Steering wheel autonomous control method and system applied to intelligent driving

Abstract

The application provides a steering wheel autonomous control method and system applied to intelligent driving, and relates to the technical field of intelligent driving. First, the real-time driving scene state is perceived, covering road form, surrounding traffic participant action and environmental influence factors and the like information. The historical driving operation record under the similar scene is called. The steering wheel control reference parameter is generated based on the above information. Then, the vehicle real-time driving state data is collected, the reference parameter is dynamically corrected to obtain the real-time control parameter. The autonomous control instruction is generated based on the real-time control parameter and is sent to the steering execution system to drive the steering wheel to steer. The application comprehensively integrates multi-source information, combines historical experience and real-time state, realizes accurate, stable and adaptive control of the steering wheel, and improves the intelligent driving safety.

Description

Technical Field

[0001] This invention relates to the field of intelligent driving technology, and more specifically, to a method and system for autonomous steering wheel control applied to intelligent driving. Background Technology

[0002] As intelligent driving technology continues to develop, autonomous steering wheel control is a crucial element in ensuring safe and stable vehicle operation. Currently, the methods for controlling the steering wheel in intelligent driving systems have many limitations.

[0003] Some existing technologies control the steering wheel based solely on limited environmental information perceived at the moment, failing to comprehensively consider road conditions, the actions of surrounding road users, and environmental factors. For example, when faced with complex and ever-changing road conditions, such as sudden changes in curve curvature or sudden lane changes by vehicles ahead, the lack of comprehensive analysis of various information makes it difficult to make accurate and reasonable steering wheel control decisions, which can easily lead to unstable vehicle trajectories and even safety accidents.

[0004] Some technologies, while referencing historical driving data, simply extract past operations without deeply correlating and matching them with the current real-time driving scenario. This results in steering wheel control parameters that are not well adapted to the actual driving conditions and cannot be flexibly adjusted according to the characteristics of different scenarios, reducing the safety and reliability of intelligent driving. Therefore, existing intelligent driving steering wheel control methods urgently need improvement to meet the increasingly complex needs of intelligent driving. Summary of the Invention

[0005] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide an autonomous steering wheel control method for intelligent driving, the method comprising: The real-time driving scenario status during the intelligent driving process is perceived, and the real-time driving scenario status includes road morphology information, the action information of surrounding traffic participants, and environmental influencing factors. Retrieve the historical driving operation records of the intelligent driving vehicle, which include steering wheel adjustment operation information in scenarios similar to the real-time driving scenario. Based on the real-time driving scenario state and the historical driving operation record, steering wheel control reference parameters are generated, which include steering angle adjustment range and steering rate constraint information. Real-time driving status data of intelligent driving vehicles is collected. The real-time driving status data includes vehicle driving posture information and wheel steering feedback information. The steering wheel control reference parameters are dynamically corrected by combining the real-time driving status data to obtain real-time steering wheel control parameters. Based on the real-time control parameters of the steering wheel, an autonomous steering control command is generated and sent to the vehicle steering execution system to drive the steering wheel to complete the corresponding steering operation.

[0006] Furthermore, embodiments of the present invention also provide an autonomous steering wheel control system for intelligent driving, comprising: A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the aforementioned autonomous steering wheel control method for intelligent driving by executing the machine-executable instructions.

[0007] In another aspect, embodiments of the present invention also provide a computer program product, the computer program product including machine-executable instructions stored in a computer-readable storage medium, a processor of an autonomous steering wheel control system for intelligent driving reading the machine-executable instructions from the computer-readable storage medium, the processor executing the machine-executable instructions, causing the autonomous steering wheel control system for intelligent driving to execute the above-described autonomous steering wheel control method for intelligent driving.

[0008] Based on the above, by sensing the real-time driving scenario state during intelligent driving, encompassing information such as road morphology, actions of surrounding traffic participants, and environmental influencing factors, historical driving operation records from scenarios similar to the real-time driving scenario are retrieved. This historical experience is then deeply integrated with the current scenario to generate steering wheel control baseline parameters. This fully utilizes past driving experience, making the control parameters more rational and adaptable, better able to handle driving needs in similar scenarios. Next, real-time driving status data of the intelligent driving vehicle is collected, including vehicle driving posture and wheel steering feedback information. Based on this, the steering wheel control baseline parameters are dynamically corrected to obtain real-time steering wheel control parameters. This enables real-time optimization of steering wheel control, allowing for timely adjustments to control parameters according to the actual driving conditions of the vehicle, ensuring that steering wheel control always matches the vehicle's state. Autonomous control commands are generated based on the real-time steering wheel control parameters, driving the steering wheel to complete steering operations. The entire process is logically coherent and highly efficient, greatly improving the accuracy, stability, and adaptability of intelligent driving steering wheel control, effectively enhancing the safety and reliability of intelligent driving. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the execution flow of the autonomous steering wheel control method for intelligent driving provided in an embodiment of the present invention.

[0010] Figure 2 This is a schematic diagram of exemplary hardware and software components of an autonomous steering wheel control system for intelligent driving provided in an embodiment of the present invention. Detailed Implementation

[0011] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating an embodiment of the autonomous steering wheel control method for intelligent driving provided by the present invention. The following is a detailed description of the autonomous steering wheel control method for intelligent driving.

[0012] Step S110: Perceive the real-time driving scene status during the intelligent driving process. The real-time driving scene status includes road morphology information, surrounding traffic participant action information, and environmental influencing factor information.

[0013] During the operation of an intelligent driving vehicle, it is necessary to continuously acquire the real-time driving scene status of the vehicle's surroundings to provide a basis for subsequent steering wheel control. This embodiment uses the example of an intelligent driving vehicle driving on urban roads, about to pass through an intersection with a pedestrian crossing. At this time, the perception of the real-time driving scene status needs to comprehensively cover the road conditions themselves, the dynamics of other traffic participants, and environmental factors that may affect driving.

[0014] Step S111: Activate the multi-source environmental perception device on the vehicle. The multi-source environmental perception device includes a visual perception device, a distance perception device, and an environmental state perception device. Collect raw environmental data around the intelligent driving vehicle through the multi-source environmental perception device.

[0015] In the aforementioned intersection scenario, the vehicle's multi-source environmental perception equipment is activated. This includes visual perception devices such as a high-definition camera mounted inside the windshield, side-view cameras below the left and right rearview mirrors, and a rear-view camera. These cameras collect image data of the vehicle's surroundings at a rate of 30 frames per second. Distance perception devices primarily consist of LiDAR and millimeter-wave radar. The LiDAR, mounted on the roof, scans targets within a 360-degree radius around the vehicle, acquiring information such as distance, orientation, and height. The millimeter-wave radar, installed in the front and rear bumpers, detects nearby targets. Environmental condition perception devices include external sensors such as temperature, humidity, light, and rain sensors, used to collect information on current ambient temperature and humidity, light intensity, and rainfall. Through these devices, the raw environmental data collected includes image data of the intersection and its surroundings, distance and motion data of various targets, and data on ambient temperature, humidity, light intensity, and rainfall.

[0016] Step S112: Parse and process the road-related data in the original environmental data, extract the road direction distribution information, pavement structure information and road sign information, and integrate the road direction distribution information, pavement structure information and road sign information to form road morphology information.

[0017] From the collected raw environmental data, road-related data is selected, mainly including image data containing road areas captured by visual perception devices and point cloud data of the road surface scanned by LiDAR. This data is then analyzed to extract road alignment, pavement structure, and road signage information.

[0018] Step S1121: Select road-related raw data segments from the raw environmental data, remove non-road area data contained therein, and perform edge detection-based image enhancement processing on the selected road-related raw data segments.

[0019] In a crossroads scenario, image data captured by a high-definition camera is segmented using a pre-defined image segmentation algorithm to distinguish road areas from non-road areas (such as buildings, green belts, and the sky), filtering out image data fragments containing only the road area. For point cloud data acquired by LiDAR, point cloud data belonging to non-road areas (such as pedestrians, vehicles, and trees) is removed based on features such as height and reflection intensity, retaining only the point cloud data of the road surface. Then, edge detection is performed on the filtered road-related image data fragments using the Canny edge detection algorithm. By setting an appropriate threshold, the contour information of the road edges is extracted. The images are then contrast-enhanced and sharpened to make the road details clearer for subsequent feature extraction.

[0020] Step S1122: The enhanced road-related original data fragments are processed using an image segmentation algorithm to divide the pixel connected regions. Based on the texture, color, and relative position features of each connected region with respect to the vehicle, they are classified and labeled as the main road region, the road edge region, or the road auxiliary region.

[0021] For road-related image data segments after image enhancement, a deep learning-based image segmentation algorithm (such as the U-Net model) is used for processing. This algorithm takes the enhanced image data as input, downsamples the image through an encoder to extract feature information, and then upsamples the image through a decoder to map the feature information back to the original image size, outputting the class probability of each pixel. Based on the class probability of each pixel, different connected components are defined. Next, the texture features (e.g., the main road area usually has a relatively uniform texture, while the road edge area may have obvious line textures), color features (e.g., the main road area is mostly gray or black, while road auxiliary areas such as traffic signs may have bright colors), and relative position features with vehicles (e.g., the main road area is located directly in front of and to the sides of the vehicle's direction of travel, while the road edge area is located outside the main road area) of each connected component are classified and labeled as the main road area, road edge area, or road auxiliary area.

[0022] Step S1123: Analyze the data of the main road area, extract the road's extension direction information, curvature information and bifurcation structure information, and integrate the road's extension direction information, curvature information and bifurcation structure information to form the road's orientation distribution information.

[0023] Detailed analysis is performed on the data of the marked main road areas to extract the road orientation and distribution information.

[0024] For example, step S1123-1: perform pixel-level analysis on the data of the main road area, identify and fit the geometric center line of the main road area as a digital representation of the road extension trajectory.

[0025] In a crossroads scenario, pixel-level analysis is performed on the image data of the main road area. Straight lines are detected using Hough transform, and combined with road width information, the approximate direction of the road is determined. Then, the least squares method is used to fit the detected straight and curved segments to obtain the geometric centerline of the main road area. For example, for straight sections at a crossroads, the fitted geometric centerline is a straight line; for turning sections, the fitted geometric centerline is a smooth curve. This geometric centerline accurately reflects the road's extension trajectory and is the foundation for subsequent extraction of extension direction, curvature, and bifurcation structure information.

[0026] Step S1123-2: Select multiple evenly distributed feature points along the center line, record the coordinate information of each feature point, and determine the extension direction of the road by the trend of the change of the coordinate information.

[0027] Along the fitted geometric centerline, multiple feature points are selected at preset intervals (e.g., every 1 meter). The horizontal and vertical coordinates of each feature point are recorded in the image coordinate system. The direction of road extension is determined by analyzing the changing trends of these feature point coordinates. For example, on an east-west straight section of a crossroads, the vertical coordinate of the feature points remains relatively constant, while the horizontal coordinate gradually increases or decreases, indicating that the road extends in the east-west direction. On a north-south straight section, the horizontal coordinate of the feature points remains relatively constant, while the vertical coordinate gradually increases or decreases, indicating that the road extends in the north-south direction. For curved sections, both the horizontal and vertical coordinates of the feature points change simultaneously. By calculating the direction angle of the line connecting adjacent feature points, the direction of road extension at that location can be determined.

[0028] Step S1123-3: Calculate the angle between the line connecting adjacent feature points and the preset reference direction, determine the change pattern of the road extension direction based on the change of the angle, and form the road extension direction information.

[0029] The default reference direction is due north, defined by the positive direction of the vertical axis of the image coordinate system. The angle between the line connecting adjacent feature points and due north is calculated, ranging from 0 to 360 degrees. For example, on a straight road segment running east-west, the angle between the line connecting adjacent feature points and due north is 90 or 270 degrees; on a straight road segment running north-south, the angle is 0 or 180 degrees. For curved road segments, as feature points move, the angle between the line connecting adjacent feature points and due north gradually changes. By recording the values ​​of these angles and their corresponding feature point positions, the road's extension direction information is formed. This extension direction information can describe in detail the road's extension direction and its changes at different locations.

[0030] Step S1123-4: Identify the curved portion of the centerline, determine the coordinates of the starting and ending points of the curve, calculate the arc length and curvature change of the curved portion, and determine the degree of curvature of the road based on the value of the curvature change.

[0031] At intersections with curves, the road centerline will bend. By analyzing the changes in the curvature of the centerline, the starting and ending points of the curve can be identified. Specifically, the curvature of each point on the centerline is calculated. When the curvature value increases from near zero and exceeds a preset curvature threshold, that point is the starting point of the curve; when the curvature value gradually decreases and approaches zero again, that point is the ending point of the curve. The coordinates of the starting and ending points of the curve are recorded, and then the arc length of the curve is calculated, which is the length of the centerline from the starting point to the ending point. Simultaneously, the change in curvature of the curve is calculated, that is, the difference between the maximum and minimum curvature values. The magnitude of the curvature change indicates the degree of road curvature; a larger change in curvature indicates a more severe curvature.

[0032] Step S1123-5: Divide the curvature degree level according to the preset curvature threshold range, assign a corresponding curvature degree level to each curved part according to the calculated curvature change result, record the position of all curved parts and their corresponding curvature degree level, and form the curvature degree information of the road.

[0033] The preset curvature threshold range categorizes curvature into three levels: slight curvature, moderate curvature, and severe curvature. For example, a curvature change value between 0 and 0.05 indicates slight curvature, between 0.05 and 0.1 indicates moderate curvature, and greater than 0.1 indicates severe curvature. Based on the calculated curvature change of the curved section, a corresponding curvature level is assigned to each curved section. For example, the right-turn section at an intersection has a curvature change value of 0.08, which falls under the category of moderate curvature. The coordinates of the start and end points of this curved section, along with its corresponding curvature level, are recorded to form the road's curvature information.

[0034] Step S1123-6: Scan the data of the main area of ​​the road, identify the location where the center line of the road branches, determine the coordinate information of the fork point, and analyze the extension direction of each branch at the fork point.

[0035] In a crossroads scenario, the road centerline branches at the intersection. By scanning the centerline data of the main road area, when a centerline is detected to split into two or more branches, that location is the fork point. The coordinates of the fork point are determined, i.e., its horizontal and vertical coordinates in the image coordinate system. Then, the extension direction of each branch at the fork point is analyzed, similar to steps S1123-2 and S1123-3, calculating the angle between the line connecting adjacent feature points on each branch and a preset reference direction to determine the extension direction of each branch. For example, the center of the crossroads is the fork point, with two branches extending in the east-west and north-south directions, respectively.

[0036] Step S1123-7: Determine the road width and length of each branch, distinguish between main branches and secondary branches, and record the angle relationship and connection method between the main branches and the secondary branches.

[0037] For each branch at the fork in the road, the road width of each branch is calculated by analyzing the image data of the main road area. The road width can be obtained by measuring the distance between the two edges of the branch, and the average of the measurements from multiple locations is taken as the width of that branch. Simultaneously, the length of the branch is determined based on its direction and length, i.e., the distance from the fork in the road to its end. Based on the road width and length, primary and secondary branches are distinguished; generally, the wider and longer road is the primary branch, and vice versa. For example, in a crossroads, the east-west road may be the primary branch, and the north-south road the secondary branch. The angular relationship between the primary and secondary branches, i.e., the angle between the centerlines of the two branches, and their connection method at the fork in the road (e.g., crossroads, T-junctions, etc.), is recorded.

[0038] Step S1123-8: Integrate the coordinate information of the bifurcation point, the extension direction of each branch, the angle relationship between the main branch and the secondary branch, and the connection method to form the bifurcation structure information of the road.

[0039] The determined bifurcation point coordinates, the extension directions of each branch, the angle between the main branch and the secondary branch, and the connection method are integrated to form the road bifurcation structure information. For example, in a crossroads scenario, the bifurcation structure information includes the bifurcation point coordinates as (x0, y0), the east-west main branch extension directions as 90 degrees and 270 degrees, the north-south secondary branch extension directions as 0 degrees and 180 degrees, the angle between the main branch and the secondary branch as 90 degrees, and the connection method as a cross intersection.

[0040] Step S1123-9: Perform a continuity check on the extension direction information of the road, and achieve a smooth connection of the extension directions between road segments by adjusting the description method.

[0041] After integrating the road extension direction information, a continuity check is required. This involves checking whether the extension directions of adjacent road segments transition smoothly. If the change in extension direction between adjacent road segments is too large, exceeding a preset continuity threshold (e.g., the angle between the extension directions of adjacent road segments exceeds 30 degrees), the description of the extension direction information needs to be adjusted. For example, if there is a small curve between two straight road segments, the original extension direction information might show the straight segment's direction as 90 degrees, while the straight segment after the curve's direction is 120 degrees, with an angle of 30 degrees, which does not exceed the threshold and requires no adjustment. However, if the angle is 40 degrees, the description of the curved road segment's extension direction needs to be refined, adding a description of the intermediate transition direction to achieve a smooth connection between the extension directions of the road segments.

[0042] Step S1123-10: The extension direction information, curvature information and bifurcation structure information of the road are associated and encoded in the order of road mileage to construct structured road direction distribution information data.

[0043] Based on the road mileage sequence of the vehicle's journey, the road's extension direction information, curvature information at each location, and bifurcation structure information are associated. For example, starting from the vehicle's current position, there is first a 100-meter-long east-west straight road segment (extension direction information), followed by a moderately curved right-turn segment (curvature information), and then the bifurcation point at the intersection (bifurcation structure information). This information is encoded using a structured data format, such as JSON, storing the start and end mileage, extension direction, curvature level (if any), and bifurcation structure (if any) of each road segment in corresponding fields to construct structured road alignment distribution information data.

[0044] Step S1124: Extract details from the road surface data, identify the road surface material type information, smoothness information and damage information, and integrate the road surface material type information, smoothness information and damage information to form road surface structure information.

[0045] In intersection scenarios, image data and LiDAR point cloud data of the main road area are processed to extract detailed road surface information. For road material type information, by analyzing the texture and color features of the road surface in the image, combined with the reflection intensity information of the LiDAR point cloud, the road surface is identified as asphalt, cement, or other materials. For example, asphalt roads are typically dark black with a relatively uniform texture and low reflection intensity; cement roads are lighter in color, may have obvious joint textures, and have higher reflection intensity. For smoothness information, the elevation change of the road surface is calculated using LiDAR point cloud data, and the smoothness of the road surface is assessed by statistically analyzing the standard deviation of the elevation data; the smaller the standard deviation, the smoother the road surface. For damage information, image recognition algorithms are used to detect whether there are cracks, potholes, repair marks, etc., and the location, type, and size of the damage are recorded. The identified road material type information, smoothness information, and damage information are integrated to form road structure information.

[0046] Step S1125: The data of the road-adjacent area is identified and processed to capture traffic sign information, lane line information and road marking text information on the road, and the traffic sign information, lane line information and road marking text information on the road are integrated to form road identification information.

[0047] For data on roadside areas, object detection algorithms (such as YOLO) are used to process the image data, identifying traffic signs such as traffic lights, stop signs, speed limit signs, and turn indicators, recording the location, type, and content of each sign. For lane line information, edge detection and line fitting algorithms are used to identify the position, color (white or yellow), type (solid or dashed), and number of lane lines. For example, an intersection entrance may have three lane lines: a left-turn lane, a straight-ahead lane, and a right-turn lane. For road marking text information, Optical Character Recognition (OCR) technology is used to identify text markings on the road surface, such as "left turn," "straight," "right turn," and "bus lane," recording the text content and its location. The traffic sign information, lane line information, and road marking text information are then integrated to form road signage information.

[0048] Step S1126: Traverse the road segment extension direction sequence in the road direction distribution information. If the rate of change of the extension direction between adjacent road segments exceeds the preset continuity threshold, generate transition direction information by smooth interpolation based on the curvature of the preceding and following road segments, and replace the original contradictory information.

[0049] The system iterates through the sequence of road segment extension directions in the road alignment distribution information, calculating the rate of change of the extension direction of adjacent road segments. For example, if the extension direction of one road segment is 90 degrees and the extension direction of the next road segment is 130 degrees, the rate of change is 40 degrees. The preset continuity threshold is 30 degrees. Since 40 degrees exceeds the threshold, it indicates that the change in the extension direction of adjacent road segments is too large, creating a contradiction. In this case, a smooth interpolation method is used to generate transition direction information based on the curvature of the preceding and following road segments. For example, several transition direction points are inserted between preceding and following road segments, gradually transitioning the extension direction from 90 degrees to 130 degrees. The direction of each transition point is calculated using the curvature of the preceding and following road segments to ensure a smooth change in direction. Then, the generated transition direction information replaces the original contradictory information.

[0050] Step S1127: Perform data fusion processing on the structural data of the same road surface area collected by different sensing devices. When the difference between the data exceeds the preset fault tolerance threshold, use the preset confidence weighting algorithm to calculate the calibrated road surface structure information.

[0051] In a crossroads scenario, visual sensing devices and LiDAR may collect structural data of the same road surface area. For example, a visual camera might capture road surface smoothness information, while LiDAR might measure road surface elevation data. To fuse this data, the differences between the data collected by different devices are first calculated. If the difference is within a preset tolerance threshold (e.g., 5%), the data is considered consistent, and the average value is taken as the final result. If the difference exceeds the tolerance threshold, a confidence-weighted algorithm is used for calibration. Different confidence weights are assigned to different sensing devices based on their performance and the current environmental conditions. For example, in good lighting conditions, the visual camera has a higher confidence weight, while the LiDAR has a lower confidence weight; in rainy or foggy weather, the LiDAR has a higher confidence weight. Then, the data collected by each device is multiplied by its corresponding confidence weight, summed, and then divided by the total weight to obtain the calibrated road surface structure information.

[0052] Step S1128: Compare the identified road sign information with a preset list of key signs. If any key sign in the list of key signs is not identified within a preset sensing range, the corresponding sensing device is triggered to re-collect and identify the area adjacent to the road.

[0053] The pre-set list of key signs includes important traffic signs such as traffic lights, stop lines, and lane direction arrows in intersection scenarios. The identified road sign information is compared with this list to check for any unidentified key signs. For example, there should be a traffic light sign within the pre-set perception range (e.g., within 100 meters in front of the vehicle), but it is not present in the current identification results. In this case, the corresponding visual perception device (e.g., a forward-facing camera) is triggered to re-acquire image data of the road's surrounding area and use a higher-precision object detection algorithm for identification, ensuring that key signs are accurately identified.

[0054] Step S1129: Integrate the road orientation distribution information, road surface structure information and road sign information after verification, validation and verification to form road morphology information that reflects the actual road conditions.

[0055] The road alignment distribution information, pavement structure information, and road sign information processed through the above steps undergo final verification, validation, and inspection. This includes verifying the completeness and accuracy of each piece of information, confirming consistency between information (e.g., the number of lanes in road sign information should match the road width in road alignment distribution information), and verifying that the data format meets requirements. All verified and inspected information is then integrated and stored using a unified data structure to form road morphology information that comprehensively reflects the actual road conditions at the intersection.

[0056] Step S113: Identify and process the traffic participant-related data in the original environmental data, distinguish different types of traffic participants, track the location movement information and behavior information of various types of traffic participants, and integrate the location movement information and behavior information of various types of traffic participants to form the surrounding traffic participant action information.

[0057] In a crossroads scenario, the raw environmental data related to traffic participants includes image data captured by visual cameras containing pedestrians, vehicles, bicycles, and other traffic participants, as well as target data detected by LiDAR and millimeter-wave radar. The above data undergoes identification processing. First, target detection algorithms are used to distinguish different types of traffic participants, such as pedestrians, cars, buses, and bicycles. Then, multi-target tracking algorithms (such as Kalman filtering) are employed to track each type of traffic participant, obtaining their position coordinates at different time points (in the vehicle coordinate system or the world coordinate system), thus obtaining position movement information. Simultaneously, by analyzing the posture changes and movement trajectories of traffic participants, their behavioral information is identified, such as whether pedestrians are crossing the road, whether vehicles are slowing down, or turning. The position movement information and behavioral information of various traffic participants are integrated to form the surrounding traffic participant action information. For example, at an intersection, a pedestrian is crossing the crosswalk from north to south. Their position movement information is from (x1, y1) to (x2, y2), and their action information is walking. A car is traveling from east to west and preparing to turn right. Its position movement information is moving along a curved trajectory, and its action information is slowing down and turning right.

[0058] Step S114: Collect and process the environmental-related data in the raw environmental data, capture information on natural environmental changes and artificial environmental interference that affect driving, and integrate the natural environmental change information and the artificial environmental interference information to form environmental influencing factor information.

[0059] The environmental data in the raw environmental data comes from environmental state sensing devices, including data on temperature, humidity, light intensity, and rainfall. This data is collected and processed to capture information on changes in the natural environment, such as whether the current light intensity is too strong or too weak, whether there is rainfall and its intensity, and whether the temperature is too high or too low. Simultaneously, it captures information on artificial environmental disturbances, such as whether there are construction areas around the road, whether traffic lights are functioning properly, and whether there are temporary traffic controls. The information on natural environmental changes and artificial environmental disturbances is integrated to form environmental impact factor information. For example, in the current intersection scenario, the environmental impact factor information includes moderate light intensity, no rainfall, a construction area on the east side of the road, and traffic lights functioning properly.

[0060] Step S115: The road morphology information, the action information of surrounding traffic participants, and the environmental influencing factors are correlated and integrated to form initial driving scenario state data.

[0061] The road morphology information, surrounding traffic participant action information, and environmental influencing factor information obtained in the previous steps are linked and integrated according to timestamps. For example, at a certain point in time, the road morphology information includes the road direction, pavement structure, and road signage at the intersection; the surrounding traffic participant action information includes the location and behavior of each traffic participant at that time; and the environmental influencing factor information includes the current environmental conditions. This information is integrated into a single data structure to form initial driving scenario state data, which reflects the driving scenario state at that specific point in time.

[0062] Step S116: Establish a data update queue, and write the supplementary information parsed from the newly collected environmental raw data into the corresponding information segment of the initial driving scenario state data in the order of timestamps, overwriting or appending it to the outdated information segment.

[0063] A data update queue is established to store supplementary information parsed from newly collected environmental raw data. When new environmental raw data is collected and parsed, the resulting supplementary information is added to the data update queue in timestamp order. Then, the supplementary information in the queue is written to the corresponding information segment of the initial driving scenario state data in timestamp order. Outdated information segments (such as previous traffic participant location information) are overwritten with new supplementary information; newly added information (such as newly appearing traffic participants) is appended to the initial driving scenario state data. This method ensures that the initial driving scenario state data is updated in real time, reflecting the latest driving scenario conditions.

[0064] Step S117: Remove redundant information unrelated to steering wheel control from the initial driving scenario state data, retain key information that directly affects steering operation, perform feature encoding processing on the retained key information, and convert various types of information into a unified feature vector or data structure expression to form a standardized driving scenario state fragment.

[0065] The initial driving scenario state data is filtered to remove redundant information unrelated to steering wheel control, such as detailed information about surrounding buildings and distant traffic participants unrelated to the vehicle's direction of travel. Key information directly affecting steering operations is retained, such as the current road curvature, lane position, the position and behavior of traffic participants ahead, traffic light status, and environmental rainfall conditions. This retained key information undergoes feature encoding, converting different types of information into unified feature vectors or data structures. For example, the road curvature level is converted into a numerical vector, the position coordinates of traffic participants are converted into relative coordinate vectors relative to the vehicle, and the traffic light status is converted into a binary vector. Through feature encoding, standardized driving scenario state fragments are formed, facilitating subsequent processing and analysis.

[0066] Step S118: Connect multiple consecutive standardized driving scenario state segments in chronological order to form a real-time driving scenario state. Verify the completeness of the information in the real-time driving scenario state. If there is missing information, return to re-collect the corresponding original environmental data to ensure that the road morphology information, the action information of the surrounding traffic participants, and the information of environmental influencing factors are all complete.

[0067] Multiple consecutive, standardized driving scenario state segments are concatenated in chronological order to form a time-series data set, i.e., the real-time driving scenario state. This time-series data reflects the dynamic changes of the driving scenario over a period of time. The completeness of the real-time driving scenario state information is verified, checking whether road morphology information, surrounding traffic participant action information, and environmental influencing factor information are all present. For example, it checks whether there are any periods without traffic participant action information, or whether a key marker in the road morphology information is missing. If information is missing, the process returns to step S111, restarts the multi-source environmental perception device to collect the corresponding raw environmental data, and performs parsing and processing until all information in the real-time driving scenario state is complete and without any missing elements.

[0068] Step S120: Retrieve the historical driving operation records of the intelligent driving vehicle, which include steering wheel adjustment operation information under scenarios similar to the real-time driving scenario.

[0069] The storage system of intelligent driving vehicles contains a large number of historical driving operation records. These records include steering wheel adjustment information under different driving scenarios, such as steering angle and steering rate. In the current intersection scenario, it is necessary to retrieve historical driving operation records from scenarios similar to the current driving situation. For example, driving operation records from previous instances at the same intersection with similar traffic participants and environmental conditions. By querying the index in the storage system and matching based on scenario features (such as road shape, traffic participant actions, and environmental factors), historical driving operation records for similar scenarios can be found and retrieved.

[0070] Step S130: Generate steering wheel control reference parameters based on the real-time driving scenario state and the historical driving operation record. The steering wheel control reference parameters include steering angle adjustment range and steering rate constraint information.

[0071] By combining the real-time driving scenario status and retrieved historical driving operation records, steering wheel control reference parameters are generated. These steering wheel control reference parameters provide an initial reference range and constraints for subsequent steering wheel control.

[0072] Step S131: Perform scene classification processing on the historical driving operation records. Divide the historical driving operation records into multiple scene categories according to the scene characteristics in the historical driving operation records. Each scene category corresponds to a specific driving scenario.

[0073] The retrieved historical driving operation records are classified into scenarios. First, scene features are extracted from each historical driving operation record, including road morphology features (such as road type, curvature, and branching structure), traffic participant features (such as traffic participant type, number, and behavior), and environmental features (such as weather, lighting, and whether there is construction). Then, a clustering algorithm (such as K-means clustering) is used to perform cluster analysis on the above scene features, grouping historical driving operation records with similar scene features into the same scene category. Each scene category corresponds to a specific driving scenario, such as urban road intersection scenarios, highway straight road scenarios, and rural road curved road scenarios. For example, all historical driving operation records containing scene features such as intersections, pedestrian crossings, and traffic lights are classified into the "urban road intersection scenario" category.

[0074] Step S132: Extract the core scene features of the real-time driving scene state, which include core features of road morphology, core features of traffic participant actions, and core features of environmental impact.

[0075] Core scene features are extracted from the real-time driving scenario. Core road morphology features include the current road type (intersection), main direction of extension, degree of curvature, and branching structure; core traffic participant actions features include the types (pedestrians, cars), numbers, and key behaviors (pedestrians crossing the street, vehicles turning); core environmental impact features include current weather conditions (no rain), light intensity (moderate), and the presence of construction areas (construction on the east side). These core scene features summarize the main characteristics of the real-time driving scenario and are used for matching with historical scene categories.

[0076] Step S133: Compare the core scene features with the typical scene features of each scene category, and select the target scene category with the highest similarity to the real-time driving scene state.

[0077] The core scene features extracted from the real-time driving scene status are compared with the typical scene features of each scene category to find the target scene category with the highest similarity.

[0078] Step S1331: Extract typical scene features for each scene category. The typical scene features for each scene category include typical road morphology features, typical traffic participant action features, and typical environmental impact features for that scene category.

[0079] For each scenario category, typical scenario features are extracted by statistically analyzing the scenario characteristics of all historical driving operation records under that category. For example, the typical road morphology features of the "urban road intersection scenario" category are crossroads, traffic lights, and clear lane lines; the typical features of traffic participant actions are pedestrians crossing the street, vehicles turning, and vehicles going straight; and the typical features of environmental impact are moderate lighting and no or little rain.

[0080] Step S1332: Assign corresponding weight coefficients to the core features of road morphology, traffic participant actions, and environmental impact in the core scene features. The weight coefficients are set according to the degree of influence of different features on driving operations.

[0081] Based on the degree of influence of different core features on steering wheel control operations, weight coefficients are assigned to the core features of road morphology, traffic participant actions, and environmental impact. For example, the core feature of road morphology has the greatest impact on steering operations, with a weight coefficient of 0.5; the core feature of traffic participant actions has the second greatest impact, with a weight coefficient of 0.3; and the core feature of environmental impact has a relatively smaller impact, with a weight coefficient of 0.2.

[0082] Step S1333: Assign the same weight coefficient to the typical road morphology features, typical traffic participant action features, and typical environmental impact features in the typical scene features of each scene category.

[0083] In line with the weighting coefficients of the core scene features, the typical features of road morphology, typical features of traffic participant actions, and typical features of environmental impact in each scene category are also assigned weighting coefficients of 0.5, 0.3, and 0.2, respectively.

[0084] Step S1334: Calculate the similarity between the feature vector of the core road morphology feature in the core scene features and the feature vector of the typical road morphology feature in the typical scene features of each scene category, and obtain the road morphology matching degree value; the matching degree value of traffic participant actions and the matching degree value of environmental impact are obtained using the same similarity calculation method.

[0085] The core features of road morphology and the typical features of road morphology for each scene category are represented as feature vectors. For example, the core feature vector of road morphology can include numerical representations of dimensions such as road type, curvature level, and bifurcation structure. The cosine similarity algorithm is used to calculate the similarity between the core scene feature vector and the typical feature vector of each scene category. The cosine similarity is calculated by dividing the dot product of two vectors by the product of their magnitudes; the closer the value is to 1, the higher the similarity. Following the same method, the similarity between the core features of traffic participant actions and the typical features of traffic participant actions for each scene category is calculated to obtain the traffic participant action matching degree; the similarity between the core features of environmental impact and the typical features of environmental impact for each scene category is calculated to obtain the environmental impact matching degree.

[0086] Step S1335: Calculate the matching degree between the core features of traffic participant actions in the core scene features and the typical features of traffic participant actions in the typical scene features of each scene category, and obtain the traffic participant action matching degree value.

[0087] The specific calculation process is the same as the calculation method of the road form matching degree value in step S1334. After converting the core features of traffic participants' actions and the typical features of traffic participants' actions in each scene category into feature vectors, the cosine similarity algorithm is used to calculate the similarity to obtain the traffic participant action matching degree value.

[0088] Step S1336: Calculate the matching degree between the core environmental impact features in the core scene features and the typical environmental impact features in the typical scene features of each scene category, and obtain the environmental impact matching degree value.

[0089] Similarly, the core features of environmental impact and the typical features of environmental impact for each scenario category are converted into feature vectors, and the similarity is calculated using the cosine similarity algorithm to obtain the environmental impact matching degree value.

[0090] Step S1337: Multiply the road morphology matching degree value, the traffic participant action matching degree value, and the environmental impact matching degree value by the corresponding weight coefficients to obtain the weighted matching degree value of each feature.

[0091] The road morphology matching degree values, traffic participant action matching degree values, and environmental impact matching degree values ​​obtained in steps S1334, S1335, and S1336 are multiplied by their corresponding weighting coefficients of 0.5, 0.3, and 0.2, respectively, to obtain the weighted matching degree values ​​for each feature. For example, a road morphology matching degree value of 0.8, multiplied by a weighting coefficient of 0.5, yields a weighted matching degree value of 0.4; a traffic participant action matching degree value of 0.7, multiplied by 0.3, yields 0.21; and an environmental impact matching degree value of 0.6, multiplied by 0.2, yields 0.12.

[0092] Step S1338: Sum the weighted matching degree values ​​of each feature of each scene category to obtain the comprehensive matching degree value of each scene category. The comprehensive matching degree value comprehensively reflects the similarity relationship between the typical scene features and the core scene features of the scene category.

[0093] The weighted matching scores for road morphology, traffic participant actions, and environmental impact for each scene category are summed to obtain the overall matching score for that scene category. For example, in the example above, the sum of the weighted matching scores is 0.73, meaning the overall matching score for that scene category is 0.73.

[0094] Step S1339: Sort the comprehensive matching degree values ​​of all scene categories, arrange each scene category in descending order of the comprehensive matching degree values, and select the scene category with the largest comprehensive matching degree value in the sorting results as the target scene category. The typical scene features of the target scene category have the highest similarity to the core scene features of the real-time driving scene state.

[0095] The overall match scores of all scene categories are sorted in descending order. The scene category with the highest overall match score at the top of the list is selected as the target scene category. For example, after sorting, the "city road intersection scene" category has the highest overall match score of 0.85, so it is selected as the target scene category.

[0096] Step S134: Retrieve all historical driving operation records under the target scenario category, and extract the steering wheel adjustment operation information, which includes steering angle data and steering rate data.

[0097] After determining the target scenario category as "urban road intersection scenario", retrieve all historical driving operation records under this category. Extract steering wheel adjustment operation information from these records, including steering angle data for each steering operation (such as 30 degrees to the left, 25 degrees to the right, etc.) and steering rate data (such as 10 degrees per second, 15 degrees per second, etc.).

[0098] Step S135: Perform statistical analysis on the steering angle data to determine the common variation range of steering angle under different driving conditions, and adjust the boundary range of the common variation range based on the preset steering strategy corresponding to the road morphology information of the real-time driving scenario.

[0099] Statistical analysis is performed on the extracted steering angle data to calculate the average, standard deviation, maximum, and minimum values ​​of the steering angle under different driving conditions (such as low-speed passage through intersections, high-speed passage through intersections, and turning), thereby determining the common variation range of the steering angle. For example, when turning right at a low speed through an intersection, the common variation range of the steering angle is 15 to 45 degrees. Then, based on the preset steering strategy corresponding to the road morphology information of the real-time driving scenario (such as a smaller turning radius at the current intersection requiring a larger steering angle), the boundary range of the common variation range is adjusted. For example, the original common variation range of 15 to 45 degrees is adjusted to 20 to 50 degrees.

[0100] Step S136: Perform trend analysis processing on the steering rate data to identify the pattern of steering rate change over time, and optimize the description of the steering rate change pattern based on the driving safety model corresponding to the action information of surrounding traffic participants in the real-time driving scenario.

[0101] Trend analysis is performed on steering rate data by plotting a curve of steering rate versus time to identify its changing patterns, such as a gradual increase in speed at the start of a turn, a stable speed during the turn, and a gradual decrease at the end of the turn. Then, by combining real-time driving scenario information on the actions of surrounding traffic participants (e.g., the need to slow down when pedestrians are crossing the street), the description of the steering rate's changing patterns is optimized based on a driving safety model. For example, at the start of a turn, the rate increase is slowed down to ensure sufficient time to observe the dynamics of traffic participants; during the turn, the stable speed is reduced to avoid excessively fast turns that could endanger surrounding traffic participants.

[0102] Step S137: Determine the steering angle adjustment range based on the adjusted common variation range, wherein the steering angle adjustment range covers the angle range corresponding to the steering operation under the target scene category.

[0103] The common range of steering angle changes after adjustment in step S135 is determined as the steering angle adjustment range. For example, if the common range of changes after adjustment is 20 degrees to 50 degrees, then the steering angle adjustment range is 20 degrees to 50 degrees. This steering angle adjustment range covers the angle range corresponding to the right turn operation under the target scene category.

[0104] Step S138: Determine steering rate constraint information based on the optimized steering rate variation pattern, wherein the steering rate constraint information regulates the rate variation of steering operation.

[0105] Based on the optimized steering rate variation pattern obtained in step S136, steering rate constraint information is determined. This steering rate constraint information includes the maximum and minimum speeds at each stage of the steering process, as well as the acceleration limits for speed changes. For example, the maximum speed at the beginning of the steering process is 10 degrees per second, the stable speed during the steering process is 15 degrees per second, the minimum speed at the end of the steering process is 5 degrees per second, and the acceleration for speed changes does not exceed the square of 5 degrees per second.

[0106] Step S139: Associate and bind the steering angle adjustment range and the steering rate constraint information to form initial steering wheel control reference parameters.

[0107] The determined steering angle adjustment range (e.g., 20 to 50 degrees) and steering rate constraint information (e.g., rate limits at each stage) are linked and bound together to form the initial steering wheel control reference parameters. For example, when the steering angle is 20 degrees, the corresponding steering rate constraint is 8 degrees per second in the initial stage, 12 degrees per second in the stable stage, and 4 degrees per second in the final stage; when the steering angle is 50 degrees, the corresponding steering rate constraint is 10 degrees per second in the initial stage, 15 degrees per second in the stable stage, and 5 degrees per second in the final stage.

[0108] Step S1310: Based on the environmental influencing factor information of the real-time driving scenario state, query the preset environment-parameter adjustment mapping table, obtain the parameter adjustment amount, update the value in the initial steering wheel control reference parameter according to the parameter adjustment amount, and finally obtain the steering wheel control reference parameter.

[0109] An environment-parameter adjustment mapping table is pre-defined, recording the steering wheel control baseline parameter adjustments corresponding to different environmental influencing factors. For example, when rainfall is included in the environmental influencing factors, the steering angle adjustment range should be reduced by 10%, and the rate values ​​in the steering rate constraints should be reduced by 20%. In the current intersection scenario, the environmental influencing factors are moderate light intensity, no rainfall, and a construction area on the east side of the road. Querying the environment-parameter adjustment mapping table, the parameter adjustments for a construction area are found to be a 5% reduction in the steering angle adjustment range and a 10% reduction in the rate values ​​in the steering rate constraints. Based on these parameter adjustments, the values ​​of the steering angle adjustment range and steering rate constraints in the initial steering wheel control baseline parameters are updated. For example, the initial steering angle adjustment range is 20 to 50 degrees, which becomes 19 to 47.5 degrees after a 5% reduction; the initial stable rate in the steering rate constraints is 15 degrees per second, which becomes 13.5 degrees per second after a 10% reduction. After the update, the final steering wheel control baseline parameters are obtained.

[0110] Step S140: Collect real-time driving status data of the intelligent driving vehicle. The real-time driving status data includes vehicle driving posture information and wheel steering feedback information. Combine the real-time driving status data to dynamically correct the steering wheel control reference parameters to obtain real-time steering wheel control parameters.

[0111] During the operation of an intelligent driving vehicle, real-time driving status data is collected, and this data is used to dynamically correct the steering wheel control reference parameters to adapt to changes in the actual driving status of the vehicle and obtain more accurate real-time steering wheel control parameters.

[0112] Step S141: Start the driving status acquisition device on the vehicle. The driving status acquisition device includes an attitude sensor and a steering feedback sensor. The driving status acquisition device continuously collects raw data of the vehicle's driving status.

[0113] The vehicle's driving status acquisition equipment is activated. The attitude sensors, including gyroscopes and accelerometers, are installed near the vehicle's center of gravity to collect angular velocity and acceleration data. Steering feedback sensors include angle and torque sensors mounted on the steering column to collect steering wheel angle and torque data, and wheel speed and steering angle sensors mounted on the wheels to collect wheel speed and actual steering angle data. These sensors continuously collect raw data on the vehicle's driving status, such as yaw rate, longitudinal acceleration, lateral acceleration, steering wheel angle, steering torque, wheel speed, and actual wheel steering angle.

[0114] Step S142: Analyze and process the attitude-related data in the original driving state data, extract the vehicle's tilt angle information, pitch angle information and rollover trend information, and integrate the tilt angle information, pitch angle information and rollover trend information to form vehicle driving attitude information.

[0115] The angular velocity and acceleration data collected by the attitude sensors are analyzed and processed. The rotation angles of the vehicle around each axis are calculated using gyroscope data, yielding tilt angle information (rotation angle around the longitudinal axis) and pitch angle information (rotation angle around the lateral axis). Using accelerometer data and a vehicle dynamics model, the vehicle's rollover coefficient is calculated. This rollover coefficient reflects the vehicle's rollover tendency; a higher rollover coefficient indicates a higher risk of rollover. The tilt angle, pitch angle, and rollover tendency information are integrated to form the vehicle's driving attitude information. For example, if the current vehicle tilt angle is 2 degrees, pitch angle is 1 degree, and rollover coefficient is 0.3, it indicates that the vehicle's driving attitude is relatively stable and the risk of rollover is low.

[0116] Step S143: Process the steering feedback related data in the original driving state data, capture the actual steering angle information and steering response delay information of the wheel, and integrate the actual steering angle information and the steering response delay information to form wheel steering feedback information.

[0117] The data collected by the steering feedback sensor is processed. The actual steering angle information of the wheel is obtained from the wheel steering angle sensor, reflecting the current steering angle of the wheel. Steering response delay information, i.e., the time delay before the wheel begins to turn after the steering wheel is turned, is calculated by comparing the time difference between the steering wheel steering angle and the actual wheel steering angle. The actual steering angle information and the steering response delay information are integrated to form the wheel steering feedback information. For example, if the current steering wheel steering angle is 30 degrees and the actual wheel steering angle is 28 degrees, the steering response delay is 0.2 seconds.

[0118] Step S144: Encapsulate the vehicle driving posture information and the wheel steering feedback information according to a preset data structure and align them with timestamps to generate real-time driving status data.

[0119] According to a pre-defined data structure, vehicle driving posture information and wheel steering feedback information are encapsulated. This data structure contains multiple fields, such as timestamp, tilt angle, pitch angle, rollover coefficient, actual wheel steering angle, and steering response delay. The encapsulated information is timestamped to ensure that the vehicle driving posture information and wheel steering feedback information correspond at the same point in time, thereby generating real-time driving status data. For example, at timestamp t1, the vehicle driving posture information shows a tilt angle of 2 degrees, a pitch angle of 1 degree, and a rollover coefficient of 0.3; the wheel steering feedback information shows an actual wheel steering angle of 28 degrees and a steering response delay of 0.2 seconds.

[0120] Step S145: Extract the steering angle adjustment range from the steering wheel control reference parameters, and evaluate the vehicle driving stability under the current steering angle adjustment range by combining the vehicle driving posture information and the vehicle dynamics model.

[0121] Extract the steering angle adjustment range from the steering wheel control reference parameters, such as 19 degrees to 47.5 degrees. Input vehicle driving posture information (such as tilt angle, pitch angle, and rollover coefficient) into the vehicle dynamics model. This model, based on parameters such as vehicle mass, wheelbase, track width, and center of gravity height, can simulate the vehicle's driving state under different steering angles. Through simulation calculations, assess whether the vehicle will experience instability phenomena such as sideslip or rollover when steering within the current steering angle adjustment range, thereby evaluating vehicle driving stability.

[0122] Step S146: If the assessment results indicate a risk of instability, the upper and lower limits of the steering angle adjustment range are calculated and adjusted based on the vehicle driving posture information and the vehicle dynamics model, so that the adjusted steering angle adjustment range can maintain vehicle driving stability.

[0123] If the assessment results indicate a risk of instability (such as a rollover coefficient exceeding a preset threshold), the maximum and minimum steering angles required to maintain vehicle stability are calculated based on the vehicle's driving posture information and the vehicle dynamics model. For example, if the current vehicle rollover coefficient is 0.6, exceeding the preset threshold of 0.5, a rollover risk exists. The vehicle dynamics model calculates that, under the current driving posture, the maximum safe steering angle is 40 degrees, and the minimum safe steering angle is 20 degrees. Therefore, the upper limit of the steering angle adjustment range is adjusted from 47.5 degrees to 40 degrees, while the lower limit remains unchanged at 19 degrees (or adjusted according to actual conditions), ensuring that the adjusted steering angle range (19 to 40 degrees) maintains vehicle driving stability.

[0124] Step S147: Extract the steering rate constraint information from the steering wheel control reference parameters, and combine it with the wheel steering feedback information and the vehicle dynamic response model to evaluate the steering execution smoothness and timeliness under the current steering rate constraint information.

[0125] Steering rate constraint information is extracted from the steering wheel control reference parameters, such as a maximum rate of 10 degrees per second at the start of steering, 13.5 degrees per second in the stable phase, and a minimum rate of 5 degrees per second at the end of steering. Steering response delay information from the wheel steering feedback is input into the vehicle dynamic response model, which simulates the vehicle's dynamic response process under different steering rates. Through simulation, the smoothness of steering operation execution (e.g., whether there are obvious shocks or vibrations) and whether the steering operation can be completed within a specified time (timeliness) are evaluated under the current steering rate constraint.

[0126] Step S148: If the evaluation result does not meet the requirements, then based on the steering response delay information, rollover trend information and vehicle dynamic response model, the rate change gradient of the steering rate constraint information is replanned, and the boundary value of the adjusted steering angle adjustment range and the parameter value of the optimized steering rate constraint information are encapsulated according to the preset parameter combination format to generate the intermediate steering wheel control parameters.

[0127] If the evaluation results show that the smoothness or timeliness of steering execution does not meet the requirements (e.g., steering response delay leads to excessively long steering completion time), the rate change gradient of the steering rate constraint information is replanned based on the steering response delay information, rollover trend information, and vehicle dynamic response model. For example, to shorten the steering completion time, the rate increase gradient at the beginning of the steering phase is appropriately increased; considering the rollover trend, the stable rate during the steering process is reduced. The replanned steering rate constraint information may be: the rate at the beginning of the steering phase gradually increases from 5 degrees per second to 12 degrees per second, remains at 12 degrees per second in the stable phase, and gradually decreases to 4 degrees per second at the end of the steering phase. The boundary values ​​of the adjusted steering angle adjustment range (19 degrees to 40 degrees) and the parameter values ​​of the optimized steering rate constraint information are encapsulated according to a preset parameter combination format (such as a data structure containing fields such as upper and lower limits of steering angle and rate ranges for each stage) to generate intermediate steering wheel control parameters.

[0128] Step S149: In each data processing cycle, the latest real-time driving status data is input into the generation rule of the intermediate steering wheel control parameters, the parameter values ​​in the intermediate steering wheel control parameter set are recalculated and updated, and finally the real-time steering wheel control parameters are obtained.

[0129] Set a fixed data processing cycle, such as 100 milliseconds. Within each data processing cycle, repeat steps S141 to S148 above, inputting the latest collected real-time driving status data into the generation rules of the center steering wheel control parameters, and recalculating and updating the parameter values ​​in the center steering wheel control parameter set. Through continuous dynamic correction, the center steering wheel control parameters can adapt to changes in the vehicle's driving status in real time. When the center steering wheel control parameters remain stable over multiple consecutive data processing cycles (e.g., the change in parameter values ​​is less than a preset threshold), they are determined as the real-time steering wheel control parameters.

[0130] Step S1491: Set a fixed data processing cycle, and read the latest real-time driving status data when each data processing cycle arrives.

[0131] The data processing cycle is set to 100 milliseconds, and the system triggers the data processing flow periodically according to this cycle. At the end of each cycle, the latest real-time driving status data is read from the data buffer to ensure that the most up-to-date vehicle driving status information is used.

[0132] Step S1492: In each data processing cycle, extract the changes in vehicle driving posture information and wheel steering feedback information from the updated real-time driving status data.

[0133] Within each data processing cycle, the currently read real-time driving status data is compared with the real-time driving status data of the previous cycle, and the changes in vehicle driving posture information (such as changes in tilt angle, pitch angle, and rollover coefficient) and wheel steering feedback information (such as changes in actual steering angle and steering response delay) are calculated.

[0134] Step S1493: Analyze the relationship between the change in the vehicle driving posture information and the preset posture change threshold. If the change in the vehicle driving posture information exceeds the preset posture change threshold, then make targeted adjustments to the center steering wheel control parameters.

[0135] Preset attitude change thresholds, such as a tilt angle change threshold of 1 degree and a rollover coefficient change threshold of 0.1. Compare the calculated changes in vehicle driving attitude information with the corresponding thresholds. If the changes exceed the thresholds (e.g., a tilt angle change of 1.5 degrees, which is greater than the 1-degree threshold), it indicates that the vehicle driving attitude has changed significantly, and the central steering wheel control parameters need to be adjusted accordingly, such as further narrowing the steering angle adjustment range.

[0136] Step S1494: Analyze the relationship between the change in the wheel steering feedback information and the preset feedback change threshold. If the change in the wheel steering feedback information exceeds the preset feedback change threshold, optimize the relevant content of the center steering wheel control parameters.

[0137] Preset feedback change thresholds, such as a 5-degree threshold for actual steering angle change and a 0.1-second threshold for steering response delay change. Compare the change in wheel steering feedback information with the corresponding thresholds. If it exceeds the threshold (e.g., a 0.15-second change in steering response delay is greater than the 0.1-second threshold), it indicates that the response characteristics of the steering system have changed, and the steering rate constraint information in the central steering wheel control parameters needs to be optimized, such as adjusting the rate change gradient.

[0138] Step S1495: If the change in the vehicle driving posture information exceeds the preset posture change threshold, readjust the steering angle adjustment range in the center steering wheel control parameters according to the updated vehicle driving posture information in the real-time driving status data, so that the adjusted steering angle adjustment range is adapted to the new driving posture.

[0139] When the change in vehicle driving posture information exceeds the threshold, the upper and lower limits of the steering angle adjustment range are recalculated based on the updated vehicle driving posture information (such as the current tilt angle and rollover coefficient) and the vehicle dynamics model, so that the adjusted range can adapt to the new driving posture and ensure vehicle driving stability.

[0140] Step S1496: If the change in the wheel steering feedback information exceeds the preset feedback change threshold, the steering rate constraint information in the intermediate steering wheel control parameters is re-optimized based on the updated wheel steering feedback information in the real-time driving status data, so that the optimized steering rate constraint information matches the new steering response.

[0141] When the change in wheel steering feedback information exceeds the threshold, the rate change gradient of steering rate constraint information is replanned based on the updated wheel steering feedback information (such as the current steering response delay) and the vehicle dynamic response model, so that the optimized steering rate constraint information can match the new steering response characteristics and ensure the smoothness and timeliness of steering execution.

[0142] Step S1497: If the change in vehicle driving posture information and the change in wheel steering feedback information both exceed the corresponding thresholds, then simultaneously adjust the steering angle adjustment range in the intermediate steering wheel control parameters and optimize the steering rate constraint information in the intermediate steering wheel control parameters, so that each part of the intermediate steering wheel control parameters is adapted to the updated real-time driving status data.

[0143] If the changes in vehicle driving posture information and wheel steering feedback information both exceed their respective thresholds, then the steering angle adjustment range and steering rate constraint information in the center steering wheel control parameters need to be adjusted and optimized simultaneously to ensure that each part of the center steering wheel control parameters can adapt to the updated real-time driving status data.

[0144] Step S1498: After adjustment and optimization, the adjusted intermediate steering wheel control parameters are input into the vehicle dynamics simulation model. Combined with the updated real-time driving state data, steering wheel control commands are simulated and the predicted trajectory of the vehicle is calculated. The deviation between the predicted trajectory and the expected trajectory is evaluated.

[0145] After adjusting and optimizing the central steering wheel control parameters, they are input into the vehicle dynamics simulation model. Simultaneously, updated real-time driving state data is used as the initial conditions for the model to simulate and generate steering wheel control commands, and the predicted trajectory of the vehicle under these commands is calculated. The predicted trajectory is then compared with the desired trajectory (such as the trajectory planned by navigation), and the deviation between the two is calculated.

[0146] Step S1499: Repeat the above monitoring, analysis, adjustment, optimization and verification process, and dynamically adjust the center steering wheel control parameters in each data processing cycle so that the center steering wheel control parameters continuously adapt to the changes in the vehicle driving state.

[0147] Within each data processing cycle, the process of continuously monitoring changes in vehicle driving status, analyzing whether the changes exceed thresholds, adjusting and optimizing the center steering wheel control parameters, and verifying whether the adjusted parameters meet the requirements is repeated, so that the center steering wheel control parameters can continuously adapt to the dynamic changes in vehicle driving status.

[0148] Step S14910: Continuously monitor the variance of each parameter value in the intermediate steering wheel control parameter set within a preset number of cycles. If the variance is less than its corresponding stability threshold, and the deviation of the test steering command generated based on the intermediate steering wheel control parameter set in the vehicle model simulation is less than the preset tolerance, then the intermediate steering wheel control parameter set of the current cycle is determined as the real-time steering wheel control parameter.

[0149] The variance of each parameter value within the central steering wheel control parameter set is continuously monitored over a preset number (e.g., 10) data processing cycles. If the variance of each parameter value is less than the corresponding stability threshold (e.g., variance less than 0.5), it indicates that the parameter value tends to be stable. Simultaneously, test steering commands are generated based on this central steering wheel control parameter set, and the deviation is calculated in the vehicle model simulation. If the deviation is less than a preset tolerance (e.g., trajectory deviation less than 0.5 meters), then the central steering wheel control parameter set is considered to meet the control requirements and is determined as the real-time steering wheel control parameter.

[0150] Step S150: Generate autonomous steering wheel control commands based on the real-time steering wheel control parameters, and send the autonomous steering wheel control commands to the vehicle steering execution system to drive the steering wheel to complete the corresponding steering operation.

[0151] Based on the obtained real-time steering wheel control parameters, specific autonomous steering wheel control commands are generated and sent to the vehicle steering execution system to control the steering wheel to perform corresponding steering operations.

[0152] Step S151: Analyze the steering angle adjustment range in the real-time steering wheel control parameters, determine the target steering angle required in the current driving scenario, and the target steering angle is within the steering angle adjustment range and adapts to the steering requirements of road driving.

[0153] The system analyzes the steering angle adjustment range in the real-time steering wheel control parameters, such as 19 to 40 degrees. Considering the current real-time driving scenario, such as turning right at an intersection, and factors like road curvature and lane width, it determines the required target steering angle. The target steering angle must be within the adjustment range and meet the road's steering requirements, allowing the vehicle to pass smoothly through the intersection. For example, a target steering angle of 30 degrees might be determined.

[0154] Step S152: Analyze the steering rate constraint information in the real-time steering wheel control parameters, determine the target steering rate corresponding to the target steering angle, and the target steering rate conforms to the steering rate constraint information and achieves smooth steering.

[0155] The system analyzes the steering rate constraint information in the real-time steering wheel control parameters. For example, the steering rate gradually increases from 5 degrees per second to 12 degrees per second at the beginning of the steering phase, remains at 12 degrees per second during the stable phase, and gradually decreases to 4 degrees per second at the end of the steering phase. Based on the target steering angle of 30 degrees and the steering rate constraint information, the system determines the target steering rate change curve over time. For instance, at the beginning of the steering phase, the rate is increased from 0 to 12 degrees per second in 0.5 seconds; during the stable phase, the rate is maintained at 12 degrees per second for 1 second; and at the end of the steering phase, the rate is decreased from 12 degrees per second to 0 in 0.5 seconds, for a total steering time of 2 seconds, achieving smooth steering.

[0156] Step S153: Associate the target steering angle and the target steering rate with encoding to generate basic control command information containing steering angle command code and steering rate command code.

[0157] The target steering angle of 30 degrees and the target steering rate over time are correlated and encoded. Using a preset encoding rule, the target steering angle is converted into a steering angle command code (such as binary encoding or a specific format of digital encoding), and the target steering rate curve is converted into a steering rate command code (such as encoding containing rate values ​​and time for each stage). The steering angle command code and the steering rate command code are combined to generate basic control command information.

[0158] Step S154: Extract the emergency driving status identifier information from the real-time driving scenario status. When the emergency driving status identifier information exists, add an emergency steering priority identifier to the basic control command information to increase the execution priority of the control command.

[0159] Emergency driving status information is extracted from the real-time driving scenario, such as whether there is a sudden traffic incident (a pedestrian suddenly running across the road, a vehicle braking suddenly in front, etc.). If emergency driving status information exists, an emergency steering priority flag (such as a specific binary bit or flag field) is added to the basic control command information. This emergency steering priority flag can notify the vehicle steering execution system to prioritize the execution of this control command in order to deal with emergency situations.

[0160] Step S155: Combining the wheel steering feedback information in the real-time driving status data, add a steering compensation command code to the basic control command information. The steering compensation command code corrects the deviation between the actual steering and the target steering.

[0161] Based on the wheel steering feedback information in the real-time driving status data, the deviation between the actual steering and the target steering is calculated, and a steering compensation instruction code is generated and added to the basic control instruction information.

[0162] Step S1551: Extract the actual steering angle information and the target steering angle information from the wheel steering feedback information, and calculate the angle difference between the actual steering angle information and the target steering angle information. The angle difference reflects the deviation in steering angle.

[0163] Extract the actual steering angle information (e.g., the current actual steering angle of the wheel is 28 degrees) and the target steering angle information (30 degrees) from the wheel steering feedback information. Calculate the angle difference between the two, which is 30 degrees minus 28 degrees equals 2 degrees. This angle difference reflects the deviation in steering angle.

[0164] Step S1552: Extract the actual steering rate information and the target steering rate information from the wheel steering feedback information, and calculate the rate difference between the actual steering rate information and the target steering rate information. The rate difference reflects the deviation in steering rate.

[0165] Extract the actual steering rate information (e.g., the current actual steering rate is 10 degrees per second) and the target steering rate information (the current target steering rate is 12 degrees per second). Calculate the rate difference as 12 degrees per second minus 10 degrees per second equals 2 degrees per second. This rate difference reflects the deviation in steering rate.

[0166] Step S1553: Analyze the positive and negative attributes of the angle difference to determine the direction of the deviation. If the angle difference is positive, the actual steering angle is greater than the target steering angle. If the angle difference is negative, the actual steering angle is less than the target steering angle.

[0167] The angle difference is 2 degrees, which is a positive value, indicating that the actual steering angle is less than the target steering angle, and the direction of the deviation is that the steering angle needs to be further increased.

[0168] Step S1554: Analyze the magnitude of the rate difference to determine the degree of influence of the deviation in steering rate. The larger the rate difference, the more significant the impact on steering performance.

[0169] The speed difference is 2 degrees per second, which is relatively small, indicating that the deviation in steering speed has a low impact on steering performance.

[0170] Step S1555: Based on a preset angle control strategy, an angle compensation coefficient is generated by mapping according to the magnitude and direction of the angle difference. The value of the angle compensation coefficient is related to the magnitude and direction of the angle difference and is used to compensate for the steering angle deviation.

[0171] The preset angle control strategy stipulates that when the angle difference is positive, the angle compensation coefficient is positive, and the larger the angle difference, the larger the angle compensation coefficient. For example, when the angle difference is 2 degrees, the angle compensation coefficient generated by the mapping is 0.1.

[0172] Step S1556: Based on the preset rate control strategy, a basic rate compensation coefficient is generated according to the magnitude of the rate difference and the time progress of the steering process. The basic rate compensation coefficient is then dynamically adjusted in conjunction with the current steering stage to obtain a rate compensation coefficient used to compensate for the steering rate deviation.

[0173] In the preset rate control strategy, a base rate compensation coefficient is generated based on the magnitude of the rate difference and the stage of the steering process (such as the start stage, stable stage, or end stage). For example, in the stable stage, when the rate difference is 2 degrees per second, the base rate compensation coefficient is 0.05. Considering that the current steering stage is the stable stage, the base rate compensation coefficient is dynamically adjusted (e.g., kept unchanged), resulting in a rate compensation coefficient of 0.05.

[0174] Step S1557: Encode the angle compensation coefficient and the rate compensation coefficient into an encoding format that meets the requirements of the control command format, forming the basic compensation command code.

[0175] The angle compensation coefficient 0.1 and the rate compensation coefficient 0.05 are encoded according to a preset encoding format, and converted into binary or a specific number sequence encoding form to form the basic compensation instruction code.

[0176] Step S1558: Query the compensation strategy table corresponding to the current road form and the actions of surrounding traffic participants, obtain the scene compensation factor, and use the scene compensation factor to perform weighted correction on the parameters in the basic compensation instruction code.

[0177] The preset compensation strategy table is queried, which records the scene compensation factors corresponding to different road shapes and the actions of surrounding traffic participants. For example, if the current road shape is a right turn at an intersection and there are pedestrians crossing the street, the corresponding scene compensation factor is 1.2. Multiplying the angle compensation coefficient and speed compensation coefficient in the basic compensation command code by the scene compensation factor 1.2 respectively, we get a corrected angle compensation coefficient of 0.12 and a speed compensation coefficient of 0.06.

[0178] Step S1559: Combining the tilt angle information and rollover trend information in the vehicle driving posture information, the optimized compensation command code is modified a second time to make the compensation operation adapt to the vehicle driving posture. The modified compensation command code is added to the basic control command information to form a complete basic control command information containing steering compensation function, so that the actual steering and the target steering are accurately matched.

[0179] The vehicle's tilt angle in the driving posture information is 2 degrees, and the rollover coefficient is 0.3, which is within a stable range. According to preset rules, when the tilt angle is small and the rollover tendency is low, the compensation command code is fine-tuned, such as increasing the angle compensation coefficient by 0.01 and the rate compensation coefficient by 0.005, resulting in a second-corrected angle compensation coefficient of 0.13 and a rate compensation coefficient of 0.065. The second-corrected compensation command code is added to the basic control command information to form complete basic control command information. This basic control command information can correct the deviation between the actual steering and the target steering, achieving precise steering.

[0180] Step S156: Perform format normalization processing on the basic control command information containing the steering angle command code, the steering rate command code, the emergency steering priority identifier, and the steering compensation command code to adapt it to the command receiving format of the vehicle steering execution system.

[0181] The vehicle steering system can receive control commands in specific formats, such as the CAN bus protocol format. The basic control command information, including steering angle command code, steering rate command code, emergency steering priority identifier, and steering compensation command code, is formatted and standardized. Each command code and identifier is encapsulated according to the requirements of the CAN bus protocol, and check bits, address bits, and other information are added to ensure that the control commands can be correctly received and parsed by the vehicle steering system.

[0182] Step S157: Generate autonomous steering wheel control commands, which contain all the normalized command information and provide comprehensive guidance for steering wheel operation.

[0183] After formatting, a complete set of autonomous steering wheel control commands is generated. These commands include all instruction information such as steering angle, steering rate, emergency priority, and steering compensation, providing comprehensive guidance for steering wheel operations.

[0184] Step S158: Send the autonomous steering wheel control command to the vehicle steering execution system, and receive the command reception receipt from the vehicle steering execution system to confirm that the autonomous steering wheel control command has been successfully received and will be executed, thus completing the process of sending the autonomous steering wheel control command.

[0185] The autonomous steering wheel control command is sent to the vehicle steering execution system via an internal vehicle communication bus (such as the CAN bus). Upon receiving the command, the vehicle steering execution system parses and verifies it. If the command format is correct and the content is valid, it returns a command reception receipt. After receiving the receipt, the system confirms that the autonomous steering wheel control command has been successfully received and will be executed, thus completing the transmission process. The vehicle steering execution system then drives the steering wheel to perform the corresponding steering operation according to the command, enabling the intelligent driving vehicle to make a safe and smooth right turn at intersections.

[0186] In one exemplary embodiment, an autonomous steering wheel control system for intelligent driving is provided. This autonomous steering wheel control system can be a terminal, server, etc., and its internal structure diagram can be as follows: Figure 2As shown, the autonomous steering wheel control system for intelligent driving includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, near-field communication, or other technologies. When the computer program is executed by the processor, it implements an autonomous steering wheel control method for intelligent driving. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, or it can be a button, trackball, or touchpad set on the housing of the autonomous steering wheel control system for intelligent driving, or it can be an external keyboard, touchpad, or mouse, etc.

[0187] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for autonomous steering wheel control applied to intelligent driving, characterized in that, The method includes: The real-time driving scenario status during the intelligent driving process is perceived, and the real-time driving scenario status includes road morphology information, the action information of surrounding traffic participants, and environmental influencing factors. Retrieve the historical driving operation records of the intelligent driving vehicle, which include steering wheel adjustment operation information in scenarios similar to the real-time driving scenario. Based on the real-time driving scenario state and the historical driving operation record, steering wheel control reference parameters are generated, which include steering angle adjustment range and steering rate constraint information. Real-time driving status data of intelligent driving vehicles is collected. The real-time driving status data includes vehicle driving posture information and wheel steering feedback information. The steering wheel control reference parameters are dynamically corrected by combining the real-time driving status data to obtain real-time steering wheel control parameters. Based on the real-time control parameters of the steering wheel, an autonomous control command for the steering wheel is generated and sent to the vehicle steering execution system to drive the steering wheel to complete the corresponding steering operation. The generation of steering wheel control reference parameters based on the real-time driving scenario state and the historical driving operation records includes: The historical driving operation records are classified into multiple scene categories based on the scene characteristics in the historical driving operation records. Each scene category corresponds to a specific driving scenario. Extract the core scene features of the real-time driving scenario state, which include core features of road morphology, core features of traffic participant actions, and core features of environmental impact. The core scene features are compared with the typical scene features of each scene category to select the target scene category with the highest similarity to the real-time driving scene state. Retrieve all historical driving operation records under the target scenario category, and extract the steering wheel adjustment operation information, which includes steering angle data and steering rate data; The steering angle data is statistically analyzed to determine the common variation range of steering angle under different driving conditions. Based on the preset steering strategy corresponding to the road morphology information of the real-time driving scenario, the boundary range of the common variation range is adjusted. The steering rate data is subjected to trend analysis to identify the pattern of steering rate change over time. Based on the driving safety model corresponding to the action information of surrounding traffic participants in the real-time driving scenario, the description of the steering rate change pattern is optimized. The steering angle adjustment range is determined based on the adjusted common variation range, and the steering angle adjustment range covers the angle range corresponding to the steering operation under the target scene category. Based on the optimized variation pattern of the steering rate, steering rate constraint information is determined, which regulates the rate variation of steering operations. The steering angle adjustment range and the steering rate constraint information are associated and bound to form initial steering wheel control reference parameters; Based on the environmental influencing factors information of the real-time driving scenario, a preset environment-parameter adjustment mapping table is queried to obtain the parameter adjustment amount. The values ​​in the initial steering wheel control reference parameters are updated according to the parameter adjustment amount to finally obtain the steering wheel control reference parameters.

2. The autonomous steering wheel control method for intelligent driving according to claim 1, characterized in that, The real-time driving scenario state during the perception-based intelligent driving process includes: The vehicle is equipped with a multi-source environmental perception device, which includes a visual perception device, a distance perception device, and an environmental state perception device. The multi-source environmental perception device collects raw environmental data around the intelligent driving vehicle. The road-related data in the original environmental data are parsed and processed to extract the road direction distribution information, pavement structure information and road sign information, and the road direction distribution information, pavement structure information and road sign information are integrated to form road morphology information; The traffic participant-related data in the original environmental data are identified and processed to distinguish different types of traffic participants, track the location movement information and behavior information of various types of traffic participants, and integrate the location movement information and behavior information of various types of traffic participants to form the surrounding traffic participant action information; The environmental data in the raw environmental data is collected and processed to capture information on natural environmental changes and artificial environmental interference that affect driving. The natural environmental change information and the artificial environmental interference information are then integrated to form information on environmental influencing factors. The road morphology information, the action information of surrounding traffic participants, and the environmental influencing factors are correlated and integrated to form initial driving scenario state data; Establish a data update queue, and write the supplementary information parsed from the newly collected environmental raw data into the corresponding information segment of the initial driving scenario state data in timestamp order, overwriting or appending it to the outdated information segment. Redundant information unrelated to steering wheel control is removed from the initial driving scenario state data, while key information that directly affects steering operation is retained. The retained key information is then processed by feature encoding to convert various types of information into a unified feature vector or data structure expression, forming a standardized driving scenario state fragment. Multiple consecutive standardized driving scenario state segments are sequentially linked to form a real-time driving scenario state. The completeness of the information in the real-time driving scenario state is verified. If any information is missing, the corresponding original environmental data is collected again to ensure that the road morphology information, the action information of the surrounding traffic participants, and the information of the environmental influencing factors are all complete.

3. The autonomous steering wheel control method for intelligent driving according to claim 1, characterized in that, The method involves collecting real-time driving status data of the intelligent driving vehicle, including vehicle driving posture information and wheel steering feedback information. The steering wheel control reference parameters are then dynamically corrected based on this real-time driving status data to obtain real-time steering wheel control parameters, including: The vehicle's driving status acquisition device is activated. The driving status acquisition device includes an attitude sensor and a steering feedback sensor. The driving status acquisition device continuously collects raw data on the vehicle's driving status. The attitude-related data in the original driving state data are parsed and processed to extract the vehicle's tilt angle information, pitch angle information and rollover trend information, and the tilt angle information, pitch angle information and rollover trend information are integrated to form the vehicle driving attitude information. The steering feedback-related data in the original driving state data are processed to capture the actual steering angle information and steering response delay information of the wheel, and the actual steering angle information and steering response delay information are integrated to form wheel steering feedback information; The vehicle driving posture information and the wheel steering feedback information are encapsulated according to a preset data structure and aligned with a timestamp to generate real-time driving status data. Extract the steering angle adjustment range from the steering wheel control reference parameters, and combine it with the vehicle driving posture information and vehicle dynamics model to evaluate the vehicle driving stability under the current steering angle adjustment range; If the assessment results indicate a risk of instability, the upper and lower limits of the steering angle adjustment range are calculated and adjusted based on the vehicle driving posture information and vehicle dynamics model, so that the adjusted steering angle adjustment range can maintain vehicle driving stability. Extract the steering rate constraint information from the steering wheel control reference parameters, and combine it with the wheel steering feedback information and the vehicle dynamic response model to evaluate the steering execution smoothness and timeliness under the current steering rate constraint information; If the evaluation results do not meet the requirements, the rate change gradient of the steering rate constraint information is replanned based on the steering response delay information, rollover trend information and vehicle dynamic response model. The boundary values ​​of the adjusted steering angle adjustment range and the parameter values ​​of the optimized steering rate constraint information are encapsulated according to the preset parameter combination format to generate the intermediate steering wheel control parameters. In each data processing cycle, the latest real-time driving status data is input into the generation rule of the center steering wheel control parameters, the parameter values ​​in the center steering wheel control parameter set are recalculated and updated, and finally the real-time steering wheel control parameters are obtained.

4. The autonomous steering wheel control method for intelligent driving according to claim 1, characterized in that, The process of generating autonomous steering wheel control commands based on the real-time steering wheel control parameters and sending the autonomous steering wheel control commands to the vehicle steering execution system to drive the steering wheel to complete the corresponding steering operation includes: The steering angle adjustment range in the real-time steering wheel control parameters is analyzed to determine the target steering angle required in the current driving scenario. The target steering angle is within the steering angle adjustment range and is adapted to the steering requirements of road driving. The steering rate constraint information in the real-time steering wheel control parameters is analyzed to determine the target steering rate corresponding to the target steering angle. The target steering rate conforms to the steering rate constraint information and achieves smooth steering. The target steering angle and the target steering rate are correlated and encoded to generate basic control command information containing steering angle command code and steering rate command code; Extract emergency driving status identification information from the real-time driving scenario status. When the emergency driving status identification information exists, add an emergency steering priority identifier to the basic control command information to increase the execution priority of the control command. Combining the wheel steering feedback information in the real-time driving status data, a steering compensation command code is added to the basic control command information. The steering compensation command code corrects the deviation between the actual steering and the target steering. The basic control command information, which includes the steering angle command code, the steering rate command code, the emergency steering priority identifier, and the steering compensation command code, is formatted to adapt it to the command receiving format of the vehicle steering execution system. Generate autonomous steering wheel control commands, which contain all the normalized command information to comprehensively guide the steering wheel operation; The system sends the autonomous steering wheel control command to the vehicle steering execution system and receives a command reception receipt from the vehicle steering execution system to confirm that the autonomous steering wheel control command has been successfully received and will be executed, thus completing the process of sending the autonomous steering wheel control command.

5. The autonomous steering wheel control method for intelligent driving according to claim 2, characterized in that, The process involves parsing and processing road-related data from the raw environmental data to extract road alignment information, pavement structure information, and road signage information. This information is then integrated to form road morphology information, including: From the raw environmental data, road-related raw data segments are selected, non-road area data contained therein are removed, and the selected road-related raw data segments are subjected to image enhancement processing based on edge detection. The enhanced road-related original data fragments are processed using an image segmentation algorithm to divide the pixel connected regions. Based on the texture, color, and relative position features of each connected region with respect to vehicles, they are classified and labeled as the main road region, the road edge region, or the road auxiliary region. The data of the main area of ​​the road are analyzed to extract the road's extension direction information, curvature information and bifurcation structure information. The extension direction information, curvature information and bifurcation structure information of the road are integrated to form the road's orientation distribution information. Detailed data of the road surface are extracted to identify the road surface material type, smoothness, and damage information. The road surface material type, smoothness, and damage information are then integrated to form road surface structure information. The data of the road's adjacent area is identified and processed to capture traffic sign information, lane line information, and road marking text information on the road. The traffic sign information, lane line information, and road marking text information on the road are then integrated to form road identification information. The road segment extension direction sequence in the road direction distribution information is traversed. If the rate of change of the extension direction between adjacent road segments exceeds the preset continuity threshold, transition direction information is generated by smooth interpolation based on the curvature of the preceding and following road segments to replace the original contradictory information. Data fusion processing is performed on structural data of the same road surface area collected by different sensing devices. When the difference between the data exceeds the preset fault tolerance threshold, the preset confidence weighting algorithm is used to calculate the calibrated road surface structural information. The identified road sign information is compared with a preset list of key signs. If any key sign in the list is not identified within the preset sensing range, the corresponding sensing device is triggered to re-collect and identify the road's adjacent area. The road alignment information, road surface structure information, and road signage information, after verification, validation, and inspection, are integrated to form road morphology information that reflects the actual road conditions.

6. The autonomous steering wheel control method for intelligent driving according to claim 1, characterized in that, The process of comparing the core scene features with typical scene features of each scene category to select the target scene category with the highest similarity to the real-time driving scene state includes: Typical scene features of each scene category are extracted. The typical scene features of each scene category include typical road morphology features, typical traffic participant action features, and typical environmental impact features under that scene category. The road morphology core feature, traffic participant action core feature and environmental impact core feature in the core scene features are respectively assigned corresponding weight coefficients, and the weight coefficients are set according to the degree of influence of different features on driving operation. The same weighting coefficients are assigned to the typical road morphology features, typical traffic participant action features, and typical environmental impact features in the typical scene features of each scene category; The similarity between the feature vector of the core road morphology feature in the core scene features and the feature vector of the typical road morphology feature in the typical scene features of each scene category is calculated to obtain the road morphology matching degree value; the matching degree value of traffic participant actions and the matching degree value of environmental impact are obtained using the same similarity calculation method; Calculate the matching degree between the core features of traffic participant actions in the core scene features and the typical features of traffic participant actions in the typical scene features of each scene category, and obtain the traffic participant action matching degree value. Calculate the matching degree between the core environmental impact features in the core scene features and the typical environmental impact features in the typical scene features of each scene category, and obtain the environmental impact matching degree value. The road morphology matching degree value, the traffic participant action matching degree value, and the environmental impact matching degree value are multiplied by their respective weight coefficients to obtain the weighted matching degree value of each feature. The weighted matching degree values ​​of each feature of each scene category are summed to obtain the comprehensive matching degree value of each scene category. The comprehensive matching degree value reflects the similarity relationship between the typical scene features and the core scene features of the scene category. The overall matching degree values ​​of all scene categories are sorted and arranged in descending order of the overall matching degree values. The scene category with the largest overall matching degree value in the sorting results is selected as the target scene category. The typical scene features of the target scene category have the highest similarity to the core scene features of the real-time driving scene state.

7. The autonomous steering wheel control method for intelligent driving according to claim 3, characterized in that, In each data processing cycle, the latest real-time driving status data is input into the generation rule of the center steering wheel control parameters, the parameter values ​​in the center steering wheel control parameter set are recalculated and updated, and finally the real-time steering wheel control parameters are obtained, including: A fixed data processing cycle is set, and the latest real-time driving status data is read when each data processing cycle arrives; Within each data processing cycle, the changes in vehicle driving posture information and wheel steering feedback information are extracted from the updated real-time driving status data. The relationship between the change in the vehicle's driving posture information and a preset posture change threshold is analyzed. If the change in the vehicle's driving posture information exceeds the preset posture change threshold, the control parameters of the center steering wheel are adjusted accordingly. Analyze the relationship between the change in wheel steering feedback information and a preset feedback change threshold. If the change in wheel steering feedback information exceeds the preset feedback change threshold, optimize the relevant content of the center steering wheel control parameters. If the change in the vehicle driving posture information exceeds the preset posture change threshold, the steering angle adjustment range in the middle steering wheel control parameters is readjusted according to the vehicle driving posture information in the updated real-time driving status data, so that the adjusted steering angle adjustment range is adapted to the new driving posture. If the change in the wheel steering feedback information exceeds the preset feedback change threshold, the steering rate constraint information in the center steering wheel control parameters is re-optimized based on the updated wheel steering feedback information in the real-time driving status data, so that the optimized steering rate constraint information matches the new steering response. If the change in vehicle driving posture information and the change in wheel steering feedback information both exceed the corresponding threshold, then the steering angle adjustment range in the center steering wheel control parameters and the steering rate constraint information in the center steering wheel control parameters are adjusted simultaneously, so that each part of the center steering wheel control parameters is adapted to the updated real-time driving status data. After adjustment and optimization, the adjusted intermediate steering wheel control parameters are input into the vehicle dynamics simulation model. Combined with the updated real-time driving status data, steering wheel control commands are simulated and the predicted trajectory of the vehicle is calculated. The deviation between the predicted trajectory and the expected trajectory is evaluated. Repeat the above process of monitoring, analysis, adjustment, optimization and verification, and dynamically adjust the center steering wheel control parameters in each data processing cycle so that the center steering wheel control parameters continuously adapt to changes in the vehicle's driving state. If the variance of each parameter value in the intermediate steering wheel control parameter set is less than its corresponding stability threshold within a preset number of consecutive monitoring cycles, and the deviation of the test steering command generated based on the intermediate steering wheel control parameter set in the vehicle model simulation is less than the preset tolerance, then the intermediate steering wheel control parameter set of the current cycle is determined as the real-time steering wheel control parameter.

8. The autonomous steering wheel control method for intelligent driving according to claim 4, characterized in that, The method involves combining the wheel steering feedback information from the real-time driving status data with the addition of steering compensation command codes to the basic control command information, including: Extract the actual steering angle information and the target steering angle information from the wheel steering feedback information, and calculate the angle difference between the actual steering angle information and the target steering angle information. The angle difference reflects the deviation in steering angle. Extract the actual steering rate information and the target steering rate information from the wheel steering feedback information, and calculate the rate difference between the actual steering rate information and the target steering rate information. The rate difference reflects the deviation in steering rate. Analyze the positive and negative attributes of the angle difference to determine the direction of the deviation. If the angle difference is positive, the actual steering angle is greater than the target steering angle. If the angle difference is negative, the actual steering angle is less than the target steering angle. By analyzing the magnitude of the rate difference, the degree of influence of the deviation in steering rate can be determined. The larger the rate difference, the more significant the impact on steering performance. Based on a preset angle control strategy, an angle compensation coefficient is generated by mapping the magnitude and direction of the angle difference. The value of the angle compensation coefficient is related to the magnitude and direction of the angle difference and is used to compensate for steering angle deviation. Based on a preset rate control strategy, a basic rate compensation coefficient is generated according to the magnitude of the rate difference and the time progress of the steering process. The basic rate compensation coefficient is then dynamically adjusted in conjunction with the current steering stage to obtain a rate compensation coefficient used to compensate for steering rate deviation. The angle compensation coefficient and the rate compensation coefficient are encoded and converted into an encoding format that meets the requirements of the control command format to form the basic compensation command code; Query the compensation strategy table corresponding to the current road condition and the actions of surrounding traffic participants, obtain the scene compensation factor, and use the scene compensation factor to perform weighted correction on the parameters in the basic compensation instruction code; By combining the tilt angle information and rollover trend information in the vehicle driving posture information, the optimized compensation command code is modified a second time to make the compensation operation adapt to the vehicle driving posture. The modified compensation command code is then added to the basic control command information to form a complete basic control command information that includes steering compensation function, so that the actual steering and the target steering are precisely matched.

9. An autonomous steering wheel control system for intelligent driving, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the autonomous steering wheel control method for intelligent driving as described in any one of claims 1 to 8 by executing the machine-executable instructions.

Images