Vehicle parking posture correction method and device, electronic equipment and medium

Abstract

The invention relates to a posture correction method and device for vehicle parking, electronic equipment and a medium, and the method comprises the steps: obtaining a current coordinate of a current parking space angular point and a current posture of a vehicle in a vehicle parking process; a parking space corner point set is inquired, reference coordinates matched with the current parking space corner point are determined, the parking space corner point set is used for storing reference coordinates of different parking space corner points, and the reference coordinates are position coordinates of the parking space corner points in a world coordinate system; and under the condition that the coordinate difference between the current coordinate and the reference coordinate is larger than a preset value, the current pose of the vehicle is corrected based on the coordinate difference, the corrected pose is obtained, and the corrected pose is used for updating the parking path of the vehicle. According to the scheme, the pose of the vehicle is corrected according to the position difference of the same parking space angular point under the world coordinate system, the accumulative error of the pose of the vehicle is counteracted, the corrected high-precision pose is finally output to be used for updating the parking path, and accurate positioning of the automatic parking auxiliary system is achieved.

Description

Technical Field

[0001] This disclosure relates to the field of autonomous driving technology, and in particular to a method, apparatus, electronic device, and medium for correcting the pose of a vehicle when parking. Background Technology

[0002] With the continuous development of intelligent driving technology in automobiles, Automatic Parking Assist (APA) has become one of the core functions for measuring the level of vehicle intelligence. High-precision positioning is a core prerequisite for the practical application of APA systems. The core working scenarios of APA systems are concentrated in enclosed or semi-enclosed spaces such as underground parking lots and multi-level parking garages. The industry generally uses Dead Reckoning (DR) technology as the core positioning method to meet the parking positioning needs of enclosed or semi-enclosed spaces.

[0003] Studies have shown that the error of DR (Driving Directional Alert) technology accumulates with increasing driving distance. In complex parking scenarios, the error can exceed 10 centimeters in a short period of time, far exceeding the centimeter-level accuracy required for parking. To address this issue, some manufacturers have attempted to improve perception accuracy by adding hardware such as LiDAR, but this significantly increases the overall vehicle cost. Other solutions optimize DR parameters using complex algorithms, but their effectiveness in suppressing hardware noise is limited. Therefore, how to achieve real-time correction of accumulated DR errors based on existing sensor configurations, thereby improving the positioning accuracy and reliability of the APA (Automatic Parking Assist) system, has become a critical problem that urgently needs to be solved in the field of intelligent parking technology. Summary of the Invention

[0004] In order to solve the above-mentioned technical problems, or at least partially solve the above-mentioned technical problems, at least one embodiment of the present disclosure provides a vehicle parking posture correction method, device, electronic device and medium.

[0005] In a first aspect, this disclosure provides a method for correcting the position and orientation of a vehicle when parking, including: Obtain the current coordinates of the corner point of the current parking space and the current pose of the vehicle during the parking process; Query the parking space corner point set to determine the reference coordinates that match the current parking space corner point. The parking space corner point set is used to store the reference coordinates of different parking space corner points. The reference coordinates are the position coordinates of the parking space corner point in the world coordinate system. If the coordinate difference between the current coordinates and the reference coordinates is greater than a preset value, the current pose of the vehicle is corrected based on the coordinate difference to obtain a corrected pose, which is used to update the parking path of the vehicle.

[0006] Secondly, this disclosure provides a vehicle parking posture correction device, including: The acquisition module is used to acquire the current coordinates of the corner point of the current parking space and the current pose of the vehicle during the parking process; The determination module is used to query the set of parking space corner points and determine the reference coordinates that match the current parking space corner point. The set of parking space corner points is used to store the reference coordinates of different parking space corner points. The reference coordinates are the position coordinates of the parking space corner point in the world coordinate system. The correction module is used to correct the current pose of the vehicle based on the coordinate difference when the coordinate difference between the current coordinate and the reference coordinate is greater than a preset value, so as to obtain the corrected pose. The corrected pose is used to update the parking path of the vehicle.

[0007] Thirdly, this disclosure provides an electronic device, including: a processor and a memory; The processor executes the vehicle parking pose correction method as described in the first aspect by calling the program or instructions stored in the memory.

[0008] Fourthly, this disclosure provides a computer-readable storage medium storing a program or instructions that cause a computer to perform the vehicle parking pose correction method as described in the first aspect.

[0009] Fifthly, this disclosure provides a computer program product for executing the vehicle parking pose correction method as described in the first aspect.

[0010] The technical solution provided in this disclosure has at least the following advantages compared with the prior art: In this embodiment, a parking space corner point set is constructed to store the reference coordinates of different parking space corner points. The reference coordinates are the position coordinates of the parking space corner points in the world coordinate system. During the vehicle parking process, the current coordinates of the current parking space corner point and the current vehicle pose are obtained, and the parking space corner point set is queried to determine the reference coordinates that match the current parking space corner point. The current coordinates are compared with the reference coordinates. If the coordinate difference between the current coordinates and the reference coordinates is greater than a preset value, the current vehicle pose is corrected based on the coordinate difference to obtain the corrected pose. The corrected pose is then used to update the vehicle's parking path. Thus, the vehicle pose is corrected based on the position difference of the same parking space corner point in the world coordinate system, which offsets the cumulative error of the vehicle pose. Finally, the corrected high-precision pose is output for updating the parking path, which provides a guarantee for the accurate positioning and parking of the automatic parking assistance system. Moreover, this solution can be implemented using the vehicle's existing sensor configuration without additional cost. It has both low cost and good robustness, and can overcome most of the positioning accuracy reduction caused by ground bumps, gear changes, etc. Attached Figure Description

[0011] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0012] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A schematic flowchart of a vehicle parking pose correction method provided as an exemplary embodiment of this disclosure; Figure 2 A flowchart illustrating the process of constructing a corner coordinate mapping library provided for an exemplary embodiment of this disclosure. Figure 1 ; Figure 3 A flowchart illustrating the process of constructing a corner coordinate mapping library provided for an exemplary embodiment of this disclosure. Figure 2 ; Figure 4 This is a schematic diagram illustrating the process of parking a vehicle in a parking space, which is an exemplary embodiment of the present disclosure. Figure 5 This is a schematic diagram of the vehicle parking posture correction device provided in one embodiment of the present disclosure. Detailed Implementation

[0014] To better understand the above-described objectives, features, and advantages of this disclosure, the present disclosure will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the described embodiments are only some, not all, of the embodiments of this disclosure. The specific embodiments described herein are merely for explaining this disclosure and not for limiting it. Unless otherwise specified, the embodiments of this disclosure and the features within them can be combined with each other. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure are within the scope of protection of this disclosure.

[0015] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0016] The vehicle positioning system, acting as the "eyes" of the entire APA system, must acquire the vehicle's position and attitude (or "pose") in the real environment in real time and accurately to ensure the rationality of subsequent path planning and the precision of control execution. Therefore, high-precision positioning is a core prerequisite for the practical application of APA systems. In open road scenarios, vehicles can typically rely on global satellite navigation systems such as GPS and BeiDou to achieve centimeter-level positioning. However, the core operating scenarios of APA systems are concentrated in enclosed or semi-enclosed spaces such as underground parking lots and multi-level parking garages. Satellite signals are easily blocked or reflected by buildings, leading to positioning signal interruption or a sharp drop in accuracy, failing to meet the positioning requirements for parking. To address this, the industry commonly uses DR (Depth Recognition) technology as the core positioning method. This technology, based on data from inertial components such as wheel speed sensors, three-axis accelerometers, and gyroscopes, calculates the vehicle's pose information in real time through integration calculations. Its advantage lies in its independence from external signals and its ability to continuously output data in enclosed environments. However, the positioning accuracy of DR technology accumulates errors over time. During vehicle operation, factors such as road bumps causing measurement deviations of inertial components, tire wear and changes in the ground adhesion coefficient leading to wheel speed errors, cumulative errors during integration calculations, and interference from complex environments (bumpy ground, slope, frequent steering and gear shifting) all cause the pose calculated by DR to gradually deviate from the actual pose over time and distance. Studies have shown that the error of DR technology accumulates with increasing distance traveled, and in complex parking scenarios, the error may exceed 10 centimeters in a short period of time, far exceeding the centimeter-level accuracy required for parking.

[0017] To address the aforementioned issues, some manufacturers have attempted to improve perception accuracy by adding hardware such as LiDAR, but this significantly increases the overall vehicle cost. Other solutions optimize DR parameters using complex algorithms, but their effectiveness in suppressing hardware noise is limited. Therefore, how to achieve real-time correction of accumulated DR errors based on existing sensor configurations, thereby improving the positioning accuracy and reliability of the APA system, has become a critical problem urgently needing to be solved in the current field of intelligent parking technology.

[0018] To address the problems of traditional solutions, this disclosure provides a vehicle parking pose correction scheme, aiming to solve the problem of error accumulation in DR (Device Recognition) in complex scenarios. In indoor parking lots where high-precision positioning via real-time dynamic positioning is lacking, this scheme provides centimeter-level positioning for the APA (Action Parking Assist) system. This scheme corrects DR pose errors by observing the position changes of the same location twice. Specifically, it first determines the visual information of the parking space corner based on a bird's-eye view generated by inverse perspective transformation from four fisheye cameras (surround-view cameras) on the vehicle body. Then, it establishes a real-world coordinate system based on the pose provided by the DR. The same corner is observed at two different times, and the position values ​​of the same corner in the real-world coordinate system under different spatiotemporal conditions are compared. Because DR error accumulation causes a deviation between the two position values, if this deviation exceeds a threshold, it is determined that the DR has accumulated errors. At this point, the pose parameters output by the DR are adjusted and optimized in reverse to make the two position values ​​coincide, thereby offsetting the accumulated error of the DR during this period. Finally, a corrected high-precision pose is output, ensuring accurate positioning and parking for the APA system. Since the cost of surround view cameras is much lower than that of high-precision sensors such as lidar, this solution has a low-cost advantage. Therefore, this solution has good robustness while being low in cost, and can overcome most of the positioning accuracy reduction caused by ground bumps, gear changes, etc.

[0019] The following detailed explanation, in conjunction with the accompanying drawings, illustrates the specific implementation methods, devices, electronic equipment, and storage media for vehicle parking posture correction disclosed herein.

[0020] Figure 1 This is a flowchart illustrating a vehicle parking pose correction method provided in an exemplary embodiment of the present disclosure. The method can be executed by a vehicle parking pose correction device provided in an embodiment of the present disclosure. The vehicle parking pose correction device can be implemented in software and / or hardware and can be integrated into an electronic device, such as an in-vehicle infotainment system.

[0021] like Figure 1 As shown, the vehicle parking posture correction method may include the following steps: Step 101: Obtain the current coordinates of the corner point of the current parking space and the current pose of the vehicle during the parking process.

[0022] The current parking space corner point can be any corner point within the parking lot. When the corner point of a parking space appears within a specific area in the image captured by the vehicle's surround-view camera, that corner point is considered the current parking space corner point. This specific area can be, for example, the optimal observation area determined by the center of the vehicle's rear axle in a bird's-eye view generated from the images captured by the surround-view camera. This could be a circular area with a radius of 1.5 meters and the center of the rear axle as the center point, or a rectangular area with a long side of 4 meters and a short side of 2.5 meters and the center of the rear axle as the center point.

[0023] During the parking process, surround-view cameras around the vehicle capture images of the surrounding environment in real time. Semantic segmentation algorithms can be used to extract core features such as parking space lines and corner points from the images. When a corner point of a parking space is detected to appear in a specific area of ​​the image, the current coordinates of that corner point are obtained, and the vehicle's pose at the current moment is also obtained (denoted as the vehicle's current pose). The current coordinates of the current corner point refer to the actual position coordinates of the corner point in the world coordinate system. The vehicle's current pose can be the vehicle pose information calculated in real time by the DR system based on wheel speed and inertial measurement unit (IMU) data.

[0024] As an example, when obtaining the current coordinates of the current parking space corner point, a semantic segmentation algorithm can be used to identify the image captured by the surround-view camera. The semantic segmentation algorithm identifies the visual coordinates (pixel coordinates) of the current parking space corner point, and then transforms the visual coordinates to the world coordinate system to obtain its real-world position coordinates, i.e., the current coordinates. For example, multiplying the visual coordinates by the vehicle's current pose yields the real position coordinates of a parking space corner point in the world coordinate system. Following this transformation method, the current coordinates of the current parking space corner point can be obtained. By using a semantic segmentation algorithm to extract the information of interest, the computational load is greatly reduced. At the same time, because this solution constrains the information to the ground, given the camera's extrinsic parameters, relatively accurate geometric information can be obtained by observing the ground through surround-view observation, thus achieving high positioning accuracy.

[0025] It should be noted that, in this embodiment, when obtaining the current coordinates of the current parking space corner point, a single frame image of the current parking space corner point can be acquired, and its current coordinates in the world coordinate system can be determined based on the visual coordinates of the current parking space corner point identified from the image; alternatively, multiple frames of images of the current parking space corner point can be acquired, and the visual coordinates of the current parking space corner point can be identified from each frame image and converted into the real position coordinates in the world coordinate system to obtain multiple real position coordinates. The multiple real position coordinates are then optimized by averaging, weighted summation, etc., to obtain the current coordinates.

[0026] Step 102: Query the parking space corner point set to determine the reference coordinates that match the current parking space corner point. The parking space corner point set is used to store the reference coordinates of different parking space corner points. The reference coordinates are the position coordinates of the parking space corner point in the world coordinate system.

[0027] The parking space corner point set is pre-built. When a vehicle enters the parking lot, the parking space corner point set can be determined based on the actual position coordinates of the parking space corner points in the world coordinate system observed during the vehicle's movement. The parking space corner point set includes the actual positions of all identified parking space corner points.

[0028] As an example, a parking space corner point set can record the correspondence between the corner point identifiers of different parking space corner points and their actual position coordinates in the world coordinate system. When a vehicle enters the parking lot, a parking space corner point set is constructed based on the corner point identifiers and actual position coordinates of the parking space corner points observed for the first time during the vehicle's journey. The actual position coordinates are the position coordinates of the parking space corner points in the world coordinate system, and the parking space corner point set stores the correspondence between the corner point identifiers and the reference coordinates.

[0029] As an example, a parking space corner point set can record the location information of different parking spaces. The location information of a parking space is represented by the reference coordinates of its four corner points. That is, a parking space in the parking space corner point set can be represented as (reference coordinates of corner point 1, reference coordinates of corner point 2, reference coordinates of corner point 3, and reference coordinates of corner point 4). When a vehicle enters the parking lot, the actual position coordinates of these corner points in the world coordinate system can be determined based on the visual coordinates of the parking space corner points identified during the vehicle's movement. Then, the reference coordinates are determined based on the actual position coordinates. If fewer than four corner points are identified, the reference coordinates of the other parking space corner points can be determined based on the standard parking space size, thereby determining the actual location information of the parking space.

[0030] It should be noted that in this embodiment, the reference coordinates are determined based on the actual position coordinates of the parking space corner point. These coordinates represent the position coordinates of the parking space corner point in the world coordinate system. The actual position coordinates of a parking space corner point determined once can be stored as the reference coordinates in the parking space corner point set. Alternatively, optimized position coordinates obtained by averaging or weighted summing the actual position coordinates of the same parking space corner point determined multiple times can also be stored as the reference coordinates in the parking space corner point set. Furthermore, when a parking space corner point is observed subsequently, the parking space corner point set can be updated. The actual position coordinates of the first observed parking space corner point can be stored in the parking space corner point set, and the actual position coordinates of parking space corner points already existing in the set but whose actual position coordinates have changed can be updated to the newly determined actual position coordinates.

[0031] In this embodiment, for the current parking space corner point, the parking space corner point set can be queried to determine whether there is a matching reference coordinate in the parking space corner point set, and if there is, the matching reference coordinate can be obtained.

[0032] As an example, the parking space corner point set records the correspondence between the corner point identifiers of different parking space corner points and their actual position coordinates in the world coordinate system. When querying the parking space corner point set, the set can be queried based on the corner point identifier of the current parking space corner point (referred to as the target corner point identifier for ease of description and distinction). If the corresponding target corner point identifier is found in the parking space corner point set, the current observation is considered to be a secondary observation, and the reference coordinates corresponding to the target corner point identifier can be obtained from the parking space corner point set.

[0033] As an example, a parking space corner point set can record the location information of different parking spaces. In this case, after obtaining the current coordinates of the current parking space corner point, the location information of the storage space to which the current parking space corner point belongs can be determined based on the current coordinates and the standard storage space size (i.e., the actual location coordinates of the four parking space corner points of that storage space). Since the four parking space corner points can define a rectangle, the Intersection of Union (IoU) principle can be used to search for whether there is a storage space in the parking space corner point set that matches the current storage space, thereby determining the reference coordinates that match the current parking space corner point. Specifically, the IoU ratio of the storage space to which the current parking space corner point belongs to and each storage space in the parking space corner point set can be calculated. If there is a storage space in the parking space corner point set whose IoU ratio is greater than or equal to a preset threshold, then it can be determined that the current parking space corner point is a corner point observed twice, rather than a corner point observed for the first time. Thus, the reference coordinates of the successfully matched storage space that are closest to the current coordinates of the current parking space corner point are obtained as the reference coordinates that match the current parking space corner point.

[0034] In one optional embodiment of this disclosure, if no corresponding target corner identifier is found in the parking space corner identifier set, or if the intersection-union ratio (IUU) of any parking space in the parking space corner identifier set with the parking space to which the current parking space corner belongs is less than a preset threshold, it can be determined that no reference coordinates matching the current parking space corner cannot be found in the parking space corner identifier set. This current parking space corner is the first observed parking space corner. In this case, the current coordinates can be stored in the parking space corner identifier set. For example, the correspondence between the target corner identifier and the current coordinates of the current parking space corner can be stored in the parking space corner identifier set. Alternatively, the location information of the parking space to which the current parking space corner belongs can be determined based on the current coordinates, and the location information can be stored in the parking space corner identifier set. In other words, whenever the current coordinates of a parking space corner point are obtained, the current parking space corner point is matched with all parking space corner points already stored in the parking space corner point set. If no match is found with any stored parking space corner points, it is considered the first observation, and the obtained current coordinates are stored as the reference coordinates of the parking space corner point in the parking space corner point set. This realizes the acquisition and storage of the reference coordinates of newly observed parking space corner points, providing data support for pose correction based on the actual position difference when the parking space corner point is observed again in the future.

[0035] Step 103: If the coordinate difference between the current coordinates and the reference coordinates is greater than a preset value, the current pose of the vehicle is corrected based on the coordinate difference to obtain the corrected pose. The corrected pose is used to update the parking path of the vehicle.

[0036] The preset values ​​can be set according to actual needs, and this disclosure does not restrict their specific values.

[0037] In this embodiment, after obtaining the reference coordinates of the current parking space corner point, the coordinate difference between the current coordinates and the reference coordinates can be calculated. This coordinate difference is then compared with a preset value. If the coordinate difference is greater than the preset value, it can be considered that the DR system has a significant cumulative error. Therefore, the vehicle's current pose can be corrected based on this coordinate difference to obtain a corrected pose. The parking path of the vehicle is then updated based on the corrected pose so that the vehicle parks according to the updated path. The coordinate difference includes the difference corresponding to different coordinate axes. When the difference on any coordinate axis is greater than the preset value corresponding to that axis, pose correction is performed based on the coordinate difference.

[0038] In one optional embodiment of this disclosure, when correcting the vehicle's current pose based on the coordinate difference to obtain the corrected pose, the goal is to ensure that the corrected position coordinates of the current parking space corner point in the world coordinate system are consistent with the reference coordinates. The pose error increment is determined based on the coordinate difference, and then the vehicle's current pose is corrected based on the pose error increment to obtain the corrected pose. In this embodiment, when the coordinate difference between the current coordinates of the current parking space corner point and the reference coordinates is greater than a preset value, pose correction is required. At this time, the pose error increment of the DR pose can be calculated in reverse based on the coordinate difference, and then the vehicle's current pose output by the DR system is corrected based on the pose error increment, so that the corrected position coordinates of the current parking space corner point in the world coordinate system (i.e., the corrected current coordinates) coincide with the reference coordinates. Specifically, when calculating the pose error increment in reverse, the pose error increment can be calculated in reverse using the coordinate difference, based on the visual coordinates of the parking space corner point and the transformation relationship from the DR pose to the world coordinate system.

[0039] For example, assuming the visual coordinates of the current parking space corner point are A, the current vehicle pose is B, and the current coordinates of the current parking space corner point in the world coordinate system are C, and A×B=C, then with the goal of correcting the position coordinates to be consistent with the reference coordinates, A×(B+B')=C-C', where B' represents the pose error increment to be solved, and C' represents the coordinate difference between the current coordinates and the reference coordinates. With other parameters known, the pose error increment B' can be obtained through mathematical algorithms, and then the current vehicle pose can be corrected to obtain the corrected pose.

[0040] In this embodiment, by aiming to make the corrected position coordinates of the current parking space corner point in the world coordinate system consistent with the reference coordinates, the pose error increment is determined based on the coordinate difference. Then, the current pose of the vehicle is corrected based on the pose error increment to obtain the corrected pose. Thus, based on the difference in the real coordinates of the same parking space corner point at different times, the DR pose is corrected by reverse adjustment and optimization to make the real coordinates of the same parking space corner point coincide, thereby offsetting the accumulated error of DR during this period. Finally, the corrected high-precision pose is output, which provides a guarantee for the accurate positioning and parking of the APA system.

[0041] It is understandable that during the parking process, the process of observing the corner point of the parking space and correcting the vehicle's DR pose is carried out continuously throughout the entire parking process until the vehicle is successfully parked in the parking space.

[0042] After obtaining the corrected pose, the system updates the target parking space (i.e., the parking space to be parked) in the real-world coordinate system based on the corrected pose (the corrected DR pose). The vehicle system then calls the corrected DR pose data and the updated parking space position information to replan the optimal parking path. This path accurately matches the parking space size and vehicle posture, ensuring a smooth and deviation-free driving trajectory. The vehicle system outputs control commands to the power, steering, and braking systems via the Controller Area Network (CAN) bus, driving the vehicle along the planned path. Throughout the process, it relies on real-time observation from the surround-view camera and collaborative feedback from the DR corrected pose to complete precise parking. After parking, the system releases control of the vehicle and sends a completion notification to the user.

[0043] The vehicle parking pose correction method of this disclosure constructs a parking space corner point set to store the reference coordinates of different parking space corner points. The reference coordinates are the position coordinates of the parking space corner points in the world coordinate system. During the vehicle parking process, the current coordinates of the current parking space corner point and the current vehicle pose are obtained, and the reference coordinates matching the current parking space corner point are determined by querying the parking space corner point set. The current coordinates are compared with the reference coordinates. If the coordinate difference between the current coordinates and the reference coordinates is greater than a preset value, the current vehicle pose is corrected based on the coordinate difference to obtain the corrected pose, so as to facilitate... The vehicle's parking path is updated using the corrected pose. This allows the vehicle's pose to be corrected based on the positional differences of the corner points of the same parking space in the world coordinate system, thus offsetting the cumulative error of the vehicle's pose. Finally, the corrected high-precision pose is output for updating the parking path, ensuring accurate positioning and parking for the automatic parking assistance system. This solution can be implemented using the vehicle's existing sensor configuration without additional cost. It is low-cost and robust, and can overcome most of the positioning accuracy reduction caused by ground bumps, gear changes, etc.

[0044] In one alternative embodiment of this disclosure, such as Figure 2 As shown, based on the aforementioned embodiments, the set of parking space corner points can be constructed through the following steps: Step 201: During vehicle operation, detect whether any parking space corners enter the target observation area of ​​the vehicle surround view camera.

[0045] The target observation area can be the optimal observation area determined by the center of the vehicle's rear axle as a reference point in the bird's-eye view generated from images captured by the surround-view camera. This area can be circular or rectangular. Extensive real-vehicle verification has revealed that the APA system based on the surround-view camera has an optimal observation area. The principle behind this is that the wide-angle lens of the surround-view camera causes barrel distortion at the image edges, while the center of the image is within the coverage area of ​​the lens's principal optical axis. The incident angle of light is close to perpendicular, resulting in high imaging linearity and accurate reproduction of object size and position. When the vehicle body is perpendicular to the parking space, the side camera has an orthographic projection view of the parking space, with the parking space line being a standard straight line and the corners being clear right angles, which reduces the misjudgment rate. At the same time, when the pixels in this area are converted into a bird's-eye view through inverse perspective mapping (IPM), the mapping to real-world coordinates is closer to linear, significantly improving the conversion accuracy. Therefore, the environmental position information observed at this time is the most accurate. Based on the above findings, in this embodiment of the disclosure, after the vehicle activates the APA function, the surround-view camera acquires images and identifies available parking spaces, extracts the corner points and center lines of the parking spaces. The system determines the optimal observation area based on the rule that "the longitudinal axis of the vehicle body is perpendicular to the center line of the parking space, and the main optical axis of the side camera is directly opposite the center of the parking space." Subsequently, it combines wheel speed and steering angle sensor data to obtain the vehicle body pose, generates a guidance path, and controls the power and steering system through the CAN bus to precisely guide the vehicle body to the target area. After the vehicle body is in place, it extracts the precise pixel coordinates of the corner points of the parking spaces, combines the camera calibration parameters and DR pose information, and maps them to the real-world coordinate system. Finally, based on these precise coordinates, a segmented path planning strategy is used to generate a parking path.

[0046] In this embodiment, during the vehicle's movement, as the vehicle approaches the parking space, it can maintain the vehicle body perpendicular to the parking space. The vehicle's surround-view camera collects images in real time and converts them into a bird's-eye view. Combined with a semantic segmentation algorithm, the corner points of the parking space are identified, and it is determined whether a parking space corner point appears within the target observation area in the bird's-eye view.

[0047] Step 202: If a parking space corner point is detected to enter the target observation area, obtain the first visual coordinates of the first parking space corner point that enters the target observation area.

[0048] In this embodiment, when a parking space corner point (referred to as the first parking space corner point for ease of description and distinction) is detected within the target observation area, the visual coordinates of the first parking space corner point (referred to as the first visual coordinates for ease of description and distinction) are obtained. The first visual coordinates are pixel coordinates identified from the bird's-eye view using a semantic segmentation algorithm.

[0049] Step 203: Based on the vehicle's trajectory at the current moment, calculate the pose and convert the first visual coordinates into the first position coordinates in the world coordinate system.

[0050] Among them, the trajectory estimation pose is the vehicle pose calculated and output by the DR system based on the current wheel speed, steering angle and other sensor data.

[0051] In this embodiment, after obtaining the first visual coordinates of the first parking space corner point, the first visual coordinates can be mapped to the real world coordinate system by combining the camera calibration parameters of the vehicle surround view camera and the pose calculated from the trajectory. This yields the real position coordinates of the first parking space corner point in the world coordinate system (referred to as the first position coordinates for ease of description and distinction). By constraining the information to the ground, and given the known camera extrinsic parameters, relatively accurate geometric information can be obtained through surround view observation of the ground. Therefore, it has high positioning accuracy and ensures the accuracy of the position coordinates.

[0052] It should be noted that the conversion from pixel coordinates to world coordinates is a relatively mature algorithm, and this publication will not elaborate on it.

[0053] Step 204: Determine the first reference coordinates corresponding to the first parking space corner point based on the first position coordinates, and construct a set of parking space corner points based on the first reference coordinates.

[0054] In this embodiment, after determining the first position coordinates of the first parking space corner point in the world coordinate system, the reference coordinates corresponding to the first parking space corner point (referred to as the first reference coordinates for ease of description and distinction) can be determined based on the first position coordinates. For example, the obtained first position coordinates can be directly used as the reference coordinates of the first parking space corner point, and then the set of parking space corner points can be determined based on the first reference coordinates.

[0055] As an example, the parking space corner point set records the correspondence between the corner point identifiers of different parking space corner points and their actual position coordinates in the world coordinate system. After obtaining the first reference coordinates of the first parking space corner point, the correspondence between the corner point identifier of the first parking space corner point (referred to as the first corner point identifier for ease of description and distinction) and the first reference coordinates is established, thus constructing the parking space corner point set.

[0056] As an example, the parking space corner point set can record the location information of different parking spaces. In this case, after determining the first reference coordinate of the first parking space corner point, the reference coordinates of the four parking space corner points (including the first parking space corner point) of the parking space to which the first parking space corner point belongs can be further determined, that is, the location information of the parking space is obtained, the location information of the parking space is stored, and the parking space corner point set is obtained.

[0057] The vehicle parking pose correction method of this disclosure obtains the visual coordinates of a parking space corner point when a parking space corner point is detected in the target observation area, and maps the visual coordinates to the world coordinate system based on the current trajectory to obtain the corresponding position coordinates. Then, the reference coordinates are determined based on the position coordinates, and a set of parking space corner points is constructed based on the reference coordinates. Thus, the reference coordinates of the parking space corner points are collected and stored, providing data support for subsequent correction of the DR pose based on the difference between the observed real coordinates of the current parking space corner point and the reference coordinates.

[0058] To ensure the accuracy of the reference coordinates and thus improve the pose correction accuracy, in one optional embodiment of this disclosure, multiple true coordinates of the same parking space corner point can be determined based on multiple frames of images. The reference coordinates of the parking space corner point, determined based on these multiple true coordinates, are used to construct a set of parking space corner points. Thus, as... Figure 3 As shown, based on the aforementioned embodiments, the set of parking space corner points can be constructed through the following steps: Step 301: During vehicle operation, detect whether any parking space corners enter the target observation area of ​​the vehicle surround view camera.

[0059] It should be noted that, in this embodiment, the description of step 301 can be found in the relevant explanation of step 201 in the foregoing embodiment, and the implementation principle is similar, so it will not be repeated here.

[0060] Step 302: If a parking space corner point is detected to enter the target observation area, acquire a multi-frame bird's-eye view of the second parking space corner point that has entered the target observation area within the target observation area.

[0061] In this embodiment, when a parking space corner point (referred to as the second parking space corner point for ease of description and distinction) appears within the target observation area, multiple frames of bird's-eye view images of the second parking space corner point within the target observation area can be acquired. For example, if the vehicle surround-view camera acquires 10 frames of images per second, then 10 frames of bird's-eye view images are obtained per second. When the second parking space corner point is first observed to appear in the target observation area, 10 consecutive frames of bird's-eye view images, including the current frame, can be acquired. From these, all bird's-eye view images of the second parking space corner point within the target observation area can be selected, and some or all of the selected bird's-eye view images can be used as multiple frames of bird's-eye view images. For example, 3 or 5 frames of bird's-eye view images can be selected from these.

[0062] Step 303: Based on the timestamps corresponding to the multiple bird's-eye view images, obtain the trajectory estimation pose corresponding to each bird's-eye view image.

[0063] In this embodiment, the timestamp generated for each frame of bird's-eye view is obtained. Based on the timestamps of each bird's-eye view, the DR pose with the same timestamp or the DR pose whose timestamp is closest to the timestamp of the bird's-eye view can be obtained, thus obtaining the trajectory estimation pose corresponding to each frame of bird's-eye view.

[0064] Step 304: Based on the trajectory corresponding to each frame of the bird's-eye view, calculate the pose and convert the second visual coordinates of the second parking space corner point in the corresponding bird's-eye view into the second position coordinates in the world coordinate system.

[0065] In this embodiment, for each frame of the bird's-eye view, the visual coordinates of the second parking space corner point in the bird's-eye view can be obtained through a semantic segmentation algorithm (referred to as the second visual coordinates for ease of description and differentiation). Then, combined with the pose calculated from the flight path corresponding to the bird's-eye view and the camera calibration parameters, the second visual coordinates are mapped to the real world coordinate system to obtain the real coordinates of the second parking space corner point in the world coordinate system (referred to as the second position coordinates for ease of description and differentiation). Thus, multiple second position coordinates corresponding to the second parking space corner point can be obtained. By constraining the information to the ground, and given the camera's extrinsic parameters, relatively accurate geometric information can be obtained through panoramic observation of the ground, resulting in high positioning accuracy and ensuring the accuracy of the position coordinates.

[0066] Step 305: Determine the second reference coordinates corresponding to the corner point of the second parking space based on multiple second position coordinates, and construct a set of parking space corner points based on the second reference coordinates.

[0067] In this embodiment, after obtaining multiple second position coordinates corresponding to the second parking space corner point, the reference coordinates (referred to as second reference coordinates for ease of description and distinction) corresponding to the second parking space corner point can be determined based on these multiple second position coordinates. For example, the average of the multiple second position coordinates can be calculated, and the average coordinates can be used as the reference coordinates of the second parking space corner point; or, for another example, the median of the multiple second position coordinates can be calculated, and the median coordinates can be used as the reference coordinates of the second parking space corner point. Furthermore, the set of parking space corner points is determined based on the second reference coordinates.

[0068] As an example, the parking space corner point set records the correspondence between the corner point identifiers of different parking space corner points and their actual position coordinates in the world coordinate system. After obtaining the second reference coordinates of the second parking space corner point, the correspondence between the corner point identifier of the second parking space corner point (referred to as the second corner point identifier for ease of description and distinction) and the second reference coordinates is established, thus constructing the parking space corner point set.

[0069] As an example, the parking space corner point set can record the location information of different parking spaces. In this case, after determining the second reference coordinates of the second parking space corner point, the reference coordinates of the four parking space corner points (including the second parking space corner point) of the parking space to which the second parking space corner point belongs can be further determined, that is, the location information of the parking space is obtained, the location information of the parking space is stored, and the parking space corner point set is obtained.

[0070] The vehicle parking pose correction method of this disclosure, when a parking space corner point is detected in the target observation area, acquires multiple frames of bird's-eye view images of the parking space corner point located in the target observation area, and obtains the corresponding trajectory-based pose estimation based on the timestamp of each frame of bird's-eye view image. For each frame of bird's-eye view image, the visual coordinates of the parking space corner point in the corresponding bird's-eye view image are mapped to the world coordinate system based on the corresponding trajectory-based position estimation to obtain the real position coordinates. Then, the reference coordinates of the parking space corner point are determined based on multiple position coordinates to construct a set of parking space corner points. This realizes the use of multiple determined real coordinates to determine the reference coordinates of the parking space corner point, which can improve the accuracy of the reference coordinates and thus improve the pose correction accuracy.

[0071] In one optional embodiment of this disclosure, based on the foregoing embodiments, when obtaining the current coordinates of the current parking space corner point during the vehicle parking process, the visual coordinates of the current parking space corner point entering the target observation area during the vehicle parking process can be obtained first (referred to as the third visual coordinates for ease of description and distinction), and the vehicle's current trajectory estimation pose can be obtained. For example, the pose output by the DR system at the current moment can be obtained as the current trajectory estimation pose. Then, based on the current trajectory estimation pose, the third visual coordinates are transformed to the world coordinate system to obtain the true position coordinates of the current parking space corner point in the world coordinate system, which is recorded as the current coordinates. When performing coordinate transformation, the camera calibration parameters of the surround-view camera can be combined for coordinate transformation. Since the visual coordinates are the coordinates of the parking space corner point in the bird's-eye view, constraining the information to the ground, and given the known camera calibration parameters, relatively accurate geometric information can be obtained by observing the ground through surround-view observation. Therefore, the transformed true position coordinates have high positioning accuracy, achieving high-precision positioning without configuring high-cost sensors such as LiDAR and positioning systems, without increasing additional costs while also having good robustness.

[0072] In one optional embodiment of this disclosure, after obtaining the corrected pose, the fourth visual coordinates corresponding to the parking space to be parked can be converted into position coordinates in the world coordinate system (referred to as third position coordinates for ease of description and distinction) based on the corrected pose. The parking space to be parked is the parking space selected by the user. The fourth visual coordinates of the parking space to be parked can be extracted and saved by the vehicle perception system from the images captured by the vehicle surround view camera. There can be multiple fourth visual coordinates, and there are also multiple corresponding third position coordinates. Next, the target position of the parking space to be parked is updated based on the third position coordinates, that is, the target position of the parking space to be parked is updated to the third position coordinates. Furthermore, based on the corrected pose and the target position, the parking path of the vehicle is updated. For example, the vehicle system can call the corrected pose and the updated parking space position information to replan the optimal parking path. This path can accurately match the parking space size and vehicle posture to ensure a smooth and deviation-free driving trajectory. By obtaining the modified pose, further updating the target position of the parking space after the modified pose is obtained, and then updating the parking path based on the modified pose and the updated position of the parking space, the path accuracy can be improved, thereby ensuring that the vehicle is accurately parked in the parking space.

[0073] Figure 4 This is a schematic diagram illustrating the process of parking a vehicle in a parking space, as an exemplary embodiment of this disclosure. Figure 4 As shown, when a user drives their vehicle into the parking lot, the vehicle perception system activates. Surround-view cameras around the vehicle capture real-time images of the surrounding environment. Using semantic segmentation algorithms, core features such as parking space lines and corner points are extracted from the images to identify and search for available parking spaces. This generates a candidate parking space list, which is then sent to the vehicle's infotainment system. The system displays this list to the user for selection. Once the user selects a parking space from the candidate list, the APA (Automatic Parking Assist) function is activated, transferring vehicle control from the user to the infotainment system. The infotainment system then uses the initially perceived location of the parking space (i.e., the parking space's position) to determine the parking space's location. Figure 4The system confirms the target location in the world coordinate system and plans an initial parking route. Simultaneously, it completes two core initialization operations: zeroing the origin of the real-world coordinate system and starting the trajectory prediction (DR) system to continuously update the vehicle's pose based on wheel speed and IMU data. Through preset control codes, the vehicle's infotainment system controls the vehicle's movement based on the initial parking route, adjusting the vehicle's attitude in real time. The system also simultaneously activates the "optimal observation area monitoring mechanism." When any parking space corner point enters the optimal observation area of ​​the side-view camera during vehicle movement (the vehicle's longitudinal axis is perpendicular to the baseline of the parking space corner point, and the parking space corner point is located in the distortion-free area at the center of the camera image), observation preparation is triggered. The vehicle accurately extracts the visual coordinate system pose of the feature corner point (i.e., the parking space corner point) that is currently entering the optimal area. Combined with the vehicle body pose output by the DR system, this is converted into absolute coordinates D1 in the real-world coordinate system (labeling the corner point's identifier and observation time T1). The system uses D1 as the reference coordinate to establish a "feature corner point - reference coordinate" mapping library (i.e., the corner point coordinate mapping library). This operation is performed on all feature corner points passing through the optimal observation area, completing the acquisition and storage of multiple sets of reference coordinates. The vehicle system updates the target position's world coordinates (i.e., updates the pre-parking space position) and the parking path, using the multi-feature reference established during the initial observation as a reference. The system continuously matches the corner point identifier of the currently observed corner point with the corner point coordinate mapping library. Whenever a feature corner point enters the optimal observation area, it is matched against all stored corner points. If no match is found with a stored corner point, it is considered the first observation; if a stored feature corner point is detected, it is considered a second observation, and the process proceeds to the next step (labeling the observation time T2).

[0074] For each feature corner point that triggers a secondary observation, its visual coordinate system pose at time T2 is extracted using the same method as the initial observation. The real-world coordinates D2 are then calculated by combining this with the vehicle body pose output by the current DR (Digital Recognition) system. The coordinate difference between the corner point and D2 is determined. If the coordinate difference exceeds a preset threshold, it is marked as an "error point," indicating a significant accumulated error in the DR system, requiring correction. During pose correction, the error increment of the DR pose is calculated in reverse based on the coordinate difference between D1 and D2, thereby correcting the pose parameters currently output by the DR system, ensuring that the corrected D2 perfectly coincides with D1. Simultaneously, using the corrected DR pose as a reference, the position information of the pre-parking space in the real-world coordinate system is updated. The vehicle system calls the corrected DR pose data and the updated parking space position information to replan the optimal parking path. This path accurately matches the parking space size and vehicle posture, ensuring a smooth and deviation-free driving trajectory. The vehicle system outputs control commands to the power, steering, and braking systems via the CAN bus, driving the vehicle along the planned path. When a corner point is marked as an error point again, the above process is repeated to correct the DR pose, update the position of the pre-parking space, and replan the parking path. The entire process relies on the real-time observation of the surround view camera and the collaborative feedback of the DR pose correction to complete accurate parking. After parking, the system releases control of the vehicle and sends a completion prompt to the user.

[0075] To achieve the above embodiments, this disclosure also provides a vehicle parking posture correction device, which can be implemented in software and / or hardware and can be integrated into an electronic device.

[0076] Figure 5 This is a schematic diagram of the vehicle parking posture correction device provided in one embodiment of the present disclosure, as shown below. Figure 5 As shown, the vehicle parking posture correction device 40 may include: an acquisition module 410, a determination module 420, and a correction module 430.

[0077] The acquisition module 410 is used to acquire the current coordinates of the corner point of the current parking space and the current position of the vehicle during the parking process. The determination module 420 is used to query the parking space corner point set and determine the reference coordinates that match the current parking space corner point. The parking space corner point set is used to store the reference coordinates of different parking space corner points. The reference coordinates are the position coordinates of the parking space corner point in the world coordinate system. The correction module 430 is used to correct the current pose of the vehicle based on the coordinate difference when the coordinate difference between the current coordinate and the reference coordinate is greater than a preset value, so as to obtain the corrected pose. The corrected pose is used to update the parking path of the vehicle.

[0078] Optionally, the vehicle parking posture correction device 40 further includes a first database module for: During vehicle operation, detect whether any parking space corners enter the target observation area of ​​the vehicle's surround view camera; If a parking space corner point is detected to enter the target observation area, the first visual coordinates of the first parking space corner point that enters the target observation area are obtained. Based on the vehicle's current trajectory, the pose is calculated and the first visual coordinates are converted into the first position coordinates in the world coordinate system. The first reference coordinates corresponding to the first parking space corner point are determined based on the first position coordinates, and a set of parking space corner points is constructed based on the first reference coordinates.

[0079] Optionally, the vehicle parking posture correction device 40 also includes a second database module for: During vehicle operation, detect whether any parking space corners enter the target observation area of ​​the vehicle's surround view camera; If a parking space corner point is detected to enter the target observation area, acquire a multi-frame bird's-eye view of the second parking space corner point that has entered the target observation area within the target observation area; Based on the timestamps corresponding to multiple bird's-eye view images, the pose of the flight path corresponding to each bird's-eye view image is obtained; Based on the trajectory corresponding to each frame of the bird's-eye view, the pose is calculated and the second visual coordinates of the second parking space corner point in the corresponding bird's-eye view are converted into the second position coordinates in the world coordinate system. The second reference coordinates corresponding to the corner points of the second parking space are determined based on multiple second position coordinates, and a set of parking space corner points is constructed based on the second reference coordinates.

[0080] Further optionally, the acquisition module 410 is also used for: Obtain the third visual coordinates of the current parking space corner point that enters the target observation area during the vehicle parking process; Obtain the vehicle's current trajectory and calculate its pose; Based on the current trajectory, the pose is calculated and the third-view coordinates are transformed into the world coordinate system to obtain the current coordinates.

[0081] Optionally, the vehicle parking posture correction device 40 also includes an update module for: Based on the corrected pose, the fourth visual coordinates of the parking space to be parked are converted into the third position coordinates in the world coordinate system; Update the target position of the parking space to be parked based on the third position coordinates; The parking path of the vehicle is updated based on the corrected pose and target position.

[0082] Optionally, the correction module 430 is also used for: With the goal of aligning the corrected position coordinates of the current parking space corner point in the world coordinate system with the reference coordinates, the pose error increment is determined based on the coordinate difference. The vehicle's current pose is corrected based on the pose error increment to obtain the corrected pose.

[0083] Optionally, the vehicle parking posture correction device 40 also includes a storage module for: If no matching reference coordinates are found in the parking space corner point set, the current coordinates are stored in the parking space corner point set.

[0084] The vehicle parking posture correction device provided in this disclosure, applicable to electronic devices, can execute the vehicle parking posture correction method provided in this disclosure, and has the corresponding functional modules and beneficial effects of the method. Content not described in detail in the device embodiments of this disclosure can be referred to the description in any method embodiment of this disclosure.

[0085] This disclosure also provides an electronic device, including a processor and a memory; the processor executes the steps of the aforementioned vehicle parking pose correction method embodiments by calling programs or instructions stored in the memory. To avoid repetition, these steps will not be repeated here.

[0086] This disclosure also provides a computer-readable storage medium that is non-transitory and stores a program or instructions that cause a computer to perform the steps of the aforementioned vehicle parking pose correction method embodiments. To avoid repetition, these steps will not be repeated here.

[0087] This disclosure also provides a computer program product for executing the steps of the aforementioned vehicle parking pose correction methods in various embodiments.

[0088] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0089] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for correcting the position and orientation of a vehicle when parking, characterized in that, The method includes: Obtain the current coordinates of the corner point of the current parking space and the current pose of the vehicle during the parking process; Query the parking space corner point set to determine the reference coordinates that match the current parking space corner point. The parking space corner point set is used to store the reference coordinates of different parking space corner points. The reference coordinates are the position coordinates of the parking space corner point in the world coordinate system. If the coordinate difference between the current coordinates and the reference coordinates is greater than a preset value, the current pose of the vehicle is corrected based on the coordinate difference to obtain a corrected pose, which is used to update the parking path of the vehicle.

2. The method according to claim 1, characterized in that, The set of parking space corner points is determined in the following way: During the vehicle's operation, it is detected whether any parking space corners enter the target observation area of ​​the vehicle's surround-view camera. If the presence of a parking space corner point entering the target observation area is detected, the first visual coordinates of the first parking space corner point entering the target observation area are obtained; Based on the vehicle's current trajectory, the pose is calculated, and the first visual coordinates are converted into the first position coordinates in the world coordinate system. The first reference coordinates corresponding to the first parking space corner point are determined based on the first position coordinates, and the parking space corner point set is constructed based on the first reference coordinates.

3. The method according to claim 1, characterized in that, The set of parking space corner points is determined in the following way: During the vehicle's operation, it is detected whether any parking space corners enter the target observation area of ​​the vehicle's surround-view camera. If a parking space corner point is detected to enter the target observation area, a multi-frame bird's-eye view of the second parking space corner point entering the target observation area is obtained within the target observation area. Based on the timestamps corresponding to the multiple bird's-eye view images, the trajectory pose is calculated for each bird's-eye view image. Based on the trajectory corresponding to each frame of the bird's-eye view, the pose is calculated, and the second visual coordinates of the second parking space corner point in the corresponding bird's-eye view are converted into the second position coordinates in the world coordinate system. The second reference coordinates corresponding to the corner points of the second parking space are determined based on multiple second position coordinates, and the set of parking space corner points is constructed based on the second reference coordinates.

4. The method according to claim 2 or 3, characterized in that, The process of obtaining the current coordinates of the corner point of the current parking space during vehicle parking includes: Obtain the third visual coordinates of the current parking space corner point that enters the target observation area during the vehicle parking process; Obtain the current trajectory and estimated pose of the vehicle; Based on the current trajectory, the pose is calculated and the third visual coordinates are transformed into the world coordinate system to obtain the current coordinates.

5. The method according to claim 1, characterized in that, After correcting the vehicle's current pose based on the coordinate difference to obtain the corrected pose, the method further includes: Based on the corrected pose, the fourth visual coordinates of the parking space to be parked are converted into the third position coordinates in the world coordinate system. The target position of the parking space to be parked is updated based on the third position coordinates; Based on the corrected pose and the target position, the parking path of the vehicle is updated.

6. The method according to claim 1, characterized in that, The step of correcting the vehicle's current pose based on the coordinate difference to obtain the corrected pose includes: With the goal of aligning the corrected position coordinates of the current parking space corner point in the world coordinate system with the reference coordinates, the pose error increment is determined based on the coordinate difference. The vehicle's current pose is corrected based on the pose error increment to obtain the corrected pose.

7. The method according to any one of claims 1-3, characterized in that, The method further includes: If no reference coordinates matching the current parking space corner point are found from the parking space corner point set, the current coordinates are stored in the parking space corner point set.

8. A vehicle parking posture correction device, characterized in that, include: The acquisition module is used to acquire the current coordinates of the corner point of the current parking space and the current pose of the vehicle during the parking process; The determination module is used to query the set of parking space corner points and determine the reference coordinates that match the current parking space corner point. The set of parking space corner points is used to store the reference coordinates of different parking space corner points. The reference coordinates are the position coordinates of the parking space corner point in the world coordinate system. The correction module is used to correct the current pose of the vehicle based on the coordinate difference when the coordinate difference between the current coordinate and the reference coordinate is greater than a preset value, so as to obtain the corrected pose. The corrected pose is used to update the parking path of the vehicle.

9. An electronic device, characterized in that, include: Processor and memory; The processor executes the vehicle parking pose correction method as described in any one of claims 1 to 7 by calling the program or instructions stored in the memory.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program or instructions that cause a computer to perform the vehicle parking pose correction method as described in any one of claims 1 to 7.

Images