A parking method and system

Abstract

The application provides a parking method and system, the parking method is applied to a controller of a vehicle to be parked, and the parking method comprises the following steps: obtaining a vehicle parking parameter of the vehicle to be parked; determining a target parking space of the vehicle to be parked based on a preset parking judgment algorithm and the vehicle parking parameter; obtaining an initial trajectory of the vehicle to be parked based on the target parking space; controlling the vehicle to be parked to park based on the initial trajectory; in the process of parking into the target parking space, obtaining a real-time parking angle point of the target parking space in real time, updating the initial trajectory based on the real-time parking angle point to obtain a real-time dynamic trajectory, and controlling the vehicle to be parked to park into the target parking space based on the real-time dynamic trajectory. The parking method and system provided by the application can dynamically correct the initial trajectory in the parking process by using the real-time parking angle point obtained dynamically, so that the vehicle can be parked into the target parking space more accurately.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of parking, in particular to a parking method and system. BACKGROUND

[0002] With the increasing number of cars, the problem of parking difficulty is increasingly prominent, and the corresponding parking assistance system emerges as the times require and gradually attracts widespread attention. The traditional automatic parking technology mainly perceives vehicle, parking space and map information through ultrasonic radar and camera and other sensors, then plans a path point set from the current position to the parking space, and finally controls the steering and vehicle speed according to the dynamics and kinematics model of the vehicle body, so that the vehicle travels on the planned path point trajectory to complete parking. However, the prior art has some deficiencies. For example, in the path planning stage, the planned path point set is directly taken as the input of the control execution layer, which causes the control parameters to be closely combined with the planned path and mutually restricted, thereby causing the mismatch between the control parameters and the parameters of the vehicle itself, resulting in the problem of inaccurate parking. In addition, the parking space environment is complex and changeable, and the size, shape and surrounding obstacle position of the parking space will all affect parking, and the success rate and accuracy of parking in the prior art when facing complex parking space environment need to be improved. SUMMARY

[0003] The present application aims to provide a parking method and system to solve the above technical problems and improve the accuracy of parking.

[0004] In order to solve the above technical problems, the present application provides a parking method applied to a controller of a vehicle to be parked, the method comprising:

[0005] obtaining vehicle parking parameters of the vehicle to be parked; determining a target parking space of the vehicle to be parked based on a preset parking judgment algorithm and the vehicle parking parameters; obtaining an initial trajectory of the vehicle to be parked based on the target parking space; controlling the vehicle to be parked to park based on the initial trajectory, in the process of parking into the target parking space, obtaining real-time parking angle points of the target parking space in real time, updating the initial trajectory based on the real-time parking angle points to obtain a real-time dynamic trajectory, and controlling the vehicle to be parked to park into the target parking space based on the real-time dynamic trajectory.

[0006] In the above scheme, the vehicle parking-in parameter of the vehicle to be parked is acquired, thereby providing a data basis for subsequent steps. The target parking-in slot of the vehicle to be parked is determined based on a preset parking-in judgment algorithm and the vehicle parking-in parameter, thereby accurately acquiring the target parking-in slot that can be parked through the parking-in judgment algorithm. The initial trajectory of the vehicle to be parked is acquired based on the target parking-in slot, thereby enabling the target vehicle to be parked to start parking in the target parking-in slot based on the initial trajectory. The vehicle to be parked is controlled to park based on the initial trajectory, and in the process of parking in the target parking-in slot, the real-time slot corner point of the target parking-in slot is acquired in real time, the initial trajectory is updated based on the real-time slot corner point to obtain a real-time dynamic trajectory, and the vehicle to be parked is controlled to park in the target parking-in slot based on the real-time dynamic trajectory. Because the initial trajectory is rough, the real-time slot corner point of the target parking-in slot is acquired in real time in the process of parking in the target parking-in slot, and the initial trajectory is dynamically corrected in the parking process by using the dynamically acquired real-time slot corner point, thereby enabling the vehicle to be parked to be more accurately parked in the target parking-in slot.

[0007] Further, the vehicle parking-in parameter includes a parking-in state flag bit, and the target parking-in slot of the vehicle to be parked is determined based on the preset parking-in judgment algorithm and the vehicle parking-in parameter, including: when the parking-in state flag bit of the vehicle to be parked is in a parking-in state, a set of optional target parking-in slots is acquired; a set of optional slot obstacle corner points and a slot map corner point of the set of optional target parking-in slots are acquired; and the target parking-in slot is selected from the set of optional target parking-in slots according to a comparison result of the set of optional slot obstacle corner points and the slot map corner point.

[0008] In the above scheme, by judging whether the vehicle to be parked is in a parking-in state, the target parking-in slot corresponding to the current vehicle to be parked is accurately identified by comparing the slot map corner point with the real-time detection corner point, interference is excluded from a set of optional target parking-in slots, and the parking-in target matching accuracy is ensured.

[0009] Further, the initial trajectory is controlled based on the initial trajectory to control the vehicle to be parked to park, and in the process of parking in the target parking space, the real-time parking angle point of the target parking space is acquired in real time, the initial trajectory is updated based on the real-time parking angle point to obtain a real-time dynamic trajectory, and the vehicle to be parked is controlled to park in the target parking space based on the real-time dynamic trajectory, comprising: acquiring a real-time parking obstacle point set, a preset obstacle contour line segment and an initial parking angle point of the target parking space; fitting the point set in the real-time parking obstacle point set by a preset fitting algorithm to obtain a fitted line segment; acquiring an included angle between the fitted line segment and the preset obstacle contour line segment as a real-time included angle; based on the real-time included angle, the real-time parking obstacle point set and the target parking space type of the target parking space, a target updating algorithm is selected from a preset updating algorithm, so that the initial parking angle point is updated based on the target updating algorithm to obtain a real-time parking angle point.

[0010] In the above scheme, in the process of parking the vehicle, the initial trajectory is dynamically updated by acquiring the real-time parking angle point, so that the real-time dynamic trajectory can be adjusted in real time with the change of the environment, the flexibility and accuracy of parking are improved, and the geometric change of different parking scenes is adapted.

[0011] Further, the initial trajectory is controlled based on the initial trajectory to control the vehicle to be parked to park, and in the process of parking in the target parking space, the real-time parking angle point of the target parking space is acquired in real time, the initial trajectory is updated based on the real-time parking angle point to obtain a real-time dynamic trajectory, and the vehicle to be parked is controlled to park in the target parking space based on the real-time dynamic trajectory, comprising: acquiring a real-time parking obstacle point set, a preset obstacle contour line segment and an initial parking angle point of the target parking space; fitting the point set in the real-time parking obstacle point set by a preset fitting algorithm to obtain a fitted line segment; acquiring an included angle between the fitted line segment and the preset obstacle contour line segment as a real-time included angle; based on the real-time included angle, the real-time parking obstacle point set and the target parking space type of the target parking space, a target updating algorithm is selected from a preset updating algorithm, so that the initial parking angle point is updated based on the target updating algorithm to obtain a real-time parking angle point.

[0012] In the above scheme, the ultrasonic detection data set is purified and converted by coordinate transformation and a preset screening algorithm, and the real-time parking obstacle point set representing the parking edge is further extracted by a Hough transformation algorithm, which significantly improves the robustness and accuracy of the fitted line segment and provides a stable boundary reference for subsequent angle point updating.

[0013] Further, the initial parking space angle point includes a first angle point and a second angle point, and the target updating algorithm is selected from the preset updating algorithms based on the real-time included angle, the real-time parking space obstacle point set, and a target parking space type of the target parking space, so as to update the initial parking space angle point based on the target updating algorithm to obtain a real-time parking space angle point. When the target parking space type is a vertical parking space, the updating object is the first angle point, and the category of the real-time parking space obstacle point set is a UPA point set, the first updating algorithm is selected as the target updating algorithm from the preset updating algorithms based on the category of the real-time parking space obstacle point set. When the target parking space type is a vertical parking space, the updating object is the first angle point, and the category of the real-time parking space obstacle point set is an APA point set, or when the target parking space type is a vertical parking space, and the updating object is the second angle point, a vertical comparison result of the real-time included angle and a preset vertical included angle range threshold is obtained, and the target updating algorithm is selected from the preset updating algorithms based on the vertical comparison result. When the target parking space type is a horizontal parking space, a horizontal comparison result of the real-time included angle and a preset horizontal included angle range threshold is obtained, and the target updating algorithm is selected from the preset updating algorithms based on the horizontal comparison result.

[0014] In the above scheme, the most suitable target updating algorithm is adaptively selected according to the parking space type (vertical or horizontal), the current updating object (first or second angle point), the category of the real-time parking space obstacle point set (APA or UPA), and the real-time included angle, thereby enhancing the adaptability and intelligent processing capability of the scheme to different parking scenarios.

[0015] Further, the preset vertical included angle range threshold includes a first included angle interval, and when the target parking space type is a vertical parking space, the updating object is the first angle point, and the category of the real-time parking space obstacle point set is an APA point set, or when the target parking space type is a vertical parking space, and the updating object is the second angle point, a vertical comparison result of the real-time included angle and a preset vertical included angle range threshold is obtained, and the target updating algorithm is selected from the preset updating algorithms based on the vertical comparison result. When the vertical comparison result is that the real-time included angle is within the first included angle interval, the target updating algorithm is a second updating algorithm. The second updating algorithm is to take the intersection point of the fitting line segment and the preset obstacle contour line segment as a target intersection point, and update the initial parking space angle point based on the target intersection point to obtain the real-time parking space angle point.

[0016] In the above scheme, when the real-time included angle of the fitting line segment and the preset obstacle contour line segment falls within the first included angle interval, the intersection point of the fitting line segment and the preset obstacle contour line segment is directly taken as the target intersection point; the initial parking space corner point is updated based on the target intersection point to obtain the real-time parking space corner point, thereby realizing efficient and accurate corner point updating with high geometric matching degree.

[0017] Further, the preset vertical included angle range threshold further includes a second included angle interval, the numerical range of the first included angle interval is smaller than the numerical range of the second included angle interval, when the target parking space type is a vertical parking space, the update object is the first corner point, and the category of the real-time parking space obstacle trajectory set is an APA trajectory, or when the target parking space type is a vertical parking space and the update object is the second corner point, a vertical comparison result of the real-time included angle and the preset vertical included angle range threshold is obtained, the target update algorithm is selected from the preset update algorithms based on the vertical comparison result, including: when the vertical comparison result is that the real-time included angle is outside the first included angle interval and within the second included angle interval, the target update algorithm is a first update algorithm; the first update algorithm is: clustering the real-time parking space obstacle trajectory set based on a preset clustering algorithm to obtain a clustering center; clustering the real-time parking space obstacle trajectory set based on the preset clustering algorithm to obtain a clustering center set; obtaining a target trajectory according to the clustering center and the fitting line segment; constructing a first vertical line perpendicular to the parking space contour line segment according to the target trajectory; obtaining a vertical intersection point of the first vertical line and the parking space contour line segment as a target intersection point; updating the initial parking space corner point based on the target intersection point to obtain the real-time parking space corner point.

[0018] In the above scheme, when the real-time included angle deviates from the ideal value but is still within the effective interval (outside the first included angle interval and within the second included angle interval), the clustering center point and the vertical projection are used to realize the corner point updating under special conditions such as unclear boundary, thereby improving the updating accuracy in complex environment.

[0019] Further, when the target parking space type is a horizontal parking space, the horizontal comparison result of the real-time included angle and the preset horizontal included angle range threshold is obtained, and the target updating algorithm is selected from the preset updating algorithms based on the horizontal comparison result, including: when the horizontal comparison result is that the real-time included angle is outside the preset horizontal included angle range threshold, the target updating algorithm is a third updating algorithm; the third updating algorithm is: clustering the real-time parking space obstacle point set based on a preset clustering algorithm to obtain a clustering center; a second vertical line perpendicular to a parking space contour line segment is constructed according to the clustering center; a vertical intersection point of the second vertical line and the parking space contour line segment is obtained as a target intersection point; the initial parking space angle point is updated based on the target intersection point to obtain the real-time parking space angle point.

[0020] In the above scheme, when the direction of the fitting line segment in the horizontal parking space deviates from the normal range (outside the preset horizontal included angle range threshold), the angle point correction logic is constructed by clustering and vertical projection, and the accuracy of the intersection point updating is improved.

[0021] Further, when the target parking space type is a horizontal parking space, the horizontal comparison result of the real-time included angle and the preset horizontal included angle range threshold is obtained, and the target updating algorithm is selected from the preset updating algorithms based on the horizontal comparison result, including: when the horizontal comparison result is that the real-time included angle is inside the preset horizontal included angle range threshold, the target updating algorithm is the second updating algorithm.

[0022] In the above scheme, when the real-time included angle of the fitting line segment and the preset obstacle contour line segment is inside the preset horizontal included angle range threshold, the angle point is updated using the intersection point, a more concise and rapid angle point correction processing flow is realized when the angle meets the expectation, and unnecessary calculation resource waste is reduced.

[0023] The application also provides a parking system, including: an acquisition module, configured to acquire vehicle parking parameters of a vehicle to be parked; a parking space confirmation module, configured to determine a target parking space of the vehicle to be parked based on a preset parking judgment algorithm and the vehicle parking parameters; a trajectory module, configured to acquire an initial trajectory of the vehicle to be parked based on the target parking space; and an updating module, configured to control the vehicle to be parked to park based on the initial trajectory, and in the process of parking in the target parking space, to acquire a real-time parking space angle point of the target parking space, to update the initial trajectory based on the real-time parking space angle point to obtain a real-time dynamic trajectory, and to control the vehicle to be parked to park in the target parking space based on the real-time dynamic trajectory.

[0024] The system provided by the scheme has simple structure and can well implement the parking method, which skillfully obtains the vehicle parking parameter of the to-be-parked vehicle, and provides data basis for subsequent steps. The target parking space of the to-be-parked vehicle is determined based on the preset parking judgment algorithm and the vehicle parking parameter, and the target parking space that can be parked is accurately obtained through the parking judgment algorithm. The initial trajectory of the to-be-parked vehicle is obtained based on the target parking space, so that the target to-be-parked vehicle can start parking based on the initial trajectory. The to-be-parked vehicle is controlled to park based on the initial trajectory, and in the process of parking in the target parking space, the real-time parking angle point of the target parking space is obtained in real time, the initial trajectory is updated based on the real-time parking angle point to obtain a real-time dynamic trajectory, and the to-be-parked vehicle is controlled to park in the target parking space based on the real-time dynamic trajectory. Because the initial trajectory is rough, the real-time parking angle point of the target parking space is obtained in real time in the process of parking in the target parking space, and the real-time parking angle point obtained dynamically is used to dynamically correct the initial trajectory in the parking process, so that the vehicle can be more accurately parked in the target parking space. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 A parking method flowchart is provided for an embodiment of the application.

[0026] Figure 2 A parking system architecture diagram is provided for an embodiment of the application.

[0027] Figure 3 A target parking space selection diagram is provided for an embodiment of the application.

[0028] Figure 4 A coordinate system at a start parking position is provided for an embodiment of the application.

[0029] Figure 5 An updated second angle point (C point) diagram in a vertical parking space is provided for an embodiment of the application.

[0030] Figure 6 A real-time included angle fitting line segment and vertical parking straight line CD diagram when the included angle is in a first included angle interval is provided for an embodiment of the application.

[0031] Figure 7 A clustering center between vertical parking straight lines CD is provided for an embodiment of the application.

[0032] Figure 8 A clustering center between B and C angle points is provided for an embodiment of the application.

[0033] Figure 9A schematic diagram for updating the first corner point (B point) when the category of the real-time parking space obstacle point set is UPA point under the vertical parking provided by an embodiment of the present application;

[0034] Figure 10 A schematic diagram for updating the first corner point (B point) when the category of the real-time parking space obstacle point set is APA point under the vertical parking provided by an embodiment of the present application;

[0035] Figure 11 A schematic diagram for updating the real-time parking space corner point under the horizontal parking provided by an embodiment of the present application. DETAILED DESCRIPTION

[0036] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0037] Please refer to Figure 1 The present embodiment provides a parking method applied to a controller of a vehicle to be parked, and the method comprises the following steps:

[0038] Step S1: acquiring vehicle parking-in parameters of the vehicle to be parked;

[0039] Step S2: determining a target parking-in parking space of the vehicle to be parked based on a preset parking-in judgment algorithm and the vehicle parking-in parameters;

[0040] Step S3: acquiring an initial trajectory of the vehicle to be parked based on the target parking-in parking space;

[0041] Step S4: controlling the vehicle to be parked to park based on the initial trajectory, acquiring real-time parking space corner points of the target parking-in parking space in real time in the process of parking into the target parking-in parking space, updating the initial trajectory based on the real-time parking space corner points to obtain a real-time dynamic trajectory, and controlling the vehicle to be parked to park into the target parking-in parking space based on the real-time dynamic trajectory.

[0042] In the scheme, the vehicle parking-in parameter of the vehicle to be parked is acquired, thereby providing a data basis for subsequent steps. The target parking space of the vehicle to be parked is determined based on a preset parking-in judgment algorithm and the vehicle parking-in parameter. The target parking space in which parking can be performed is accurately acquired through the parking-in judgment algorithm. The initial trajectory of the vehicle to be parked is acquired based on the target parking space, thereby enabling the target vehicle to be parked to start parking in the target parking space based on the initial trajectory. The vehicle to be parked is controlled to park based on the initial trajectory. In the process of parking in the target parking space, the real-time parking space corner point of the target parking space is acquired in real time. The initial trajectory is updated based on the real-time parking space corner point, and the real-time dynamic trajectory is obtained, so as to control the vehicle to be parked to park in the target parking space based on the real-time dynamic trajectory. Because the initial trajectory is rough, the real-time parking space corner point of the target parking space is acquired in real time in the process of parking in the target parking space. The real-time parking space corner point acquired dynamically is used to dynamically correct the initial trajectory in the parking process, thereby enabling the vehicle to be parked to park in the target parking space more accurately.

[0043] In another embodiment, the vehicle parking-in parameter includes a parking-in state flag. The target parking space of the vehicle to be parked is determined based on a preset parking-in judgment algorithm and the vehicle parking-in parameter, including:

[0044] When the parking-in state flag of the vehicle to be parked is in a parking-in state, a set of optional target parking spaces is acquired.

[0045] A set of optional parking space obstacle corner points and a set of parking space map corner points of the set of optional target parking spaces are acquired.

[0046] The target parking space is selected from the set of optional target parking spaces according to a comparison result of the set of optional parking space obstacle corner points and the set of parking space map corner points.

[0047] It should be noted that the vehicle parking-in parameter includes a parking-in state flag. The parking-in state flag indicates whether the vehicle to be parked is in a parking state (such as a parking-in state / non-parking-in state). When the parking-in state flag of the vehicle to be parked is in a parking-in state, a set of optional target parking spaces can be acquired by searching for nearby parking spaces through a control ultrasonic wave recognition module. Then, a set of parking space map corner points and a set of optional target parking space corner points of the set of optional target parking spaces are acquired. The set of optional target parking space corner points can also be obtained through the ultrasonic wave recognition module. For details, refer to Figure 3, assuming that the parking space map corner points of a parking space include p1, p3 (parking space head placement), p2, p4 (parking space tail placement), if there is an obstacle in each parking space, the parking space obstacle corner points on the left side of the parking space are set as A, B (obstacle head), and the parking space obstacle corner points on the right side of the parking space are set as C, D (obstacle head), it can be understood that the corner point close to the parking space on the left side is the B point, and the corner point far from the parking space is the A point, and the corner point close to the parking space on the right side is the C point, and the corner point far from the parking space is the D point. The front is the head placed in the parking space, and the rear is the tail placed in the parking space to determine the left and right directions. The p1 and p3 in the parking space map corner points and the B and C in the optional obstacle corner point set are compared, if the p1 and p3 are between the B and C two corner points, it is determined that the parking space is the target parking-in parking space. If there are multiple optional parking-in parking spaces that meet the comparison condition, the optional parking-in parking space closest to the to-be-parked vehicle is selected as the target parking-in parking space. Further, based on the target parking-in parking space, the initial parking space corner point is determined from the optional parking space obstacle corner point set (for example Figure 3 In the middle, P2 is selected as the target parking-in parking space from P1, P2, P3, and then the B and C points next to it are the initial parking space corner points). Further, the radar ultrasonic wave and other identification modules can also identify the type of the optional parking-in parking space (such as a vertical parking space / horizontal parking space) in the process of searching and identifying the parking space that can be parked in, or it can be preset. After the to-be-parked vehicle enters the corresponding parking lot, the related information of the parking lot is obtained based on the position, and then the parking space type of the optional parking-in parking space set is obtained.

[0048] In another embodiment, the to-be-parked vehicle is controlled to park based on the initial trajectory, and in the process of parking in the target parking-in parking space, the real-time parking space corner points of the target parking-in parking space are obtained in real time, the initial trajectory is updated based on the real-time parking space corner points to obtain a real-time dynamic trajectory, and the to-be-parked vehicle is controlled to park in the target parking-in parking space based on the real-time dynamic trajectory, including:

[0049] Obtaining a real-time parking space obstacle point trajectory set of the target parking-in parking space, a preset obstacle contour line segment, and an initial parking space corner point;

[0050] Fitting the point trajectories in the real-time parking space obstacle point trajectory set by a preset fitting algorithm to obtain a fitting line segment;

[0051] Obtaining an included angle between the fitting line segment and the preset obstacle contour line segment as a real-time included angle;

[0052] Based on the real-time included angle, the real-time parking space obstacle point trajectory set, and the target parking-in parking space type of the target parking-in parking space, a target updating algorithm is selected from a preset updating algorithm, so that the initial parking space corner point is updated based on the target updating algorithm to obtain a real-time parking space corner point.

[0053] It should be noted that the to-be-parked vehicle moves based on the initial trajectory when it is at the initial position, and the initial trajectory at this time is rough, so the initial trajectory needs to be updated and corrected according to the real-time obtained parking space angle points during the movement, to obtain a real-time dynamic trajectory, and the to-be-parked vehicle moves along the real-time dynamic trajectory to enter the target parking space. In this process, how to obtain the real-time parking space angle points, specifically: obtaining a real-time parking space obstacle point set, a preset obstacle contour line segment and an initial parking space angle point of the target parking space. It can be understood that, in the parking process, it is not the existence of the entire obstacle (such as other parked vehicles) beside the target parking space that affects the parking path of the to-be-parked vehicle, but only a small part. For a vertical parking space, attention should be paid to the front part of the vehicle parked on the left and right sides of the target parking space, and for a horizontal parking space, attention should be paid to the front and rear parts of the vehicle parked on the front and rear sides of the target parking space. Therefore, it can be understood that the real-time parking space obstacle point set refers to the part of the point set of the obstacle (i.e. the vehicle parked beside the target parking space) beside the target parking space that will affect the parking trajectory of the to-be-parked vehicle. Similarly, the preset obstacle contour line segment is the contour line segment of the part of the obstacle beside the target parking space that will affect the parking trajectory of the to-be-parked vehicle. Further, the real-time parking space obstacle point set can be obtained by an ultrasonic radar recognition module and processed. The initial parking space angle point is a point on the preset obstacle contour line segment.

[0054] In another embodiment, the points in the real-time parking space obstacle point set are fitted by a preset fitting algorithm to obtain a fitted line segment, including:

[0055] Obtaining an ultrasonic detection data set of the to-be-parked vehicle;

[0056] Selecting an ultrasonic detection data subset from the ultrasonic detection data set according to the target parking space;

[0057] Converting the ultrasonic detection data subset into an initial unfiltered parking space obstacle point set according to a preset coordinate conversion algorithm and vehicle parking parameters;

[0058] Filtering the initial unfiltered parking space obstacle point set based on a preset filtering algorithm to obtain an initial parking space obstacle point set;

[0059] Filtering the initial parking space obstacle point set by a Hough transform algorithm to obtain a real-time parking space obstacle point set.

[0060] It should be noted that in the process of determining the target parking space, the target parking space corresponding to the parking space angle point is obtained from the selectable parking space angle point set. For the output of the target parking space, the rear axle center of the parking vehicle is taken as the origin, the front of the vehicle is taken as the y axis, and the x axis is perpendicular to the right side of the vehicle. If the target parking space is on the right side of the parking vehicle, the coordinate x value of the target parking space corresponding to the parking space angle point is positive (reference Figure 4 ), if the target parking space is on the left side of the parking vehicle, the coordinate x value of the target parking space corresponding to the parking space angle point is negative. Since the ultrasonic detection data is generated in real time, the ultrasonic radar probe and other identification modules detect obstacles in real time during parking, generate an ultrasonic detection data set, and then select a target parking space corresponding to an ultrasonic detection data subset from the ultrasonic detection data set. But the process of parking the target parking space is mobile, and the ultrasonic detection data detected by the radar ultrasonic identification module is measured in the coordinate system with the rear axle center of the deflected parking vehicle as the origin. Therefore, in order to unify the subsequent processing, it is necessary to map these ultrasonic detection data back to the coordinate system with the rear axle center of the parking vehicle at the start of parking as the origin. Therefore, it is necessary to convert the detected distance data (i.e. the ultrasonic detection data subset) into the starting parking position trace (i.e. the initial unfiltered parking space obstacle trace set), specifically:

[0061] The real-time detected ultrasonic detection data subset is converted to the coordinate system at the start of parking to obtain the initial unfiltered parking space obstacle trace set in the corresponding coordinate system at the start of parking:

[0062] First, the probe of the ultrasonic radar identification module of the detected ultrasonic detection data subset is converted from the probe coordinate to the DR coordinate system (with the rear axle center of the parking vehicle as the origin, the front of the parking vehicle as the positive direction of the y axis, and the x axis perpendicular to the right side of the front of the parking vehicle as the positive direction of the x axis):

[0063] SensorPtX=SnsPtX•cos(DR.A)-SnsPtY•sin(DR.A)+DR.X;

[0064] SensorPtY=SnsPtX•sin(DR.A)+SnsPtY•cos(DR.A)+DR.Y;

[0065] SnsPtX and SnsPtY, SnsPtA represent the installation parameters of the probe, SnsPtX is the lateral installation distance of the probe based on the center of the rear axle of the vehicle, SnsPtY is the longitudinal distance of the probe based on the center of the rear axle of the vehicle, and SnsPtA is the installation outward angle of the probe. DR.X, DR.Y, DR.A are respectively the X coordinate of the center of the rear axle of the vehicle to be parked in the DR information, the Y coordinate of the center of the rear axle of the vehicle to be parked in the DR information, and the heading angle of the vehicle to be parked. SensorPtX is the X coordinate data of the probe in the DR coordinate system, and SensorPtY is the Y coordinate data of the probe in the DR coordinate system. The above parameters are all vehicle parking parameters.

[0066] Then calculate the initial unfiltered parking space obstacle point set curTarget:

[0067] curTarget(1,1) = SensorPtX - dis*sin(DR.A + SnsPtA);

[0068] curTarget(2,1) = SensorPtY + dis*cos(DR.A + SnsPtA);

[0069] Where dis is the target ultrasonic detection data (distance data) detected by the radar probe and the like identification module, and curTarget is the point coordinates corresponding to the initial unfiltered parking space obstacle point set calculated. curTarget(1,1) and curTarget(2,1) represent the X axis coordinate and Y axis coordinate of the point in the initial unfiltered parking space obstacle point set at the beginning of the parking state.

[0070] Further, because the initial parking angle point is a point on the preset obstacle contour line segment. In the subsequent movement of the vehicle to be parked into the target parking space, it is also necessary to map the updated real-time parking angle point to the coordinate system of the vehicle to be parked in the beginning of the parking state, to facilitate subsequent calculation. The process of converting the real-time parking angle point to the coordinate system of the beginning of the parking position is described in detail below with the example of "the target parking space type is a vertical parking space, and the updated object is the second angle point (i.e. the C point under the target parking space type of vertical parking space)".

[0071] C_x = Slot_PT.C.x*cos(BC_A) - Slot_PT.C.y*sin(BC_A) + DR.X;

[0072] C_y = Slot_PT.C.x*sin(BC_A) + Slot_PT.C.y*cos(BC_A) + DR.Y ;

[0073] Wherein Slot PT.C.x, Slot PT.C.y is the C point position detected by the radar ultrasonic wave identification module probe (Slot PT.C.x is the x coordinate, Slot PT.C.y is the y coordinate), BC_A is the angle between the vehicle and the start parking position of the vehicle in the parking process, such as Figure 5 , Figure 5 The point trace diagram generated in the parking process, Figure 5 The dotted line is the front of the vehicle starting to park, and C_x, C_y is the coordinate value of the real-time parking angle point converted to the coordinate system under the starting parking position. DR.X, DR.Y represents the motion trajectory of the origin point at the reset moment of the rear axle center of the vehicle to be parked. It can be understood that the conversion method of the detected parking process point trace (i.e. the real-time parking obstacle point trace set after the initial unfiltered parking obstacle point trace set curTarget is subjected to Hough transform and screening algorithm) is the same as the C point coordinate conversion method.

[0074] It should be noted that, because in the parking process, the vehicle to be parked will slowly drive into the target parking space, the radar ultrasonic wave identification module will continuously generate ultrasonic detection data in real time. However, not all ultrasonic detection data is useful, some are abnormal, detection position is wrong, etc. Therefore, it is necessary to find out those ultrasonic detection data which accurately reflect the actual position of the obstacle on the front side boundary of the parking space (i.e. CD line). After converting the ultrasonic detection data subset into the initial unfiltered parking obstacle point trace set, the initial parking obstacle point trace set is obtained by screening the initial unfiltered parking obstacle point trace set through the preset screening algorithm, and the screening process is as follows:

[0075] Taking the position close to the C point as an example, the distance from the vertical parking straight line CD in the parking process is calculated, and all point traces with a distance greater than a preset threshold (such as 20 cm) and distance increment are selected and saved. (1) The vertical distance from the CD line needs to be greater than 20 cm because the bumper has a certain curvature, and the points close to the C point may fall inside the obstacle vehicle body, and the distance is slightly far, which can reflect the real boundary line in front of the parking space. Therefore, the point traces with a distance greater than 20 cm from the CD line are screened, which means that only the point traces on the outer edge of the parking space and the front obstacles are retained. It can be understood that the ultrasonic detection data subset of the obstacle has been converted into the initial unfiltered parking obstacle point trace set under the coordinate system of the starting parking position by the coordinate conversion algorithm, so that the spatial distance between the initial unfiltered parking obstacle point trace set and the line segment CD can be directly calculated. (2) The distance needs to be increasing because the vehicle to be parked updates the C point in the backward parking process, and because the obstacle vehicle body has a certain curvature, the middle position is the closest distance to the line segment CD, so the point traces from the line segment CD are the point traces of the right side of the obstacle vehicle, i.e. the position point traces at the corner point C.

[0076] After that, the initial parking space obstacle point set is screened by the Hough transform algorithm to obtain a real-time parking space obstacle point set. Specifically, taking the C point of the target parking-in parking space type as a vertical parking space as an example: if the number of point traces of the initial un-screened parking space obstacle point set is greater than 2, the Hough transform is performed on the initial parking space obstacle point set, a large number of unreasonable point traces are removed by the Hough transform, and a real-time parking space obstacle point set is obtained. The point traces in the real-time parking space obstacle point set are fitted by a preset fitting algorithm to obtain a fitting line segment. The Hough transform mainly finds a relatively concentrated point trace in a straight line, then calculates the perpendicular distance of all point traces to the straight line, and removes the point traces with a distance greater than a preset value, such as 30 cm, thereby removing a large number of unreasonable point traces.

[0077] In another embodiment, the initial parking space angle point includes a first angle point and a second angle point, and based on the real-time included angle, the real-time parking space obstacle point set, and the target parking-in parking space type of the target parking-in parking space, a target updating algorithm is selected from the preset updating algorithms, so that the initial parking space angle point is updated based on the target updating algorithm to obtain a real-time parking space angle point, including:

[0078] When the target parking-in parking space type is a vertical parking space, the updating object is the first angle point, and the category of the real-time parking space obstacle point set is a UPA point trace, a first updating algorithm is selected as the target updating algorithm from the preset updating algorithms based on the category of the real-time parking space obstacle point set.

[0079] When the target parking-in parking space type is a vertical parking space, the updating object is the first angle point, and the category of the real-time parking space obstacle point set is an APA point trace, or when the target parking-in parking space type is a vertical parking space and the updating object is the second angle point, a vertical comparison result of the real-time included angle and a preset vertical included angle range threshold is obtained, and a target updating algorithm is selected from the preset updating algorithms based on the vertical comparison result.

[0080] When the target parking-in parking space type is a horizontal parking space, a horizontal comparison result of the real-time included angle and a preset horizontal included angle range threshold is obtained, and a target updating algorithm is selected from the preset updating algorithms based on the horizontal comparison result.

[0081] It should be noted that for different target parking-in parking space types, updating objects, and categories of real-time parking space obstacle point sets, a target updating algorithm most suitable for the current situation is dynamically selected from the preset updating algorithms, so that the initial parking space angle point is corrected to obtain a more accurate real-time parking space angle point.

[0082] In another embodiment, the preset vertical angle range threshold value includes a first angle interval, when the target parking space type is a vertical parking space, and the update object is a first corner point, and the category of the real-time parking space obstacle point set is an APA point, or when the target parking space type is a vertical parking space, and the update object is a second corner point, a vertical comparison result of the real-time angle and the preset vertical angle range threshold value is obtained, a target update algorithm is selected from the preset update algorithms based on the vertical comparison result, including:

[0083] When the vertical comparison result is that the real-time angle is within the first angle interval, the target update algorithm is a second update algorithm;

[0084] The second update algorithm is:

[0085] The intersection of the fitting line segment and the preset obstacle contour line segment is taken as a target intersection point;

[0086] The initial parking space corner point is updated based on the target intersection point to obtain a real-time parking space corner point.

[0087] In another embodiment, the preset vertical angle range threshold value further includes a second angle interval, the numerical range of the first angle interval is smaller than the numerical range of the second angle interval, when the target parking space type is a vertical parking space, and the update object is a first corner point, and the category of the real-time parking space obstacle point set is an APA point, or when the target parking space type is a vertical parking space, and the update object is a second corner point, a vertical comparison result of the real-time angle and the preset vertical angle range threshold value is obtained, a target update algorithm is selected from the preset update algorithms based on the vertical comparison result, including:

[0088] When the vertical comparison result is that the real-time angle is outside the first angle interval and within the second angle interval, the target update algorithm is a first update algorithm;

[0089] The first update algorithm is:

[0090] The real-time parking space obstacle point set is clustered based on a preset clustering algorithm to obtain a clustering center;

[0091] The real-time parking space obstacle point set is clustered based on a preset clustering algorithm to obtain a clustering center set;

[0092] A target point is obtained according to the clustering center and the fitting line segment;

[0093] A vertical intersection point of a first vertical line and a parking space contour line segment is obtained as a target intersection point;

[0094] The initial parking space corner point is updated based on the target intersection point to obtain a real-time parking space corner point.

[0095] In another embodiment, when the target parking space type is a horizontal parking space, a horizontal comparison result of the real-time included angle and a preset horizontal included angle range threshold is obtained, a target update algorithm is selected from preset update algorithms based on the horizontal comparison result, and the target update algorithm includes:

[0096] When the horizontal comparison result is that the real-time included angle is outside the preset horizontal included angle range threshold, the target update algorithm is a third update algorithm.

[0097] The third update algorithm is:

[0098] Based on a preset clustering algorithm, the real-time parking space obstacle point set is clustered to obtain a clustering center.

[0099] A second vertical line perpendicular to the parking space contour line segment is constructed according to the clustering center.

[0100] A vertical intersection point of the second vertical line and the parking space contour line segment is obtained as a target intersection point.

[0101] Based on the target intersection point, the initial parking space angle point is updated to obtain a real-time parking space angle point.

[0102] In another embodiment, when the target parking space type is a horizontal parking space, a horizontal comparison result of the real-time included angle and a preset horizontal included angle range threshold is obtained, a target update algorithm is selected from preset update algorithms based on the horizontal comparison result, and the target update algorithm includes:

[0103] When the horizontal comparison result is that the real-time included angle is within the preset horizontal included angle range threshold, the target update algorithm is a second update algorithm.

[0104] It should be noted that different target update algorithms are determined in multiple cases, which are explained one by one as follows:

[0105] First, the case of the target parking space type being a vertical parking space is explained, in which the initial parking space angle point includes a first angle point (point B) and a second angle point (point C), and the specific cases include:

[0106] (I) The following is an explanation of the update of the second angle point (point C) when the target parking space type is a vertical parking space:

[0107] Firstly, it is judged whether the target parking space is on the left side or the right side. Taking the center of the rear axle of the vehicle to be parked as the origin, the front of the vehicle to be parked as the y-axis, and the right side perpendicular to the front of the vehicle as the x-axis, if the target parking space is on the right side relative to the vehicle to be parked, the coordinate x value of the target parking space corresponding to the parking space angle point is positive, and if the target parking space is on the left side relative to the vehicle to be parked, the coordinate x value of the target parking space corresponding to the parking space angle point is negative. The main purpose is to determine whether to select the left radar ultrasonic wave recognition module point or the right radar ultrasonic wave recognition module point (left parking selects the left radar ultrasonic wave recognition module point, and right parking selects the right radar ultrasonic wave recognition module point). Then, the ultrasonic wave detection data subset is processed to obtain a real-time parking obstacle point set. It can be understood that the left side here mainly refers to the left rear radar ultrasonic wave of the vehicle to be parked, and the right side mainly refers to the right rear radar ultrasonic wave of the vehicle to be parked, because during the reversing process, the main concern is the tail part of the vehicle to be parked.

[0108] (1) When the target parking space type is a vertical parking space, and the updating object is the second angle point, a vertical comparison result of the real-time included angle and the preset vertical included angle range threshold is obtained, and a target updating algorithm is selected from the preset updating algorithm based on the vertical comparison result, including:

[0109] When the vertical comparison result is that the real-time included angle is in the first included angle interval, the target updating algorithm is the second updating algorithm.

[0110] The second updating algorithm is: taking the intersection point of the fitting line segment and the preset obstacle contour line segment as a target intersection point; updating the initial parking space angle point based on the target intersection point to obtain a real-time parking space angle point.

[0111] It should be noted that when the target parking space type is a vertical parking space, and the updating object is the second angle point (i.e., the C point in Figure 5 When the target parking space type is a vertical parking space, and the updating object is the second angle point (i.e., the C point in Figure 5 The real-time included angle of the fitting line segment of the corresponding second angle point and the preset obstacle contour line segment (when the target parking space type is a vertical parking space, and the updating object is the second angle point (i.e., the C point in Figure 5 The real-time included angle of the fitting line segment of the corresponding second angle point and the preset obstacle contour line segment (when the target parking space type is a vertical parking space, and the updating object is the second angle point (i.e., the C point in Figure 5, the green dots represent the last screening of the vertical parking space, the C point corresponds to the real-time parking obstacle point set, the yellow / orange dots represent the screened out other point traces, the green line segment represents the fitting line segment corresponding to the C point in the parking obstacle point set, BC_A is the angle between the parking vehicle in the parking process and the vehicle in the starting parking position), so when judging the angle, the acute and obtuse angles need to be considered. First, judge whether the vertical comparison result is in the real-time angle in the first angle interval, wherein the preset vertical angle range threshold includes the first angle interval and the second angle interval, and the first angle interval is 85° to 95°, for example. If it is in the first angle interval, it means that the fitting line segment and the vertical parking straight line CD are close to vertical (as shown in Figure 6 , the intersection point of the two line segments is the corner point closest to the real parking position C of the vehicle, so the target updating algorithm for the C point is the second updating algorithm: the intersection point of the fitting line segment and the preset obstacle contour line segment (the vertical parking straight line CD) is taken as the target intersection point; the initial parking corner point C is updated based on the target intersection point to obtain the real-time parking corner point C.

[0112] (2) When the target parking space type is a vertical parking space, and the update object is the second corner point, the vertical comparison result of the real-time angle and the preset vertical angle range threshold is obtained, and the target updating algorithm is selected from the preset updating algorithm based on the vertical comparison result, including:

[0113] When the vertical comparison result is that the real-time angle is outside the first angle interval and is in the second angle interval, the target updating algorithm is the first updating algorithm;

[0114] The first updating algorithm is:

[0115] The real-time parking obstacle point set is clustered based on the preset clustering algorithm to obtain a clustering center; the real-time parking obstacle point set is clustered based on the preset clustering algorithm to obtain a clustering center set; the clustering center located on the fitting line segment is taken as a target point trace; a first vertical line perpendicular to the parking contour line segment is constructed according to the target point trace; the vertical intersection point of the first vertical line and the parking contour line segment is taken as a target intersection point; the initial parking corner point is updated based on the target intersection point to obtain a real-time parking corner point.

[0116] It should be noted that preferably, the second angle interval is 45° to 135°, if the real-time included angle between the fitting line segment and the vertical parking straight line CD is outside the first angle interval of 85°~95° and within the second angle interval of 45°~135°, the target updating algorithm is the first updating algorithm: clustering the real-time parking obstacle point set based on a preset clustering algorithm to obtain a clustering center set, because only one clustering center needs to be clustered here. Then, the vertical distance Dis_C of the clustering center to the vertical parking straight line CD is calculated as the target distance. Since the bumper of the obstacle (vehicle beside the parking space) has a certain curvature, if the value of Dis_C is too small (less than the first preset distance threshold), the clustered point will be between the vertical parking straight lines CD (such as the red dot in Figure 7 ), and if the distance value is too large (greater than the second preset distance threshold), the clustering center will be between the two corner points B and C (such as the red dot in Figure 8 ), therefore, according to the slope of the fitting line segment, a clustering center with a preset vertical distance (such as 25 cm) to the vertical parking straight line CD and a point on the fitting line segment is found as the target point (such as the red dot in Figure 5 ). According to the target point, a first vertical line perpendicular to the vertical parking straight line CD passing through the target point is found (such as the black line segment perpendicular to CD in Figure 5 ), and the vertical intersection point of the first vertical line and the vertical parking straight line CD (such as the black point on CD in Figure 5 ) is taken as the target intersection point, and the target intersection point is output as the updated second corner point (C point).

[0117] Further, it can be judged whether the C point needs to be updated: whether the included angle between the fitting line segment and the vertical parking straight line CD is within the second angle interval (preferably, the second angle interval is 45° to 135°), if not, it means that the angle is too small or too large, indicating that the fitting line segment is close to and parallel to the vertical parking straight line CD, and for this case, the next step is not needed.

[0118] (II) The following is the update of the first corner point (B point) for the case that the target parking space type is a vertical parking space:

[0119] First, determine whether the target parking space is on the left or right side. The determination method here is the same as that in the C point update method, and the main purpose is to determine whether to select the left rear radar ultrasonic recognition module or the right rear radar ultrasonic recognition module (left parking selects the right rear radar ultrasonic recognition module, right parking selects the left rear radar ultrasonic recognition module). It can be understood that here the left side mainly refers to the left rear radar ultrasonic wave of the vehicle to be parked, and the right side mainly refers to the right rear radar ultrasonic wave of the vehicle to be parked, because during the reversing process, the main concern is the tail part of the vehicle to be parked. After determining which side the target parking space is on, a series of processes are performed on the ultrasonic detection data subset corresponding to the target parking space to obtain a real-time parking obstacle point set. Secondly, determine the category of the real-time parking obstacle point set. The ultrasonic detection data subset is usually detected by a radar recognition module, which is a radar. For the updating process of the C angle point of the vertical parking space and the updating process of the initial parking angle point of the horizontal parking space, the APA long-range radar can meet the needs of the radar point subset required for real-time parking angle point updating during the parking process. Therefore, almost all the data used in the C point updating process are based on the ultrasonic detection data subset detected by the APA long-range radar. However, for the updating process of the B angle point of the vertical parking space, it is impossible to detect the ultrasonic detection data subset corresponding to the APA long-range radar at some positions, and even if it can be detected, the number of corresponding point sets is very small. At this time, the ultrasonic detection data subset detected by the UPA short-range radar is obtained. It can be understood that the short-range ultrasonic wave (UPA) and the long-range ultrasonic wave sensor (APA) each have a specific detection range, performance advantage, and applicable scenario. The UPA radar detection distance is generally between 15-250 cm, the sensing distance is shorter, but the frequency is higher and the accuracy is high. The APA radar detection distance is generally between 30-500 cm, the sensing distance is longer, but the frequency is lower and the accuracy is generally lower. The working principle of ultrasonic radar is to emit and receive ultrasonic waves outward, and to measure the distance according to the return time of the ultrasonic waves.More specifically, when the to-be-parked vehicle starts to back into the vertical parking space, the tail of the to-be-parked vehicle is close to the state of facing the B point at the beginning of parking. Considering the accuracy, the tail of the to-be-parked vehicle is provided with a plurality of (for example, 4) UPA short-range radars for obtaining the point traces corresponding to the UPA short-range radar detection. When the to-be-parked vehicle approaches the vertical parking, the tail cannot detect the position of the B point at this time, and therefore the APA long-range radar needs to be used to detect the point traces corresponding to the obstacles on the side of the to-be-parked vehicle for updating the B point. During the process of the to-be-parked vehicle back into the vertical parking space, the obstacles are not located in the position that can be detected by the tail of the to-be-parked vehicle for updating the C point, and therefore the APA long-range radar is used to obtain the point traces corresponding to the obstacles throughout the process. For the horizontal parking space, the position of the to-be-parked vehicle is mainly determined by the accuracy of the positions of the head and the tail, and therefore the point traces detected by the front and rear UPA are selected. Further, for the tail of the to-be-parked vehicle, the B point is directly updated by using the point traces detected by the UPA short-range radar because the tail of the to-be-parked vehicle faces the rear obstacle during the horizontal parking. For the head of the to-be-parked vehicle, although the side of the to-be-parked vehicle faces the obstacle during part of the parking process, the angle of the point traces fitted by the APA long-range radar is very large at this time, and therefore the APA long-range radar cannot be used as the updating condition for parking in the horizontal parking space. Therefore, the C point is updated by using the point traces detected by the UPA short-range radar.

[0120] (1) When the target parking space type is a vertical parking space, the updating object is the first corner point, and the category of the real-time parking obstacle point trace set is a UPA point trace, a first updating algorithm is selected from the preset updating algorithms based on the category of the real-time parking obstacle point trace set.

[0121] It should be noted that the first corner point is the B point. In the case that there is no APA point trace meeting the condition, the subset of the ultrasonic detection data detected by the rear UPA is converted into the real-time parking obstacle point trace set. When the number of point traces in the real-time parking obstacle point trace set is greater than 2, the point traces in the real-time parking obstacle point trace set corresponding to the first corner point are fitted to obtain a fitted line segment (for example, L1). Figure 9 At this time, only the subset of the ultrasonic detection data detected by the UPA is converted into the real-time parking obstacle point trace set, Figure 9The black line segment that passes through the vertical parking space straight line AB at an angle is the fitting line segment, and the real-time included angle between the fitting line segment and the preset obstacle profile line segment (when the target parking space type is a vertical parking space, the update object is the first corner point, and the category of the real-time parking space obstacle point set is a UPA point, the preset obstacle profile line segment is the vertical parking space straight line AB) is calculated. At this time, the first update algorithm is selected as the target update algorithm, corresponding to the update of point B, which is specifically: clustering the real-time parking space obstacle point set based on the preset clustering algorithm, which only needs to cluster one cluster center. Then calculate the vertical distance Dis_B from the cluster center to the vertical parking space straight line AB as the target distance. Since the vehicle bumper has a certain curvature, if the value of Dis_B is too small (less than the first preset distance threshold), the clustered point will be between the vertical parking space straight line AB, and if the distance value is too large (greater than the second preset distance threshold), the cluster center will be between the two corner points A and B, so the point (such as Figure 9 the red dot in the middle of the fitting line segment) that is a preset vertical distance (such as 25 cm) from the vertical parking space straight line AB and on the fitting line segment is found as the target point. According to the target point, a vertical line (such as Figure 9 the black line segment in the middle of the vertical parking space straight line AB) passing through the target point is found, and the intersection point (such as Figure 9 the black point on the vertical parking space straight line AB) of the vertical line and the vertical parking space straight line AB is taken as the target intersection point, and the target intersection point is output as the updated first corner point (point B).

[0122] (2) When the target parking space type is a vertical parking space, the update object is the first corner point, and the category of the real-time parking space obstacle point set is an APA point, the method comprises the following steps:

[0123] When the vertical comparison result is that the real-time included angle is in the first included angle interval, the target update algorithm is the second update algorithm;

[0124] The second update algorithm is:

[0125] The intersection point of the fitting line segment and the preset obstacle profile line segment is taken as the target intersection point; the initial parking space corner point is updated based on the target intersection point to obtain the real-time parking space corner point.

[0126] It should be noted that the first corner point is point B, and for the case where a certain number of APA points exist, due to the parking position relationship of the vehicle to be parked, there may be UPA detected points, but the accuracy of the target intersection point obtained subsequently using UPA is lower than that obtained using APA points. Therefore, only APA detected points are selected as the real-time parking space obstacle point set (such as Figure 10, the yellow line segment and passing through the vertical parking straight line AB, that is, the fitting line segment fitted by taking the point traces detected by the APA as the real-time parking obstacle point trace set). First, judge the vertical comparison result: the real-time included angle (such as Figure 10 , the real-time included angle includes P1 and P2) is in the first included angle interval, preferably, the first included angle interval is 85° to 95°, if it is in the first included angle interval, it represents that the fitting line segment and the vertical parking straight line AB are close to vertical, at this time, the intersection point of the two line segments is the corner point closest to the real position B of the vehicle to be parked, at this time, the second update algorithm is selected as the target update algorithm, corresponding to the update of the B point, specifically: taking the intersection point of the fitting line segment and the preset obstacle contour line segment (the vertical parking straight line AB) as the target intersection point; updating the initial parking corner point B based on the target intersection point to obtain the real-time parking corner point B.

[0127] (3) When the target parking space type is a vertical parking space, the update object is the first corner point, and the category of the real-time parking obstacle point trace set is an APA trace, comprising:

[0128] When the vertical comparison result is that the real-time included angle is outside the first included angle interval and is in the second included angle interval, the target update algorithm is the first update algorithm;

[0129] The first update algorithm is:

[0130] Based on the preset clustering algorithm, the real-time parking obstacle point trace set is clustered to obtain a clustering center; based on the preset clustering algorithm, the real-time parking obstacle point trace set is clustered to obtain a clustering center set; according to the clustering center and the fitting line segment, a target trace is obtained; a first vertical line perpendicular to the parking contour line segment is constructed according to the target trace; a vertical intersection point of the first vertical line and the parking contour line segment is obtained as a target intersection point; the initial parking corner point is updated based on the target intersection point to obtain a real-time parking corner point.

[0131] It should be noted that the first corner point is B, and for the case where there are a certain number of APA traces, at this time, due to the parking position relationship of the vehicle to be parked, there may also be UPA detected traces, but the accuracy of the target intersection point obtained subsequently is lower than that obtained when using APA traces, therefore, only APA detected traces are selected as the real-time parking obstacle point trace set (such as Figure 10, the yellow line segment and passing through the vertical parking straight line AB, that is, the fitting line segment fitted by taking the point trail detected by the APA as the real-time parking obstacle point trail set). Preferably, the second angle interval is 45° to 135°, if the real-time angle between the fitting line segment and the vertical parking straight line AB is outside the first angle interval of 85°~95° and within the second angle interval of 45°~135°, then the first updating algorithm is selected as the target updating algorithm at this time, corresponding to the update of the B point, specifically: clustering the real-time parking obstacle point trail set based on a preset clustering algorithm to obtain a clustering center set, here only one clustering center needs to be clustered (such as Figure 10 the red dot on the yellow line segment). Therefore, the vertical distance Dis_B of the clustering center to the vertical parking straight line AB is calculated as the target distance. Since the vehicle bumper has a certain curvature, if the value of Dis_B is too small (less than the first preset distance threshold), the point position of the clustering center will be between the vertical parking straight line AB, and if the distance value is too large (greater than the second preset distance threshold), the clustering center will be between the two corner points A and B, therefore, the clustering center with a preset vertical distance (such as 25 cm) to the vertical parking straight line AB and on the fitting line segment is found as the target point trail according to the slope of the fitting line segment. According to the target point trail, the first vertical line (such as Figure 10 the black line segment passing through the red dot on the yellow line segment and perpendicular to the vertical parking straight line AB) perpendicular to the vertical parking straight line AB is found, and the vertical intersection point (such as Figure 10 the black dot on the vertical parking straight line AB) of the first vertical line and the vertical parking straight line AB is taken as the target intersection point, and the target intersection point is output as the updated first corner point (B point).

[0132] Further, it can be judged whether the B point needs to be updated: whether the real-time angle between the fitting line segment and the vertical parking straight line AB is within the second angle interval (preferably, the second angle interval is 45° to 135°), if not, it means that the angle is too small or too large, indicating that the fitting line segment is close to parallel to the vertical parking straight line AB, and for this case, the next step is not needed.

[0133] Secondly, the case of the target parking space type being a horizontal parking space is described. In this case, the initial parking space corner point also includes a first corner point (point B) and a second corner point (point C). It should be noted that in the updating process of the initial parking space corner point for a horizontal parking space, almost all the data used in the updating process is based on the ultrasonic detection data subset detected by the UPA radar. Based on the characteristics of the vertical parking space, all the radars used are rear radars of the vehicle (left rear side or right rear side). Based on the characteristics of the horizontal parking space, the radars used are front radars or rear radars of the vehicle. Since each radar probe has an ID, it is easy to distinguish whether the UPA detection distance belongs to the front radar of the vehicle or the rear radar of the vehicle. The ultrasonic detection data detected by the front radar is used to update the corner point C (the C point of the initial parking space corner point corresponding to the horizontal parking space). The rear radar is used to update the corner point B (the B point of the initial parking space corner point corresponding to the horizontal parking space). Then, a series of processes are performed on the ultrasonic detection data subset detected by the UPA to obtain an initial parking space obstacle point set. When the number of point traces in the real-time parking space obstacle point set is greater than 2, the Hough transform is used to screen the point traces in the initial parking space obstacle point set to obtain a real-time parking space obstacle point set. The line segment fitting is performed on the real-time parking space obstacle point set to obtain a fitted line segment.

[0134] (Three) The following describes the updating of the first corner point (point B) and the second corner point (point C) when the target parking space type is a horizontal parking space.

[0135] (1) When the target parking space type is a horizontal parking space, a horizontal comparison result of the real-time included angle and a preset horizontal included angle range threshold is obtained. Based on the horizontal comparison result, a target updating algorithm is selected from a preset updating algorithm, including:

[0136] When the horizontal comparison result is that the real-time included angle is outside the preset horizontal included angle range threshold, the target updating algorithm is a third updating algorithm.

[0137] The third updating algorithm is:

[0138] The real-time parking space obstacle point set is clustered based on a preset clustering algorithm to obtain a clustering center. A second vertical line perpendicular to the parking space contour line segment is constructed according to the clustering center. A vertical intersection point of the second vertical line and the parking space contour line segment is obtained as a target intersection point. The initial parking space corner point is updated based on the target intersection point to obtain a real-time parking space corner point.

[0139] It should be noted that for the update of the second corner point (C point), the preset obstacle contour line segment is selected as the horizontal parking space straight line CD, at this time, the fitting line segment corresponding to the second corner point under the horizontal parking space and the horizontal parking space straight line CD form a real-time included angle. For the update of the first corner point (B point), the preset obstacle contour line segment is selected as the horizontal parking space straight line AB, at this time, the fitting line segment corresponding to the first corner point under the horizontal parking space and the horizontal parking space straight line AB form a real-time included angle. It can be understood that the horizontal parking space straight line CD and the vertical parking space straight line CD are not the same, and the horizontal parking space straight line AB and the vertical parking space straight line AB are also not the same. The selection of the preset obstacle contour line segment is also not the same, so the specific position of the initial parking space corner point is also not the same as that of the initial parking space corner point under the vertical parking space. Preferably, the preset horizontal included angle range threshold is 80°-100°. It is judged whether the real-time included angle (such as Figure 11 R1 and R2 in the formula) is between 80°-100°, if not, the target update algorithm is the third update algorithm: the clustering center is obtained by clustering the real-time parking space obstacle point set (the real-time parking space obstacle point of the B point or the real-time parking space obstacle point of the C point is selected according to the actual situation) based on the preset clustering algorithm. It can be understood that when obtaining the clustering center of the horizontal parking space, it does not need to consider the problem that the vehicle bumper has a certain arc as when obtaining the clustering center of the vertical parking space, so it needs to adopt "finding the clustering center with a preset vertical distance from the straight line AB / straight line CD and the point position on the fitting line segment according to the slope of the fitting line segment". Because the position of the probe of the ultrasonic radar and other recognition modules of the horizontal parking space is mainly the front and rear of the obstacle / obstacle vehicle, it only needs to be judged according to the real-time included angle.

[0140] For the update of the C point, a second vertical line perpendicular to the horizontal parking space straight line CD is constructed through the clustering center, and the vertical intersection point of the second vertical line and the horizontal parking space straight line CD is taken as the target intersection point, which is taken as the updated C point (i.e. the C point in the real-time parking space corner point). Similarly, for the update of the B point, a second vertical line perpendicular to the horizontal parking space straight line AB is constructed through the clustering center, and the vertical intersection point of the second vertical line and the horizontal parking space straight line AB is taken as the target intersection point, which is taken as the updated B point (i.e. the B point in the real-time parking space corner point).

[0141] (2) When the target parking space type is a horizontal parking space, the horizontal comparison result of the real-time included angle and the preset horizontal included angle range threshold is obtained, and the target update algorithm is selected from the preset update algorithm based on the horizontal comparison result, including:

[0142] When the horizontal comparison result is that the real-time included angle is within the preset horizontal included angle range threshold, the target update algorithm is the second update algorithm.

[0143] It should be noted that firstly, it is judged whether the horizontal comparison result is within a preset horizontal angle range threshold, preferably, the preset horizontal angle range threshold is 80°-100°, if it is within the preset horizontal angle range threshold, it means that the fitting line segment corresponding to the B point and the vertical parking straight line AB are close to vertical, at this time, the intersection point of the two line segments is the corner point closest to the real position B of the vehicle to be parked, or the fitting line segment corresponding to the C point and the vertical parking straight line CD are close to vertical, at this time, the intersection point of the two line segments is the corner point closest to the real position C of the vehicle to be parked, at this time, the second update algorithm is selected as the target update algorithm, corresponding to the update of the B point, specifically: the intersection point of the fitting line segment corresponding to the B point and the preset obstacle contour line segment (horizontal parking straight line AB) is taken as the target intersection point; the initial parking corner point B point is updated based on the target intersection point, to obtain the B point in the real-time parking corner point. Corresponding to the update of the C point, specifically: the intersection point of the fitting line segment corresponding to the C point and the preset obstacle contour line segment (horizontal parking straight line CD) is taken as the target intersection point; the initial parking corner point C point is updated based on the target intersection point, to obtain the C point in the real-time parking corner point.

[0144] Reference Figure 11 , Figure 11 For the parking situation under the horizontal parking space, the blue line segment passing through the horizontal parking straight line AB is the fitting line segment corresponding to the B point, P1 and P2 are real-time angles, the red dot on the fitting line segment is the clustering center, the black line segment passing through the red dot and being perpendicular to the horizontal parking straight line AB is the second vertical line corresponding to the update of the B point, the intersection point of the second vertical line and the horizontal parking straight line AB is the target intersection point (the B point in the real-time parking corner point obtained by updating); the blue line segment passing through the horizontal parking straight line CD is the fitting line segment corresponding to the C point, R1 and R2 are real-time angles, the red dot on the fitting line segment is the clustering center, the black line segment passing through the red dot and being perpendicular to the horizontal parking straight line CD is the second vertical line corresponding to the update of the C point, the intersection point of the second vertical line and the horizontal parking straight line CD is the target intersection point (the C point in the real-time parking corner point obtained by updating).

[0145] See Figure 2 , the embodiment also provides a parking system, comprising: an acquisition module, configured to acquire vehicle parking-in parameters of a vehicle to be parked; a parking space confirmation module, configured to determine a target parking-in parking space of the vehicle to be parked based on a preset parking-in judgment algorithm and the vehicle parking-in parameters; a trajectory module, configured to acquire an initial trajectory of the vehicle to be parked based on the target parking-in parking space; and an update module, configured to control the vehicle to be parked to park based on the initial trajectory, in the process of parking into the target parking-in parking space, to acquire real-time parking corner points of the target parking-in parking space in real time, to update the initial trajectory based on the real-time parking corner points, to obtain a real-time dynamic trajectory, and to control the vehicle to be parked to park into the target parking-in parking space based on the real-time dynamic trajectory.

[0146] The system provided by the embodiment is simple in construction and can well implement the above-mentioned parking method, which ingeniously obtains the vehicle parking-in parameter of the vehicle to be parked and provides a data basis for subsequent steps. The target parking-in parking space of the vehicle to be parked is determined based on a preset parking-in judgment algorithm and the vehicle parking-in parameter, and the target parking-in parking space that can be parked is accurately obtained through the parking-in judgment algorithm. The initial trajectory of the vehicle to be parked is obtained based on the target parking-in parking space, so that the target vehicle to be parked can start parking in the target parking-in parking space based on the initial trajectory. The vehicle to be parked is controlled to park based on the initial trajectory, and in the process of parking in the target parking-in parking space, the real-time parking angle point of the target parking-in parking space is obtained in real time, the initial trajectory is updated based on the real-time parking angle point to obtain a real-time dynamic trajectory, and the vehicle to be parked is controlled to park in the target parking-in parking space based on the real-time dynamic trajectory. Because the initial trajectory is rough, the real-time parking angle point of the target parking-in parking space needs to be obtained in real time in the process of parking in the target parking-in parking space, and the initial trajectory is dynamically corrected in the parking process by using the dynamically obtained real-time parking angle point, so that the vehicle can be more accurately parked in the target parking-in parking space.

[0147] The above is the preferred embodiment of the present application. It should be noted that those skilled in the art can make several improvements and refinements without departing from the principles of the present application, and these improvements and refinements are also considered within the scope of protection of the present application.

Claims

1. A parking method characterized by, The method applied to a controller of a vehicle to be parked comprises: acquiring a vehicle parking parameter of the vehicle to be parked; determining a target parking space of the vehicle to be parked based on a preset parking judgment algorithm and the vehicle parking parameter; acquiring an initial trajectory of the vehicle to be parked based on the target parking space; controlling the vehicle to be parked to park based on the initial trajectory, and in the process of parking into the target parking space, acquiring a real-time parking angle point of the target parking space in real time, updating the initial trajectory based on the real-time parking angle point to obtain a real-time dynamic trajectory, and controlling the vehicle to be parked to park into the target parking space based on the real-time dynamic trajectory; wherein the controlling the vehicle to be parked to park based on the initial trajectory, and in the process of parking into the target parking space, acquiring a real-time parking angle point of the target parking space in real time, updating the initial trajectory based on the real-time parking angle point to obtain a real-time dynamic trajectory, and controlling the vehicle to be parked to park into the target parking space based on the real-time dynamic trajectory comprises: acquiring a real-time parking obstacle point set, a preset obstacle contour line segment and an initial parking angle point of the target parking space; fitting a point in the real-time parking obstacle point set by a preset fitting algorithm to obtain a fitting line segment; acquiring an included angle between the fitting line segment and the preset obstacle contour line segment as a real-time included angle; selecting a target updating algorithm from a preset updating algorithm based on the real-time included angle, the real-time parking obstacle point set and a target parking space type of the target parking space, and updating the initial parking angle point based on the target updating algorithm to obtain a real-time parking angle point; specifically, the fitting a point in the real-time parking obstacle point set by a preset fitting algorithm to obtain a fitting line segment comprises: acquiring an ultrasonic detection data set of the vehicle to be parked; selecting an ultrasonic detection data subset from the ultrasonic detection data set according to the target parking space; converting the ultrasonic detection data subset into an initial unfiltered parking obstacle point set according to a preset coordinate conversion algorithm and the vehicle parking parameter; filtering the initial unfiltered parking obstacle point set based on a preset filtering algorithm to obtain an initial parking obstacle point set; filtering the initial parking obstacle point set by a Hough transformation algorithm to obtain the real-time parking obstacle point set.

2. The parking method according to claim 1, characterized in that, The vehicle parking parameter comprises a parking state flag bit, and the determining a target parking space of the vehicle to be parked based on a preset parking judgment algorithm and the vehicle parking parameter comprises: when the parking state flag bit of the vehicle to be parked is in a parking state, acquiring a selectable target parking space set; acquiring a selectable parking obstacle angle point set and a parking map angle point of the selectable target parking space set; selecting a target parking space from the selectable target parking space set according to a comparison result of the selectable parking obstacle angle point set and the parking map angle point.

3. The parking method according to claim 1, characterized in that, The initial parking space angle point comprises a first angle point and a second angle point, the target updating algorithm is selected from the preset updating algorithm based on the real-time included angle, the real-time parking space obstacle point set and a target parking space type of the target parking space, and the initial parking space angle point is updated based on the target updating algorithm to obtain a real-time parking space angle point, comprising: When the target parking space type is a vertical parking space, the updating object is the first angle point, and the category of the real-time parking space obstacle point set is a UPA point, a first updating algorithm is selected as the target updating algorithm from the preset updating algorithm based on the category of the real-time parking space obstacle point set; When the target parking space type is a vertical parking space, the updating object is the first angle point, and the category of the real-time parking space obstacle point set is an APA point, or when the target parking space type is a vertical parking space and the updating object is the second angle point, a vertical comparison result of the real-time included angle and a preset vertical included angle range threshold is obtained, and the target updating algorithm is selected from the preset updating algorithm based on the vertical comparison result; When the target parking space type is a horizontal parking space, a horizontal comparison result of the real-time included angle and a preset horizontal included angle range threshold is obtained, and the target updating algorithm is selected from the preset updating algorithm based on the horizontal comparison result.

4. The parking method according to claim 3, characterized in that, The preset vertical included angle range threshold comprises a first included angle interval, and the vertical comparison result of the real-time included angle and the preset vertical included angle range threshold is obtained when the target parking space type is a vertical parking space, the updating object is the first angle point, and the category of the real-time parking space obstacle point set is an APA point, or when the target parking space type is a vertical parking space and the updating object is the second angle point, and the target updating algorithm is selected from the preset updating algorithm based on the vertical comparison result, comprising: When the vertical comparison result is that the real-time included angle is within the first included angle interval, the target updating algorithm is a second updating algorithm; The second updating algorithm is: The intersection point of the fitting line segment and the preset obstacle contour line segment is taken as a target intersection point; The initial parking space angle point is updated based on the target intersection point to obtain the real-time parking space angle point.

5. The parking method according to claim 4, characterized in that, The preset vertical included angle range threshold further comprises a second included angle interval, the numerical range of the first included angle interval is smaller than the numerical range of the second included angle interval, the vertical comparison result of the real-time included angle and the preset vertical included angle range threshold is obtained when the target parking space type is a vertical parking space, the updating object is the first angle point, and the category of the real-time parking space obstacle point set is an APA point, or when the target parking space type is a vertical parking space and the updating object is the second angle point, and the target updating algorithm is selected from the preset updating algorithm based on the vertical comparison result, comprising: When the vertical comparison result is that the real-time included angle is outside the first included angle interval and within the second included angle interval, the target updating algorithm is a first updating algorithm; The first updating algorithm is: cluster the real-time parking space obstacle point set based on a preset clustering algorithm to obtain a cluster center set; obtain a target point based on the cluster center and the fitting line segment; construct a first vertical line perpendicular to the parking space contour line segment based on the target point; obtain a vertical intersection point of the first vertical line and the parking space contour line segment as a target intersection point; update the initial parking space angle point based on the target intersection point to obtain the real-time parking space angle point.

6. The parking method according to claim 3, wherein, when the target parking space type is a horizontal parking space, obtain a horizontal comparison result of the real-time included angle and a preset horizontal included angle range threshold, and select the target update algorithm from the preset update algorithms based on the horizontal comparison result, including: when the horizontal comparison result is that the real-time included angle is outside the preset horizontal included angle range threshold, the target update algorithm is a third update algorithm; the third update algorithm is: cluster the real-time parking space obstacle point set based on a preset clustering algorithm to obtain a cluster center set; construct a second vertical line perpendicular to the parking space contour line segment based on the cluster center; obtain a vertical intersection point of the second vertical line and the parking space contour line segment as a target intersection point; update the initial parking space angle point based on the target intersection point to obtain the real-time parking space angle point.

7. The parking method according to claim 4, wherein, when the target parking space type is a horizontal parking space, obtain a horizontal comparison result of the real-time included angle and a preset horizontal included angle range threshold, and select the target update algorithm from the preset update algorithms based on the horizontal comparison result, including: when the horizontal comparison result is that the real-time included angle is within the preset horizontal included angle range threshold, the target update algorithm is the second update algorithm.

8. A parking system, characterized in that including: an acquisition module, configured to acquire vehicle parking parameters of a vehicle to be parked; a parking space confirmation module, configured to determine a target parking space of the vehicle to be parked based on a preset parking judgment algorithm and the vehicle parking parameters; a trajectory module, configured to acquire an initial trajectory of the vehicle to be parked based on the target parking space; an update module, configured to control the vehicle to be parked to park based on the initial trajectory, acquire a real-time parking space angle point of the target parking space in real time in the process of parking into the target parking space, update the initial trajectory based on the real-time parking space angle point to obtain a real-time dynamic trajectory, and control the vehicle to be parked to park into the target parking space based on the real-time dynamic trajectory; wherein the update module is further configured to: acquire a real-time parking space obstacle point set of the target parking space, a preset obstacle contour line segment, and an initial parking space angle point; fit a point in the real-time parking space obstacle point set based on a preset fitting algorithm to obtain a fitting line segment; obtain an included angle between the fitting line segment and the preset obstacle contour line segment as a real-time included angle; select a target update algorithm from preset update algorithms based on the real-time included angle, the real-time parking space obstacle point set, and a target parking space type of the target parking space, and update the initial parking space angle point based on the target update algorithm to obtain a real-time parking space angle point; Specifically, the updating module is configured to fit the trajectories in the set of real-time parking space obstacle trajectories by a preset fitting algorithm to obtain a fitted line segment, and specifically configured to: obtain a set of ultrasonic detection data of the vehicle to be parked; select a subset of ultrasonic detection data from the set of ultrasonic detection data according to the target parking space; convert the subset of ultrasonic detection data into an initial set of unfiltered parking space obstacle trajectories according to a preset coordinate conversion algorithm and the vehicle parking parameters; filter the initial set of unfiltered parking space obstacle trajectories based on a preset filtering algorithm to obtain an initial set of parking space obstacle trajectories; filter the initial set of parking space obstacle trajectories by a Hough transform algorithm to obtain the set of real-time parking space obstacle trajectories.

Images