vehicle, the three-dimensional space R that the rail transit vehicle will pass through in the future during the period of traveling the first distance L is confirmed; and the left and right widths of the three-dimensional space R are laterally expanded according to the preset width △w of the vehicle's single-side lateral safety area;

[0072] Among them, the cross-sectional area of ​​the three-dimensional space R that completes the horizontal expansion is (W+2△w)×H, and the length is L;

[0073] Here, the driving path of each rail transit vehicle is fixed, and the vehicle route high-precision map is a high-precision map along the road corresponding to the current vehicle driving path. The real-time first position of the vehicle can be used to know the unique positioning of the vehicle on its own driving path, and then combined with the vehicle route high-precision map, the map coordinates or world coordinate system coordinates of the point can be obtained; in addition, the vehicle route high-precision map can also be used to know the track curvature, slope and other information of each point on the vehicle's driving path; under the premise of knowing the current vehicle position, future driving distance (first distance L), and future path shape (such as curvature and slope of each point), it is natural to confirm the three-dimensional space R that will be passed within the future driving period of the first distance L; Figure 2 Taking the schematic diagram of the three-dimensional space R provided in the embodiment of the present invention as an example, the shapes of the two three-dimensional spaces R can be understood. Figure 2 The upper middle part gives a three-dimensional space R with a curve, while the lower middle part gives a three-dimensional space R that is straight;

[0074] Step A2, performing target detection according to the three-dimensional space R and the current scene point cloud to obtain a corresponding first detection target sequence;

[0075] Wherein, when the first detection target sequence is not empty, it is composed of one or more first detection targets; the first detection target includes the first target size and the first target ego-vehicle coordinates;

[0076] Here, in another specific implementation of the embodiment of the present invention, the main control module 5 is specifically used to perform target detection according to the three-dimensional space R and the current scene point cloud to obtain a corresponding first detection target sequence:

[0077] Step A2-1, extracting the point cloud in the three-dimensional space R from the current scene point cloud as the corresponding first space point cloud;

[0078] Here, by filtering out the point cloud outside the three-dimensional space R, the cloud computing volume can be effectively reduced and the efficiency of point cloud target detection can be improved;

[0079] Step A2-2, filtering the first spatial point cloud by ground points to obtain a corresponding second spatial point cloud;

[0080] Here, by filtering out the ground points within the three-dimensional space R, the computing volume of the point cloud can be effectively reduced and the efficiency of point cloud target detection can be improved. The processing method for identifying the ground points in the point cloud can be implemented by a variety of ground point recognition methods. All ground point recognition methods are not enumerated here, and only one typical processing method is described. The typical processing method is: the lowest height coordinate in the point cloud in the three-dimensional space R, that is, the first space point cloud is identified, and the lowest height coordinate is used as the minimum value of the ground point height range, and the sum of the lowest height coordinate and a preset small height value is used as the maximum value of the ground point height range to construct a corresponding ground point height range [lowest height coordinate, lowest height coordinate + small height value], and all point cloud data in the first space point cloud whose height coordinates meet the ground point height range are deleted;

[0081] Step A2-3, performing target detection on the second spatial point cloud to obtain a corresponding first target detection frame set;

[0082] Wherein, when the first target detection frame set is not empty, it is composed of one or more first target detection frames; the first target detection frame includes a first detection frame size and a first detection frame center point coordinate;

[0083] Here, the embodiment of the present invention can perform target detection on the second space point cloud through a variety of conventional point cloud target detection methods; one of the point cloud target detection methods is to perform point cloud clustering on the second space point cloud based on a point cloud clustering algorithm (such as KD-Tree clustering algorithm, adaptive Euclidean clustering algorithm, etc.) to obtain a plurality of clustered point cloud clusters, and then based on a preset cluster point number threshold, delete the point cloud clusters whose total number of points in the cluster is less than the cluster point number threshold, if the number of remaining point cloud clusters is not 0, a corresponding first target detection frame can be constructed based on each remaining point cloud cluster, and the first detection frame size and the first detection frame center point coordinates of each first target detection frame are given; another point cloud target detection method is to use a pre-trained point cloud target detection model (such as Voxel Net model, SECOND model, YOLO model, etc.) to perform target detection on the second space point cloud;

[0084] Step A2-4, and identifying whether the first target detection frame set is empty;

[0085] Step A2-5: if the first target detection frame set is empty, set the corresponding first detection target sequence to empty;

[0086] Step A2-6, if the first target detection frame set is not empty, then based on the preset radar internal and external parameters, the coordinates of the center points of each first detection frame are converted from the radar coordinate system to the self-vehicle coordinate system of the rail transit vehicle to obtain the corresponding first detection frame self-vehicle coordinates; and the first detection frame size and the first detection frame self-vehicle coordinates corresponding to each first target detection frame are used as the corresponding first target size and the first target self-vehicle coordinates to form a corresponding first detection target; and all the obtained first detection targets are sequentially sorted to form a corresponding first detection target sequence;

[0087] Here, it is known from the previous text that the current scene point cloud is the latest scene point cloud, that is, the current scene point cloud can be regarded as the latest scene point cloud at the current time (set as t 0 ), then based on the radar internal and external parameters, the point cloud coordinates in the current scene point cloud (i.e., the coordinates of the center point of the first detection frame) are transformed from the radar coordinate system to the vehicle coordinate system of the rail transit vehicle, and the vehicle coordinate system coordinates (i.e., the first detection frame vehicle coordinates) can be regarded as a coordinate system corresponding to the current time t 0 The matched ego-vehicle coordinate system coordinates;

[0088] Step A3, performing target detection according to the three-dimensional space R, the current scene video, the current motion parameter group and the current time period parameter group sequence to obtain a corresponding second detection target sequence;

[0089] Wherein, when the second detection target sequence is not empty, it is composed of one or more second detection targets; the second detection target includes the second target size, the second target ego-vehicle coordinates, and the second target predicted trajectory;

[0090] Here, in another specific implementation of the embodiment of the present invention, the main control module 5 is specifically used to perform target detection according to the three-dimensional space R, the current scene video, the current motion parameter group and the current time period parameter group sequence to obtain a corresponding second detection target sequence:

[0091] Step A3-1, performing frame-by-frame image extraction processing on the current scene video to obtain a corresponding first image sequence;

[0092] The first image sequence consists of a plurality of first images P i Sorted in chronological order, index i is an integer greater than 1; each first image P i Corresponds to a video time point;

[0093] Step A3-2, and compare the storage time of the parameter group in the current time period parameter group sequence with each first image P i The first position and the first pose of the first motion parameter group closest to the video time point are extracted to form a corresponding first parameter group C i; and extract the first position and the first pose of the current motion parameter group to form a corresponding current parameter group C 0 ; and for each first parameter group C i With the current parameter group C 0 Perform position and posture differential calculation to obtain the corresponding differential parameter group △c i ;

[0094] For example, it is known that the first image sequence consists of four first images P i composition;

[0095] Current parameter group C 0 {First position s 0 , the first position 0} and the current time t 0 correspond;

[0096] 4 first images P i The corresponding time point t i With the current time t 0 The corresponding relations are t i＝1 =t 0 -3△t,t i＝2 =t 0 -2△t,t i＝3 =t 0 -△t,t i＝4 =t 0 ;

[0097] 4 time points t i The corresponding 4 first parameter groups C i :C 1 {s 1 ,p 1}、C 2 {s 2 ,p 2}、C 3 {s 3 ,p 3}、C 4 {s 4 =s 0 ,p 4 =p 0};

[0098] Then, the four differential parameter groups △c i =△c 1 {s 1 -s 0 ,p 1 -p 0}、△c 2 {s 2 -s 0 ,p 2 -p0}、△c 3 {s 3 -s 0 ,p 3 -p 0}、△c 1 {s 4 -s 0 =0,p 4 -p 0 =0};

[0099] Step A3-3, and for each first image P i Performing target detection processing to obtain a corresponding second target detection frame set;

[0100] The second target detection frame set includes a plurality of second target detection frames; the second target detection frame includes a second detection frame size and a second detection frame center point coordinate;

[0101] Here, the embodiment of the present invention uses a conventional visual target detection model (such as a YOLO model, an SSD model, a Ret i naNet model, etc.) to perform target detection on each first image Pi;

[0102] Step A3-4, based on the preset camera internal and external parameters, the coordinates of the center points of each second detection frame are transformed from the image coordinate system to the vehicle coordinate system of the rail transit vehicle to obtain the corresponding first transformation coordinates; then all differential parameter groups △c i Traverse; and during the traversal, the current traversal difference parameter group △c i As the corresponding current differential parameter group, and the first image P corresponding to the current differential parameter group i As the corresponding current image, and based on the current differential parameter group, performing displacement rotation processing on each first transformation coordinate corresponding to the current image to obtain the corresponding second transformation coordinate; and resetting the coordinates of the center points of each second detection frame corresponding to the current image to the corresponding second transformation coordinates;

[0103] Among them, each first transformation coordinate is the first parameter group C corresponding to i The coordinates of the vehicle coordinate system at the corresponding time point; each second conversion coordinate and the current parameter group C 0 The coordinates of the vehicle coordinate system at the corresponding time point;

[0104] Here, each first image P i After the coordinate transformation from the image coordinate system to the vehicle coordinate system of the rail transit vehicle based on the corresponding camera internal and external parameters is performed, a corresponding vehicle coordinate system coordinate (i.e., the first transformation coordinate) can be generated. However, this first transformation coordinate should be the coordinate of the first image P iThe corresponding time point t i The coordinates of the ego-vehicle coordinate system on the ego-vehicle coordinate system, rather than the current time t 0 The coordinates of the vehicle coordinate system on the t, so we also need to make a transformation from t i to 0 The vehicle coordinate system transformation;

[0105] This from t i to 0 The conversion of the ego-vehicle coordinate system is actually a process of translating and rotating the origin of the ego-vehicle coordinate system. This translation and rotation calculation process requires two sets of parameters, namely translation parameters and rotation parameters. The translation parameters mentioned here are actually the position point of the vehicle corresponding to the origin of the ego-vehicle coordinate system on the high-precision map from t i to 0 The differential displacement vector of the known differential parameter set △c i △s i-0 =s i -s 0 Then, it can be obtained by querying the high-precision map of the vehicle route; the rotation parameter mentioned here is actually the rotation of a certain position point of the vehicle corresponding to the origin of the vehicle coordinate system on the high-precision map from t i to 0 The differential pose vector is actually the differential parameter group △c i △p i-0 =p i -p 0 Therefore, for each first transformation coordinate, when its corresponding differential parameter set △c is known i In the case of i The first transformed coordinates are shifted and rotated to obtain the first transformed coordinates at the current time t 0 The mapping coordinates in the vehicle coordinate system are the second transformation coordinates;

[0106] Step A3-5, and sort all the obtained second target detection frame sets in chronological order to form a corresponding second target detection frame set sequence; and perform target association processing according to the second target detection frame set sequence to obtain a plurality of corresponding first target association sequences; and perform target motion trajectory prediction processing based on each first target association sequence to obtain a corresponding first predicted trajectory;

[0107] Among them, each first target association sequence is composed of one or more second target detection frames corresponding to the same detection target, sorted in chronological order; the first predicted trajectory includes multiple first trajectory points; each first trajectory point corresponds to a first point coordinate; the first point coordinate is the coordinate of the first point corresponding to the current parameter group C 0The coordinates of the vehicle coordinate system at the corresponding time point;

[0108] Here, the embodiment of the present invention uses a conventional target tracking / association algorithm to perform target association processing on the second target detection frame set sequence; such conventional target tracking / association algorithms are such as IOU-based greedy matching algorithm, nearest neighbor algorithm, probabilistic data association algorithm, joint probabilistic data association algorithm, Hungarian algorithm, KM algorithm, data association network algorithm, etc.; they will not be described one by one here;

[0109] After obtaining the target detection frame sequence of each detected target, i.e., the first target association sequence, the center point coordinates of each detection frame in the sequence can be used as the position coordinates of a historical motion point, and the corresponding speed, acceleration and motion direction can be calculated based on the point spacing and time difference of adjacent motion points, and then the motion trajectory of a future period, i.e., the first predicted trajectory, can be predicted based on the position coordinates, speed acceleration and motion direction of the last historical motion point;

[0110] It should be noted that the trajectory duration corresponding to the first predicted trajectory can be a preset specified duration, or can be dynamically associated with the motion parameters of the rail transit vehicle; when the trajectory duration corresponding to the first predicted trajectory is dynamically associated with the motion parameters of the rail transit vehicle, a customized association can be performed based on application requirements. For example, it is known that the first distance L is the emergency braking distance of the rail transit vehicle at the current moment, and a corresponding emergency braking duration can be calculated based on the first distance L. Then, the trajectory duration corresponding to the first predicted trajectory can be set based on the emergency braking duration;

[0111] Step A3-6, and use the second detection frame size and the second detection frame center point coordinates of the last second target detection frame in each first target association sequence as the corresponding second target size and second target ego-vehicle coordinates, and use the first predicted trajectory corresponding to each first target association sequence as the corresponding second target predicted trajectory; and form a corresponding second detection target by the second target size, second target ego-vehicle coordinates and second target predicted trajectory corresponding to each first target association sequence;

[0112] Step A3-7, and delete the second detection targets whose second target ego-vehicle coordinates or second target predicted trajectory have no intersection with the three-dimensional space R; and identify whether the total number of remaining second detection targets is zero, if so, set the corresponding second detection target sequence to empty, if not, the corresponding second detection target sequence is composed of all the remaining second detection targets;

[0113] Step A4, and identify the first and second detection target sequences; if the first and second detection target sequences are both empty, set the corresponding first detection result to be no obstacle; if the first and second detection target sequences are not both empty, set the corresponding first detection result to be with an obstacle.

[0114] In another specific implementation of the embodiment of the present invention, the main control module 5 is specifically used for calling the sound and light alarm module 6 and the storage module 7 to perform sound and light warning and warning log storage processing:

[0115] Step B1, taking the preset obstacle prompting audio in front as the corresponding first audio; setting the corresponding first mode as the warning mode; and forming the corresponding first warning data by the first audio and the first mode; and sending the first warning data to the sound and light alarm module 6 through the data exchange module 4;

[0116] Here, the obstacle ahead prompt audio is a pre-recorded voice audio, which is used to prompt the driver / system operator that an obstacle has appeared ahead;

[0117] Step B2, and use the current time as the corresponding first end time; and use the time of the first end time minus the preset first duration as the corresponding first start time; and the first start time and the first end time form a corresponding first time period; and extract the first scene point cloud in the first cache queue whose storage time meets the first time period to form a corresponding first backup point cloud set; and extract the first scene video in the second cache queue whose storage time meets the first time period to form a corresponding first backup video set; and extract the first motion parameter group in the third cache queue whose storage time meets the first time period to form a corresponding first backup parameter set; and form a corresponding first warning log by the first start time, the first end time, the first backup point cloud set, the first backup video set and the first backup parameter set; and store the first warning log in the storage module 7 through the data exchange module 4.

[0118] Here, the first duration is a preset time length, such as 15 minutes, half an hour, etc.

[0119] (six) Sound and light alarm module 6:

[0120] The sound and light alarm module 6 of the embodiment of the present invention includes a voice playing component and an alarm light component.

[0121] The sound and light alarm module 6 is used to extract the corresponding first audio and first mode from the first warning data received at that time when receiving the first warning data sent by the main control module 5; and identify the first mode; if the first mode is a non-warning mode, the first audio is played through the voice playback component and the alarm light component is in a closed state; if the first mode is a warning mode, the first audio is played through the voice playback component and the alarm light component is turned on for light warning.

[0122] (VII) Storage module 7:

[0123] The storage module 7 of the embodiment of the present invention is a storage medium (such as a data hard disk, a USB flash drive, etc.), a storage server or a storage system for providing a data storage function.

[0124] The storage module 7 is used to store multiple first warning logs.

[0125] (VIII) Start-stop button module 8:

[0126] The start-stop button module 8 of the embodiment of the present invention includes a start button and a stop button.

[0127] The start-stop button module 8 is used to send a start signal to the main control module 5 when the start button is pressed; and to send a stop signal to the main control module 5 when the stop button is pressed.

[0128] The start-stop button module 8 of the example of the present invention can provide a one-button start-stop function for the driver and passengers / system users.

[0129] It should be noted that the division of the various system components / modules of the above system can be regarded as a division of logical functions. In actual implementation, they can be fully or partially integrated into one physical entity, or they can be physically separated. And these system components / modules can all be implemented in the form of software calling through processing elements; they can also be all implemented in the form of hardware; some modules can also be implemented in the form of software calling through processing elements, and some modules can be implemented in the form of hardware. For example, the main control module can be a separately established processing element, or it can be integrated in a chip of the above system for implementation. In addition, it can also be stored in the memory of the above system in the form of program code, and called and executed by a processing element of the above system. The implementation of other system components / modules is similar. In addition, these modules can be fully or partially integrated together, or they can be implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. In the implementation process, the processing steps of each module of the aforementioned system can be completed by the hardware integrated logic circuit in the processor element or the instructions in the form of software.

[0130] For example, the system components / modules of the above system may be one or more integrated circuits configured to implement the functions of the corresponding components / modules, such as: one or more specific integrated circuits (Application Specific Integrated Circuit, ASIC), or, one or more digital signal processors (Digital Signal Processor, DSP), or, one or more Field Programmable Gate Arrays (Field Programmable Gate Array, FPGA), etc. For another example, when a certain system component / module of the above system is implemented in the form of a processing element scheduling program code, the processing element may be a general-purpose processor, such as a central processing unit (Central Processing Unit, CPU) or other processors that can call program codes. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (System-on-a-chip, SOC).

[0131] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the above system are generated. The above computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The above-mentioned computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the above-mentioned computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, Bluetooth, microwave, etc.) methods. The above-mentioned computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated. The above-mentioned available medium can be a magnetic medium (such as a floppy disk, a hard disk, a tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid state drive (SSD)), etc.

[0132] In summary, an embodiment of the present invention provides a rail transit obstacle detection and warning system; it can be seen from the above content that the system includes: a laser radar, a camera, an inertial navigation module, a data exchange module, a main control module, an audio and video alarm module, a storage module and a start-stop button module; wherein, the data exchange module is a data exchange device for the entire system (similar to a switch), the laser radar, the camera and the inertial navigation module are three types of perception components of the system, the audio and video alarm module is an active warning component of the system, the storage module is an early warning log storage component of the system, the main control module is a core component of the system, and the start-stop button module is a one-button start-stop component for controlling the start / stop of the main control module; the main control module is used to actively identify whether there are obstacles in the front driving track area + the proximal area on both sides of the track according to the real-time perception data (scene point cloud, scene video, self-vehicle operation parameters) provided by the three types of perception components, and when it is confirmed that an obstacle has occurred, it is promptly warned through the audio and video alarm module and the early warning log is stored through the storage module. On the one hand, the system of the embodiment of the present invention can provide rail transit vehicles with a technical means for actively detecting obstacles, thereby improving driving safety and reducing the probability of major collision accidents; on the other hand, based on the one-button start-stop function, it also reduces the operational complexity of drivers and passengers / system users and improves the convenience of using the system.

[0133] The professionals should further realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0134] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0135] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A rail transit obstacle detection and warning system, characterized in that: The time synchronization system includes: a laser radar, a camera, an inertial navigation module, a data exchange module, a main control module, an audio and visual alarm module, a storage module and a start-stop button module; The data exchange module includes at least six Ethernet exchange ports, which are respectively connected to the laser radar, the camera, the inertial navigation module, the main control module, the sound and light alarm module and the storage module through RJ45 interface cables; the main control module is directly connected to the start-stop button module; The laser radar is installed on the front of the rail transit vehicle; the laser radar is used to perform radar scanning on the front scene of the rail transit vehicle to obtain a corresponding first scene point cloud; and regularly send the latest first scene point cloud to the main control module according to a preset first upload frequency; The camera is installed at the front of the rail transit vehicle; the camera is used to capture a video of a scene in front of the rail transit vehicle to obtain a corresponding first scene video; and regularly sends the latest captured first scene video to the main control module at a preset second upload frequency; the video length of each first scene video is the preset first video length; the second upload frequency is less than the first upload frequency; the first video length is greater than the time period corresponding to the first upload frequency; The inertial navigation module is used to measure the acceleration, speed, position and posture of the rail transit vehicle in real time to calculate the corresponding first acceleration, first speed, first position and first posture to form a corresponding first motion parameter group; and regularly send the latest first motion parameter group to the main control module according to the preset third upload frequency; the first position is the track travel distance of the rail transit vehicle after it departs from the departure station; the first posture includes pitch angle, yaw angle and roll angle; the third upload frequency is greater than the first and second upload frequencies; The data exchange module is used to forward data to the six Ethernet switch ports according to the Ethernet data exchange protocol; The main control module is used to load and run the built-in detection and warning software upon receiving the start signal sent by the start-stop button module; and when the loading and running is successful, the preset system startup prompt audio is used as the corresponding first audio; and the corresponding first mode is set to a non-warning mode; and the first audio and the first mode form the corresponding first warning data; and the first warning data is sent to the sound and light alarm module through the data exchange module; the first mode includes a non-warning mode and a warning mode; The main control module is also used to stop the running detection warning software when receiving the stop signal sent by the start-stop button module; and when the stop operation is successful, the preset system has stopped prompting audio is used as the corresponding first audio; and the corresponding first mode is set to a non-warning mode; and the first audio and the first mode form the corresponding first warning data; and the first warning data is sent to the sound and light alarm module through the data exchange module; The main control module is also used to create three cache queues as the corresponding first, second and third cache queues when the detection and early warning software is running normally; and store the first scene point cloud received at that time into the corresponding first cache queue each time the first scene point cloud is received; and store the first scene video received at that time into the corresponding second cache queue each time the first scene video is received; and store the first motion parameter group received at that time into the corresponding third cache queue each time the first motion parameter group is received; The main control module is also used to extract the latest first scene point cloud from the first cache queue as the corresponding current scene point cloud when the detection and warning software is running normally; and extract the first scene video whose storage time in the second cache queue is closest to the storage time of the current scene point cloud as the corresponding current scene video; and extract the first motion parameter group whose storage time in the third cache queue is closest to the storage time of the current scene point cloud as the corresponding current motion parameter group; and record the video period of the current scene video as the corresponding current video period; and extract the first motion parameter group whose storage time in the third cache queue meets the current video period and sort them in chronological order as the corresponding current period parameter group sequence; and perform obstacle detection based on the current scene point cloud, the current scene video, the current motion parameter group and the current period parameter group sequence to obtain the corresponding first detection result; and when the first detection result is that there is an obstacle, call the sound and light alarm module and the storage module to perform sound and light warning and warning log storage processing; the first detection result includes no obstacle and obstacle; The sound and light alarm module includes a voice playback component and an alarm light component; the sound and light alarm module is used to extract the corresponding first audio and the first mode from the first alarm data received at that time when receiving the first alarm data sent by the main control module; and identify the first mode; if the first mode is a non-alarm mode, the first audio is played by the voice playback component and the alarm light component is turned off; if the first mode is an alarm mode, the first audio is played by the voice playback component and the alarm light component is turned on for light alarm; The storage module is used to store multiple first warning logs; The start-stop button module includes a start button and a stop button; the start-stop button module is used to send the start signal to the main control module when the start button is pressed; and send the stop signal to the main control module when the stop button is pressed.

2. The rail transit obstacle detection and warning system according to claim 1, characterized in that: The main control module is specifically used for obtaining a corresponding first detection result when performing obstacle detection based on the current scene point cloud, the current scene video, the current motion parameter group and the current time period parameter group sequence. According to the first acceleration, the first speed and the first posture of the current motion parameter group, the emergency braking distance of the rail transit vehicle is estimated to obtain the corresponding first distance L; and according to the first position of the current motion parameter group, the vehicle height H and the vehicle width W of the rail transit vehicle and the preset high-precision map of the vehicle route of the rail transit vehicle, the three-dimensional space R that the rail transit vehicle will pass through in the future during the period of traveling the first distance L is confirmed; and according to the preset vehicle single-side lateral safety area width △w, the left and right widths of the three-dimensional space R are laterally expanded; the cross-sectional area of ​​the three-dimensional space R that has completed the lateral expansion is (W+2△w)×H, and the length is L; And performing target detection according to the three-dimensional space R and the current scene point cloud to obtain a corresponding first detection target sequence; when the first detection target sequence is not empty, it is composed of one or more first detection targets; the first detection target includes a first target size and a first target ego-vehicle coordinate; And according to the three-dimensional space R, the current scene video, the current motion parameter group and the current time period parameter group sequence, target detection is performed to obtain a corresponding second detection target sequence; when the second detection target sequence is not empty, it is composed of one or more second detection targets; the second detection target includes a second target size, a second target ego-vehicle coordinate and a second target predicted trajectory; The first and second detection target sequences are identified; if the first and second detection target sequences are both empty, the corresponding first detection result is set to no obstacle; if the first and second detection target sequences are not both empty, the corresponding first detection result is set to have an obstacle.

3. The rail transit obstacle detection and warning system according to claim 2, characterized in that: The main control module is specifically used for obtaining a corresponding first detection target sequence by performing target detection according to the three-dimensional space R and the current scene point cloud. Extracting the point cloud in the three-dimensional space R from the current scene point cloud as the corresponding first space point cloud; and performing ground point filtering on the first spatial point cloud to obtain a corresponding second spatial point cloud; and performing target detection on the second spatial point cloud to obtain a corresponding first target detection frame set; when the first target detection frame set is not empty, it is composed of one or more first target detection frames; the first target detection frame includes a first detection frame size and a first detection frame center point coordinate; and identifying whether the first target detection frame set is empty; If the first target detection frame set is empty, setting the corresponding first detection target sequence to empty; If the first target detection frame set is not empty, the coordinates of the center points of each first detection frame are converted from the radar coordinate system to the vehicle coordinate system of the rail transit vehicle based on the preset radar internal and external parameters to obtain the corresponding first detection frame self-vehicle coordinates; and the first detection frame size and the first detection frame self-vehicle coordinates corresponding to each first target detection frame are used as the corresponding first target size and the first target self-vehicle coordinates to form a corresponding first detection target; and all the obtained first detection targets are sequentially sorted to form a corresponding first detection target sequence.

4. The rail transit obstacle detection and warning system according to claim 2, characterized in that: The main control module is specifically used for obtaining a corresponding second detection target sequence by performing target detection according to the three-dimensional space R, the current scene video, the current motion parameter group and the current time period parameter group sequence. The current scene video is subjected to frame-by-frame image extraction processing to obtain a corresponding first image sequence; the first image sequence is composed of a plurality of first images P i The first images P are arranged in chronological order, and the index i is an integer greater than 1; each of the first images P i Corresponds to a video time point; and storing the parameter group in the parameter group sequence of the current period with each of the first images P i The first position and the first posture of the first motion parameter group closest to the video time point are extracted to form a corresponding first parameter group C i ; and extract the first position and the first posture of the current motion parameter group to form a corresponding current parameter group C0; and for each of the first parameter groups C i The position and posture difference calculation is performed with the current parameter group C0 to obtain the corresponding differential parameter group △c i ; And for each of the first images P i Performing target detection processing to obtain a corresponding second target detection frame set; the second target detection frame set includes a plurality of second target detection frames; The second object detection frame includes a second detection frame size and a second detection frame center point coordinate; Based on the preset camera internal and external parameters, the coordinates of the center points of each second detection frame are transformed from the image coordinate system to the vehicle coordinate system of the rail transit vehicle to obtain the corresponding first transformation coordinates; then all the differential parameter groups △c i traverse; and during the traversal, the differential parameter group △c currently traversed i as the corresponding current differential parameter group, and the first image P corresponding to the current differential parameter group i as the corresponding current image, and based on the current differential parameter group, the first transformation coordinates corresponding to the current image are subjected to displacement and rotation processing to obtain the corresponding second transformation coordinates; and the center point coordinates of the second detection frame corresponding to the current image are reset to the corresponding second transformation coordinates; each of the first transformation coordinates is the first parameter group C corresponding to the first transformation coordinates. i The coordinates of the vehicle coordinate system at the corresponding time point; the coordinates of the vehicle coordinate system at the time point corresponding to each of the second transformed coordinates and the current parameter group C0; And sorting all the obtained second object detection frame sets in chronological order to form a corresponding second object detection frame set sequence; And target association processing is performed according to the second target detection frame set sequence to obtain a corresponding plurality of first target association sequences; and target motion trajectory prediction processing is performed based on each of the first target association sequences to obtain a corresponding first predicted trajectory; each of the first target association sequences is formed by one or more of the second target detection frames corresponding to the same detection target, sorted in chronological order; the first predicted trajectory includes a plurality of first trajectory points; each of the first trajectory points corresponds to a first point coordinate; the first point coordinate is the coordinate of the self-vehicle coordinate system at the time point corresponding to the current parameter group C0; and taking the second detection frame size and the second detection frame center point coordinates of the last second target detection frame in each first target association sequence as the corresponding second target size and second target ego-vehicle coordinates, and taking the first predicted trajectory corresponding to each first target association sequence as the corresponding second target predicted trajectory; and forming a corresponding second detection target by the second target size, the second target ego-vehicle coordinates and the second target predicted trajectory corresponding to each first target association sequence; And delete the second detection targets whose second target self-vehicle coordinates or the second target predicted trajectory have no intersection with the three-dimensional space R; and identify whether the total number of the remaining second detection targets is zero, if so, set the corresponding second detection target sequence to empty, otherwise the corresponding second detection target sequence is composed of all the remaining second detection targets.

5. The rail transit obstacle detection and warning system according to claim 1, characterized in that: The main control module is specifically used for calling the sound and light alarm module and the storage module to perform sound and light warning and warning log storage processing. The preset obstacle-appearing-ahead warning audio is used as the corresponding first audio; and the corresponding first mode is set as the warning mode; and the first audio and the first mode form the corresponding first warning data; and sending the first warning data to the sound and light alarm module through the data exchange module; and taking the current time as the corresponding first end time; and taking the time of the first end time minus the preset first duration as the corresponding first start time; and forming a corresponding first time period by the first start time and the first end time; and extracting the first scene point cloud whose storage time in the first cache queue meets the first time period to form a corresponding first backup point cloud set; and extracting the first scene videos whose storage time in the second cache queue meets the first time period to form a corresponding first backup video set; and extracting the first motion parameter group whose storage time in the third cache queue meets the first time period to form a corresponding first backup parameter set; The first start time, the first end time, the first backup point cloud set, the first backup video set and the first backup parameter set form a corresponding first warning log; and the first warning log is stored in the storage module through the data exchange module.

Images