An intelligent road roller track planning method, electronic equipment and storage medium

Abstract

The application discloses a kind of intelligent road roller track planning method, electronic equipment and storage medium, comprising: using target detection and semantic segmentation network to obtain road boundary detection result and the category, speed and distance information of obstacle;Relative pose of road roller and road boundary is calculated based on road boundary detection result;Using multi-sensor data fusion method, the global positioning information of road roller and obstacle is obtained by fusing obstacle information, road boundary information and the global positioning information of road roller obtained from combined navigation device;Global path planning model is established;Multi-objective function considering compaction quality and multi-constraint condition considering construction time window are constructed, and the operation speed of road roller and compaction number are solved by dynamic programming method based on optimization;The coupling of road roller speed and path is realized using spline curve interpolation method, and intelligent compaction global trajectory is generated.The application can realize the best compaction quality under complex construction, while ensuring the efficiency and economy of operation.

Description

Technical Field

[0001] This invention relates to automatic driving technology for road rollers, and in particular to an intelligent road roller trajectory planning method, electronic device, and storage medium. Background Technology

[0002] With increasingly stringent quality requirements and increasingly complex and harsh operating environments in modern infrastructure construction, intelligent road roller path planning methods are a crucial technology for improving road construction quality and operational efficiency. Precise trajectory planning and control enable efficient operation of road rollers, ensuring uniformity and stability of construction quality, thereby improving the overall quality of road construction and reducing subsequent maintenance costs.

[0003] Traditional manually operated road rollers face numerous challenges in actual operation. First, operators are susceptible to fatigue and weather conditions during long working hours, leading to unstable operation. Second, the compaction process in manual operation is often uneven, prone to over-compaction, under-compaction, or insufficient compaction at road edges. These problems not only reduce construction quality but may also cause premature road damage and increase maintenance costs. Finally, the harsh working environment poses significant health risks to operators. To address these issues, driverless road rollers are gradually becoming an important development direction. Among these technologies, trajectory planning is the core of driverless road roller technology and a key to ensuring compaction quality and efficiency.

[0004] Invention patent CN116449848A discloses a path planning system for an unmanned road roller. This method determines the construction area through two-point or four-point data acquisition and then plans the roller's operating path based on pre-set parameters such as compaction width and number of compaction passes. However, this method only plans the path based on the compacted area and does not consider obstacle information, limiting its application to obstacle-free areas and restricting its scope. Invention patent CN113406954 discloses another path planning method for an unmanned road roller. This method obtains information on the boundary of the work area and the location and size of obstacles, determines the length of a continuous obstacle-free zone on each compaction lane, and determines the path for continued compaction based on a comparison between the length of the continuous obstacle-free zone and a preset length. However, this method does not consider real-time road surface temperature and cannot dynamically optimize the roller's path based on the quality of the construction process. Chinese invention patent CN115951680A discloses a control method, device, equipment, and system for an unmanned intelligent road roller. This method determines the compaction completion rate within a preset time period by acquiring multiple sets of compaction data from the road roller over that time period, and then controls the compaction trajectory for the next time period based on the compaction completion rate. However, this method does not consider vehicle speed, making it difficult to balance fuel economy with compaction quality.

[0005] In summary, existing roller path planning methods typically only consider the coverage of the compacted area, failing to take into account dynamic factors such as real-time road surface temperature and vehicle speed, making it difficult to balance compaction quality and efficiency.

[0006] Therefore, there is an urgent need for a road roller path planning method that comprehensively considers road surface compaction quality, operating efficiency, and fuel economy, so as to further improve the overall quality and economy of road construction. Summary of the Invention

[0007] Purpose of the invention: In order to overcome the shortcomings of existing technologies, such as inaccurate compaction range and low compaction quality and efficiency, this invention provides an intelligent roller trajectory planning method, equipment and storage medium, which can achieve the best compaction quality in complex construction environments, while ensuring high efficiency and economy of operation.

[0008] Technical solution: The first aspect of the present invention provides a method for intelligent road roller trajectory planning, the method comprising the following steps:

[0009] Step 1: Based on multi-sensor data from cameras and millimeter-wave radar, a target detection and semantic segmentation network is used to obtain road boundary detection results and information on the category, speed, and distance of obstacles;

[0010] Step 2: Calculate the relative pose of the road roller and the road boundary based on the road boundary detection results;

[0011] Step 3: Use multi-sensor data fusion methods such as Kalman filtering / factor graph to fuse obstacle information, road boundary information and global positioning information of the road roller obtained from the integrated navigation device to obtain global positioning information of the road roller and obstacles;

[0012] Step 4: Establish a global path planning model and plan the operation path of the road roller based on road boundary information and vehicle kinematics model;

[0013] Step 5: Construct a multi-objective function that considers compaction quality and multiple constraints that consider construction time windows, and optimize the solution of roller operating speed and number of compaction passes based on dynamic programming.

[0014] Step 6: Use spline curve interpolation to couple the speed and path of the road roller, and generate a global trajectory for intelligent compaction.

[0015] Preferably, step 1 includes the following sub-steps:

[0016] Step 1-1: Perform distortion correction on the original image input from the fisheye camera to obtain the distortion-free image;

[0017] Steps 1-2: Use an object detection and semantic segmentation network to obtain the category, location, and road pixel coordinates of obstacles in the distortion-corrected image;

[0018] Steps 1-3: Use millimeter-wave radar to obtain distance, speed, and azimuth information of obstacles;

[0019] Preferably, step 2 includes the following sub-steps:

[0020] Step 2-1: Based on the known camera intrinsic and extrinsic parameters, the image is transformed to the vehicle coordinate system through inverse perspective transformation, and the model expression of the road boundary line in the vehicle coordinate system is obtained by applying polynomial fitting.

[0021] Step 2-2: Calculate the lateral offset and direction of movement of the vehicle relative to the road boundary using the road boundary line model in the vehicle coordinate system;

[0022] Preferably, step 3 includes the following sub-steps.

[0023] Step 3-1: Project the target points detected by the millimeter-wave radar onto the vehicle coordinate system;

[0024] Step 3-2: Correlate the temporal and spatial data outputs of the camera and millimeter-wave radar;

[0025] Step 3-3: Use Kalman filtering / factor graph and other methods to fuse the detection information from the camera and millimeter-wave radar to obtain accurate information such as obstacle position, speed and category;

[0026] Step 3-4: Combine the lateral distance offset of the road boundary line obtained in Step 2-2, the obstacle information obtained in Step 3-3, and the vehicle global positioning information obtained from the integrated navigation device to obtain the corrected global position information of the road roller and obstacles.

[0027] Preferably, the kinematic model of the road roller in step 4 is as follows:

[0028] ;

[0029] in, , Let be the coordinates of the front steel wheel mass. , Let these be the coordinates of the rear steel wheel mass. , Let be the heading angles of the front and rear steel wheels, respectively. The model assumes no lateral slippage during travel and that both front and rear wheels are rigid bodies. Let the distances from the mass points of the front and rear steel wheels to the hinge points be respectively. , The hinge point rotation angle is Then the position of the rear steel wheel mass and the heading angle can be expressed as:

[0030] ;

[0031] From this, the speed of the front wheel of the road roller can be obtained. angular velocity of the hinge point :

[0032] ;

[0033] Preferably, step 4 includes the following sub-steps:

[0034] Step 4-1: Based on the road boundary information, obtain the coordinates of the vertex positions of the rectangular compaction area and the road width. and the length of the compacted area According to the width of the vehicle's rollers Determine the overlap width based on the conditions at the construction site. Calculate the number of paths for the road roller to operate in a straight line. :

[0035] ;

[0036] Step 4-2: Calculate the coordinates of the path points on each work path based on the length of the compacted area. The longitudinal spacing between each path point on each path is... rice, The value is determined based on the diameter of the compaction wheel and the length of its contact with the ground, the requirements for compaction uniformity of different materials and construction specifications, and the vibration frequency of the compaction wheel. In practice, the longitudinal spacing is gradually optimized from half the diameter of the compaction wheel to approximately the diameter of the compaction wheel based on the specific compaction effect.

[0037] ;

[0038] Step 4-3: Using the path point information obtained in Step 4-2, select the starting and ending points of the staggered lane change. Based on a fifth-order polynomial and constrained by the vehicle kinematics model, plan the staggered path for the road roller. The fifth-order polynomial fitting formula is as follows:

[0039] ;

[0040] in, These are the polynomial coefficients.

[0041] Step 5 includes the following sub-steps:

[0042] Step 5-1: Using vehicle speed, number of compaction passes, and sensor information, quantitatively evaluate the compaction quality, operating efficiency, and fuel economy of the road roller construction, and construct a multi-objective optimization function based on this evaluation.

[0043] ;

[0044] in, For fuel economy indicators, This is an indicator of work efficiency. To solidify quality indicators, These are the weighting coefficients for each objective.

[0045] Step 5-2: Construct construction time window constraints based on the road surface temperature field decay model, construct dynamic constraints based on vehicle dynamic characteristics, and determine obstacle avoidance constraints based on perception and positioning information. Combine all constraints to determine the overall constraint conditions:

[0046] Step 5-3: Based on dynamic programming, considering the constraints proposed in Step 5-2, solve the objective function obtained in Step 5-1 to obtain global velocity information and the number of compaction passes.

[0047] Preferably, step 6 includes the following sub-steps:

[0048] Step 6-1: Input the path and speed information obtained in Step 4 and Step 5, couple them together to determine the planned speed of each path point, and form global trajectory information;

[0049] Step 6-2: Based on the vehicle's current position, determine the optimal planning speed for the current position using cubic spline interpolation.

[0050] According to a second aspect of the present invention, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement the method described thereon.

[0051] According to a third aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method described thereon.

[0052] Compared with the prior art, the present invention has the following advantages:

[0053] (1) Accurate compaction range: The path planning algorithm proposed in this invention detects the road boundary in real time based on camera input and dynamically adjusts the compaction path, which effectively solves the problems of over-compaction, under-compaction and inadequate compaction at the road edge that are easy to occur with traditional road rollers.

[0054] (2) Balancing compaction quality and efficiency: This invention introduces various dynamic factors such as real-time road surface temperature and vehicle speed, and constructs a multi-objective function that considers compaction quality and multiple constraints that consider construction time windows, so as to ensure that the best compaction quality can be achieved under different construction conditions and good fuel economy can be guaranteed. Attached Figure Description

[0055] Figure 1 This is a flowchart illustrating the intelligent road roller path planning method of the present invention;

[0056] Figure 2 This is a schematic diagram showing the relative pose of the road roller and the road edge in an embodiment of the present invention;

[0057] Figure 3 This is a schematic diagram of the feature-level fusion process between the camera and the millimeter-wave radar in an embodiment of the present invention;

[0058] Figure 4 This is a flowchart illustrating the multi-sensor fusion method in an embodiment of the present invention;

[0059] Figure 5 This is a schematic diagram of the roller's operating path in an embodiment of the present invention;

[0060] Figure 6 This is a schematic diagram of the kinematic model of the road roller in an embodiment of the present invention. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0062] The intelligent road roller trajectory planning method of this embodiment has the following process: Figure 1 As shown, it includes:

[0063] Step 1: Based on multi-sensor data from cameras and millimeter-wave radar, a target detection and semantic segmentation network is used to obtain road boundary detection results and information on the category, speed, and distance of obstacles;

[0064] Step 2: Calculate the relative pose of the road roller and the road boundary based on the road boundary detection results;

[0065] Step 3: Use multi-sensor data fusion methods such as Kalman filtering / factor graph to fuse obstacle information, road boundary information and global positioning information of the road roller obtained from the integrated navigation device to obtain global positioning information of the road roller and obstacles;

[0066] Step 4: Establish a global path planning model and plan the operation path of the road roller based on road boundary information and vehicle kinematics model;

[0067] Step 5: Construct a multi-objective function that considers compaction quality and multiple constraints that consider construction time windows, and optimize the solution of roller operating speed and number of compaction passes based on dynamic programming.

[0068] Step 6: Based on the spline curve interpolation method, realize the coupling of the roller speed and path to generate a global intelligent compaction trajectory.

[0069] Step 1 is as follows:

[0070] Step 1-1: Perform distortion correction on the original image input from the fisheye camera to obtain the distortion-free image;

[0071] The distortion correction method in step 1-1 is as follows:

[0072] Fisheye camera distortion is divided into two types: radial distortion and tangential distortion. Radial distortion can be seen as the coordinate point changing along the length direction, while tangential distortion is the coordinate point changing along the tangential direction.

[0073] Radial distortion can be corrected using quadratic and higher-order polynomial functions related to the center distance:

[0074] ;

[0075] ;

[0076] in, , These are the normalized coordinates of the points after distortion. , These are the coordinates of the corrected point (after distortion correction). This is the distance between the correction point and the optical center point. , , This is the radial distortion parameter.

[0077] For tangential distortion, tangential distortion parameters can be used. , Correction:

[0078] ;

[0079] ;

[0080] After obtaining the camera's intrinsic parameters and distortion parameters using the checkerboard calibration method, the above-mentioned formula can be used to calculate the corrected coordinates of the pixels, thereby achieving image distortion correction.

[0081] Steps 1-2: After obtaining the distortion-corrected image, the UNet semantic segmentation network is first used to obtain the pixel coordinate range of the road, while the ResNet backbone network is used to extract features for obstacle detection. Then, multi-scale fusion is performed on the feature maps. Finally, the category, bounding box position, and size of the obstacles are regressed.

[0082] Steps 1-3: Use millimeter-wave radar to obtain distance, speed, and azimuth information of obstacles;

[0083] Step 2 is as follows:

[0084] Step 2-1: Based on the known camera intrinsic and extrinsic parameters, transform the image to the vehicle coordinate system through coordinate transformation, and apply polynomial fitting to obtain the geometric expression of the road boundary line in the vehicle coordinate system;

[0085] Step 2-2: Using the road boundary line model in the vehicle coordinate system, calculate the lateral offset and direction between the vehicle and the road boundary, including the lateral offset amount and direction. See the schematic diagram below. Figure 2 As shown;

[0086] Step 3 specifically involves:

[0087] Step 3-1: Project the target points detected by the millimeter-wave radar onto the vehicle coordinate system;

[0088] Step 3-2: Correlate the temporal and spatial data outputs of the camera and millimeter-wave radar.

[0089] Step 3-3: Using the Kalman filter algorithm, fuse the detection information from the camera and millimeter-wave radar to obtain accurate information such as obstacle position, velocity, and category. The fusion process is as follows: Figure 3 As shown, specifically:

[0090] (1) Before the start of the whole process, initialize the filter and set the initial state estimate and covariance matrix;

[0091] (2) Camera processing:

[0092] according to State estimation at time 1 and the predicted covariance matrix Predict the state at the current moment Kalman gain and covariance matrix This leads to an update of the state estimate. and covariance matrix ;

[0093] (3) Processing of the millimeter-wave radar section:

[0094] Similarly, according to State estimation at time 1 and predict covariance matrix Predict the state at the current moment Kalman gain and covariance matrix ;

[0095] (4) Joint tracking and correlation:

[0096] The predicted camera and radar measurements are compared, and the ellipsoidal tracking gate rule is applied. Determine if the measured value is valid. If the condition is met, it is considered a valid echo, where, It is the eigenvector residual. Let covariance matrix be the variance matrix. Satisfying the observation dimension distributed;

[0097] For valid echoes, calculate their correlation probability. ,in Indicates until time. The cumulative confirmed measurement set; Represents the cumulative set of confirmed events;

[0098] (5) Based on the correlation probability of camera and radar data, the updated Kalman gain is combined to obtain a more accurate target state estimate. The fused state estimate will be used as the output of the system for subsequent control or decision-making.

[0099] Steps 3-4: Based on the factor graph fusion algorithm, the lateral distance offset of the road boundary line obtained in Step 2-2, the obstacle information obtained in Step 3-3, and the vehicle global positioning information obtained from the integrated navigation device are fused to obtain the corrected global position information of the road roller and obstacles. The overall fusion process is as follows: Figure 4 As shown.

[0100] Step 4 specifically involves:

[0101] Step 4-1: Based on the road boundary information, obtain the coordinates of the vertex positions of the rectangular compaction area and the road width. and the length of the compacted area According to the width of the vehicle's rollers Determine the overlap width based on the conditions at the construction site. Calculate the number of paths for the road roller to operate in a straight line. ,like Figure 5 As shown:

[0102] ;

[0103] Step 4-2: Calculate the coordinates of the path points on each work path based on the length of the compacted area. The longitudinal spacing between each path point on each path is 1 meter.

[0104] ;

[0105] Step 4-3: Using the path point information obtained in Step 4-2, select the starting and ending points of the staggered lane change. Based on a fifth-order polynomial and constrained by the vehicle kinematics model, plan the staggered path for the road roller. The fifth-order polynomial fitting formula is as follows:

[0106] ;

[0107] in, The coefficients are polynomials. The structure of the kinematic model of the road roller is as follows: Figure 6 As shown, specifically:

[0108] ;

[0109] in, , Let be the coordinates of the front steel wheel mass. , Let these be the coordinates of the rear steel wheel mass. , Let be the heading angles of the front and rear steel wheels, respectively. The model assumes no lateral slippage during travel and that both front and rear wheels are rigid bodies. Let the distances from the mass points of the front and rear steel wheels to the hinge points be respectively. , The hinge point rotation angle is Then the position of the rear steel wheel mass and the heading angle can be expressed as:

[0110] ;

[0111] From this, the speed of the front wheel of the road roller can be obtained. angular velocity of the hinge point :

[0112] ;

[0113] Step 5 specifically involves:

[0114] Step 5-1: Using vehicle speed, number of compaction passes, and sensor information, quantitatively evaluate the compaction quality, operating efficiency, and fuel economy of the road roller construction, and construct a multi-objective optimization function based on this evaluation.

[0115] ;

[0116] in, For fuel economy indicators, This is an indicator of work efficiency. To solidify quality indicators, These are the weighting coefficients for each objective.

[0117] Step 5-2: Construct construction time window constraints based on the road surface temperature field decay model, construct dynamic constraints based on vehicle dynamic characteristics, and determine obstacle avoidance constraints based on perception and positioning information. Combine all constraints to determine the overall constraint conditions:

[0118] use If we represent the total task time, then the time window constraint can be expressed as:

[0119] ;

[0120] Based on the vehicle's kinematic model, the constraints that the vehicle must satisfy when changing lanes can be determined. These dynamic constraints can be expressed as:

[0121] ;

[0122] in, Let be the radius of curvature of the lane-changing path. This is the minimum turning radius of the road roller. For the steering angular velocity, For the maximum steering angular velocity, This refers to the vehicle's longitudinal acceleration.

[0123] Based on perception and localization information, the specific location of obstacles and road boundary information can be determined, thereby enabling the determination of obstacle avoidance constraints. Indicates the straight-line distance from the obstacle. If we represent the vehicle's current position, then the obstacle avoidance constraint can be expressed as:

[0124] ;

[0125] in, It is determined by the boundary coordinates of the compacted area.

[0126] Step 5-3: Based on dynamic programming, considering the constraints proposed in Step 5-2, solve the objective function obtained in Step 5-1 to obtain global velocity information and the number of compaction passes:

[0127] The vehicle's motion state is described in the ST coordinate system. Time is discretized, and dynamic programming is used to solve the problem and determine the vehicle's position at different times.

[0128] With each path point spaced 1 meter apart and the time interval 1 second, and constrained by path length and time window, the maximum length on the S-axis and the maximum time on the T-axis can be determined as follows:

[0129] ;

[0130] Next, let the node be... Establish dimensions as OK, Column matrix This is used to store the cost from the starting point to each node; a matrix of the same dimension is established. This is used to store the best predecessor node of each node. Let... As the starting point for planning, For the first Okay, number Column nodes, The cost of state transitions between two adjacent columns can be used to establish a dynamic programming solution process:

[0131] First, initialize the matrix and calculate the cost of each path point from the planning starting point to the first column:

[0132] ;

[0133] Next, calculate the shortest cost from each node in the preceding sequence to the following sequence between adjacent columns. In each column... In the middle, traverse and calculate each node. Transfer to the nodes in the next column The value of ,right and The matrix is ​​updated:

[0134] ;

[0135] Afterwards, according to By performing a forward traversal and recording the position information of each optimal node from the end point to the start point, and by performing a difference between each point, the speed between adjacent columns can be determined, thereby determining the global speed information for each operation.

[0136] After each overall operation is completed, based on the real-time compaction quality information, determine whether further compaction is needed. If so, repeat steps 5-3 and 6 until the compaction quality meets the requirements. Step 6 specifically involves:

[0137] Step 6-1: Input the path and speed information obtained in Step 4 and Step 5, couple them together to determine the planned speed of each path point, and form global trajectory information;

[0138] Step 6-2: Based on the vehicle's current position, determine the optimal planning speed for the current position using cubic spline interpolation.

[0139] The electronic device of this invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in read-only memory (ROM) or loaded from a storage unit into random access memory (RAM). The RAM may also store various programs and data required for device operation. The CPU, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0140] Multiple components in the device are connected to the I / O interface, including: input units such as keyboards and mice; output units such as various types of displays and speakers; storage units such as disks and optical discs; and communication units such as network interface cards (NICs), modems, and wireless transceivers. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0141] The processing unit executes the various methods and processes described above, such as methods S1 to S6. For example, in some embodiments, methods S1 to S6 may be implemented as computer software programs tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by the CPU, one or more steps of methods S1 to S6 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S6 by any other suitable means (e.g., by means of firmware).

[0142] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload programmable logic devices (CPLDs), and so on.

[0143] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0144] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0145] The above are merely preferred embodiments of the present invention and do not constitute any limitation on the present invention. Any equivalent substitutions or modifications made by those skilled in the art to the technical solutions and content disclosed in the present invention without departing from the scope of the present invention shall be deemed to have remained within the protection scope of the present invention.

Claims

1. A method for intelligent road roller trajectory planning, characterized in that, Includes the following steps: Step 1: Use an object detection and semantic segmentation network to obtain road boundary detection results and obstacle information; Step 2: Calculate the relative pose of the road roller and the road boundary based on the road boundary detection results; Step 3: Use a multi-sensor data fusion method to fuse obstacle information, road boundary information, and global positioning information of the road roller obtained from the integrated navigation device to obtain global positioning information of the road roller and obstacles; Step 4: Establish a global path planning model and plan the operation path of the road roller based on road boundary information and vehicle kinematics model; Step 5: Construct a multi-objective function that considers compaction quality and multiple constraints that consider construction time windows, and use dynamic programming to optimize the solution of roller operating speed and number of compaction passes; Step 6: Utilize spline curve interpolation to couple the roller speed with the path, generating a global intelligent compaction trajectory. Step 5 specifically involves: Step 5-1: Using vehicle speed, number of compaction passes, and sensor information, quantitatively evaluate the compaction quality, operating efficiency, and fuel economy of the road roller construction, and construct a multi-objective optimization function based on this evaluation. ； in, For fuel economy indicators, This is an indicator of work efficiency. To solidify quality indicators, These are the weighting coefficients for each objective. Step 5-2: Construct construction time window constraints based on the road surface temperature field decay model, construct dynamic constraints based on vehicle dynamic characteristics, determine obstacle avoidance constraints based on perception and positioning information, and combine the various constraints to determine the overall constraint conditions: use If we represent the total task time, then the time window constraint can be expressed as: ； Based on the vehicle's kinematic model, the constraints that the vehicle must satisfy when changing lanes can be determined. These dynamic constraints can be expressed as: ； in, Let be the radius of curvature of the lane-changing path. This is the minimum turning radius of the road roller. For the steering angular velocity, For the maximum steering angular velocity, For the longitudinal acceleration of the vehicle, Based on perception and positioning information, the specific location of obstacles and road boundary information can be determined, thereby enabling the determination of obstacle avoidance constraints. Indicates the straight-line distance from the obstacle. If we represent the vehicle's current position, then the obstacle avoidance constraint can be expressed as: ； in, Determined by the boundary coordinates of the compacted area. Step 5-3: Based on dynamic programming, considering the constraints proposed in Step 5-2, solve the objective function obtained in Step 5-1 to obtain global velocity information and the number of compaction passes.

2. The intelligent road roller trajectory planning method according to claim 1, characterized in that, Step 1 includes the following sub-steps: Step 1-1: Correct the distortion of the original image to obtain the distortion-free image; Steps 1-2: Use an object detection and semantic segmentation network to obtain the category, location, and road pixel coordinates of obstacles in the distortion-corrected image; Steps 1-3: Use millimeter-wave radar to obtain information on the distance, speed, and azimuth of obstacles.

3. The intelligent road roller trajectory planning method according to claim 2, characterized in that, Step 2 includes the following sub-steps: Step 2-1: Based on the known camera intrinsic and extrinsic parameters, the image is transformed to the vehicle coordinate system through inverse perspective transformation, and the geometric expression of the road boundary line in the vehicle coordinate system is obtained by applying polynomial fitting. Step 2-2: Calculate the lateral offset and direction of movement of the vehicle relative to the road boundary using the geometric representation of the road boundary line in the vehicle coordinate system.

4. The intelligent road roller trajectory planning method according to claim 3, characterized in that, Step 3 includes the following sub-steps: Step 3-1: Project the target points detected by the millimeter-wave radar onto the vehicle coordinate system; Step 3-2: Correlate the temporal and spatial data outputs of the camera and millimeter-wave radar; Step 3-3: Use Kalman filtering or factor graph methods to fuse the detection information from the camera and millimeter-wave radar to obtain accurate information such as obstacle position, speed and category; Step 3-4: Combine the lateral distance offset of the road boundary line obtained in Step 2-2, the obstacle information obtained in Step 3-3, and the vehicle global positioning information obtained from the integrated navigation device to obtain the corrected global position information of the road roller and obstacles.

5. The intelligent road roller trajectory planning method according to claim 4, characterized in that, The kinematic model of the road roller in step 4 is as follows: ； in, , Let be the coordinates of the front steel wheel mass. , Let these be the coordinates of the rear steel wheel mass. , Let be the heading angles of the front and rear steel wheels, respectively. The model assumes no lateral slippage during travel and that both front and rear wheels are rigid bodies. Let the distances from the mass points of the front and rear steel wheels to the hinge points be respectively. , The hinge point rotation angle is Then the position of the rear steel wheel mass and the heading angle can be expressed as: ； From this, the speed of the front wheel of the road roller can be obtained. angular velocity of the hinge point : 。 6. The intelligent road roller trajectory planning method according to claim 5, characterized in that, Step 4 includes the following sub-steps: Step 4-1: Based on the road boundary information, obtain the coordinates of the vertices of the rectangular compaction area and the road width. and the length of the compacted area According to the width of the vehicle's rollers Determine the overlap width based on the conditions at the construction site. Calculate the number of paths for the road roller to operate in a straight line. : ； Step 4-2: Calculate the coordinates of the path points on each work path based on the length of the compacted area. The longitudinal spacing between each path point on each path is... rice, The value is determined based on the diameter of the compaction wheel and the length of its contact with the ground, the requirements for compaction uniformity of different materials and construction specifications, and the vibration frequency of the compaction wheel. In practice, the longitudinal spacing is gradually optimized from half the diameter of the compaction wheel to approximately the diameter of the compaction wheel based on the specific compaction effect. ； Step 4-3: Using the path point information obtained in Step 4-2, select the starting and ending points of the staggered lane change. Based on a fifth-order polynomial and constrained by the vehicle kinematics model, plan the staggered path for the road roller. The fifth-order polynomial fitting formula is as follows: ； in, These are the polynomial coefficients.

7. The intelligent road roller trajectory planning method according to claim 6, characterized in that, Step 6 includes the following sub-steps: Step 6-1: Input the path and speed information obtained in Step 4 and Step 5, couple them together to determine the planned speed of each path point, and form global trajectory information; Step 6-2: Based on the vehicle's current position, determine the optimal planning speed for the current position using cubic spline interpolation.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 7.

9. A storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 7.

Images