Multi-sensor magnetic solar track layer boundary invasion detection system

Abstract

The invention discloses a multi-sensor magnetic solar track layer boundary invasion detection system, which comprises a magnetic installation module, a solar power supply module, a sensor module, a calculation module and a driver visual interface, and is characterized in that the sensor module comprises a laser radar module, an IMU module and a GPS module; and the calculation module comprises a cartographic slab algorithm module and a GICP (Generic Inductively Coupled Plasma) algorithm module. The device is installed on a track layer arm in a magnetic attraction mode, and power is supplied through solar energy. According to the system, a grid map is generated in real time according to a cartograph slab algorithm, a preset railway gauge diagram is converted into the generated grid map, point cloud data is subjected to height matching through a GICP algorithm, the posture and the map of a track layer are continuously updated, then the real-time position of the track layer and the preset railway gauge diagram are accurately analyzed through a ray guide method, and the real-time position of the track layer and the real-time position of the track layer are accurately analyzed. And judging whether the track layer is beyond the boundary or not in the operation process. According to the invention, the reliability, the high precision and the real-time performance of the rail layer invasion detection are improved.

Description

Technical Field

[0001] The invention belongs to the field of railway intelligent construction, and in particular relates to a multi-sensor magnetic solar track-laying machine intrusion detection system. Background Art

[0002] During the railway laying process, the accuracy and safety of the track laying machine are particularly important, and are the basis for ensuring the safety of railway construction and operation. The traditional track laying machine intrusion detection system relies on manual observation and single sensor operation, which has problems such as insufficient accuracy, insufficient safety, and low intrusion detection efficiency. In addition, the railway construction environment is complex, and there will be difficulties in powering the equipment and slow data processing speed. Therefore, the development of a high-precision, real-time, efficient, and all-weather track laying machine intrusion detection system is of great significance to improving the safety and reliability of railway construction. Summary of the invention

[0003] The objectives of the present invention are achieved through the following technical solutions.

[0004] In order to overcome the defects and shortcomings of the existing track paving machine boundary detection technology, the present invention provides a multi-sensor magnetic solar track paving machine boundary detection system. Aiming at the boundary detection problem of the track paving machine during operation, the SLAM algorithm is integrated to improve the reliability, safety and automation level of the boundary detection operation.

[0005] The technical solution of the present invention is: a multi-sensor magnetic solar track paving machine intrusion detection system, including a magnetic installation module, a solar power supply module, a sensor module, a computing module and a driver visualization interface, wherein the magnetic installation module is used to fix the intrusion detection system on the arm of the track paving machine; the solar power supply module is used to provide the electric energy required for the operation of the equipment; the sensor module is used to collect point cloud data around the track paving machine, provide the track paving machine's motion posture information including angular velocity and other information, and provide the track paving machine's real-time position information; the cartographer slam algorithm module is used to output a real-time grid map according to the data provided by the sensor; the GICP algorithm module is used to optimize point cloud matching and update the map; the driver interface is used to display the interface to display the grid map, intrusion risk alarm and system working status in real time.

[0006] Preferably, the magnetic mounting module selects mounting components with different adsorption forces according to the construction materials of different track laying machines and the working environment of the track laying machines. The sensor module includes a laser radar module, an IMU module, and a GPS module, which are independently integrated on the device to facilitate the maintenance and replacement of the module and improve the accuracy of the system operation.

[0007] Further preferably, the sensor module includes a laser radar module, an IMU module, and a GPS module. By combining the point cloud data provided by the laser radar with the working posture information of the track laying machine provided by the IMU, the positioning stability of the track laying machine in a dynamic and complex working environment is improved, and the overall stability of the system is improved. The calculation module includes a cartographer slam algorithm module, which uses the high-precision position information provided by the GPS to update the grid map, ensure the high accuracy of the overall positioning of the system, and is suitable for large-scale railway construction scenarios.

[0008] Further preferably, the system uses the ray method to determine whether an invasion has occurred, and on this basis adds a machine learning model to analyze multiple repeated detection results to avoid false positives and missed positives, thereby improving the reliability of the system operation.

[0009] Further preferably, the real-time high-precision grid map output based on the Cartographer SLAM algorithm is combined with the closed-loop detection function in the Cartographer SLAM algorithm to correct the accumulated error part in the grid map in real time to ensure that the position of the railway limit in the map is always accurate.

[0010] The driver's visual interface is further optimized. The display interface displays the grid map, boundary risk alarm and system working status in real time, and displays the above information in partitions to avoid information overload and show the panel content to the driver more clearly. At the same time, the visual interface supports multi-language switching to meet the needs of railway construction.

[0011] The advantages of the present invention are: The multi-sensor magnetic solar track paving machine intrusion detection system provided by the present invention integrates multiple sensors, including lidar, IMU and GPS, and can achieve centimeter-level precise positioning and map construction, ensuring high-precision detection of track paving machines.

[0012] The multi-sensor magnetic solar track paving machine intrusion detection system provided by the present invention uses magnetic installation accessories, the detection equipment is light in weight, and is easy to disassemble and move, thereby realizing the convenience of track paving machine intrusion detection.

[0013] The multi-sensor magnetic solar track-laying machine intrusion detection system provided by the present invention comprehensively utilizes solar power supply modules and lithium battery energy storage to achieve low energy consumption in detection and reduce system costs. At the same time, it can adapt to different complex construction environments and realize all-weather operation of track-laying machine intrusion detection.

[0014] The multi-sensor magnetic solar track paving machine intrusion detection system provided by the present invention uses the cartographer slam algorithm and the GICP matching algorithm to build a high-precision grid map in real time, accurately mark the location of obstacles, achieve high-precision point cloud matching, and optimize the alignment of map and sensor data. Combined with the use of the guide ray method to determine the positional relationship between the detection point and the railway limit diagram, digitally determine whether the track paving machine is intruding, and achieve system algorithm optimization and real-time and accurate detection.

[0015] The multi-sensor magnetic solar track paving machine intrusion detection system provided by the present invention has the following beneficial effects: 1. Use the GICP algorithm to highly match the point cloud data, continuously update the map, and then use the ray method to accurately analyze the real-time position of the track laying machine and the preset railway limit map to determine whether the track laying machine has violated the boundary during operation.

[0016] 2. Ensure data accuracy, remove interfering point cloud data, reduce the amount of point cloud data, continuously update maps, improve obstacle detection precision and accuracy, and reduce the missed detection rate of small obstacles; 3. Taking into account the visualization requirements of the driver's interface, the grid map, railway limit model, track laying machine working status and intrusion risk prompts generated by the system are displayed in real time on the driver's interface. When the system detects intrusion, it provides timely visual and auditory feedback, allowing the driver to respond in time to prevent danger. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings: Figure 1 This is the overall working flow chart of the multi-sensor magnetic solar track paving machine intrusion detection system of the present invention; Figure 2 A schematic diagram of a framework based on a cartographer slam algorithm combined with a GICP algorithm according to an embodiment of the present invention; Figure 3 A schematic diagram of a ray method according to an embodiment of the present invention; Figure 4 4 is a diagram showing a driver interface according to an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0019] like Figure 1 As shown, a multi-sensor magnetic solar track laying machine intrusion detection system provided by the present invention installs the intrusion detection device on the arm of the track laying machine by magnetic attraction and is powered by solar energy. The system generates a grid map in real time according to the cartographer slam algorithm, converts the preset railway limit map into the generated map, highly matches the point cloud data through the GICP algorithm, continuously updates the map, and then uses the ray method to accurately analyze the real-time position of the track laying machine and the preset railway limit map to determine whether the track laying machine intrudes during the operation. At the same time, the system will also display the grid map, the risk of intrusion alarm and the working status of the system in real time on the driver display interface, so that the driver can make a judgment during the work process. The acquisition process of the preset railway limit map is as follows: according to the outline of the building obstacles around the railway, the size and shape of the railway limit are defined, and converted into a railway limit model; according to the real-time position of the track laying machine, the position of the railway limit model in the generated grid map is determined, and the area covered by the railway limit model is mapped to the grid map generated by the system, and marked as the railway limit area.

[0020] Preferably, the device is magnetically mounted on the middle section of the track laying machine's arm, which can not only cover most of the track laying machine's construction area, but also adjust the device's viewing angle as the track laying machine works to ensure maximum coverage of the track laying machine's working area.

[0021] Preferably, high-efficiency photovoltaic solar panels are used for power supply, which has low energy consumption and supports all-weather intrusion detection. At the same time, built-in energy storage lithium batteries ensure that the system can perform detection work at night or on cloudy days.

[0022] Specifically, the specific implementation process of the multi-sensor magnetic solar track paving machine intrusion detection system described in the present invention includes the following contents.

[0023] 1. Cartographer slam algorithm module generates raster maps like Figure 2 As shown, the multi-sensor magnetic solar track paving machine intrusion detection system uses the cartographerslam algorithm to generate a real-time high-precision grid map.

[0024] First, input the sensor data, including the point cloud information of the track laying machine's surrounding environment obtained by the lidar module; the working posture of the track laying machine (such as angular velocity, acceleration, etc.) obtained by the IMU module; and the location information of the track laying machine obtained by the GPS module.

[0025] Preprocess the data: ① Time synchronization: Time alignment of all sensor data to ensure the time uniformity of multi-source data. ② Point cloud filtering: Remove points that exceed the maximum detection range of the lidar.

[0026] ③ Attitude correction: First obtain the acceleration and angular velocity data from the IMU sensor, and perform a preliminary attitude estimation of the track laying machine: use the accelerometer data to calculate the preliminary pitch angle (pitch) and roll angle (roll).

[0027] The calculation formula is as follows:

[0028] in, , , are the accelerations of the x-axis, y-axis, and z-axis respectively.

[0029] Then use the gyroscope data to get the angle change through integration and estimate the rotation of the device. The calculation formula is as follows:

[0030] Among them, θ(t) is the attitude angle at the current moment, θ(t-1) is the attitude angle at the previous moment, ω is the angular velocity measured by the gyroscope, and Δt is the time difference between the current moment and the previous moment.

[0031] After the initial estimation, the complementary filtering method is used to achieve a more accurate attitude estimation by fusing the output data of the accelerometer and gyroscope. The calculation formula is as follows:

[0032] in: θcomp is the current estimated angle, α is a weight factor (usually around 0.98) used to adjust the contribution ratio of the two sensor data. θprev is the angle estimated at the previous moment (from the integration result of the gyroscope), ω is the angular velocity read from the gyroscope, Δt is the time difference between the current moment and the previous moment, θaccel is the angle calculated from the accelerometer.

[0033] Each time new data is received, the device's attitude estimate is updated and the result is combined with the lidar data to provide the final corrected attitude for subsequent raster map generation.

[0034] Cartographer front-end processing: (1) Use IMU data for motion modeling and predict the position of the track laying machine in the current frame. The calculation formula is as follows:

[0035] Where V is the velocity at the previous moment, a is the acceleration at the current moment, P(t) is the position of the track laying machine at the current moment, P(t-1) is the position of the track laying machine at the previous moment, and △t is the time difference between the current moment and the previous moment.

[0036] (2) GICP point cloud matching Through iteration, the error between the source point cloud and the target point cloud is minimized, and the local point cloud Li currently generated in real time is matched with the previously generated point cloud C(i-1) to achieve accurate matching of the point clouds.

[0037]

[0038] in: is the i-th point in the source point cloud, is the i-th corresponding point in the target point cloud, R and t are rotation matrices and translation vectors, respectively, representing the rigid transformation from the source point cloud to the target point cloud. ∑i is the covariance matrix, which is used to measure the local geometric structure of each point. represents the weighted error metric function between the source point cloud and the target point cloud, N is the total number of points in the point cloud, indicating the number of matching point pairs between the source point cloud and the target point cloud. T is the matrix transpose operation, which represents the rotation and translation required to transform the source point cloud into the target point cloud.

[0039] (3) Local trajectory optimization In the process of point cloud matching, IMU data is used to optimize the matching results to reduce drift errors caused by device movement. The laser point cloud posture is adjusted by the least squares method: ①Define the error function f(R, t):

[0040] in: R is the rotation matrix (describing the orientation adjustment of the track laying machine), t is the translation vector (describing the position adjustment of the pose), p i is a point in the source point cloud, q i is a point in the target point cloud, n is the number of points in the source point cloud and the target point cloud, ||·|| represents the Euclidean distance.

[0041] ②Calculate the error gradient:

[0042] Calculate the gradient of the error function with respect to R and t in order to update the parameters.

[0043] ③Optimize parameters: Apply the least squares method, iteratively update R and t, reduce the error, and the update formula is as follows:

[0044] in: Δθ represents the update amount of the pose parameters, J represents the Jacobian matrix, which is usually in the form of: For each point error, T represents the transpose of the matrix, -1 represents the inverse matrix, δ is the error vector, which represents the difference between the current pose estimate and the target point cloud, and can be expressed as .

[0045] ④According to the calculated gradient, R and t are continuously adjusted until the maximum number of iterations is reached.

[0046] ⑤Finally, the optimized rotation matrix R and translation vector t are obtained, which are the adjusted point cloud pose.

[0047] (4) Subgraph construction Combined with the pre-processed sensor data, the cartographer slam algorithm divides the working area of ​​the track laying machine into several small sub-areas through spatial window division, and each sub-area corresponds to a submap. Among them, the local feature method can be used to divide the area, that is, to perform spatial division according to specific local features (such as track turns, etc.). As the track laying machine works in the area, the newly collected and processed point cloud data will be continuously added to the current sub-map. Each sub-map will be optimized through GICP point cloud matching to ensure the precise alignment of overlapping areas between different scan data, and the point cloud data in each sub-map will be locally optimized according to the real-time posture of the track laying machine to ensure the accuracy of the sub-map.

[0048] Closed loop detection uses the SIFT algorithm to extract feature points or key frames in each frame of data. These features are usually local geometric information, such as corner points, plane features, etc. Use FPFH descriptors to describe each feature point, and use Euclidean distance to calculate the similarity between the current frame and the historical frame to evaluate whether a closed loop is possible. The calculation formula is as follows:

[0049] in and is a descriptor, is the Euclidean distance.

[0050] When a potential closed loop is detected, further verification is performed using branch and bound.

[0051] ① Branching: Decompose the problem into multiple sub-problems (i.e. different path hypotheses), each sub-problem represents a different possible closed-loop position.

[0052] ② Bounding: Calculate the cost function of each path. Usually the cost function is the weighted sum of the matching error and the path constraint error. The calculation formula is as follows: Matching error formula:

[0053] ,in and are the feature points in the local map and the global map respectively, It is to convert local map feature points The position after transformation to the global coordinate system, are corresponding points in the global map.

[0054] Path constraint error formula: , Where X i and X j are two position nodes in the path (usually representing the position of the track laying machine), h(X i , X j ) is from X i To X j The expected pose difference (usually the pose difference predicted by the model), z ij is the actual observed pose difference, represents the squared error.

[0055] Cost function: , where λ1 and λ2 are weighted coefficients, E(T) is the matching error, and P(T) is the path constraint error.

[0056] ③ Pruning: If the cost function of a path is large, that is, the matching error is high, the path (branch) is discarded.

[0057] ④Optimal solution selection: Select the path with the smallest cost function as the matching position of the closed loop.

[0058] After selecting the best closed-loop match, Cartographer will build an optimized graph and minimize the constraint error, correct the map and trajectory through graph optimization, and then perform incremental updates to globally optimize all pose quantities along the entire pose journey. The corresponding map points on each pose are also corrected accordingly. This is global mapping. The objective function of graph optimization is:

[0059] in, and are two position nodes in the figure (usually representing the position of the track laying machine), is from arrive The expected pose difference of (usually the pose difference predicted by the model), is the actual observed pose difference, E represents the set of all points in the graph.

[0060] The incremental update formula is:

[0061] in, is the pose quantity calculated by graph optimization.

[0062] 2. Determine the occurrence of invasion by using the ray method According to the outline of the building obstacles around the railway, define the size and shape of the railway clearance (such as rectangle, polygon, etc.) and convert it into a digital model. Determine the position of the railway clearance model in the generated grid map according to the real-time position of the track laying machine. Map the area covered by the clearance model to the grid map generated by the system and mark it as "railway clearance area".

[0063] Ray Casting Algorithm is a common method used to determine whether a point is inside a polygon or to detect whether a path is blocked by an obstacle. The key is to count the number of intersections between the ray and the closed boundary. The main content of the Ray Casting Algorithm is that if there is an odd number of intersections, it means that the origin is inside the boundary, and if there is an even number of intersections, it means that the origin is outside the boundary.

[0064] like Figure 3As shown, the railway limit of the present invention is based on the building limit, and the red dot is the current position of the track laying machine, which is located inside the limit. For the detection points collected by the sensor, the ray method is used. Starting from the current working position, four rays are emitted from the front, left, right and rear of the track laying machine along the direction of the track laying machine, and the intersection of each ray and the building limit is calculated. And according to the railway limit model, the number of intersections of each ray and the building limit is calculated: (1) Single-sided intrusion: If the ray emitted from one side of the track laying machine has an even number of intersections with the building limit, it indicates that the detection origin on the track laying machine is outside the limit, that is, the direction where this ray is located may have intruded. If the number of intersections is an odd number, then the point is inside the limit (that is, no intrusion). (2) Bilateral intrusion: If the rays on both sides intersect with the building limit, and the number of intersections indicates that the track laying machine has crossed the track boundary in both directions, it is considered that bilateral intrusion has occurred. If none of the four rays intersect, the detection point is outside the boundary and is considered an intrusion.

[0065] 3. Driver display interface The driver display panel interface is visualized. (1) Real-time display: The relative position of the track laying machine and the railway limit and the grid map of the surrounding environment are displayed in real time on the driver's panel. (2) Alarm prompt: When the system detects an intrusion, the interface provides visual and auditory alarms to remind the driver to stop the track laying machine immediately. (3) System status monitoring: Displays the working status of the intrusion monitoring device, such as whether the solar panel has sufficient power and whether the system is working properly. (4) Track laying machine working status display: Displays the current position of the track laying machine, working status, and real-time movement trajectory of the machine arm.

[0066] like Figure 4 As shown, in order to realize driver visualization, the system of the present invention displays the generated grid map, railway clearance model, track laying machine working status and boundary intrusion risk prompts in real time on the driver interface. When the system finds that boundary intrusion occurs, it provides timely visual and auditory feedback, allowing the driver to respond in time to prevent danger. At the same time, an alarm is issued when the system operates abnormally to improve the reliability and safety of detection.

[0067] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A multi-sensor magnetic solar track paving machine intrusion detection system, characterized in that: include: Magnetic mounting module, solar power supply module, sensor module, computing module and driver visual interface module; among them, The magnetic mounting module is used to magnetically fix the intrusion detection system on the arm of the track laying machine; The solar power supply module is used to provide the electric energy required for the operation of the equipment of the system; The sensor module further includes a laser radar module, an IMU module, and a GPS module, wherein the laser radar module is used to collect point cloud data around the track laying machine, the IMU module is used to provide the motion posture of the track laying machine, and the GPS module is used to provide real-time position information of the track laying machine; The calculation module further includes a cartographer slam algorithm module and a GICP algorithm module; wherein the cartographer slam algorithm module is used to output a real-time grid map according to the data provided by the sensor module, and the GICP algorithm module is used to optimize point cloud matching and update the map; the calculation module determines whether there is an intrusion through the guide line method based on the updated map, the real-time grid map, and the preset railway clearance map; The driver visualization interface module is used to display the grid map, boundary intrusion risk alarm and system working status in real time on the display interface.

2. A multi-sensor magnetic solar track paving machine intrusion detection system according to claim 1, characterized in that: The Cartographer Slam algorithm module performs data input, data preprocessing, local mapping, closed-loop detection and global mapping based on the data from the sensor module, and outputs a real-time grid map.

3. A multi-sensor magnetic solar track paving machine intrusion detection system according to claim 2, characterized in that: The data preprocessing includes: time synchronization, point cloud filtering, and posture correction.

4. A multi-sensor magnetic solar track paving machine intrusion detection system according to claim 2, characterized in that: The local mapping includes: combining the pre-processed sensor data, dividing the working area of ​​the track laying machine into a plurality of small sub-areas through spatial window division, and each sub-area corresponds to a sub-map.

5. A multi-sensor magnetic solar track paving machine intrusion detection system according to claim 2, characterized in that: The closed loop detection includes: using the SIFT algorithm to extract feature points or key frames in each frame of data, using the FPFH descriptor to describe each feature point, and using the Euclidean distance to calculate the similarity between the current frame and the historical frame to evaluate whether a closed loop is possible.

6. A multi-sensor magnetic solar track paving machine intrusion detection system according to claim 5, characterized in that: When a potential closed loop is detected, further verification is performed using branch and bound methods, including: ① Branching: decompose the problem into multiple sub-problems, that is, different path assumptions, each sub-problem represents a different possible closed loop position; ② Bounding: Calculate the cost function of each path, which is the weighted sum of the matching error and the path constraint error; ③ Clipping: If the cost function of a path is greater than the preset range, the path is discarded; ④Optimal solution selection: Select the path with the smallest cost function as the matching position of the closed loop.

7. A multi-sensor magnetic solar track paving machine intrusion detection system according to claim 6, characterized in that: The global mapping includes: after selecting the optimal closed-loop match, constructing an optimized graph and minimizing the constraint error, correcting the map and trajectory through graph optimization, and then performing incremental updates, globally optimizing all posture quantities along the entire posture journey, and also correcting the corresponding map points on each posture accordingly.

8. The multi-sensor magnetic solar track paving machine intrusion detection system according to claim 1, characterized in that: The GICP algorithm module uses the GICP algorithm to match the local point cloud currently generated in real time with the previously generated point cloud, aligns the source point cloud to the target point cloud, and in the point cloud matching process, uses IMU data to optimize the matching results and adjusts the laser point cloud posture through the least squares method.

9. The multi-sensor magnetic solar track paving machine intrusion detection system according to claim 1, characterized in that: The process of obtaining the preset railway limit map is as follows: according to the outline of the building obstacles around the railway, the size and shape of the railway limit are defined and converted into a railway limit model; according to the real-time position of the track laying machine, the position of the railway limit model in the generated grid map is determined, and the area covered by the railway limit model is mapped to the grid map generated by the system and marked as the railway limit area.

10. A multi-sensor magnetic solar track paving machine intrusion detection system according to claim 9, characterized in that: The ray method includes: for the detection points collected by the sensor module, starting from the current working position, four rays are emitted from the front, left, right and rear of the track laying machine along the forward direction of the track laying machine, and the intersection of each ray and the railway limit is calculated; according to the railway limit model, the number of intersections is calculated, if the number of intersections between the emitted rays and the railway limit is an even number, it is considered that an invasion occurs, and if the number of intersections between the emitted rays and the railway limit is an odd number, it is considered that no invasion occurs.

Images