Method and apparatus for generating memory parking assist path

Abstract

A method and apparatus for generating a safe memory parking assist path by taking into account perception and positioning errors due to distortion of an SVM wide angle camera is disclosed. The method for generating the parking path comprises the steps of obtaining an image around a vehicle through a camera; obtaining positions of the vehicle and obstacles around the vehicle by using the image; in response to determining that the vehicle is approaching the memory path, activating a memory parking assist (MPA); predicting whether the vehicle collides with any of the obstacles if the vehicle follows the memory path; in response to determining that the vehicle is predicted to collide with the obstacle, determining whether the obstacle is evadable; and generating an avoidance path for the vehicle to avoid the obstacle and a convergence path for the vehicle to converge to the memory path.

Description

Technical Field

[0001] The present disclosure relates to a method and apparatus for generating a memory parking path. More specifically, the present disclosure relates to a method and apparatus for generating a safe memory parking path by considering perception and positioning errors caused by distortion of a wide-angle camera of a surround view monitor (SVM). Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] Parking assist technologies that can utilize autonomous driving technologies are being widely developed. Among parking assist functions, a notable function is the memory parking assist (MPA) function.

[0004] The memory parking assist function enables a vehicle to memorize a parking path when a driver parks the vehicle in a path memory mode. The memory parking assist function automatically parks the vehicle in a corresponding parking space during the next parking without driver intervention. The memory parking assist function allows a driver to remotely park the vehicle at the same location based on the recorded memory path, thereby reducing the driver's burden in the parking task.

[0005] The memory parking assist device controls the vehicle to follow the recorded memory path. The memory parking assist device utilizes information from a wide-angle camera of a surround view monitor (SVM) installed on the vehicle to sense the driving environment and the vehicle position, and then generates a safe path.

[0006] The SVM camera uses a wide-angle lens. A camera with a wide-angle lens can generate an image with a wide viewing angle. In other words, even in a limited space, the SVM wide-angle camera can obtain an image with a wide field of view, thus reducing the need for a large number of cameras to capture the entire area.

[0007] However, a wide-angle lens inherently introduces lens distortion and prevents the vehicle from accurately identifying the position and distance of surrounding objects. Therefore, positioning that estimates the vehicle position based on information of surrounding objects inevitably introduces errors.

[0008] In a parking lot environment, various factors cause uncertainties, such as limited visibility due to other vehicles and narrow driving roads. If the vehicle attempts to follow the memory path by relying on error-prone perception and positioning information, the vehicle may not faithfully follow the memory path in the parking lot environment. Sometimes, a situation may occur where the vehicle fails to converge to the memory path and collides with an obstacle.

[0009] Therefore, the memory parking assist device needs to find a path to safely converge from the initial vehicle position to the memory path by considering the kinematic behavior of the vehicle. The memory parking assist device uses an SVM wide-angle camera to capture information on the surrounding environment and estimates the position of the vehicle. Since a wide-angle camera is used, perception and positioning errors will occur. The memory parking assist device should consider the corresponding errors and design a safe path according to the driving situation and the driving environment in the parking lot. The memory parking assist device should be able to design a path that safely avoids obstacles that interfere with following the memory path and returns to the memory path using collision prediction technology that takes into account the corresponding errors. Summary of the Invention

[0010] In view of the above situation, an object of the present disclosure is to provide a method and apparatus for generating a safe memory parking assist path by considering perception and positioning errors caused by distortion of a panoramic monitor (SVM) wide-angle camera.

[0011] The technical objectives to be achieved by the present disclosure are not limited to the above technical objectives, and other technical objectives not mentioned above can also be clearly understood by those of ordinary skill in the art to which the present disclosure pertains from the following description.

[0012] In an embodiment of the present disclosure, a method for generating a parking path includes: obtaining an image of the vehicle's surroundings by using a camera. The method further includes: obtaining the positions of the vehicle and obstacles around the vehicle by using the image. The method further includes: activating memory parking assist (MPA) in response to determining that the vehicle is approaching the memory path. The method further includes: predicting whether the vehicle will collide with any of the obstacles if the vehicle follows the memory path. The method further includes: determining whether the obstacle is avoidable in response to determining that the vehicle is predicted to collide with the obstacle. The method further includes: generating an avoidance path for the vehicle to avoid the obstacle. The method further includes: generating a convergence path for the vehicle to converge to the memory path.

[0013] According to an embodiment of the present disclosure, a memory parking assist path can be generated by considering perception and positioning errors caused by using an SVM wide-angle camera, thereby enabling safer autonomous parking.

[0014] According to an embodiment of the present disclosure, the vehicle margin can be designed in more detail, thereby enabling safer autonomous parking in a parking lot environment where the vehicle must navigate along a narrow road with a high curvature.

[0015] The advantageous effects of the present disclosure are not limited to the above. Those of ordinary skill in the art can clearly understand other advantageous effects of the present disclosure not mentioned above from the following description. Brief Description of the Drawings

[0016] Figure 1 is a block diagram briefly illustrating the internal structure of a memory parking assist device according to an embodiment of the present disclosure.

[0017] Figure 2 is a flowchart illustrating the overall operation process of a memory parking assist function according to an embodiment of the present disclosure.

[0018] Figure 3 is a flowchart illustrating the detailed operation process of a memory parking assist function according to an embodiment of the present disclosure.

[0019] Figure 4 illustrates the vehicle side area used in a general road environment and the front shape of the vehicle side area adopted in an embodiment of the present disclosure.

[0020] Figure 5 illustrates a method of designing a vehicle side area according to an embodiment of the present disclosure.

[0021] Figure 6 illustrates the collision prediction result according to the vehicle side area design method for predicting whether a vehicle collides with an obstacle at a narrow intersection of a road.

[0022] Figure 7 illustrates the spiral curve path along which a vehicle follows a memory path in a memory parking assist method according to an embodiment of the present disclosure.

[0023] Figures 8A - 8D illustrates the process by which a memory parking assist device according to an embodiment of the present disclosure determines whether a vehicle collides with an obstacle or avoids the obstacle, and generates an avoidance path and a convergence path.

[0024] Figure 9A and Figure 9B illustrates the generation of a safe path by a memory parking assist device according to an embodiment of the present disclosure considering the perception error caused by a wide-angle camera.

[0025] Figure 10A and Figure 10B illustrates the operation of a memory parking assist device according to an embodiment of the present disclosure when the vehicle position contains an error. Detailed Description of the Embodiment

[0026] Hereinafter, an embodiment of the present disclosure will be described in detail with reference to the accompanying drawings. In the following description, like reference numerals preferably denote like elements, although these elements are shown in different drawings. In addition, in the following description of the embodiment, detailed descriptions of known functions and configurations incorporated therein have been omitted for the purpose of clarity and conciseness.

[0027] In addition, various terms such as first, second, A, B, (A), (B), etc. are only used to distinguish one component from another, rather than implying or suggesting the substance, order, or sequence of the components. Throughout the present disclosure, when a component 'includes' or 'comprises' a component, the component is intended to also include other components, rather than being intended to exclude other components, unless there is a contrary specific statement. Terms such as 'unit','module', etc. refer to one or more units for processing at least one function or operation, which can be implemented by hardware, software, or a combination thereof. When a controller, module, component, device, element, etc. of the present disclosure is described as having a purpose or performing an operation, function, etc., the controller, module, component, device, element, etc. should be considered herein as 'configured to' meet the purpose or perform the operation or function. Each controller, module, component, device, element, etc. can be embodied separately or include a processor and a memory, such as a non-transitory computer-readable medium, as components of the device.

[0028] The following detailed description and the accompanying drawings are intended to describe an embodiment of the present invention, rather than merely representing an embodiment in which the present disclosure can be practiced.

[0029] Memory Parking Assist (MPA) is a function in which the vehicle memorizes the parking path when the driver parks the vehicle, and the MPA automatically parks the vehicle in the same parking space the next time without driver intervention.

[0030] Figure 1 is a block diagram briefly illustrating the internal structure of a memory parking assist device according to an embodiment of the present disclosure.

[0031] As Figure 1 shown, a memory parking assist device according to an embodiment of the present disclosure may include all or part of a camera 110, an obstacle detector 120, an obstacle determiner 130, a path generator 140, and a navigation controller 150.

[0032] It should be noted that not all of the components shown Figure 1 are necessary components, and some modules in the device for generating a safe memory parking assist path can be added, modified, or deleted. On the other hand, Figure 1 the components represent functionally different elements, and at least one or more of the components can be implemented in an integrated form in an actual physical environment.

[0033] The camera 110 includes a surround view monitor (SVM) wide-angle camera. The SVM camera mounted on the vehicle captures images of the surrounding area of the vehicle. According to an embodiment of the present disclosure, the SVM camera may capture images of vehicles, pillars, and / or parking lines in a parking lot. The SVM camera may be mounted at one or more positions to capture images of the front, rear, and sides of the vehicle. However, the present disclosure is not limited to the specific description and may vary according to specific applications.

[0034] The obstacle detector 120 may identify the space within the parking lot based on the images obtained from the camera 110. The obstacle detector 120 may obtain the positions of other vehicles, pillars, and parking lines in the parking lot and may generate a map of the parking lot using the obtained positions.

[0035] The obstacle detector 120 may generate point cloud data representing obstacles based on the images of the camera 110. The obstacle detector 120 clusters the point cloud data. Here, clustering refers to a machine learning technique that classifies data with similar features into the same group. Since the specific method for clustering the points representing a single obstacle from the point cloud data is well known to those skilled in the art, a detailed description thereof is omitted. Therefore, the present disclosure does not limit the clustering method to a specific method. According to an embodiment, the obstacle detector 120 may obtain data that has been preprocessed and / or clustered in the camera 110.

[0036] The obstacle detector 120 simplifies each cluster. Each cluster contains a large number of points. The obstacle detector 120 replaces the large number of points with, for example, two or three points to simplify each cluster. The obstacle detector 120 may distinguish adjacent obstacles and may obtain the position and orientation information of the obstacles by connecting two or three points selected within each cluster.

[0037] The obstacle determiner 130 uses the data obtained by the obstacle detector 120 to search the memory path navigation area. The obstacle determiner 130 determines whether the vehicle will collide with an obstacle within the navigation area when the vehicle moves along the memory path. If it is determined that the vehicle is expected to collide with an obstacle within the navigation area, the obstacle determiner 130 determines whether the corresponding obstacle can be avoided.

[0038] The path generator 140 may generate an autonomous parking path using the information received from the camera 110, and vehicles near the memory path may converge to the memory path along this path.

[0039] The path generator 140 may generate an autonomous parking path using the information received from the obstacle detector 120 and the obstacle determiner 130, and the vehicle may navigate along this path without colliding with obstacles.

[0040] The navigation controller 150 can control the movement of the vehicle. The navigation controller 150 can generate signals for controlling the movement of the vehicle according to an autonomous navigation path. The navigation controller 150 can electrically control various vehicle driving devices within the vehicle through the generated signals. The vehicle driving devices can include a steering device, a braking device, a suspension device, and / or a power system.

[0041] Figure 2 is a flowchart illustrating the overall operation process of a memory parking assist function according to an embodiment of the present disclosure.

[0042] When the vehicle enters a parking lot, the position of the vehicle within the parking lot is estimated to activate the memory parking assist function (S201). The memory parking assist device utilizes a pre-recorded driving path. The memory parking assist device determines whether the vehicle is approaching the memory path (S202). If the vehicle is not approaching the memory path ("No" in S202), the position of the vehicle within the parking lot is periodically updated to determine whether the vehicle is approaching the memory path (S201 and S202).

[0043] If the vehicle is approaching the memory path ("Yes" in S202), the vehicle activates the memory path navigation function (S203). When following the pre-stored memory path, the vehicle determines whether a navigation problem occurs, such as a collision with an obstacle. If the vehicle determines that the pre-stored memory path is inappropriate, the vehicle creates a new safe path (S204). The vehicle uses the parking lot navigation control system to navigate to the destination parking position (S205). When the vehicle reaches the parking position, the parking control system is used to safely park the vehicle at the parking position (S206).

[0044] Figure 3 is a flowchart illustrating the detailed operation process of a memory parking assist function according to an embodiment of the present disclosure. Figure 3 Illustrates a series of processes for generating a safe path and controlling navigation in a parking lot after the vehicle determines that it has approached the memory path ( Figure 2 steps S203 to S205).

[0045] When entering the parking lot, if the vehicle determines that it is approaching the memory path, the vehicle activates the memory parking assist function (S301).

[0046] The memory parking assist device generates a navigation area on the memory path and searches for obstacles existing within the navigation area (S302).

[0047] When the vehicle moves along the memory path, the memory parking assist device determines whether the vehicle collides with an obstacle in the navigation area (S303). If the memory parking assist device determines that even if the vehicle follows the memory path, the vehicle will not collide with an obstacle (\"No\" in S303), the memory parking assist device performs memory path following control (S309). Refer to Figures 4 - 6 Describe in detail the specific method for determining whether the vehicle is expected to collide with an obstacle. Refer to Figure 7 Describe in detail the specific method for the vehicle to converge to the memory path.

[0048] If the memory parking assist device determines that when the vehicle follows the memory path, it will collide with an obstacle (\"Yes\" in S303), the memory parking assist device determines whether the collided obstacle is an avoidable obstacle (S304). An avoidable obstacle refers to an obstacle that minimally encroaches on the vehicle navigation area, allowing the memory parking assist device to generate an avoidance path. If an obstacle occupies a large part of the vehicle navigation area and the vehicle cannot move, the corresponding obstacle may not be classified as an avoidable obstacle. For example, if another vehicle is in the process of parking and obstructs the road in the parking lot, the corresponding vehicle may not be classified as an avoidable obstacle.

[0049] The memory parking assist device can determine whether the collided obstacle searched in step S302 is an avoidable obstacle (S304). In step S304, if the memory parking assist device determines that a part of the area occupied by the obstacle is large (\"No\" in S304), the memory parking assist device temporarily stops the vehicle (S307). After the vehicle is temporarily stopped, the memory parking assist device determines whether the obstacle has disappeared or has been removed (S308). When the obstacle persists (\"No\" in S308), the memory parking assist device maintains the temporarily stopped state (S307). When the memory parking assist device determines that the obstacle has been removed (\"Yes\" in S308), the memory parking assist device performs memory path following control (S309).

[0050] In step S304, if the memory parking assist device determines that the part of the area occupied by the obstacle is not large (\"Yes\" in S304), the memory parking assist device classifies the corresponding obstacle as an avoidable obstacle and generates an avoidance path and a convergence path (S305). Will refer to Figures 8A - 8D Describe in detail the specific method for generating an avoidance path and a convergence path.

[0051] When generating a avoidance path and a convergence path, the memory parking assist device stores the avoidance path and the convergence path as a new memory path (S306). Subsequently, if the memory parking assist function is activated in the same parking lot, the memory parking assist device performs memory parking assist using the updated memory path.

[0052] After updating the memory path, the memory parking assist device returns to step S302. The memory parking assist device determines whether the parking lot environment has changed during the update of the memory path, and whether a collision with an obstacle or an avoidance of an obstacle has occurred (S303 and S304). In the case of no collision with an obstacle or avoidance of an obstacle, the memory parking assist device performs path following control using the updated memory path (S309).

[0053] Figure 4 The figure illustrates a vehicle perimeter used in a general road environment and the front shape of a vehicle perimeter employed in an embodiment of the present disclosure. Figure 4 The dashed line 401 represents a conventional vehicle perimeter, while the solid line 403 represents a vehicle perimeter according to an embodiment of the present disclosure.

[0054] The vehicle perimeter refers to a reference point or reference line used to determine whether a vehicle is expected to collide with an obstacle. If the vehicle perimeter contacts an obstacle, the memory parking assist device determines a collision.

[0055] Compared with the prior art, the present disclosure applies a more specifically designed vehicle perimeter for predicting whether a vehicle will collide with an obstacle (S503), determining whether an obstacle is an avoidance obstacle (S304), and generating an avoidance path and a convergence path executed by the memory parking assist device (S305).

[0056] In Figure 4 the long straight line 402 represents the length measured diagonally from the front bumper of the vehicle when applying a conventional vehicle perimeter design. The short straight line 404 represents the length measured diagonally from the front bumper of the vehicle when applying a vehicle perimeter design according to an embodiment of the present disclosure. In the following description, the length measured diagonally from the front bumper of the vehicle is simply referred to as the length of the round perimeter.

[0057] The conventional vehicle perimeter has a rectangular shape. If a rectangular perimeter is applied, the round perimeter length 402 becomes longer.

[0058] The roads in a parking lot environment are usually narrow. In a parking lot environment, a vehicle needs to navigate along a narrow road with a high curvature. Situations involving a vehicle navigating along a path with a high curvature include full-turn navigation at a parking lot intersection, a T-shaped road, or a U-shaped road.

[0059] Since conventional vehicle margins are designed assuming a general road environment, they are not suitable for use in a parking lot environment. If a conventional vehicle margin is applied to a parking lot environment, the automatic parking system may determine that the vehicle and an obstacle will collide even if the obstacle is at a safe distance. The automatic parking system may determine to generate an overly evasive path or may not generate an evasive path. In other words, due to the characteristics of the parking lot environment (which requires navigation along narrow roads with high curvature), it is not suitable to directly apply the vehicle margin design used in a general road environment to the parking lot environment.

[0060] The vehicle margin design according to an embodiment of the present disclosure designs the vehicle margin based on various positions including FL, FR, FSL, and FSR. If the vehicle margin according to an embodiment of the present disclosure is applied, the length 404 of the rounded margin becomes shorter than the length 402 of the existing rounded margin. Based on this feature, the automatic parking system can avoid erroneously determining that an obstacle at a safe distance collides with the vehicle. The automatic parking system can generate an evasive path that does not deviate significantly from the memorized path.

[0061] Figure 5 Illustrated is a method for designing a vehicle margin according to an embodiment of the present disclosure. Figure 5 Illustrated is a case where the vehicle moves from an initial pose (0, 0, 0) to the next pose (x, y, ψ). The pose refers to information including the vehicle position (x and y coordinate values in a 3D coordinate system) and orientation information (the angle measured around the z-axis in the 3D coordinate system, i.e., the vehicle's heading angle).

[0062] In Figure 5 , the front left (FL) represents the left front point of the vehicle margin, the front right (FR) represents the right front point of the vehicle margin, the front left side (FSL) represents the left front side point of the vehicle margin, the front right side (FSR) represents the right front side point of the vehicle margin, the rear left (RL) represents the left rear point of the vehicle margin, and the rear right (RR) represents the right rear point of the vehicle margin.

[0063] In Figure 5 , w represents the horizontal width of the vehicle, FOL represents the distance from the center of the rear axle of the vehicle to the front bumper, and ROH represents the distance from the rear axle of the vehicle to the rear bumper. In Figure 5 In margin , Long margin represents the longitudinal margin of the front part of the vehicle, and Lat Figure 5In this case, L1 represents the length from the center of the rear axle of the vehicle to FR or FL; α1 represents the angle formed by the straight line L1 and the straight line FOL; L2 represents the length from the center of the rear axle to FSR or FSL; α2 represents the angle formed by the straight line L2 and the straight line FOL; L3 represents the length from the center of the rear axle of the vehicle to RR or RL; and α3 represents the angle formed by the straight line L3 and the straight line perpendicular to FOL.

[0064] In a conventional side region design, the vehicle side region is modeled as a rectangular shape. However, as Figure 5 shown, the vehicle side region according to an embodiment of the present disclosure can be modeled as a polygonal shape formed by connecting four or more points (FL, FR, FSL, FSR, RL, and RR).

[0065] As Figure 5 shown, when the vehicle attempts to move from the initial posture (0, 0, 0) to the next posture (x, y, ψ), the coordinates of each point (FL, FR, FSL, FSR, RL, and RR) can be calculated by the following formulas 1 to 3.

[0066] [Formula 1]

[0067] Lat margin = 0.3

[0068] Long margin = 0.2

[0069] F.Lat = Lat margin + 0.1

[0070] F.Long = Long margin + 0.1

[0071] In Formula 1, the unit is meters.

[0072] According to this embodiment, the lateral side region Lat margin of the vehicle is 0.3 m, and the longitudinal side region Long margin in front of the vehicle is 0.2 m.

[0073] F.Lat is a value obtained by adding a predetermined length to Lat margin , and F.Lat according to this embodiment is a value obtained by adding 0.1 m to Lat margin . F.Long is a value obtained by adding a predetermined value to Long margin , and F.Long according to this embodiment is a value obtained by adding 0.1 m to Long margin . By adding a predetermined length to F.Lat and F.Long, a circular side region length suitable for the parking lot environment can be set for predicting obstacle collisions.

[0074] The values of Formula 1 set for each variable are merely examples; those of ordinary skill in the art to which the present disclosure pertains should understand that the values of each variable according to an embodiment of the present disclosure can be adjusted in various ways without impairing their inherent characteristics.

[0075] Formula 2 calculates L1, α1, L2, α2, L3, and α3 using the result of Formula 1.

[0076] [Formula 2]

[0077]

[0078] Formula 2 can calculate the length L1 from the center of the rear axle of the vehicle to FR or FL, the angle α1 formed between the straight line L1 and the straight line FOL, the length L2 from the center of the rear axle of the vehicle to FSR or FSL, the angle α2 formed between the straight line L2 and the straight line FOL, the length L3 from the center of the rear axle of the vehicle to RR or RL, and the angle α3 formed between the straight line L3 and the straight line FOL.

[0079] Since those of ordinary skill in the art to which the present disclosure pertains can easily derive Formula 2 for calculating lengths and angles using the Pythagorean theorem and trigonometric functions, a detailed description of Formula 2 is omitted.

[0080] Formula 3 calculates the coordinates of FL, FR, FSL, FSR, RL, and RR using the result of Formula 2.

[0081] [Formula 3]

[0082] FL = [x + L1·cos(Ψ + α1), y + L1·sin(Ψ + α1)]

[0083] FR = [x + L1·cos(Ψ - α1), y + L1·sin(Ψ - α1)]

[0084] FSL = [x + L2·cos(Ψ + α2), y + L2·sin(Ψ + α2)]

[0085] FSR = [x + L2·cos(Ψ - α2), y + L2·sin(Ψ - α2)]

[0086] RL = [x + L3·cos(Ψ - α3), y + L3·sin(Ψ - α3)]

[0087] RR = [x + L3·cos(Ψ + α3), y + L3·sin(Ψ + α3)]

[0088] In Equation 3, x represents the distance that the vehicle moves along the x-axis from its current position, y represents the distance that the vehicle moves along the y-axis from its current position, and Ψ represents the heading angle at the destination position when assuming that the current heading angle of the vehicle is 0°.

[0089] When the calculation result of Equation 2 is substituted into Equation 3, the coordinates of FL, FR, FSL, FSR, RL, and RR at the next attitude (x, y, ψ) of the vehicle can be calculated.

[0090] Since those of ordinary skill in the art to which the present disclosure pertains can easily derive Equation 3 for calculating coordinates using the Cartesian coordinate system and trigonometric functions, a detailed description of Equation 3 is omitted.

[0091] In Figure 3 the step (S303) of predicting an obstacle collision, when predicting whether the vehicle collides with another vehicle or an obstacle in the parking lot, the vehicle side region design shown in Figure 5 is applied. The obstacle determiner 130 determines whether the vehicle collides with an obstacle in the parking lot when the vehicle follows the path by using the coordinates FL, FR, FSL, FSR, RL, and RR at the next attitude (x, y, ψ) of the vehicle calculated using Equations 1 to 3.

[0092] Figure 3 The step S305 of generating an avoidance path and a convergence path in Figure 5 also generates a path to avoid collision with an obstacle by applying the vehicle side region design shown in

[0093] Figure 6 illustrates the collision prediction result according to the vehicle side region design method for predicting whether a vehicle collides with an obstacle at an intersection with a narrow road. Figure 6 shows the problems encountered when performing collision prediction by applying the existing vehicle side region design in a parking lot environment. Figure 6 shows the improved effect obtained when performing collision prediction by applying the vehicle side region according to an embodiment of the present disclosure.

[0094] The vehicle predicts whether it collides with an obstacle before following the memorized path from the current position 600 to the next position 600a or 600b.

[0095] When Figure 6 the vehicle in turns left and navigates along the memorized path 602a, the vehicle can move from the current position 600 to the next position 600a. Before following the memorized path 602a, the memorized parking assistance device predicts whether the vehicle collides with the obstacle 608a if the vehicle navigates along the memorized path 602a.

[0096] In Figure 6In [description], the dashed line 604a is the boundary line based on which the memory parking assist device determines that the vehicle will collide with the obstacle 608a when applying the existing rectangular vehicle side area. Using the existing rectangular side area increases the collision area of the vehicle. The memory parking assist device determines that if the vehicle follows the memory path 602a, the vehicle will collide with the obstacle 608a. The memory parking assist device determines that an avoidance path must be generated or the vehicle must stop temporarily.

[0097] In Figure 6 [description], the solid line 606a is the boundary line based on which the memory parking assist device determines that the vehicle collides with the obstacle 608a when applying the vehicle side area according to an embodiment of the present disclosure. When using the detailed vehicle side area, the collision area of the vehicle becomes relatively narrow. The memory parking assist device determines that even if the vehicle follows the existing memory path 602a, the vehicle will not collide with the obstacle 608a. The memory parking assist device can perform the follow-up control of following the memory path instead of generating an avoidance path or making a temporary stop.

[0098] Figure 6 Another example of [description] involves the case where the vehicle turns right. In the corresponding example, if the vehicle follows the memory path 602b, the vehicle navigates along a path with a high curvature.

[0099] In Figure 6 Another example of [description], the dashed line 604b is the boundary line based on which the memory parking assist device determines that the vehicle will collide with the right obstacle when applying the existing rectangular vehicle side area. Using the existing rectangular side area increases the collision area of the vehicle. When the vehicle navigates on a narrow road with a high curvature, the memory parking assist device frequently determines that the vehicle will collide with the right obstacle 608b. The memory parking assist device determines that an avoidance path must be generated or the vehicle must stop temporarily.

[0100] In Figure 6 Another example of [description], the solid line 606b is the boundary line based on which the memory parking assist device determines that the vehicle collides with the right obstacle 608b when applying the detailed vehicle side area according to an embodiment of the present disclosure. When using the detailed vehicle side area, the collision area of the vehicle becomes relatively narrow. The memory parking assist device determines that even if the vehicle follows the memory path 602b, the vehicle will not collide with the obstacle 608b. The memory parking assist device can perform the follow-up control of following the memory path instead of generating an avoidance path or making a temporary stop.

[0101] Figure 6The figure illustrates the improved effect obtained when predicting a collision with the vehicle boundary according to an embodiment of the present disclosure. If the vehicle boundary is designed in detail, the memory parking assist device can prevent the frequent generation of avoidance paths and the update of memory paths. If the vehicle boundary is designed in detail, unnecessary calculations of the memory parking assist device can be reduced.

[0102] The path for guiding the vehicle from its current position to the memory path during step S302 of searching for the memory path navigation area and the path for guiding the vehicle from the avoidance position to the memory path during step S305 of generating the avoidance path and the convergence path are based on a clothoid curve.

[0103] A clothoid curve is a type of easement curve. An easement curve (easement section) refers to a curve (section) in which the curve curvature changes linearly with the curve length. The easement section is a path section that connects two paths so that the vehicle can travel safely and comfortably when moving between a straight line and a curve or between curves with different curvatures. Since step S305 of generating the avoidance path and the convergence path basically requires generating a path for safe vehicle navigation, the embodiments of the present disclosure default to generating a path based on a clothoid curve.

[0104] However, the path generation according to the present disclosure is not limited to the path generation based on a clothoid curve. In addition to the clothoid curve, the path according to the present disclosure can be generated based on various easement curves, including a lemniscate, a cubic parabola, etc.

[0105] Figure 7 The figure illustrates a clothoid curve path 703 for a vehicle 700 to follow a memory path 702 in a memory parking assist method according to an embodiment of the present disclosure.

[0106] The vehicle 700 generates a clothoid curve path based on its initial attitude and target attitude. More specifically, a vehicle near the memory path 702 can generate the Figure 7 illustrated clothoid curve path 703 by using G1 Hermite interpolation, where the initial attitude (-L, 0, 0) and the target attitude (x, y, ψ) at the center of the rear axle of the vehicle 700 are used as inputs. Since the specific method of calculating the clothoid curve parameters and generating the clothoid curve path by using the initial attitude and target attitude of the vehicle as inputs is well known, a detailed description thereof is omitted.

[0107] If the path generation method according to an embodiment of the present disclosure is applied, when the vehicle approaches the memory path in the parking lot, the vehicle can converge safely from the initial attitude of the vehicle to the memory path.

[0108] If the path generation method according to an embodiment of the present disclosure is applied, the vehicle can converge safely from the position generated after the vehicle avoids an obstacle to the memory path.

[0109] If the path generation method according to one embodiment of the present disclosure is applied, even if the vehicle deviates from the memory path due to the occurrence of perception error, positioning error or control error, the vehicle can safely converge to the memory path.

[0110] Figures 8A - 8D A process in which a memory parking assist device determines whether a vehicle collides with an obstacle or avoids an obstacle and generates an avoidance path and a convergence path according to an embodiment of the present disclosure is illustrated.

[0111] If an obstacle in the navigation area is determined to be an avoidance obstacle after performing obstacle collision prediction (S303) and obstacle avoidance determination (S304), the memory parking assist device generates an avoidance path for avoiding the corresponding obstacle and a convergence path returning to the memory path (S305). Figures 8A - 8D , the process of generating the avoidance path and the convergence path by the memory parking assist device will be described in more detail.

[0112] Figure 8A The result of the obstacle search in the navigation area 802a and the area of ​​interest according to the memory path 801a is shown. The vehicle can classify obstacles that are likely to collide with the vehicle among multiple obstacles. Figure 8A In the figure, dots represent obstacles with potential collision.

[0113] Figure 8B An obstacle collision prediction process is shown, which predicts whether the vehicle will collide with an obstacle within the navigation area as the vehicle navigates along the memorized path. Figure 8B A point 801b where the vehicle collides with an obstacle when navigating along the memory path is shown. When the collision point 801b occurs, the memory parking aid determines whether the obstacle at the collision point 801b is an avoidance obstacle. In this example, the corresponding obstacle is classified as an avoidance obstacle.

[0114] Figure 8C A process for generating an avoidance path 803c by a memory parking assist device is shown. The memory parking assist device draws a virtual straight line that passes through the collision point 801b and is perpendicular to the memory path 802a, and sets an avoidance point 802c on the straight line in a direction away from the avoidance obstacle. The memory parking assist device generates a clothoid path connecting the current position of the vehicle and the avoidance point 802c. The memory parking assist device determines whether a collision with an obstacle occurs on the corresponding clothoid path. Figure 8CIt shows the point 801c where the vehicle collides with another obstacle in the navigation area when generating a clothoid path using the initial avoidance point 802c. The memory parking assist device reconfigures the avoidance point 802c until an avoidance path 803c is generated, which ensures that the vehicle does not collide with the obstacles in the navigation area.

[0115] Figure 8D It shows the process of generating a convergence path 803d by the memory parking assist device. The memory parking assist device sets a return point 801d on the memory path 802a. The return point 801d is set at a position separated from the avoidance point 802d by a sufficient distance so that a vehicle that has deviated from the memory path can safely return to the memory path. The memory parking assist device generates a clothoid path connecting the avoidance point 802d and the return point 801d. The memory parking assist device predicts whether the vehicle collides with an obstacle on the corresponding clothoid path. The memory parking assist device reconfigures the return point 801d until a convergence path 803d is generated, which ensures that the vehicle does not collide with the obstacles in the navigation area.

[0116] If the avoidance path 803c and the convergence path 803d are generated, the memory parking assist device according to an embodiment of the present disclosure updates the avoidance path 803c and the convergence path 803d to a new memory path S306.

[0117] In another embodiment of the present disclosure, after updating the memory path, the memory parking assist device can reduce the vehicle margin compared to the previously set value to prevent frequent updates of the memory path. The vehicle margin can be set by adjusting one or more values of Lat margin 、Long margin 、F.Lat and F.Long.

[0118] Figure 9A and Figure 9B It illustrates the generation of a safe path by the memory parking assist device according to an embodiment of the present disclosure considering the perception error caused by the wide-angle camera.

[0119] As the distance between the wide-angle camera and the obstacle increases, the distortion of the wide-angle lens becomes severe. The position error of the obstacle measured by the SVM wide-angle camera also increases as the distance between the camera and the obstacle increases. In one embodiment, the memory parking assist device can reflect the error caused by the obstacle distance by adjusting one or more values of Lat margin 、Long margin 、F.Lat and F.Long, so as to perform vehicle margin design.

[0120] Figure 9AIllustrates the process by which a memory parking assist device determines whether a vehicle 900 collides with a nearby obstacle when navigating along a memory path 902a according to an embodiment of the present disclosure. Considering the perception error of the wide-angle camera, the vehicle border is designed to increase with distance. When predicting whether the vehicle 900 will collide with a distant obstacle, the memory parking assist device designs a wider vehicle border to address the larger obstacle perception error. However, when predicting whether the vehicle 900 will collide with a nearby obstacle, the memory parking assist device designs the vehicle border width d to be narrower, as Figure 9A shown. By setting a smaller value for Lat margin , the vehicle border width d can be designed to be narrower.

[0121] Figure 9B Illustrates the process by which a memory parking assist device generates an avoidance path 902b according to another embodiment of the present disclosure. When generating the avoidance path, the memory parking assist device designs a wider vehicle border d, as Figure 9B shown. By setting a larger value for Lat margin , the vehicle border width d can be designed to be wider. By doing so, a safer avoidance path 902b can be generated.

[0122] When predicting whether a collision will occur when navigating along the updated memory path after updating the memory path to an avoidance path, the memory parking assist device designs the vehicle border to be smaller than the vehicle border considered when generating the avoidance path. As a result, the memory parking assist device can prevent frequent updates of the memory path.

[0123] In the field of autonomous navigation, positioning refers to the technology of determining the vehicle position, speed, and path.

[0124] For autonomous navigation, positioning uses the features of the parking lot environment. The autonomous navigation vehicle uses the images obtained by the SVM wide-angle camera to calculate the features. Since the SVM wide-angle camera uses a wide-angle lens, the position of the features recognized by the autonomous navigation vehicle may not be consistent with the actual position. Therefore, errors may occur when using the feature points of the parking lot environment for positioning.

[0125] Figure 10A and Figure 10B Illustrates the operation of a memory parking assist device according to an embodiment of the present disclosure when the vehicle position contains an error.

[0126] The memory parking assist device follows the memory path using the vehicle position obtained through positioning. If the current position of the vehicle contains an error, the vehicle may follow the wrong memory path. When the vehicle navigates along a path with a high curvature, even if the initial following error is small, as the positioning error accumulates, the navigation path of the vehicle will deviate from the memory path. Figure 10AThis shows a situation where the actual navigation path 1001a of the vehicle deviates from the memory path 1002a due to a positioning error occurring at the current position 1000a of the vehicle.

[0127] When the vehicle navigates along a path with a high curvature and follows a memory path containing a positioning error, there may be a problem of the vehicle colliding with an obstacle.

[0128] In order to respond to the external environment in real time, an autonomous navigation vehicle periodically updates the characteristics of the parking lot environment during navigation. Since the positioning is performed again if the characteristics are updated, the positioning accuracy continuously improves.

[0129] As Figure 10B shown, the memory parking assist device must update the current position of the vehicle each time positioning is performed, and generate a convergence path 1002b using the updated current position 1000b so that the vehicle can safely converge to the memory path 1001b. The memory parking assist device must converge to the memory path 1001b based on the current position 1000b, and simultaneously generate an avoidance path and a convergence path 1002b where the vehicle will not collide with an obstacle.

[0130] Each element of the device or method according to the present invention can be implemented by hardware or software or a combination of hardware and software. The functions of each element can be implemented by software, and a microprocessor can be implemented to execute the software functions corresponding to each element.

[0131] Various embodiments of the systems and techniques described herein can be implemented by digital electronic circuits, integrated circuits, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), computer hardware, firmware, software, and / or combinations thereof. Various embodiments can include implementations using one or more computer programs executable on a programmable system. The programmable system includes at least one programmable processor, at least one input device, and at least one output device. The at least one programmable processor can be a dedicated processor or a general-purpose processor, and is coupled to receive data and instructions from a storage system and to send data and instructions to the storage system. A computer program (also referred to as a program, software, software application, or code) includes instructions for the programmable processor and is stored in a "computer-readable recording medium".

[0132] A computer-readable recording medium may include all types of storage devices that can store computer-readable data. The computer-readable recording medium may be a non-volatile or non-transitory medium, such as a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk, or an optical data storage device. In addition, the computer-readable recording medium may further include a temporary medium such as a data transmission medium. Further, the computer-readable recording medium may be distributed over computer systems connected through a network, and the computer-readable program code may be stored and executed in a distributed manner.

[0133] Although the operations are illustrated as being executed in sequence in the flowcharts / timelines of the present specification, this is merely a description of the technical concept of an embodiment of the present disclosure. In other words, those of ordinary skill in the art to which an embodiment of the present disclosure pertains can understand that various modifications and changes can be made without departing from the essential features of an embodiment of the present disclosure, that is, the order shown in the flowcharts / timelines can be changed, and one or more of the operations can be executed in parallel. Therefore, the flowcharts / timelines are not limited to the chronological order.

[0134] Although an embodiment of the present disclosure has been described for illustrative purposes, those of ordinary skill in the art should understand that various modifications, additions, and substitutions are possible without departing from the spirit and scope of the claimed invention. Therefore, an embodiment of the present disclosure has been described for the sake of brevity and clarity. The scope of the technical concept of this embodiment is not limited by the illustrations. Therefore, those of ordinary skill in the art should understand that the scope of the claimed invention should not be limited by the embodiments explicitly described above, but by the claims and their equivalents.

Claims

1. A method for generating a parking path, the method comprising: Obtaining images of the vehicle's surroundings through a camera; Using the image, obtain the position of the vehicle and obstacles around the vehicle; activating a memory parking assist (MPA) in response to determining that the vehicle is approaching a memory path; predicting whether the vehicle will collide with any of the obstacles if the vehicle follows the memory path; In response to determining that the vehicle is predicted to collide with the obstacle, determining whether the obstacle is avoidable; generating an avoidance path for the vehicle to avoid the obstacle; as well as A convergence path is generated for the vehicle to converge to the memory path.

2. The method according to claim 1, further comprising: updating the memory path to the avoidance path and the convergence path; as well as By using the updated memory path, it is repeatedly predicted whether the vehicle collides with any of the obstacles and it is repeatedly determined whether the obstacle is avoidable.

3. The method according to claim 1, wherein: The polygonal vehicle boundary is applied to predict whether the vehicle collides with any of the obstacles, determine whether the obstacles are avoidable, generate the avoidance path, and generate the convergence path.

4. The method according to claim 3, wherein: The area of ​​the vehicle boundary increases or decreases according to the distance between the obstacle and the vehicle.

5. The method according to claim 2, wherein: Repeatedly predicting whether the vehicle collides with any of the obstacles by using the updated memory path and repeatedly determining whether the obstacle is avoidable includes: A vehicle margin that is reduced compared to the vehicle margin before updating the memory path is applied.

6. The method according to claim 1, wherein: The avoidance path and the convergence path are clothoid paths.

7. The method according to claim 1, further comprising: The vehicle's position is periodically updated by using the current environment around the parking lot; determining whether the vehicle follows the memorized path by using the updated position of the vehicle; as well as If the vehicle deviates from the memorized path, a convergence path is generated for the vehicle to converge to the memorized path.

8. A system for generating a parking path, the system comprising: Memory devices; and a processing device, communicatively coupled to the memory device, and configured to: Obtaining images of the vehicle's surroundings through a camera; obtaining positions of the vehicle and obstacles around the vehicle by using the image; activating a memory parking assist (MPA) in response to determining that the vehicle is approaching a memory path; predicting whether the vehicle will collide with any of the obstacles if the vehicle follows the memory path; In response to determining that the vehicle is predicted to collide with the obstacle, determining whether the obstacle is avoidable; generating an avoidance path for the vehicle to avoid the obstacle; as well as A convergence path is generated for the vehicle to converge to the memory path.

9. The system according to claim 8, wherein: The processing device is further configured to: updating the memory path to the avoidance path and the convergence path; and By using the updated memory path, it is repeatedly predicted whether the vehicle collides with any of the obstacles and it is repeatedly determined whether the obstacle is avoidable.

10. The system according to claim 8, wherein: The polygonal vehicle boundary is applied to predict whether the vehicle collides with any of the obstacles, determine whether the obstacles are avoidable, generate the avoidance path, and generate the convergence path.

11. The system according to claim 10, wherein: The area of ​​the vehicle boundary increases or decreases according to the distance between the obstacle and the vehicle.

12. The system according to claim 9, wherein: When repeatedly predicting whether the vehicle collides with any of the obstacles and repeatedly determining whether the obstacle is avoidable by using the updated memory path, a vehicle margin reduced compared to the vehicle margin before updating the memory path is applied.

13. The system according to claim 8, wherein: The avoidance path and the convergence path are clothoid paths.

14. The system according to claim 8, wherein: The processing device is further configured to: Periodically update the vehicle's location by using the current environment around the parking lot; determining whether the vehicle follows the memorized path by using the updated position of the vehicle; as well as If the vehicle deviates from the memorized path, a convergence path is generated for the vehicle to converge to the memorized path.

Images