High-precision automatic driving intelligent control system for agricultural machinery

Abstract

The present invention discloses a high-precision automatic driving intelligent control system for agricultural machinery, belonging to the field of automatic driving technology for agricultural machinery vehicles. The system includes a data acquisition module, a path planning module, a control module, and an instruction execution module; the data acquisition module acquires a dynamic data set in real time and transmits it to the path planning module; the path planning module receives and analyzes the dynamic data set to generate an optimal path; the control module synchronously receives the optimal path, real-time vehicle body posture dynamic data, and environmental change data, and generates a control instruction set; the instruction execution module receives the control instruction set, synchronously adjusts the steering angle and driving speed of the agricultural machinery, generates actual execution parameters, and feeds them back to the control module. By providing a control module, a data acquisition module, a path planning module, and an instruction execution module, the present invention solves the collision risk caused by the lack of environmental adaptability of traditional fixed trajectory algorithms.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic driving of agricultural machinery vehicles, and in particular to a high-precision automatic driving intelligent control system for agricultural machinery. Background Art

[0002] With the development of science and technology, autonomous driving technology is becoming increasingly developed and popular. From the perspective of application fields, it can be divided into: autonomous driving of urban traffic road vehicles, autonomous driving of special engineering operation vehicles, autonomous driving of agricultural machinery vehicles, etc.

[0003] Agricultural machinery offers unmatched efficiency compared to manual labor. It not only reduces labor intensity but also frees up a significant portion of the workforce for other occupations. Therefore, the development of agricultural machinery benefits both the nation and its people. Agricultural machinery is primarily used for tasks such as tilling, sowing, and harvesting. The vast majority of existing agricultural machinery is manually operated, and for large fields, drivers face long hours of driving, which can lead to fatigue. This has led to the development of autonomous agricultural machinery. The core purpose of autonomous driving is to replace humans with machines, freeing up all or part of the driver's labor. Especially in the field of agricultural vehicles, autonomous driving technology, combined with its specific operating methods, can significantly improve labor efficiency and effectively reduce agricultural costs.

[0004] Although existing agricultural machinery autonomous driving systems have the advantage of improving labor efficiency, they still have the following defects: the preset path tracking algorithm in the agricultural machinery autonomous driving system cannot adapt to dynamic environments (such as the sudden appearance of pedestrians in the field), and only travels along a fixed trajectory, with a high risk of collision.

[0005] Therefore, there is an urgent need to provide a high-precision automatic driving intelligent control system for agricultural machinery to solve the above problems. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the above-mentioned disadvantages of the existing agricultural machinery automatic driving system, which has the advantage of improving labor efficiency but still has the following defects: the preset path tracking algorithm in the agricultural machinery automatic driving system cannot adapt to dynamic environments (such as sudden appearance of pedestrians in the field), and only travels along a fixed trajectory, with a high risk of collision, and provide a high-precision automatic driving intelligent control system for agricultural machinery.

[0007] In order to solve the above technical problems, a technical solution adopted by the present invention is: providing a high-precision automatic driving intelligent control system for agricultural machinery, including a data acquisition module, a path planning module, a control module and an instruction execution module;

[0008] The data acquisition module acquires the agricultural machinery position information, vehicle posture data, obstacle dynamic distance data and environmental change data in real time, and integrates them into a dynamic data set to be transmitted to the path planning module;

[0009] A path planning module receives and analyzes the dynamic data set, generates a global reference path based on the agricultural machine position information, analyzes the dynamic obstacle distance data in real time, generates a local obstacle avoidance correction path linked to the real-time vehicle posture data, and dynamically superimposes the local obstacle avoidance correction path on the global reference path to generate an optimal path;

[0010] a control module that synchronously receives the optimal path, the real-time dynamic data of the vehicle body posture, and the environmental change data, simulates the operation trajectory of the agricultural machine in the future cycle along the optimal path in real time through a preset twin unit, and generates a control instruction set based on the dynamic coupling relationship between the operation trajectory and the environmental change data;

[0011] The instruction execution module receives the control instruction set, synchronously adjusts the steering angle and driving speed of the agricultural machinery, generates actual execution parameters, and feeds back to the control module.

[0012] The present invention is further configured as follows: the data acquisition module includes a high-precision positioning unit, a sensor unit, a multimodal perception unit, an environment perception unit and a data preprocessing unit;

[0013] The high-precision positioning unit collects the position information of the agricultural machinery in real time during the autonomous driving process; the sensor unit collects the vehicle body posture data during the autonomous driving process; the multimodal perception unit detects the dynamic distance data of obstacles during the autonomous driving process; and the environmental perception unit collects environmental change data during the autonomous driving process;

[0014] The data preprocessing unit receives and integrates the agricultural machinery position information, the vehicle body posture data, the obstacle dynamic distance data and the environmental change data, performs preset data cleaning, coordinate conversion, and timestamp synchronization processing, and generates a dynamic data set to be transmitted to the path planning module.

[0015] The present invention is further configured as follows: the system also includes a database, which stores the real-time steering capability, working width, center of gravity offset and the agricultural machinery's own parameters of the agricultural machinery.

[0016] The present invention is further configured as follows: the method for generating the global reference path in the path planning module is:

[0017] S1. Sending the current agricultural machine location information in the dynamic data set and the origin coordinates, transit coordinates, and destination coordinates obtained by a preset navigation module in the agricultural machine to a preset cloud server;

[0018] S2. The cloud server generates, based on the current agricultural machine location information, the origin coordinates, the transit coordinates, and the destination coordinates, a plurality of sub-paths from the current agricultural machine location information to the origin coordinates, the origin coordinates to the transit coordinates, the transit coordinates to the transit coordinates, and the transit coordinates to the destination coordinates;

[0019] S3. Attach a continuous mark to the starting end and the ending end of each sub-path, connect multiple sub-paths in series according to the arrangement order of the continuous marks, generate a global reference path, and transmit the global reference path to the path planning module.

[0020] The present invention is further configured as follows: the steps for generating the local obstacle avoidance correction path in the path planning module are as follows:

[0021] S101: A pre-set parsing unit in the path planning module analyzes in real time the dynamic distance data of obstacles during the automatic driving of the agricultural machine along the global reference path. When the distance between any obstacle and the agricultural machine exceeds a preset safety threshold, the parsing unit simultaneously integrates the heading angle and yaw rate in the vehicle posture data to generate a dynamic collision zone between the agricultural machine trajectory and the obstacle.

[0022] S102: extracting a planar structural graphic of an obstacle within the dynamic collision area, and constructing a dynamic passage corridor based on the real-time steering capability, operating width, and center of gravity offset of the agricultural machine;

[0023] S103. Based on the geometry of the dynamic passage corridor, generate multiple curvature continuous paths, each of which connects to the current segment of the global reference path at its head end and bypasses the obstacle area and returns to the global reference path at its tail end. Comprehensively evaluate the multiple curvature continuous paths, and select the path with the best comprehensive evaluation as the local obstacle avoidance correction path.

[0024] The present invention is further configured as follows: the steps for generating the dynamic collision area in step S101 are as follows:

[0025] S1011. Projecting the trajectory of the agricultural machine during the automatic driving process into a first dynamic envelope based on the global reference path; projecting the trajectory of the agricultural machine during the automatic driving process into a second dynamic envelope based on the heading angle and yaw rate in the vehicle body posture data; superimposing the second dynamic envelope on the first dynamic envelope; and expanding the area of ​​the first dynamic envelope to generate a compensation dynamic envelope.

[0026] S1012: Synchronously projecting the outline of the obstacle into a plane structure graphic, rendering the overlapping area generated by superimposing the compensated dynamic envelope and the plane structure graphic, and generating a dynamic collision area.

[0027] The present invention is further configured as follows: the process of constructing the dynamic passage corridor in step S102 is as follows:

[0028] S1021. Performing a preset triangular meshing within the dynamic collision area based on the outline features of the planar structure of the obstacle, and dividing the area into three levels of passage areas according to the distance between the mesh vertices and the compensated dynamic envelope: a core passage area, a buffer area, and a prohibited passage area.

[0029] S1022. Combine the real-time steering capability of the agricultural machinery to map the minimum turning radius value and the operating width to map the lateral passage margin, and expand the core passage area in the width direction along the operating trajectory of the agricultural machinery. The expansion width in the expansion process = operating width + lateral passage margin × center of gravity offset compensation coefficient. At the same time, at the turning node of the core passage area, the minimum turning radius value × the preset safety curvature multiplier is used as a reference to generate a curved channel structure. The curved channel structure, the expansion width and the operating trajectory of the agricultural machinery are superimposed to generate a main passage. The main passage is superimposed with the buffer zone and the prohibited passage zone, and the overlapping part is retained as a dynamic passage corridor.

[0030] The present invention is further configured as follows: the simulation method of the running trajectory of the agricultural machinery in the future cycle in the optimal path in the control module is:

[0031] Q1. Based on the optimal path and the real-time speed and steering angle of the agricultural machine in the vehicle posture data, a virtual agricultural machine kinematic model that is completely consistent with the physical agricultural machine is constructed in the twin unit, and a preset scene map and the optimal path are loaded into the virtual agricultural machine kinematic model;

[0032] Q2. Input the heading angle, yaw rate and center of gravity offset in the vehicle body posture data into the virtual agricultural machinery kinematic model, predict the real-time posture changes of the virtual agricultural machinery in the future cycle with a preset step size, generate a continuous posture sequence, and combine the continuous posture sequence with the optimal path to obtain the operation trajectory of the agricultural machinery in the future cycle.

[0033] The present invention is further configured such that after the control module obtains the operation trajectory of the agricultural machinery in the future period, two situations are divided into the following: first, if there are no dynamic obstacles within a set range in the current environmental change data, a first control instruction is generated according to the operation trajectory and transmitted to the instruction execution module; second, if there are dynamic obstacles within the set range in the current environmental change data, a composite risk level is generated based on the dynamic coupling relationship between the operation trajectory and the environmental change data. When the composite risk level exceeds a preset tolerance threshold, a second control instruction is generated, and the first control instruction and the second control instruction are combined to generate a control instruction set;

[0034] The steps for generating the second control instruction are as follows:

[0035] Q3. Performing a spatial superposition operation on the continuous pose sequence and the optimal path. When the continuous pose sequence exceeds the boundary of the optimal path, it is marked as a posture instability risk point. Simultaneously, the real-time soil moisture content and surface friction coefficient in the environmental change data are integrated to calculate the slip rate deviation of the virtual agricultural machinery tire.

[0036] Q4. Counting the duration of the posture instability risk point and the cumulative value of the slip rate deviation, combined with the boundary distance of the prohibited zone of the dynamic passage corridor and the displacement acceleration vector of the dynamic obstacle in the current dynamic obstacle distance data, to generate a composite risk heat map;

[0037] Q5. Based on the spatial distribution, diffusion direction, and diffusion rate of the high-risk core areas in the composite risk heat map, reversely deduce the steering angle correction and driving speed adjustment required for the agricultural machinery. The steering correction angle is calculated by combining the azimuth offset and center of gravity offset of the high-risk core areas relative to the agricultural machinery. The driving speed adjustment is dynamically set based on the matching relationship between the diffusion rate of the high-risk core areas and the real-time speed of the agricultural machinery.

[0038] Q6. Generate a second control instruction by fusing the steering angle correction amount and the driving speed adjustment amount.

[0039] The beneficial effects of the present invention are as follows:

[0040] 1. The present invention utilizes a control module, data acquisition module, and path planning module to simulate the future trajectory of agricultural machinery in real time within the twin unit. Combined with the dynamic obstacle displacement acceleration vectors detected by the multimodal sensing unit, this method generates a composite risk heat map. Spatial superposition calculations are used to detect deviations between the pose sequence and the optimal path. The slip rate deviation is calculated by integrating soil moisture content and surface friction coefficient, dynamically marking risk points for posture instability. Steering corrections and speed adjustments are calculated based on the diffusion direction and velocity of high-risk core areas, generating a secondary control instruction that enables the agricultural machinery to actively deviate from the preset path in the event of an unexpected obstacle. This achieves millisecond-level dynamic obstacle avoidance and addresses the potential collision risks associated with traditional fixed-trajectory algorithms due to their limited environmental adaptability.

[0041] 2. The present invention constructs a curvature-continuous obstacle avoidance path during the path planning phase based on the three-level traffic area constraints of the dynamic traffic corridor, combined with center of gravity offset and real-time steering capabilities. The control module predicts tire slip rate deviation through a virtual agricultural machinery kinematic model and dynamically adjusts the control instruction set based on the distance to the boundary of the prohibited zone, ensuring that the agricultural machinery maintains trajectory tracking accuracy and body stability in a complex environment with a superposition of dynamic obstacles and sudden terrain changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is a system flow chart of the present invention;

[0043] Figure 2 A flow chart of steps for generating a global reference path of the present invention;

[0044] Figure 3 This is a flow chart of the steps for generating a local obstacle avoidance correction path according to the present invention. DETAILED DESCRIPTION

[0045] The preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more precise definition of the protection scope of the present invention.

[0046] See also Figure 1-Figure 3 , a high-precision automatic driving intelligent control system for agricultural machinery, including data acquisition module, path planning module, control module and instruction execution module;

[0047] The data acquisition module acquires the agricultural machinery's position information, vehicle posture data, dynamic obstacle distance data, and environmental change data in real time, and integrates them into a dynamic data set to be transmitted to the path planning module;

[0048] Among them, the data acquisition module includes a high-precision positioning unit, a sensor unit, a multimodal perception unit, an environmental perception unit and a data preprocessing unit;

[0049] The high-precision positioning unit collects real-time position information of the agricultural machinery during autonomous driving. The sensor unit collects vehicle posture data during autonomous driving. The multimodal perception unit detects dynamic obstacle distance data during autonomous driving. The environmental perception unit collects data on environmental changes during autonomous driving. The high-precision positioning unit is preferably based on the Beidou / GPS satellite positioning system, achieving centimeter-level precision in the real-time position of the agricultural machinery, providing a spatial reference for path planning. The sensor unit integrates gyroscopes, inclination sensors, and velocity sensors to collect real-time vehicle posture data such as longitudinal acceleration, yaw rate, heading angle, slope inclination, and driving speed for stability control and trajectory correction. The multimodal perception unit integrates lidar, millimeter-wave radar, and visual cameras to dynamically detect the distance, size, motion trajectory, and displacement acceleration vector of obstacles (such as trees, ridges, and moving objects), generating a dynamic obstacle distance dataset. The environmental perception unit includes soil moisture sensors, light intensity sensors, and temperature and humidity sensors to monitor farmland environmental parameters (such as soil moisture content, surface friction coefficient, and light intensity) in real time, providing environmental variables for path risk prediction.

[0050] The data preprocessing unit receives and integrates the agricultural machinery position information, vehicle posture data, obstacle dynamic distance data and environmental change data, performs preset data cleaning, coordinate conversion, and timestamp synchronization processing, and generates a dynamic data set to be transmitted to the path planning module. Among them, data cleaning, coordinate conversion, and timestamp synchronization processing are all existing technologies and will not be elaborated here.

[0051] The system also includes a database, which stores the real-time steering capability, working width, center of gravity offset and the agricultural machinery's own parameters. The agricultural machinery's own parameters include the width and other dimensions of the agricultural machinery.

[0052] The path planning module receives and analyzes dynamic data sets, generating a global reference path based on the position information of the agricultural machinery. It analyzes the dynamic distance data of obstacles in real time, generates a local obstacle avoidance correction path linked to the real-time vehicle posture data, and dynamically superimposes the local obstacle avoidance correction path on the global reference path to generate the optimal path.

[0053] Among them, the method for generating the global reference path in the path planning module is:

[0054] S1. Send the current agricultural machinery location information in the dynamic data set and the origin, transit, and destination coordinates obtained by the navigation module preset in the agricultural machinery to a preset cloud server. The cloud server is a remote computing platform based on cloud computing technology. It integrates distributed physical server resources through virtualization technology to provide elastic and scalable path planning computing power support for the agricultural machinery automatic driving system.

[0055] S2. The cloud server generates, based on the current agricultural machine location information, the coordinates of the origin, the coordinates of the transit points, and the coordinates of the destination, a plurality of sub-paths from the current agricultural machine location information to the coordinates of the origin, the coordinates of the origin to the coordinates of the transit points, the coordinates of the transit points to the coordinates of the transit points, and the coordinates of the transit points to the coordinates of the destination, preferably each sub-path is the shortest path;

[0056] S3. Attach a continuous mark to the starting end and the ending end of each sub-path, and connect multiple sub-paths in series according to the arrangement order of the continuous marks to generate a global reference path, and transmit the global reference path to the path planning module.

[0057] The steps for generating the local obstacle avoidance correction path in the path planning module are as follows:

[0058] S101. A pre-set parsing unit in the path planning module analyzes in real time the dynamic distance data of obstacles encountered by the agricultural machine during its automatic driving along the global reference path. When the distance between any obstacle and the agricultural machine exceeds a preset safety threshold, the parsing unit simultaneously integrates the heading angle and yaw rate in the vehicle posture data to generate a dynamic collision zone between the agricultural machine trajectory and the obstacle. The safety threshold is calculated as half the width of the space where the agricultural machine's plow is mounted plus a safety margin of 0.3 to 0.8 meters, and is dynamically increased in proportion to the real-time speed of the agricultural machine.

[0059] S102. Extract a planar structural diagram of an obstacle (this obstacle is a static obstacle) within the dynamic collision area, and construct a dynamic passage corridor based on the real-time steering capability, working width, and center of gravity offset of the agricultural machine.

[0060] S103. Based on the geometry of the dynamic corridor, generate multiple curvature-continuous paths that connect to the current segment of the global reference path at the beginning and bypass the obstacle area and return to the global reference path at the end. Comprehensively evaluate the multiple curvature-continuous paths and select the path with the best comprehensive evaluation as the local obstacle avoidance correction path.

[0061] The comprehensive assessment method includes calculating the real-time distance between the curvature-continuous path and the boundary of the dynamic collision zone based on a safety margin, dynamically adjusting the safety threshold based on the center of gravity offset to ensure that the curvature-continuous path still has a buffer space under adverse conditions such as slippery soil. Furthermore, a preset crop protection dimension is introduced. By integrating operating width and crop row distribution data, the area of ​​unharvested crops crushed by the curvature-continuous path is predicted, and agronomic damage is factored into the assessment weight. A weighted evaluation generates a comprehensive performance index for each curvature-continuous path, and the curvature-continuous path with the highest index is ultimately selected as the local obstacle avoidance correction path.

[0062] Preferably, the local obstacle avoidance correction path in the path planning module includes the correction of the sub-path from the current agricultural machinery position information to the coordinates of the starting point. Since the accuracy of the navigation module in the agricultural machinery is not high, the position positioning of the current agricultural machinery is inaccurate, and there may be obstacles when traveling from the current agricultural machinery position information to the coordinates of the starting point.

[0063] The steps for generating the dynamic collision area in step S101 are as follows:

[0064] S1011. Based on the global reference path, the trajectory of the agricultural machine during the automatic driving process is projected into a first dynamic envelope (the bandwidth of the first dynamic envelope = the width of the agricultural machine × the safety expansion factor. The value of the safety expansion factor is a dynamically adjusted parameter, and its reference range is generally set to 1.2–1.8). Based on the heading angle and yaw rate in the vehicle posture data, the trajectory of the agricultural machine during the automatic driving process is projected into a second dynamic envelope. The second dynamic envelope is superimposed on the first dynamic envelope, and the area of ​​the first dynamic envelope is expanded to generate a compensation dynamic envelope.

[0065] S1012. Synchronously project the outline of the obstacle into a plane structure graphic, render the overlapping area generated by superimposing the compensated dynamic envelope and the plane structure graphic, and generate a dynamic collision area.

[0066] The process of constructing the dynamic passage corridor in step S102 is as follows:

[0067] S1021. Within the dynamic collision zone, perform a pre-set triangular mesh based on the outline features of the obstacle's planar structural diagram. Divide the area into three levels of traffic zones based on the spacing between the mesh vertices and the compensated dynamic envelope: a core traffic zone (spacing ≥ safety expansion factor × width of the agricultural machinery), a buffer zone (spacing ∈ [0.5 × safety expansion factor, 1 × safety expansion factor] × width of the agricultural machinery), and a prohibited traffic zone (spacing < 0.5 × safety expansion factor × width of the agricultural machinery).

[0068] S1022. Combine the real-time steering capability of the agricultural machinery to map the minimum turning radius value, and the operating width to map the lateral passage margin, and expand the core passage area in the width direction along the operating trajectory of the agricultural machinery. The expansion width in the expansion process = operating width + lateral passage margin × center of gravity offset compensation coefficient. At the same time, at the turning nodes of the core passage area, the minimum turning radius value × the preset safety curvature multiplier is used as a basis to generate a curved channel structure. The curved channel structure, the expanded width and the operating trajectory of the agricultural machinery are superimposed to generate a main passage. The main passage is superimposed with the buffer zone and the prohibited passage zone, and the overlapping part is retained as a dynamic passage corridor.

[0069] In Example 1, during corn harvesting on a farm, a 3-meter-wide autonomous harvester (with a safety expansion coefficient set to 1.6) obtained centimeter-level position information in real time (positioning error ≤ 2 cm) based on the Beidou high-precision positioning unit. The gyroscope of the sensor unit detected a heading deviation of 3° and a yaw rate of 0.15 rad / s. At the same time, the multimodal perception unit detected a fallen tree (a static obstacle with an outline size of 2.1 m × 0.8 m) 8 meters ahead through the lidar. The environmental perception unit monitored the The soil moisture content is measured to be 28% (the surface friction coefficient is 0.32). After the data preprocessing unit integrates the above information to generate a dynamic data set, the path planning module first generates the shortest sub-path from the current position (E126.73°, N47.28°) to the starting point (E126.74°, N47.29°) through the cloud server; when the obstacle distance exceeds the safety threshold (the basic value is 3m / 2+0.5m=2.0m, which increases to 2.3m due to the vehicle speed of 1.8m / s), the analysis unit combines the heading angle The dynamic collision zone is generated by the yaw rate - the bandwidth of the first dynamic envelope is 3m×1.6=4.8m, and the second envelope with yaw compensation is expanded to 5.2m, overlapping with the obstacle outline projection to form a collision zone; then triangulation is performed in the collision zone to divide the core passage area (spacing ≥4.8m) and the buffer area (spacing 2.4-4.8m). Combined with the minimum turning radius of agricultural machinery of 4.5m (real-time steering capability mapping value) and the working width of 3.2m, the core area is expanded (expansion width = 3 .2m+0.3×center of gravity offset compensation coefficient 1.2=3.56m), a curved channel with a curvature radius of 8.5m is generated at the turning node, and finally a dynamic traffic corridor is constructed; the system generates three continuous curvature paths (with curvature radii of 8.5m / 7.2m / 6.0m respectively). After comprehensive evaluation, the path with the highest safety margin (2.1m from the collision boundary) and the smallest crop crushing area (only 0.8㎡) is selected (comprehensive efficiency index 92.5) as the local obstacle avoidance correction path, achieving return to the global reference path after detouring.

[0070] This embodiment significantly improves the obstacle avoidance safety and operating efficiency of agricultural machinery in slippery soil conditions: through the dynamic envelope band superposition mechanism, it effectively adapts to sudden changes in heading angle and yaw disturbances, ensuring that the collision risk area is fully covered even when the vehicle posture is unstable; the three-level traffic area division and curved channel structure design accurately adapt to the minimum turning radius and operating width of agricultural machinery, opening up a continuous and smooth traffic corridor in complex obstacle terrain, and greatly reducing the risk of roll caused by sharp turns; the multi-path comprehensive evaluation model deeply integrates the dynamic correction of safety margin and crop protection weight, while avoiding obstacles, minimizing the crushing loss of unharvested crops; the collaborative mechanism of cloud-based global planning and local obstacle avoidance correction takes into account long-distance path optimality and real-time response capabilities to sudden obstacles, significantly shortening obstacle avoidance decision-making time and reducing empty mileage; the center of gravity offset compensation and slippery road control strategy work together to ensure steering stability on low-friction roads, completely avoiding the hidden dangers of slipping and loss of control.

[0071] The control module synchronously receives the optimal path, real-time vehicle posture dynamic data, and environmental change data. It simulates the operation trajectory of the agricultural machinery in the future cycle along the optimal path in real time through the preset twin unit. Based on the dynamic coupling relationship between the operation trajectory and environmental change data, it generates a control instruction set.

[0072] Among them, the simulation method of the operation trajectory of agricultural machinery in the future cycle in the optimal path in the control module is:

[0073] Q1. Based on the real-time speed and steering angle of the agricultural machinery from the optimal path and body posture data, a virtual agricultural machinery kinematic model that is completely consistent with the physical agricultural machinery is constructed within the twin unit. The preset scenario map and optimal path are then loaded into the virtual agricultural machinery kinematic model. Preferably, the twin unit mainly includes a dynamic mapping model of the physical agricultural machinery, a multi-source data fusion interface, and a high-fidelity virtual simulation environment. The virtual agricultural machinery kinematic model is driven by real-time synchronization of the physical agricultural machinery status data to deduce the future trajectory.

[0074] Q2. Input the heading angle, yaw rate and center of gravity offset in the vehicle posture data into the kinematic model of the virtual agricultural machinery, predict the real-time posture changes of the virtual agricultural machinery in the future cycle with a preset step size, generate a continuous posture sequence, and combine the continuous posture sequence with the optimal path to obtain the operation trajectory of the agricultural machinery in the future cycle. Preferably, the step size is 10 milliseconds and the future cycle is within 3 seconds.

[0075] Among them, after the control module obtains the operation trajectory of the agricultural machinery in the future cycle, it is divided into the following two situations: First, if there are no dynamic obstacles within the set range in the current environmental change data, a first control instruction is generated according to the operation trajectory and transmitted to the instruction execution module; Second, if there are dynamic obstacles within the set range in the current environmental change data, a composite risk level is generated based on the dynamic coupling relationship between the operation trajectory and the environmental change data. When the composite risk level exceeds the preset tolerance threshold, a second control instruction is generated, and the first control instruction and the second control instruction are combined to generate a control instruction set;

[0076] The steps for generating the second control instruction are as follows:

[0077] Q3. Perform spatial superposition operations on the continuous pose sequence and the optimal path. When the continuous pose sequence exceeds the boundary of the optimal path, it is marked as a pose instability risk point. At the same time, the real-time soil moisture content and surface friction coefficient of the environmental change data are integrated to calculate the slip rate deviation of the virtual agricultural machinery tire.

[0078] In step Q3, the spatial superposition operation is as follows: in the twin unit, the system first aligns the continuous pose sequence with the optimal path center baseline. It then calculates the vertical distance between each node in the continuous pose sequence and the optimal path center baseline. Based on the real-time steering capability and working width, a third dynamic envelope is generated (bandwidth = agricultural machine width × safety expansion coefficient). The system then checks whether the continuous pose nodes exceed the third envelope boundary. If the projection point of the continuous pose sequence in a turning area or slippery road section exceeds the boundary threshold, it is marked as a posture instability risk point, and its spatial coordinates and timestamp are recorded.

[0079] The slip rate deviation is calculated as follows: The slip rate calculation model is constructed by integrating the real-time soil moisture content and surface friction coefficient from environmental change data with the tire-ground interaction mechanical parameters (including tire contact pressure, tread depth, and sidewall stiffness) from the virtual agricultural machinery kinematic model. The theoretical tire slip rate is inferred based on the longitudinal acceleration and yaw rate from the vehicle body posture data. The actual slip rate is calculated by simulating the relative motion speed between the tire and the ground in future posture sequences using the twin unit. Finally, the deviation between the theoretical and actual values ​​is mapped into a normalized slip rate deviation coefficient, which is then dynamically compensated and corrected in conjunction with the center of gravity offset to generate the slip rate deviation.

[0080] Q4. Calculate the duration of the attitude instability risk point and the cumulative value of the slip rate deviation. Combined with the boundary distance of the prohibited zone of the dynamic passage corridor and the displacement acceleration vector of the dynamic obstacle in the current dynamic distance data of the obstacle, a composite risk heat map is generated.

[0081] The steps for generating a composite risk heat map are as follows: The duration distribution of attitude instability risk points and the cumulative value of slip rate deviation are statistically analyzed, combined with the boundary distance of the restricted zone of the dynamic corridor (a three-level division of core zone, buffer zone, and restricted zone); the displacement acceleration vector of the dynamic obstacle in the dynamic distance data is simultaneously analyzed to calculate the spatial intersection probability of the obstacle's motion trajectory and the future motion trajectory of the agricultural machinery; the above parameters are weighted and superimposed through a preset multi-source data fusion engine to generate a three-dimensional space-time composite risk heat map; this map is centered on the high-risk core area and renders risk gradient color bands along the diffusion direction and diffusion rate to quantify the comprehensive risk level of each area;

[0082] Q5. Based on the spatial distribution, diffusion direction, and diffusion rate of the high-risk core areas in the composite risk heat map, reversely deduce the steering angle correction and driving speed adjustment required for agricultural machinery. The steering correction angle is calculated by combining the azimuth offset and center of gravity offset of the high-risk core area relative to the agricultural machinery. The driving speed adjustment is dynamically set based on the matching relationship between the diffusion rate of the high-risk core area and the real-time speed of the agricultural machinery.

[0083] Q6. Generate a second control instruction by fusing the steering angle correction amount and the driving speed adjustment amount.

[0084] Preferably, the steps for generating the first control instruction are mostly the same as those for generating the second control instruction, and only the displacement acceleration vector of the dynamic obstacle in the front environment data in step Q4 needs to be removed, which will not be elaborated here.

[0085] The instruction execution module receives the control instruction set, synchronously adjusts the steering angle and driving speed of the agricultural machinery, generates actual execution parameters, and feeds them back to the control module.

[0086] Example 2: In the corn harvesting operation scenario, the control module synchronizes the vehicle posture data (heading angle 3°, yaw rate 0.15 rad / s, real-time speed 1.8 m / s, center of gravity offset 0.12 m) and environmental change data (soil moisture content 28%, surface friction coefficient 0.32) of the twin unit in real time, and deduces the posture changes of the virtual agricultural machinery kinematic model in the next 3 seconds with a step size of 10 milliseconds to generate a continuous posture sequence; when the multimodal perception unit detects a dynamic obstacle (hare displacement acceleration vector 2.4 m / s², 4.5 meters away from the agricultural machinery), the composite risk level calculation process is triggered: first, through the air The inter-superposition operation aligns each node of the posture sequence with the coordinate system of the optimal path center baseline, calculates the vertical distance (maximum offset 0.25m), and generates the third dynamic envelope (bandwidth 4.8m) based on the width of the agricultural machinery of 3 meters and the safety expansion coefficient of 1.6. It is detected that the displacement sequence exceeds the envelope boundary by 0.18m at the turning node and is marked as a posture instability risk point; the soil parameters and tire mechanical model (ground pressure 85kPa, tread pattern depth 15mm) are synchronously integrated to reversely infer the theoretical slip rate of 12.3%, and the actual slip rate of 15.7% is simulated through the twin unit to generate a normalized slip rate deviation of +3.4%; then the system is integrated into the tire mechanical model (ground pressure 85kPa, tread pattern depth 15mm). The duration of the risk point (1.2 seconds) and the accumulated value of the slip rate are calculated, combined with the boundary distance of the prohibited area of ​​the dynamic corridor (1.2m) and the hare displacement vector, and a composite risk thermal map is generated by weighting through a multi-source data fusion engine (the diffusion rate of the high-risk core area is 1.8m / s); based on the map, the steering angle correction amount of -1.8° (calculated by the core area azimuth offset of 0.7m and the center of gravity offset) and the driving speed adjustment amount of -0.3m / s (matching the diffusion rate and the real-time speed) are reversely calculated, and the second control instruction is generated by fusion; after the instruction execution module receives the instruction set, the steering correction amount is converted into The system dynamically distributes inter-wheel torque based on the differential torque command (-32 N·m for the left wheel and +28 N·m for the right wheel) by integrating ground adhesion and steering delay coefficient with a BP neural network. The system also adjusts the fuel injection rate of the agricultural machinery's engine ECU and the opening of the electronically controlled hydraulic valve to 42% to match the output torque to the target speed of 1.5 m / s. The system collects yaw rate (0.12 rad / s) and preset wheel speed pulse frequency feedback in real time. After closed-loop compensation using a preset sliding mode observer, the actual execution parameter set (steering tracking error 0.7°, speed fluctuation rate 4.1%) is generated and transmitted to the control module via the preset CAN bus for iterative optimization.

[0087] This embodiment significantly enhances the robustness and response efficiency of the system in dynamic obstacle scenarios: the twin unit predicts the risk of posture instability in advance through millisecond-level trajectory deduction and multi-source data fusion, and combines the soil-tire dynamics model to accurately quantify the slip deviation, so that the control instructions under slippery conditions have physical authenticity; the composite risk thermal map dynamically couples the obstacle motion vector, path geometric constraints and mechanism dynamic parameters, and realizes the quantitative positioning and diffusion trend visualization of high-risk areas through space-time domain risk gradient rendering; the steering and speed correction instructions based on the reverse solution of the risk map deeply integrate the kinematic characteristics of agricultural machinery and the center of gravity offset compensation mechanism to ensure steering stability on low-adhesion roads.

[0088] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention's description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. High-precision automatic driving intelligent control system for agricultural machinery, characterized by: Including data acquisition module, path planning module, control module and instruction execution module; The data acquisition module acquires the agricultural machinery position information, vehicle posture data, obstacle dynamic distance data and environmental change data in real time, and integrates them into a dynamic data set to be transmitted to the path planning module; a path planning module, receiving and analyzing the dynamic data set, and generating a global reference path based on the agricultural machine position information; Real-time analysis of the dynamic distance data of the obstacle, generating a local obstacle avoidance correction path linked to the real-time vehicle body posture data, and dynamically superimposing the local obstacle avoidance correction path onto the global reference path to generate an optimal path; The steps for generating the local obstacle avoidance correction path in the path planning module are as follows: S101: A pre-set parsing unit in the path planning module analyzes in real time the dynamic distance data of obstacles during the automatic driving of the agricultural machine along the global reference path. When the distance between any obstacle and the agricultural machine exceeds a preset safety threshold, the parsing unit simultaneously integrates the heading angle and yaw rate in the vehicle posture data to generate a dynamic collision zone between the agricultural machine trajectory and the obstacle. S102: extracting a planar structural graphic of an obstacle within the dynamic collision area, and constructing a dynamic passage corridor based on the real-time steering capability, operating width, and center of gravity offset of the agricultural machine; S103: Based on the geometry of the dynamic corridor, generate multiple curvature-continuous paths that connect to the current segment of the global reference path at their first end and bypass the obstacle area and return to the global reference path at their last end; comprehensively evaluate the multiple curvature-continuous paths and select the path with the best comprehensive evaluation as the local obstacle avoidance correction path; The steps for generating the dynamic collision area in step S101 are as follows: S1011. Projecting the trajectory of the agricultural machine during the automatic driving process into a first dynamic envelope based on the global reference path; projecting the trajectory of the agricultural machine during the automatic driving process into a second dynamic envelope based on the heading angle and yaw rate in the vehicle body posture data; superimposing the second dynamic envelope on the first dynamic envelope; and expanding the area of ​​the first dynamic envelope to generate a compensation dynamic envelope. S1012. Synchronously projecting the outline of the obstacle into a plane structure graphic, rendering the overlapping area generated by superimposing the compensated dynamic envelope and the plane structure graphic, and generating a dynamic collision area; The process of constructing the dynamic passage corridor in step S102 is as follows: S1021. Performing a preset triangular meshing within the dynamic collision area based on the outline features of the planar structure of the obstacle, and dividing the area into three levels of passage areas according to the distance between the mesh vertices and the compensated dynamic envelope: a core passage area, a buffer area, and a prohibited passage area. S1022. Based on the real-time steering capability of the agricultural machinery, the minimum turning radius is mapped, and the working width is mapped to the lateral clearance margin. The core passage area is expanded in the width direction along the operating trajectory of the agricultural machinery. The expanded width in the expansion process is calculated as follows: the working width + the lateral clearance margin × the center of gravity offset compensation coefficient. At the turning nodes of the core passage area, a curved channel structure is generated based on the minimum turning radius × a preset safety curvature multiplier. The curved channel structure, the expanded width, and the operating trajectory of the agricultural machinery are superimposed to generate a main passage. The main passage is then superimposed with the buffer zone and the prohibited passage zone, and the overlapping portion is retained as a dynamic passage corridor. a control module that synchronously receives the optimal path and the dynamic data set, simulates in real time the operation trajectory of the agricultural machinery in future cycles along the optimal path through a preset twin unit, and generates a control instruction set based on a dynamic coupling relationship between the operation trajectory and the dynamic data set; The instruction execution module receives the control instruction set, synchronously adjusts the steering angle and driving speed of the agricultural machinery, generates actual execution parameters, and feeds back to the control module.

2. The high-precision automatic driving intelligent control system for agricultural machinery according to claim 1 is characterized by: The data acquisition module includes a high-precision positioning unit, a sensor unit, a multimodal perception unit, an environment perception unit and a data preprocessing unit; The high-precision positioning unit collects the position information of the agricultural machinery in real time during the autonomous driving process; the sensor unit collects the vehicle body posture data during the autonomous driving process; the multimodal perception unit detects the dynamic distance data of obstacles during the autonomous driving process; and the environmental perception unit collects environmental change data during the autonomous driving process; The data preprocessing unit receives and integrates the agricultural machinery position information, the vehicle body posture data, the obstacle dynamic distance data and the environmental change data, performs preset data cleaning, coordinate conversion, and timestamp synchronization processing, and generates a dynamic data set to be transmitted to the path planning module.

3. The high-precision automatic driving intelligent control system for agricultural machinery according to claim 2, characterized in that: The system also includes a database, which stores the real-time steering capability, working width, center of gravity offset and parameters of the agricultural machine itself.

4. The high-precision automatic driving intelligent control system for agricultural machinery according to claim 3 is characterized by: The method for generating the global reference path in the path planning module is: S1. Sending the current agricultural machine location information in the dynamic data set and the origin coordinates, transit coordinates, and destination coordinates obtained by a preset navigation module in the agricultural machine to a preset cloud server; S2. The cloud server generates, based on the current agricultural machine location information, the origin coordinates, the transit coordinates, and the destination coordinates, a plurality of sub-paths from the current agricultural machine location information to the origin coordinates, the origin coordinates to the transit coordinates, the transit coordinates to the transit coordinates, and the transit coordinates to the destination coordinates; S3. Attach a continuous mark to the starting end and the ending end of each sub-path, connect multiple sub-paths in series according to the arrangement order of the continuous marks, generate a global reference path, and transmit the global reference path to the path planning module.

5. The high-precision automatic driving intelligent control system for agricultural machinery according to claim 4 is characterized in that: The simulation method of the running trajectory of the agricultural machinery in the future cycle in the optimal path in the control module is: Q1. Based on the optimal path and the real-time speed and steering angle of the agricultural machine in the vehicle posture data, a virtual agricultural machine kinematic model that is completely consistent with the physical agricultural machine is constructed in the twin unit, and a preset scene map and the optimal path are loaded into the virtual agricultural machine kinematic model; Q2. Input the heading angle, yaw rate and center of gravity offset in the vehicle body posture data into the virtual agricultural machinery kinematic model, predict the real-time posture changes of the virtual agricultural machinery in the future cycle with a preset step size, generate a continuous posture sequence, and combine the continuous posture sequence with the optimal path to obtain the operation trajectory of the agricultural machinery in the future cycle.

6. The high-precision automatic driving intelligent control system for agricultural machinery according to claim 5, characterized in that: After the control module obtains the operation trajectory of the agricultural machinery in the future cycle, it is divided into the following two situations: first, if there is no dynamic obstacle within the set range in the current environmental change data, a first control instruction is generated according to the operation trajectory and transmitted to the instruction execution module; second, if there is a dynamic obstacle within the set range in the current environmental change data, a composite risk level is generated based on the dynamic coupling relationship between the operation trajectory and the environmental change data. When the composite risk level exceeds a preset tolerance threshold, a second control instruction is generated, and the first control instruction and the second control instruction are combined to generate a control instruction set; The steps for generating the second control instruction are as follows: Q3. Performing a spatial superposition operation on the continuous pose sequence and the optimal path. When the continuous pose sequence exceeds the boundary of the optimal path, it is marked as a posture instability risk point. Simultaneously, the real-time soil moisture content and surface friction coefficient in the environmental change data are integrated to calculate the slip rate deviation of the virtual agricultural machinery tire. Q4. Counting the duration of the posture instability risk point and the cumulative value of the slip rate deviation, combined with the boundary distance of the prohibited zone of the dynamic passage corridor and the displacement acceleration vector of the dynamic obstacle in the current dynamic obstacle distance data, to generate a composite risk heat map; Q5. Based on the spatial distribution, diffusion direction, and diffusion rate of the high-risk core areas in the composite risk heat map, reversely deduce the steering angle correction and driving speed adjustment required for the agricultural machinery. The steering correction angle is calculated by combining the azimuth offset and center of gravity offset of the high-risk core areas relative to the agricultural machinery. The driving speed adjustment is dynamically set based on the matching relationship between the diffusion rate of the high-risk core areas and the real-time speed of the agricultural machinery. Q6. Generate a second control instruction by fusing the steering angle correction amount and the driving speed adjustment amount.

Images