Autonomous parking method and device based on pole load bird's-eye view

Abstract

This invention discloses an autonomous parking method and device based on a pole-mounted bird's-eye view, comprising: identifying and segmenting target vehicles, pedestrians, parking spaces, and vehicle-accessible areas based on a bird's-eye view of the parking space obtained by a pole-mounted sensing module; locating pedestrians and vehicles using an adjustment function; planning an unmanned vehicle parking path based on the identified target information, the vehicle-accessible area, and geometric connections; tracking and controlling the parking path and adjusting the vehicle's pose to achieve autonomous parking. This invention achieves collaborative perception and collaborative decision-making control through information interaction, ignoring interference from complex ground environments, efficiently acquiring location information, and realizing unmanned parking planning in outdoor parking lots under different environments. It greatly expands the vehicle's perception range and improves its perception capabilities, achieving fully autonomous unmanned parking without requiring vehicle environmental perception, thus improving the safety and stability of autonomous driving and accelerating its large-scale deployment.

Description

Technical Field

[0001] This invention relates to the technical field of automatic parking, and more particularly to an autonomous parking method and device based on a pole-mounted bird's-eye view. Background Technology

[0002] Against the backdrop of continuous development of new-generation information technology, "communication" between automobiles and urban roads has become possible. Autonomous driving technology, as a product of the deep integration of the automotive industry with new-generation information technologies such as artificial intelligence, big data, and the Internet of Things, has become the development direction of the automotive industry.

[0003] Currently, some long-tail technical issues in autonomous driving remain unresolved, limiting its application scenarios. Parking is an important urban application scenario, and existing research mostly uses vehicle-mounted radar or cameras for perception and leverages vehicle-mounted computing power for path planning. This method suffers from limitations such as a small perception area, significant susceptibility to ground conditions, and high cost per vehicle, exhibiting certain limitations and instability in parking scenarios. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, this invention provides an autonomous parking method and device based on a pole-mounted bird's-eye view to solve the problems of existing parking methods, such as small sensing area, large influence of environmental factors, high cost, resulting in strong limitations and poor stability in car parking.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] In a first aspect, embodiments of the present invention provide an autonomous parking method based on a pole-mounted bird's-eye view, comprising:

[0009] Based on the bird's-eye view of the parking space obtained by the pole-mounted sensing module, the target vehicle, pedestrian, parking space and vehicle driving area are identified and segmented, and the pedestrian and vehicle targets are located by adjusting the function.

[0010] Based on the identified target information, the vehicle's drivable area, and geometric connections, a parking path for the unmanned vehicle is planned.

[0011] The parking path is tracked and controlled, and the vehicle's pose is adjusted to achieve autonomous parking for unmanned vehicles. As a preferred embodiment of the autonomous parking method based on a pole-mounted bird's-eye view described in this invention, the adjustment function includes:

[0012] The models of different vehicles in the obtained bird's-eye view of the parking space are simplified, and the standard pose of the vehicles in the original image is set.

[0013] Establish the first mapping f c1 Based on the difference between the current pose and the standard pose, the center of the current bounding box is mapped to the center of the bounding box under the standard pose.

[0014] Establish the second mapping f c2 The center point of the recognition box in the standard pose is mapped to the coordinates on the transformed bird's-eye view, and the coordinates are the position of the car in the vertical overhead view.

[0015] Establish a mapping of pedestrian locations from the original bird's-eye view to the transformed bird's-eye view. p , is represented as:

[0016] (x Bp ,y Bp )=(x p ,y p +h p / 2)+o

[0017] Among them, (x Bc ,y Bc ),l Bc ,w Bc These represent the vehicle's coordinates, length, and width in the transformed bird's-eye view; (x p ,y p h is the center of the pedestrian recognition bounding box in the original image. p Let x be the height of the pedestrian recognition bounding box. Bp ,y Bp ) represents the coordinates of the pedestrian in the bird's-eye view, and o is a small vector.

[0018] As a preferred embodiment of the autonomous parking method based on pole-mounted bird's-eye view described in this invention, the identification and segmentation of the target vehicle, pedestrians, parking spaces, and the vehicle's drivable area includes:

[0019] The target recognition uses YOLOv5s to identify vehicles, parking space corners, and pedestrians, and sets the pixel coordinate system of the transformed bird's-eye view as the basic coordinate system to obtain the coordinates of key objects for localization.

[0020] The drivable area segmentation is based on the collected overhead view of the parking lot, and the fine semantic segmentation label data is obtained by manual annotation using a labeling tool. The data is used as the dataset of the parking lot scene to train the BiseNetV2 model.

[0021] The bird's-eye view is inferred as a batch using bisenenetV2 to obtain free space segmentation;

[0022] The free space segmentation mainly refers to the boundary between the road surface and objects with height.

[0023] As a preferred embodiment of the autonomous parking method based on pole-mounted bird's-eye view described in this invention, it further includes:

[0024] Parking space identification and positioning only considers single parking space scenarios;

[0025] Four corner points are identified: a quadrilateral parking space is generated; two corner points are identified: the distance between the corner points is used to determine whether the parking space is parallel or perpendicular, and two possible quadrilateral parking spaces are generated.

[0026] The selection of parking spaces is based on their relative position to the vehicle. By calculating the difference in grayscale values ​​within the parking spaces, it is determined whether the selected parking spaces are occupied.

[0027] As a preferred embodiment of the autonomous parking method based on pole-mounted bird's-eye view described in this invention, the planning of the unmanned vehicle parking path includes: planning the unmanned vehicle parking path based on a map with target information and a simplified two-degree-of-freedom model of the vehicle. The path planning method is based on the geometric connection of arcs and straight lines, and is mainly divided into vertical and parallel parking according to the actual situation.

[0028] As a preferred embodiment of the autonomous parking method based on pole-mounted bird's-eye view described in this invention, it further includes:

[0029] The planning method determines the choice between parallel and perpendicular parking based on the length-width relationship of the parking space, as follows:

[0030] d=((Rl Bc / 2-δ1) 2 -(Rw p / 2) 2 ) 1 / 2 h min =Rd;

[0031] D min =((R+l Bc / 2) 2 +(w Bc -h ro ) 2 ) 1 / 2 -d+δ2

[0032] Among them, l Bc ,w Bc These represent the vehicle's length and width in the transformed bird's-eye view, respectively, where R is the minimum turning radius at the rear axle center, and h is the distance between the two values. ro For rear overhang, w p D is the width of the parking space. min The minimum lane width is given by δ1, the first safety distance is given by δ2, the second safety distance is given by δ2, d is the distance from the turning center to the two corner points (i.e., the line connecting the nearest vehicles), and h is the distance from the turning center to the two corner points. min The minimum distance between the car and the nearest corner point;

[0033] C-shaped parking is represented as:

[0034] h∈[h min ,h min +h f -D min ]

[0035] Among them, h f The feasible region height is h, and h is the distance between the car and the nearest corner point.

[0036] The A-frame parking method is represented as follows:

[0037] h∈(h f -R-δ2,h min )

[0038] The heading angle for a V-shaped parking maneuver is expressed as:

[0039] θ = arccos[(h+R) / 2R]

[0040] Where θ is the heading angle;

[0041] The parallel berthing heading angle is expressed as:

[0042] θ=arccos[2R-hw Bc / 2] / 2R

[0043] As a preferred embodiment of the autonomous parking method based on pole-mounted bird's-eye view described in this invention, the method includes: tracking and controlling the parking path and adjusting the vehicle's pose, comprising:

[0044] Based on the path planning, a reference point is set, and the reference front wheel steering angle δ is calculated. r Reference speed v r Reference heading angle θ r ;

[0045] The state-space equation is established based on the target vehicle information and is expressed as follows:

[0046] X(k+1) = AX(k) + Bu(k)

[0047] Where X(k) is the error of the spatial state at time k, u(k) is the difference between the control quantity and the reference value at time k, k = 1, 2, ..., N, N is the number of reference points set, A is the first matrix representing the sampling interval, reference speed, reference heading angle and wheelbase at time k, and B is the second matrix representing the sampling interval, reference speed, reference heading angle and wheelbase at time k.

[0048] The optimal control for multi-objective optimization is expressed as:

[0049] J=Σ(X T QX(k)+u T Gu(k))

[0050] Where J is the objective function, which is the weighted sum of the cumulative tracking deviation and the cumulative control input during the tracking process; Q is the first weight matrix; and G is the second weight matrix.

[0051] The optimal control rate is obtained by solving LQR:

[0052] u = -KX

[0053] K is determined by A, B and the solution P of the Riccati equation;

[0054] Based on the control law, path tracking control is performed, and the vehicle status is updated in real time to achieve autonomous parking of unmanned vehicles.

[0055] Secondly, the present invention provides an autonomous parking device based on a pole-mounted bird's-eye view, for applying the autonomous parking method based on a pole-mounted bird's-eye view as described in any one of claims 1-7, wherein the autonomous parking device based on a pole-mounted bird's-eye view comprises:

[0056] The pole-mounted bird's-eye view perception module is used for overhead perception of parking spaces and vehicles, providing perception data for dynamic planning of parking routes. The pole-mounted bird's-eye view perception module takes into account the height of the light pole. From the bird's-eye view, vehicles are large while parking space corners and pedestrians are small. In the Neck structure of YOLOv5s, Bi-FPN is used to replace PAN-Net. Bi-FPN introduces weights to balance feature information at different scales.

[0057] The edge intelligent planning module performs parking path planning for unmanned vehicles based on a map rich in perceived target information generated by the perception module and a simplified two-degree-of-freedom model of the vehicle.

[0058] As a preferred embodiment of the autonomous parking device based on pole-mounted bird's-eye view described in this invention, it further includes:

[0059] The V2L wireless communication module is used to transmit the dynamically planned path to the autonomous vehicle's motion control module, enabling vehicle-road cooperative interaction.

[0060] The V2L wireless communication module is based on cellular communication C-V2X wireless communication technology and establishes a stable two-way wireless encrypted communication channel with the on-board unit (OBU).

[0061] The V2L wireless communication module communicates with the on-board unit (OBU) to obtain parking demand events and initiate parking collaborative perception and path planning decisions based on a bird's-eye view.

[0062] Based on the communication protocol, the collaborative sensing information is compressed and transmitted to the on-board unit (OBU);

[0063] The collaborative perception information consists of the planned path from the edge intelligent planning module and the map information generated by the edge perception module.

[0064] The On-Board Unit (OBU) decompresses the collaborative perception information and transmits it to the autonomous vehicle's motion control module via the vehicle network. The autonomous vehicle's motion control module then executes intelligent decision-making and motion planning tasks through the Vehicle Control Unit (VCU).

[0065] As a preferred embodiment of the autonomous parking device based on pole-mounted bird's-eye view described in this invention, it further includes:

[0066] The unmanned vehicle motion control module receives instructions from the V2L wireless communication module to perform parking path tracking control and vehicle posture adjustment control.

[0067] The unmanned vehicle motion control module receives cooperative sensing information from the light pole via V2L and transmits speed and front wheel angle signals to the VCU via vehicle communication to correct the vehicle's posture.

[0068] Parking path tracking control is based on vehicle kinematics and uses a linear quadratic regulator method to achieve path tracking control.

[0069] Compared with existing technologies, the beneficial effects of this invention are as follows: By achieving collaborative perception and collaborative decision-making control through information interaction, the vehicle and the site are combined, which can ignore the interference of complex ground environment, efficiently obtain location information, realize unmanned parking planning in outdoor parking lots in different environments, greatly expand the perception range of the car, improve the perception capability of the car, realize fully autonomous unmanned parking without the need for vehicle environmental perception, improve the safety and stability of autonomous driving and the speed of large-scale deployment, making it more feasible. Attached Figure Description

[0070] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0071] Figure 1 This is a flowchart illustrating an autonomous parking method and device based on a pole-mounted bird's-eye view, according to an embodiment of the present invention.

[0072] Figure 2 This is a schematic diagram of the module connection of an autonomous parking method and device based on a pole-mounted bird's-eye view according to an embodiment of the present invention;

[0073] Figure 3 This is a schematic diagram of the dual-end initialization of an autonomous parking method and device based on a pole-mounted bird's-eye view according to an embodiment of the present invention;

[0074] Figure 4 This is a schematic diagram of a pole-mounted bird's-eye view perception module of an autonomous parking method and device based on a pole-mounted bird's-eye view, according to an embodiment of the present invention.

[0075] Figure 5 This is a schematic diagram of the edge intelligent planning module of an autonomous parking method and device based on a pole-mounted bird's-eye view, according to an embodiment of the present invention.

[0076] Figure 6 This is a schematic diagram of a V2L wireless communication module for an autonomous parking method and device based on a pole-mounted bird's-eye view, according to an embodiment of the present invention. Detailed Implementation

[0077] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0078] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0079] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0080] This invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not adhering to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0081] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0082] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0083] Example 1

[0084] Reference Figure 1 This is one embodiment of the present invention, which provides an autonomous parking method based on a pole-mounted bird's-eye view, including:

[0085] S102: Based on the bird's-eye view of the parking space obtained by the pole-mounted sensing module, the target vehicle, pedestrian, parking space and vehicle driving area are identified and segmented, and the pedestrian and vehicle are located by adjusting the function.

[0086] Furthermore, the adjustment functions include:

[0087] The models of different vehicles in the obtained bird's-eye view of the parking space are simplified, and the standard pose of the vehicles in the original image is set.

[0088] Establish the first mapping f c1Based on the difference between the current pose and the standard pose, the center of the current bounding box is mapped to the center of the bounding box under the standard pose.

[0089] Establish the second mapping f c2 The center point of the recognition box in the standard pose is mapped to the coordinates on the transformed bird's-eye view, and the coordinates are the position of the car in the vertical overhead view.

[0090] Establish a mapping of pedestrian locations from the original bird's-eye view to the transformed bird's-eye view. p , is represented as:

[0091] (x Bp ,y Bp )=(x p ,y p +h p / 2)+o

[0092] Among them, (x Bc ,y Bc ),l Bc ,w Bc These represent the vehicle's coordinates, length, and width in the transformed bird's-eye view; (x p ,y p h is the center of the pedestrian recognition bounding box in the original image. p Let x be the height of the pedestrian recognition bounding box. Bp ,y Bp ) represents the coordinates of the pedestrian in the bird's-eye view, and o is a small vector.

[0093] It should be noted that the original bird's-eye view is used for vehicle and pedestrian identification and location, as well as for segmentation of drivable areas. After being transformed into a three-dimensional perspective, the original bird's-eye view is used for the identification and location of parking space corners. Finally, the key information gathered on the transformed image is used to generate a raster image, which facilitates the planning and parking training of autonomous parking.

[0094] It should also be noted that by using the bird's-eye view of the light pole to obtain an aerial view of vehicles and parking spaces, the aerial view has a wider field of view, greatly increasing the perception range and making the detection of feasible areas, parking spaces and pedestrians more immediate and accurate, with less interference from ground visual obstacles. The aerial view's characteristics avoid the distortion problem caused by converting 3D images to 2D over a larger area. After adjusting the constructor, the location information of the bird's-eye view can be used for path planning over a larger area, increasing parking efficiency and stability and providing support for autonomous parking.

[0095] Furthermore, the system identifies and segments target vehicles, pedestrians, parking spaces, and drivable areas, including:

[0096] Target recognition uses YOLOv5s to identify vehicles, parking space corners, and pedestrians. The pixel coordinate system of the transformed bird's-eye view is set as the basic coordinate system to obtain the coordinates of key objects and perform localization.

[0097] Driving area segmentation is based on collected overhead views of parking lots, which are manually annotated using annotation tools to obtain fine semantic segmentation label data. This data is used as a dataset for parking lot scenes to train the BiseNetV2 model.

[0098] The bird's-eye view is inferred as a batch using bisenenetV2 to obtain free space segmentation;

[0099] Free space segmentation mainly refers to the boundary between the road surface and objects with height.

[0100] It should be noted that, in order to overcome the latency caused by YOLOv5s detection, free space semantic segmentation, planning calculation, and wireless communication modules, a latency parameter t and an interval time T are introduced. The vehicle position is predicted based on two consecutive samples and used as the target recognition output (x). o ,y o );

[0101] The specific formula is as follows:

[0102] t = t r +t c +t t

[0103] (x o ,y o )=(x Bc2 -x Bc1 ,y Bc2 -y Bc1 )t / T+(x Bc2 ,y Bc2 )

[0104] Among them, t r For image processing latency, t c To calculate the delay, t t This refers to transmission delay.

[0105] It should also be noted that YOLOv5s's fast recognition feature can minimize latency, and combined with the prediction of discrete systems to overcome latency, it can quickly and accurately provide feedback on the planned route.

[0106] Specifically, parking space identification and positioning only considers the case of a single parking space;

[0107] Four corner points are identified: a quadrilateral parking space is generated; two corner points are identified: the distance between the corner points is used to determine whether the parking space is parallel or perpendicular, and two possible quadrilateral parking spaces are generated.

[0108] The selection of parking spaces is based on their relative position to the vehicle. By calculating the difference in grayscale values ​​within the parking spaces, it is determined whether the selected parking spaces are occupied.

[0109] It should be noted that the calculation of the grayscale difference value within the parking space marking area is necessary because the grayscale value changes significantly differ between when the parking space is occupied and when it is not occupied. Therefore, the grayscale difference value within the parking space marking area must first be calculated. Then, by setting an average threshold for the grayscale difference when the parking space is not occupied, the grayscale difference value within the parking space area is compared with this average threshold to preliminarily determine whether the parking space is occupied.

[0110] S104: Based on the identified target information, the vehicle's drivable area, and geometric connections, plan the parking path for the unmanned vehicle;

[0111] Furthermore, planning parking paths for autonomous vehicles includes: planning parking paths for autonomous vehicles based on maps with target information and simplified two-degree-of-freedom models of vehicles. The path planning method is based on the geometric connection of arcs and straight lines, and is mainly divided into perpendicular and parallel parking according to the actual situation.

[0112] Specifically, the planning method determines whether to use parallel or perpendicular parking based on the length and width relationship of the parking space, as shown below:

[0113] d=((Rl Bc / 2-δ1) 2 -(Rw p / 2) 2 ) 1 / 2 h min =Rd;

[0114] D min =((R+l Bc / 2) 2 +(w Bc -h ro ) 2 ) 1 / 2 -d+δ2

[0115] Among them, l Bc ,w Bc These represent the vehicle's length and width in the transformed bird's-eye view, respectively, where R is the minimum turning radius at the rear axle center, and h is the distance between the two values. ro For rear overhang, w p D is the width of the parking space. min The minimum lane width is given by δ1, the first safety distance is given by δ2, the second safety distance is given by δ2, d is the distance from the turning center to the two corner points (i.e., the line connecting the nearest vehicles), and h is the distance from the turning center to the two corner points. min The minimum distance between the car and the nearest corner point;

[0116] C-shaped parking is represented as:

[0117] h∈[h min ,hmin +h f -D min ]

[0118] Among them, h f The feasible region height is h, and h is the distance between the car and the nearest corner point.

[0119] The A-frame parking method is represented as follows:

[0120] h∈(h f -R-δ2,h min )

[0121] The heading angle for a V-shaped parking maneuver is expressed as:

[0122] θ = arccos[(h+R) / 2R]

[0123] Where θ is the heading angle;

[0124] The parallel berthing heading angle is expressed as:

[0125] θ=arccos[2R-hw Bc / 2] / 2R

[0126] S106: Track and control the parking path and adjust the vehicle's position to achieve autonomous parking for unmanned vehicles;

[0127] Furthermore, tracking and controlling the parking path and adjusting the vehicle's position includes:

[0128] Based on the path planning, a reference point is set, and the reference front wheel steering angle δ is calculated. r Reference speed v r Reference heading angle θ r ;

[0129] The state-space equation is established based on the target vehicle information and is expressed as follows:

[0130] X(k+1) = AX(k) + Bu(k)

[0131] Where X(k) is the error of the spatial state at time k, u(k) is the difference between the control quantity and the reference value at time k, k = 1, 2, ..., N, N is the number of reference points set, A is the first matrix representing the sampling interval, reference speed, reference heading angle and wheelbase at time k, and B is the second matrix representing the sampling interval, reference speed, reference heading angle and wheelbase at time k.

[0132] The optimal control for multi-objective optimization is expressed as:

[0133] J=Σ(X T QX(k)+u T Gu(k))

[0134] Where J is the objective function, which is the weighted sum of the cumulative tracking deviation and the cumulative control input during the tracking process; Q is the first weight matrix; and G is the second weight matrix.

[0135] The optimal control rate is obtained by solving LQR:

[0136] u = -KX

[0137] K is determined by A, B and the solution P of the Riccati equation;

[0138] Based on the control law, path tracking control is performed, and the vehicle status is updated in real time to achieve autonomous parking of unmanned vehicles.

[0139] Example 2

[0140] Reference Figures 2-6 As one embodiment of the present invention, an autonomous parking device based on a pole-mounted bird's-eye view is provided, comprising:

[0141] The pole-mounted bird's-eye view perception module is used for overhead perception of pedestrians, feasible areas, parking spaces and vehicles, providing perception data for dynamic planning of parking paths. The pole-mounted bird's-eye view perception module takes into account the height of the light pole. From the bird's-eye view, vehicles are large while parking space corners and pedestrians are small. In the Neck structure of YOLOv5s, Bi-FPN is used to replace PAN-Net. Bi-FPN introduces weights to balance feature information at different scales.

[0142] It should be noted that the pole-mounted bird's-eye view perception module includes an image processor and a target recognition and locator, which are used to acquire feasible domain, target vehicle and parking space, and pedestrian coordinate information under pixel coordinates;

[0143] Basic parameters are obtained from the initialization of the light pole. A bilateral filter is applied to preserve edges and reduce noise. The acquired bird's-eye view image is sharpened and target recognition and localization are performed. The positions of vehicles, parking spaces, dynamic pedestrians, and feasible areas are overlaid and mapped onto a map for motion trajectory planning.

[0144] It should also be noted that the pedestrian position is updated in real time. During the parking process, the pole-mounted bird's-eye view perception module can be used to identify and locate moving pedestrians, which is then transmitted to the vehicle via V2L, which is beneficial for subsequent vehicle movement decisions.

[0145] The edge intelligent planning module performs parking path planning for unmanned vehicles based on a map rich in perceived target information generated by the perception module and a simplified two-degree-of-freedom model of the vehicle.

[0146] It should be noted that the edge intelligent planning module and the autonomous vehicle motion control module fully utilize parameters from other modules and combine them with the actual parking scenario to make decisions, determine the parking method, plan the parking path, and achieve reverse parking. Specifically, path planning is simplified to determining the parking method based on the vehicle's lateral distance and planning the route based on the heading angle; motion control simplifies the vehicle model and implements path tracking based on the LQR method, exhibiting good performance in the transition process.

[0147] Furthermore, this also includes:

[0148] The V2L wireless communication module is used to transmit the dynamically planned path to the autonomous vehicle's motion control module, enabling vehicle-road cooperative interaction.

[0149] The V2L wireless communication module is based on the C-V2X cellular wireless communication technology and establishes a stable two-way wireless encrypted communication channel with the on-board unit (OBU).

[0150] The V2L wireless communication module communicates with the on-board unit (OBU) to obtain parking demand events and initiate parking collaborative perception and path planning decisions based on a bird's-eye view.

[0151] Based on the communication protocol, the collaborative sensing information is compressed and transmitted to the on-board unit (OBU);

[0152] Collaborative sensing information includes the planned path of the edge intelligent planning module and the map information generated by the edge sensing module.

[0153] The On-Board Unit (OBU) decompresses the collaborative perception information and transmits it to the autonomous vehicle's motion control module via the vehicle network. The autonomous vehicle's motion control module then executes intelligent decision-making and motion planning tasks through the Vehicle Control Unit (VCU).

[0154] Furthermore, this also includes:

[0155] The unmanned vehicle motion control module receives instructions from the V2L wireless communication module to perform parking path tracking control and vehicle posture adjustment control.

[0156] The autonomous vehicle motion control module receives cooperative sensing information from the light pole via V2L and transmits speed and front wheel steering angle signals to the VCU via vehicle communication to correct the vehicle's posture.

[0157] Parking path tracking control is based on vehicle kinematics and uses a linear quadratic regulator method to achieve path tracking control.

[0158] It should be noted that by making reasonable use of the bird's-eye view of the light pole, image recognition and positioning technology, V2X technology, and path planning and control technology, collaborative decision-making and planning between the vehicle and the site can be achieved, avoiding the problem of excessively high costs for a single vehicle or a single pole, so as to achieve more feasible unmanned parking.

[0159] In one embodiment, lamppost initialization includes...

[0160] Obtain vehicle size, rear overhang, minimum turning radius, standard parking space data, and mappings in two different coordinate systems;

[0161] Load deep learning models;

[0162] Build a pole-mounted bird's-eye perception module, an edge intelligent planning module, and a V2L wireless communication module;

[0163] Car initialization, including,

[0164] Obtain reference parking speed;

[0165] Build a V2L wireless communication module and an unmanned vehicle motion control module.

[0166] It should be noted that the technical solution of the autonomous parking device based on the pole-mounted bird's-eye view is the same concept as the above-mentioned autonomous parking method based on the pole-mounted bird's-eye view. The technical solution of the autonomous parking device based on the pole-mounted bird's-eye view in this embodiment can be found in the description of the technical solution of the autonomous parking method based on the pole-mounted bird's-eye view in the above embodiment.

[0167] The storage medium proposed in this embodiment and the autonomous parking method based on pole-mounted bird's-eye view proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0168] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0169] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0170] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0171] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0172] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0173] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0174] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. An autonomous parking method based on pole-mounted bird's-eye view, characterized in that, include: Based on the bird's-eye view of the parking space obtained by the pole-mounted sensing module, the target vehicle, pedestrian, parking space and vehicle driving area are identified and segmented, and the pedestrian and vehicle targets are located by adjusting the function. Based on the identified target information, the vehicle's drivable area, and geometric connections, a parking path for the unmanned vehicle is planned. The tracking control of the parking path and the control of the adjustment of the vehicle pose realize autonomous parking of the unmanned vehicle, and specifically include: setting a reference point and calculating a reference front wheel steering angle δ r , a reference speed v r , and a reference heading angle θ r according to path planning; a state space equation is established based on target vehicle information and is expressed as: X(k+1) = AX(k) + Bu(k) Where X(k) is the error of the spatial state at time k, u(k) is the difference between the control quantity and the reference value at time k, k = 1, 2, ..., N, N is the number of reference points set, A is the first matrix representing the sampling interval, reference speed, reference heading angle and wheelbase at time k, and B is the second matrix representing the sampling interval, reference speed, reference heading angle and wheelbase at time k. The optimal control for multi-objective optimization is expressed as: J=Σ(X T QX(k)+u T Gu(k)) Where J is the objective function, which is the weighted sum of the cumulative tracking deviation and the cumulative control input during the tracking process; Q is the first weight matrix; and G is the second weight matrix. The optimal control rate is obtained by solving LQR: u = -KX K is determined by A, B and the solution P of the Riccati equation; Based on the control law, path tracking control is performed, and the vehicle status is updated in real time to achieve autonomous parking of unmanned vehicles.

2. The autonomous parking method based on pole-mounted bird's-eye view as described in claim 1, characterized in that, The adjustment function includes: The models of different vehicles in the obtained bird's-eye view of the parking space are simplified, and the standard pose of the vehicles in the original image is set. establishing a first mapping f c1 , according to the difference between the current pose and the standard pose, mapping the center of the current recognition box to the center of the recognition box under the standard pose; establishing a second mapping f c2 mapping the center point of the recognition frame in the standard pose to a coordinate on the transformed bird's-eye view, the coordinate being the position of the vehicle in the vertical downward view. Establish a mapping of pedestrian locations from the original bird's-eye view to the transformed bird's-eye view. p , is represented as: (x Bp ,y Bp ) = (x p ,y p +h p / 2)+o Among them, (x p ,y p h is the center of the pedestrian recognition bounding box in the original image. p Let x be the height of the pedestrian recognition bounding box. Bp ,y Bp Let be the coordinates of the pedestrian in the bird's-eye view, and o be a small vector.

3. The autonomous parking method based on pole-mounted bird's-eye view as described in claim 2, characterized in that, Identify and segment target vehicles, pedestrians, parking spaces, and vehicle-accessible areas, including: By identifying vehicles, parking space corners, and pedestrians using YOLOv5s, and setting the pixel coordinate system of the transformed bird's-eye view as the basic coordinate system, the coordinates of key objects are obtained for localization. The drivable area segmentation is based on the collected overhead view of the parking lot, and the fine semantic segmentation label data is obtained by manual annotation using a labeling tool. The data is used as the dataset of the parking lot scene to train the BiseNetV2 model. The bird's-eye view is inferred as a batch using bisenenetV2 to obtain free space segmentation; The free space segmentation mainly refers to the boundary between the road surface and objects with height.

4. The autonomous parking method based on pole-mounted bird's-eye view as described in claim 3, characterized in that, Also includes: Parking space identification and positioning only considers single parking space scenarios; Four corner points are identified: a quadrilateral parking space is generated; two corner points are identified: the distance between the corner points is used to determine whether the parking space is parallel or perpendicular, and two possible quadrilateral parking spaces are generated. The selection of parking spaces is based on their relative position to the vehicle. By calculating the difference in grayscale values ​​within the parking spaces, it is determined whether the selected parking spaces are occupied.

5. The autonomous parking method based on pole-mounted bird's-eye view as described in claim 3 or 4, characterized in that, Planning parking paths for unmanned vehicles includes: planning parking paths for unmanned vehicles based on a map with target information and a simplified two-degree-of-freedom model of the vehicle. The path planning method is based on the geometric connection of circular arcs and straight lines, and is mainly divided into perpendicular and parallel parking according to the actual situation.

6. The autonomous parking method based on pole-mounted bird's-eye view as described in claim 5, characterized in that, Also includes: The path planning method selects between parallel and perpendicular parking methods based on the length and width relationship of the parking space, as follows: d=((R-l Bc / 2-δ1) 2 -(R-w p / 2) 2 ) 1 / 2 ; h min =R-d; D min =((R+l Bc / 2) 2 +(w Bc -h ro ) 2 ) 1 / 2 -d+δ2 Among them, l Bc ,w Bc These represent the vehicle's length and width in the transformed bird's-eye view, respectively, where R is the minimum turning radius at the rear axle center, and h is the distance between the two values. ro For rear overhang, w p D is the width of the parking space. min The minimum lane width is given by δ1, the first safety distance is given by δ2, the second safety distance is given by δ2, d is the distance from the turning center to the two corner points (i.e., the line connecting the nearest vehicles), and h is the distance from the turning center to the two corner points. min The minimum distance between the car and the nearest corner point; C-shaped parking is represented as: h∈[h min ,h min +h f -D min ] Among them, h f The feasible region height is h, and h is the distance between the car and the nearest corner point. The A-frame parking method is represented as follows: h∈(h f -R-δ2,h min ) The heading angle for a V-shaped parking maneuver is expressed as: θ = arccos[(h+R) / 2R] Where θ is the heading angle; The parallel berthing heading angle is expressed as: θ=arccos[2R-h-w Bc / 2] / 2R。 7. An autonomous parking device based on a pole-mounted bird's-eye view, characterized in that, The application of the autonomous parking method based on pole-mounted bird's-eye view as described in any one of claims 1-6, wherein the autonomous parking device based on pole-mounted bird's-eye view comprises: The pole-mounted bird's-eye view perception module is used for overhead perception of parking spaces and vehicles, providing perception data for dynamic planning of parking routes. The pole-mounted bird's-eye view perception module takes into account the height of the light pole. From the bird's-eye view, vehicles are large while parking space corners and pedestrians are small. In the Neck structure of YOLOv5s, Bi-FPN is used to replace PAN-Net. Bi-FPN introduces weights to balance feature information at different scales. The edge intelligent planning module performs parking path planning for unmanned vehicles based on a map rich in perceived target information generated by the perception module and a simplified two-degree-of-freedom model of the vehicle.

8. The autonomous parking device based on pole-mounted bird's-eye view as described in claim 7, characterized in that, Also includes: The V2L wireless communication module is used to transmit the dynamically planned path to the autonomous vehicle's motion control module, enabling vehicle-road cooperative interaction. The V2L wireless communication module is based on cellular communication C-V2X wireless communication technology and establishes a stable two-way wireless encrypted communication channel with the on-board unit (OBU). The V2L wireless communication module communicates with the on-board unit (OBU) to obtain parking demand events and initiate parking collaborative perception and path planning decisions based on a bird's-eye view. Based on the communication protocol, the collaborative sensing information is compressed and transmitted to the on-board unit (OBU); The collaborative perception information consists of the planned path from the edge intelligent planning module and the map information generated by the edge perception module. The On-Board Unit (OBU) decompresses the collaborative perception information and transmits it to the autonomous vehicle's motion control module via the vehicle network. The autonomous vehicle's motion control module then executes intelligent decision-making and motion planning tasks through the Vehicle Control Unit (VCU).

9. The autonomous parking device based on pole-mounted bird's-eye view as described in claim 8, characterized in that, Also includes: The unmanned vehicle motion control module receives instructions from the V2L wireless communication module to perform parking path tracking control and vehicle posture adjustment control. The unmanned vehicle motion control module receives cooperative sensing information from the light pole via V2L and transmits speed and front wheel angle signals to the VCU via vehicle communication to correct the vehicle's posture. Parking path tracking control is based on vehicle kinematics and uses a linear quadratic regulator method to achieve path tracking control.

Images