A road barrier recognition method, a path planning method, a device and equipment

Abstract

The application relates to the technical field of vehicle navigation, and discloses a road barrier identification method, a path planning method, a device and equipment, which comprises the following steps: acquiring millimeter wave radar detection point traces in real time, and processing the point traces to obtain static point traces; mapping the static point traces to a grid coordinate system to obtain corresponding grid confidence; screening effective grids according to a set confidence threshold; processing the effective grids to obtain continuous effective grids; updating the effective grid coordinates in real time according to vehicle body displacement; fitting the effective grid coordinates through a dynamic fitting model to obtain a barrier fitting curve; acquiring a road fitting curve; and weighting and fusing the barrier fitting curve and the road fitting curve to obtain a barrier curve. The application obtains the barrier fitting curve through the dynamic fitting model based on the static point traces, simultaneously fuses the road fitting curve to obtain the barrier curve, improves the barrier identification precision, and provides accurate boundary basis for danger early warning.

Description

Technical Field

[0001] This invention relates to the field of vehicle navigation technology, specifically to a road barrier recognition method, a route planning method, a device, and equipment. Background Technology

[0002] With the development of autonomous driving technology, barrier detection has become one of the core functional modules of millimeter-wave radar. As a key marker of road boundaries, the detection accuracy of barriers directly affects the path planning and driving safety of autonomous driving systems. Especially in high-speed scenarios, millimeter-wave radar needs to identify effective barriers at long distances to allow sufficient calculation time for the decision-making system and avoid accidents caused by reaction delays.

[0003] Existing barrier detection solutions mainly employ visual recognition and traditional millimeter-wave radar. Visual recognition relies on cameras to capture images and algorithms to identify barriers, but it is highly susceptible to environmental factors. In adverse conditions such as heavy fog, heavy rain, and nighttime, image contrast decreases, leading to a sharp drop in barrier recognition accuracy, which cannot meet the all-weather requirements of autonomous driving. Traditional millimeter-wave radar solutions construct barrier models directly from raw points output by millimeter-wave radar through filtering, or perform fitting based on raw points. However, fitting based on raw points easily produces false barriers, which are misjudged by the system as real road boundaries, causing path planning deviations or interfering with the identification of dangerous targets, increasing safety risks. Furthermore, the large number of raw points results in high algorithm complexity, high memory consumption, and significant time consumption. In addition, the fitted model does not incorporate the actual road conditions, leading to a large deviation between the fitted barrier model and the actual road, failing to accurately reflect the road direction, and causing a disconnect between perception and decision-making in the autonomous driving system. Summary of the Invention

[0004] To address the problems of high complexity, large memory consumption, and significant time consumption in existing millimeter-wave radar for detecting guardrails, as well as the tendency to generate false guardrails, leading to large identification errors and inaccurate reflection of road direction, this invention provides a road guardrail identification method, path planning method, device, and equipment.

[0005] According to an embodiment of the present invention, a method for identifying road railings is provided, comprising the following steps: The points detected by millimeter-wave radar are acquired in real time and processed to obtain stationary points. The stationary point traces are mapped to a grid coordinate system to obtain the corresponding grid confidence level; Valid grids are filtered based on the set confidence threshold; The effective mesh is processed to obtain a continuous effective mesh; The effective grid coordinates are updated in real time based on the vehicle body displacement; The effective grid coordinates are fitted using a dynamic fitting model to obtain the railing fitting curve; Obtain the road fitting curve, and then perform a weighted fusion of the railing fitting curve and the road fitting curve to obtain the railing curve.

[0006] In some optional implementations, obtaining grid confidence specifically includes: Construct a 3D grid coordinate system around the vehicle body and set the grid parameters; Traverse all stationary points and map their coordinates onto the grid to obtain the corresponding grid confidence.

[0007] In some optional implementations, valid grids are filtered based on a set confidence threshold, specifically including: Set the confidence threshold; The grid confidence level is compared with the confidence threshold. Grids with a confidence level greater than the confidence threshold are considered valid grids.

[0008] In some optional implementations, the effective mesh is processed to obtain a continuous effective mesh, specifically including: Calculate the effective grid coordinates, and sort the effective grids according to the effective grid coordinates; The sorted valid grids are then eliminated by applying an Euclidean distance threshold to obtain a continuous valid grid.

[0009] In some optional implementations, the acquisition of the railing fitting curve specifically includes: The quadratic curve model was chosen as the dynamic fitting model. The effective grid coordinates are fitted using a quadratic curve model to obtain the railing fitting curve.

[0010] In some optional implementations, the road fitting curve is obtained, and the railing fitting curve is weighted and fused with the road fitting curve to obtain the railing curve; Based on the vehicle's trajectory, a curve model of the road centerline is fitted to obtain the road fitting curve. Weight values ​​are set for the railing fitting curve and the road fitting curve respectively, so as to perform weighted fusion of the railing fitting curve and the road fitting curve to obtain the railing curve.

[0011] In some optional implementations, the points detected by millimeter-wave radar are acquired and processed to obtain stationary points, specifically including: Real-time acquisition of all raw data points detected by millimeter-wave radar; The target clustering algorithm is used to distinguish all original point traces into moving point traces and stationary point traces; The identified moving points are filtered to obtain stationary points.

[0012] According to another objective of embodiments of the present invention, a path planning method is provided, comprising the following steps: The guardrail curves obtained by any of the road guardrail recognition methods described above are used as road boundaries for vehicle travel path planning.

[0013] According to another objective of the present invention, a road barrier recognition device is provided, comprising: The data acquisition and preprocessing module is used to acquire the point traces detected by the millimeter-wave radar and process the acquired point traces to obtain stationary point traces. A grid confidence construction module is used to map the obtained stationary point traces to a grid coordinate system to obtain the corresponding grid confidence. The effective grid filtering module is used to filter out effective grids according to a set confidence threshold, and process the effective grids to obtain continuous effective grids; The railing dynamic fitting module is used to fit the effective grid coordinates through a dynamic fitting model to obtain the railing fitting curve; The road fusion output module is used to obtain the road fitting curve and perform weighted fusion of the railing fitting curve and the road fitting curve to obtain the railing curve.

[0014] According to another objective of the present invention, a computer device is provided, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus; The memory is used to store at least one executable instruction that causes the processor to perform the operation of a road barrier recognition method as described in any of the preceding claims.

[0015] Compared with the prior art, the present invention has the following advantages: This invention provides a road railing identification method. It processes points detected by millimeter-wave radar, retaining only stationary points to reduce the likelihood of false railings at the data source. A constructed grid combined with a confidence threshold is used to eliminate false grids, completely preventing interfering points from participating in the fitting process. Continuity verification of the effective grid further eliminates abnormal grids, retaining only continuous, high-density effective points, thus significantly reducing data dimensionality. A dynamic fitting model is used to obtain the railing fitting curve, reducing computation time and improving computational efficiency. Simultaneously, the road fitting curve is fused to obtain the railing curve, improving railing identification accuracy and providing precise boundary information for hazard warning.

[0016] The above description is merely an overview of the technical solutions of the embodiments of the present invention. In order to better understand the technical means of the embodiments of the present invention and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0017] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 The diagram shows a flowchart of a road railing recognition method provided by an embodiment of the present invention.

[0018] Figure 2 The diagram shows a structural block diagram of a road railing recognition device provided in an embodiment of the present invention.

[0019] Figure 3 A structural block diagram of a computer device provided by an embodiment of the present invention is shown. Detailed Implementation

[0020] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0021] To address the problems of existing millimeter-wave radar detection of guardrails, such as high algorithm complexity, large memory consumption, and significant time consumption, as well as the tendency to generate false guardrails, leading to large identification deviations and inaccurate reflection of road direction, this invention provides a road guardrail identification method.

[0022] This embodiment provides a method for identifying road railings, such as... Figure 1 As shown, it includes the following steps: S10. Acquire the points detected by millimeter-wave radar in real time and process them to obtain stationary points.

[0023] In this step, the points detected by the millimeter-wave radar are filtered to initially remove interfering points.

[0024] This step can initially eliminate most of the interfering points, reducing the amount of data for subsequent processing and laying the foundation for reducing complexity; at the same time, it avoids moving points being misjudged as railing points, reducing the possibility of false railings from the source.

[0025] S20. Map the stationary point traces to the grid coordinate system to obtain the corresponding grid confidence.

[0026] In this step, a grid coordinate system is constructed, and the stationary target is quantitatively filtered by grid confidence to accurately distinguish between valid railing marks and interference marks.

[0027] S30. Filter out valid grids based on the set confidence threshold.

[0028] In this step, valid grids are selected by comparing the set confidence threshold with the grid confidence obtained in step S30.

[0029] S40. Process the effective mesh to obtain a continuous effective mesh.

[0030] In this step, the effective grid is further filtered, and false grids are further eliminated based on the continuity check of effective points, thereby reducing the false detection rate of false railings.

[0031] S50. Update the effective grid coordinates in real time based on the vehicle body displacement.

[0032] In this step, given that the vehicle's position changes dynamically during travel, the absolute coordinates of the effective grid are updated based on the vehicle's displacement to ensure that the guardrail fitting is based on the real-time position.

[0033] S60. Fit the effective grid coordinates using a dynamic fitting model to obtain the railing fitting curve.

[0034] In this step, considering that the railings are mostly gentle curves or straight lines, a dynamic fitting model is used to fit the effective grid coordinates to obtain the railing fitting curve.

[0035] Compared to traditional straight-line models or high-order curve models, the dynamic fitting model used in this embodiment balances accuracy and complexity, with smaller fitting errors. It avoids the fitting deviation of traditional straight-line models for curved guardrails, and the dynamic update ensures that the fitted guardrail is always synchronized with the actual road when the vehicle moves, avoiding the accumulation of deviations caused by static fitting.

[0036] S70. Obtain the road fitting curve, and perform a weighted fusion of the railing fitting curve and the road fitting curve to obtain the railing curve.

[0037] In this step, a road fitting curve is obtained based on the vehicle's driving trajectory. Weights are set for the guardrail fitting curve and the road fitting curve, and the two curves are then weighted and fused to obtain the guardrail curve. This process correlates the guardrail fitting with the road trajectory, ensuring the fitting result closely matches the actual road and improving the guardrail fitting accuracy.

[0038] This embodiment provides a road railing identification method. It processes the points detected by millimeter-wave radar, retaining only stationary points to reduce the likelihood of false railings at the data source. False grids are eliminated by constructing a grid and using a confidence threshold, completely preventing interfering points from participating in the fitting process. Continuity verification is performed on the effective grid, further eliminating abnormal grids and retaining only continuous, high-density effective points, thus achieving significant data dimensionality reduction. A dynamic fitting model is used to obtain the railing fitting curve, reducing computation time and improving computational efficiency. Simultaneously, the road fitting curve is fused to obtain the railing curve, improving railing identification accuracy and providing precise boundary information for hazard warning.

[0039] This embodiment is a preferred embodiment, and the process of obtaining stationary points in step S10 has been optimized.

[0040] In this embodiment, step S10, acquiring the point traces detected by the millimeter-wave radar and processing them to obtain stationary point traces, specifically includes: S101. Real-time acquisition of all raw data points detected by millimeter-wave radar.

[0041] In this step, the original data points include moving targets such as other vehicles and stationary targets such as railings and utility poles.

[0042] S102. Use the target clustering algorithm to distinguish all original dots into moving dots and stationary dots.

[0043] In this step, a target clustering algorithm is used to distinguish between moving and stationary points. Since the railing is a static target, moving points need to be filtered out, and only stationary points are retained.

[0044] S103. Filter the identified moving points to obtain stationary points.

[0045] In this step, the moving points identified in step S102 are filtered out, and only the stationary points are retained.

[0046] This embodiment is a preferred embodiment, and the process of obtaining grid confidence in step S20 has been optimized.

[0047] In this embodiment, step S20, mapping the stationary point trace to a grid coordinate system to obtain the corresponding grid confidence score, specifically includes: S201. Construct a 3D grid coordinate system around the vehicle body and set the grid parameters.

[0048] In this step: with the vehicle body as the origin, the x-axis range is set to -5m~5m, and the y-axis range is set to -60m~60m, a two-dimensional grid coordinate system is constructed around the vehicle body; Simultaneously define the mesh parameters and set the mesh step size: xStep=1m, yStep=3m.

[0049] In this step, the step size and the range of the grid coordinate system can be adjusted according to the actual scenario.

[0050] S202. Traverse all stationary points and map their coordinates to the grid to obtain the corresponding grid confidence.

[0051] In this step, the coordinates of all stationary points are mapped to a grid. If a stationary point falls within a grid, the confidence level of that grid is incremented by 1.

[0052] This embodiment is a preferred embodiment, and the process of selecting effective grids in step S30 has been optimized.

[0053] In this embodiment, step S30, filtering out valid grids based on a set confidence threshold, specifically includes: S301. Set the confidence threshold.

[0054] In this step, since the density of traces decreases at long distances, the threshold needs to be adjusted accordingly to avoid missed detections. Therefore, as the distance along the y-axis increases, the confidence threshold decreases, thus achieving dynamic adjustment of the confidence threshold.

[0055] S302. Compare the grid confidence level with the confidence threshold.

[0056] S303. The grids corresponding to the grid confidence scores that are greater than the confidence threshold are determined to be valid grids.

[0057] In this step, grids with a confidence level greater than the confidence threshold are identified as valid grids, while others are identified as invalid grids.

[0058] This embodiment is a preferred embodiment, and the process of selecting continuous effective grids in step S40 has been optimized.

[0059] In this embodiment, step S40, processing the effective mesh to obtain a continuous effective mesh, specifically includes: S401. Calculate the effective grid coordinates and sort the effective grids according to the effective grid coordinates.

[0060] In this step, given that each valid grid contains multiple stationary points, the average x and y coordinates of all stationary points within each valid grid are obtained and used as the coordinates of that valid grid.

[0061] S402. The sorted effective grids are eliminated by using an Euclidean distance threshold to obtain continuous effective grids.

[0062] In this step, the valid grids are sorted according to the y-axis coordinate, and abrupt grids are removed by using the Euclidean distance threshold. That is, if the Euclidean distance between adjacent valid grids is a preset distance, they are determined to be discontinuous grids and are removed, thus obtaining continuous valid grids.

[0063] This embodiment is a preferred embodiment, and the process of obtaining the railing fitting curve in step S60 has been optimized.

[0064] In this embodiment, step S60, fitting the effective grid coordinates using a dynamic fitting model to obtain the railing fitting curve, specifically includes: S601. Select the quadratic curve model as the dynamic fitting model.

[0065] In this step, given that the railings are mostly gentle curves or straight lines, a quadratic curve model is selected.

[0066] The model of this conic section is: y1 = a1x 2 +c1; Where a1 and c1 are constants; When a1≈0, the model degenerates into a straight line, which is suitable for straight guardrails on municipal roads; When a1≠0, the model is a gentle curve, which is suitable for curved guardrails on highways.

[0067] S602. Fit the effective grid coordinates using a quadratic curve model to obtain the railing fitting curve.

[0068] The effective grid coordinates are fitted using the quadratic curve model selected in step S601 to obtain the railing fitting curve.

[0069] This embodiment is a preferred embodiment, and the process of obtaining the railing curve in step S70 has been optimized.

[0070] In this embodiment, step S70, obtaining the road fitting curve and weightedly fusing the guardrail fitting curve with the road fitting curve to obtain the guardrail curve, specifically includes: S701. Based on the vehicle's driving trajectory, fit a curve model of the road centerline to obtain the road fitting curve.

[0071] In this step, the road fitting curve is: y2=a2x 2 +c2; Where a2 and c2 are constants.

[0072] S702. Set the weight values ​​for the railing fitting curve and the road fitting curve respectively, so as to perform weighted fusion of the railing fitting curve and the road fitting curve to obtain the railing curve.

[0073] In this step, the obtained railing curve is: y final =A*(a1x 2 +c1)+B*(a2x 2 +c2); Where A represents the weight of the railing fitting curve and B represents the weight of the road fitting curve.

[0074] In this embodiment, the weight values ​​of the railing fitting curve and the road fitting curve can be adjusted according to the scenario.

[0075] In some optional embodiments, the present invention also discloses a path planning method.

[0076] This embodiment discloses a path planning method, which includes the following steps: The guardrail curves obtained through the above-described embodiment of the road guardrail recognition method are used as road boundaries for planning vehicle travel paths.

[0077] In this embodiment, using the railing curve as a hard boundary constraint for path planning can solve the route deviation problem caused by fuzzy road boundaries in traditional path planning. In addition, in this embodiment, the autonomous driving system fuses the guardrail curves with the lane line detection results to determine the effective driving area of ​​the lane, and calculates the lane centerline based on the perpendicular bisectors of the guardrail curves on both sides, thereby controlling the vehicle to always drive along the centerline.

[0078] In high-speed scenarios, millimeter-wave radar can detect guardrails at long distances and output the guardrail curves. The autonomous driving system can plan a smooth, centered driving route in advance to avoid route deviation caused by lane lines being obscured.

[0079] In addition, the railing curve can accurately reflect the road curvature. The path planning module can calculate the minimum turning radius and the speed on the slope based on the curvature change of the railing curve, thereby avoiding excessive centrifugal force or slippage on the slope when turning.

[0080] In some optional embodiments, the present invention also discloses a road railing recognition device.

[0081] This embodiment discloses a road railing recognition device, such as Figure 2 As shown, it includes: a data acquisition and preprocessing module 100, a grid confidence construction module 200, an effective grid screening module 300, a guardrail dynamic fitting module 400, and a road fusion output module 500.

[0082] The data acquisition and preprocessing module 100 is used to acquire the point traces detected by the millimeter-wave radar and process the acquired point traces to obtain stationary point traces.

[0083] In this embodiment, the specific implementation process of the data acquisition and preprocessing module 100 acquiring stationary point traces is described in the relevant steps of the above-mentioned embodiment of a road railing recognition method, and this embodiment does not repeat the limitation.

[0084] The grid confidence construction module 200 is used to map the obtained stationary point traces to the grid coordinate system to obtain the corresponding grid confidence.

[0085] In this embodiment, the specific implementation process of the grid confidence construction module 200 obtaining grid confidence is described in the relevant steps of the above-mentioned embodiment of a road railing recognition method, and this embodiment does not repeat the limitation.

[0086] The effective grid filtering module 300 is used to filter out effective grids according to the set confidence threshold, and process the effective grids to obtain continuous effective grids.

[0087] In this embodiment, the specific implementation process of the effective grid filtering module 300 obtaining continuous effective grids is described in the relevant steps of the above-mentioned embodiment of a road railing recognition method, and this embodiment does not repeat the limitation.

[0088] The railing dynamic fitting module 400 is used to fit the effective grid coordinates through a dynamic fitting model to obtain the railing fitting curve.

[0089] In this embodiment, the specific implementation process of the railing dynamic fitting module 400 acquiring the railing fitting curve is described in the relevant steps of the above-mentioned embodiment of a road railing recognition method, and this embodiment does not repeat the limitation.

[0090] The road fusion output module 500 is used to acquire the road fitting curve and perform weighted fusion of the railing fitting curve and the road fitting curve to obtain the railing curve.

[0091] In this embodiment, the specific implementation process of the road fusion output module 500 acquiring the guardrail curve is described in the relevant steps of the above-mentioned embodiment of a road guardrail recognition method, and this embodiment does not repeat the limitation.

[0092] In some alternative embodiments, the present invention also discloses a computer device.

[0093] Figure 3 A schematic diagram of the structure of a computer device provided in an embodiment of the present invention is shown.

[0094] like Figure 3 As shown, the computer device includes a processor 610, a memory 620, a communication interface 630, and a communication bus 640. The processor 610, the memory 620, and the communication interface 630 communicate with each other through the communication bus 640.

[0095] In this embodiment, the memory 620 stores at least one executable instruction, which causes the processor 620 to perform the operation of a road barrier recognition method as described in any of the above embodiments; the communication interface 630 is used to communicate with other devices, such as clients or other server network elements. The processor 610 is used to execute the program 650, which can specifically execute the relevant steps in the above embodiments of the road barrier recognition method.

[0096] Specifically, program 650 may include program code, which includes computer-executable instructions.

[0097] The processor 610 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The computer device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0098] Memory 620 is used to store program 650. Memory 620 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0099] Specifically, program 650 can be called by processor 610 to cause the computer device to perform the operation of a road barrier recognition method as described in any of the above embodiments.

[0100] In some optional embodiments, the present invention also discloses a computer-readable storage medium storing at least one executable instruction that, when executed on a computer device, causes the computer device to perform the steps of a road barrier recognition method in any of the above method embodiments.

[0101] The specific implementation process of the road railing recognition method described in this embodiment can be found in any of the above method embodiments, and will not be repeated in this embodiment.

[0102] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. Similarly, for the sake of brevity and to aid in understanding one or more aspects of the invention, in the description of exemplary embodiments of the invention above, various features of the embodiments are sometimes grouped together in a single embodiment, figure, or description thereof. The claims, which follow the detailed description, are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of the invention.

[0103] Those skilled in the art will understand that the modules in the device of the embodiment can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiment can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components, except that at least some of such features and / or processes or units are mutually exclusive.

[0104] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

Claims

1. A method for identifying road railings, characterized in that, Includes the following steps: The points detected by millimeter-wave radar are acquired in real time and processed to obtain stationary points. The stationary point traces are mapped to a grid coordinate system to obtain the corresponding grid confidence level; Valid grids are filtered based on the set confidence threshold; The effective mesh is processed to obtain a continuous effective mesh; The effective grid coordinates are updated in real time based on the vehicle body displacement; The effective grid coordinates are fitted using a dynamic fitting model to obtain the railing fitting curve; Obtain the road fitting curve, and then perform a weighted fusion of the railing fitting curve and the road fitting curve to obtain the railing curve.

2. The method for identifying road railings according to claim 1, characterized in that, The acquisition of grid confidence specifically includes: Construct a 3D grid coordinate system around the vehicle body and set the grid parameters; Traverse all stationary points and map their coordinates onto the grid to obtain the corresponding grid confidence.

3. The method for identifying road railings according to claim 1, characterized in that, Valid grids are filtered based on the set confidence threshold, specifically including: Set the confidence threshold; The grid confidence level is compared with the confidence threshold. Grids with a confidence level greater than the confidence threshold are considered valid grids.

4. The method for identifying road railings according to claim 1, characterized in that, The effective mesh is processed to obtain a continuous effective mesh, specifically including: Calculate the effective grid coordinates, and sort the effective grids according to the effective grid coordinates; The sorted valid grids are then eliminated by applying an Euclidean distance threshold to obtain a continuous valid grid.

5. The method for identifying road railings according to claim 1, characterized in that, Obtaining the railing fitting curve specifically includes: The quadratic curve model was chosen as the dynamic fitting model. The effective grid coordinates are fitted using a quadratic curve model to obtain the railing fitting curve.

6. The method for identifying road railings according to claim 1, characterized in that, Obtain the road fitting curve, and then weight and fuse the railing fitting curve with the road fitting curve to obtain the railing curve; Based on the vehicle's trajectory, a curve model of the road centerline is fitted to obtain the road fitting curve. Weight values ​​are set for the railing fitting curve and the road fitting curve respectively, so as to perform weighted fusion of the railing fitting curve and the road fitting curve to obtain the railing curve.

7. The method for identifying road railings according to claim 1, characterized in that, Acquire the point traces detected by millimeter-wave radar and process them to obtain stationary point traces, specifically including: Real-time acquisition of all raw data points detected by millimeter-wave radar; The target clustering algorithm is used to distinguish all original point traces into moving point traces and stationary point traces; The identified moving points are filtered to obtain stationary points.

8. A path planning method, characterized in that, Includes the following steps: The guardrail curves obtained by the road guardrail recognition method according to any one of claims 1-7 are used as road boundaries for planning vehicle driving paths.

9. A road railing identification device, characterized in that, include: The data acquisition and preprocessing module is used to acquire the point traces detected by the millimeter-wave radar and process the acquired point traces to obtain stationary point traces. A grid confidence construction module is used to map the obtained stationary point traces to a grid coordinate system to obtain the corresponding grid confidence. The effective grid filtering module is used to filter out effective grids according to a set confidence threshold, and process the effective grids to obtain continuous effective grids; The railing dynamic fitting module is used to fit the effective grid coordinates through a dynamic fitting model to obtain the railing fitting curve; The road fusion output module is used to obtain the road fitting curve and perform weighted fusion of the railing fitting curve and the road fitting curve to obtain the railing curve.

10. A computer device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction that causes the processor to perform the operation of a road barrier recognition method as described in any one of claims 1-7.

Images