Parking evaluation method, device, equipment and storage medium

Abstract

The present invention relates to the field of autonomous driving, and discloses a parking evaluation method, device, equipment and storage medium, the method comprising: obtaining first annotation data, wherein the first annotation data is the vertex coordinates of the manually annotated dockable area; calculating the location information of the dockable area based on the vertex coordinates and the map information of the semantic map where the dockable area is located, and performing secondary annotation of the dockable area using the location information as the second annotation data; when there is an autonomous driving vehicle parked in the dockable area, the parking position of the autonomous driving vehicle is evaluated based on the first annotation data and the second annotation data of the dockable area. The method proposes to use semantic maps to assist in manual annotation in the docking station scenario, and only needs to manually annotate the rectangular outline of the allowable docking area to ensure that the docking station evaluation system can obtain sufficient information to complete the evaluation of the automatic parking performance level of the autonomous driving system in the scenario.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving, and in particular to a parking assessment method, device, equipment and storage medium. Background Art

[0002] With the development of artificial intelligence technology, autonomous driving has become a hot topic of concern and research. Automatic parking is an important process of autonomous driving. In the field of autonomous driving, the scene of picking up and dropping off passengers at a stop is an important part of the commercial operation of autonomous driving, and it is also one of the most complex scenes. In order to evaluate the performance of the autonomous driving system in the stop scene, a stable and reliable evaluation system needs to be established. The basis of the evaluation system usually needs to be input through manual annotation. Manual annotation usually requires a lot of manpower. In order to maximize the quality and efficiency of annotation, the annotation content should be simplified as much as possible. However, for the evaluation system, the more comprehensive the input information, the better. For example, in the stop scene, manual annotation is expected to only mark the allowable stop area. However, the stop evaluation system needs to know at least the relative position of the stop area and the station, the driving distance to the station, and the direction before it can be evaluated. This leads to the incomplete evaluation of autonomous driving parking under manual annotation and low evaluation accuracy. If all the above annotation data are manually annotated, the annotation content is complex, the accuracy is difficult to guarantee, and the probability of annotation errors increases. Summary of the invention

[0003] The main purpose of the present invention is to solve the technical problem of low precision and accuracy of existing manual marking parking assessment.

[0004] A first aspect of the present invention provides a parking assessment method, comprising: obtaining first annotation data, wherein the first annotation data is the vertex coordinates of a manually annotated dockable area; calculating the position information of the dockable area based on the vertex coordinates and map information of a semantic map where the dockable area is located, and performing secondary annotation on the dockable area using the position information as second annotation data; when an autonomous driving vehicle is parked in the dockable area, assessing the parking position of the autonomous driving vehicle based on the first annotation data and the second annotation data of the dockable area.

[0005] Optionally, in a first implementation method of the first aspect of the present invention, before calculating the location information of the dockable area based on the vertex coordinates and the map information of the semantic map where the dockable area is located, it also includes: determining a site corresponding to the dockable area in the semantic map, wherein the site is in the same layer as the dockable area in the semantic map; and obtaining map information of the layer in the semantic map.

[0006] Optionally, in a second implementation manner of the first aspect of the present invention, the map information includes a lane centerline, which is composed of multiple road points; the calculating the location information of the dockable area based on the vertex coordinates and the map information of the semantic map where the dockable area is located, and using the location information as the second annotation data to perform secondary annotation on the dockable area includes: calculating the direction of the dockable area based on the vertex coordinates and the road points on the lane centerline, and determining the order of vertices on the dockable area based on the direction of the dockable area; determining a parking mode of the dockable area based on the station and the road points on the lane centerline; calculating the optimal docking position in the dockable area based on the parking mode, the vertex order of the dockable area, and the road points on the lane centerline, and using the direction of the dockable area and the optimal docking position as the location information; and performing secondary annotation on the dockable area using the location information as the second annotation data.

[0007] Optionally, in a third implementation method of the first aspect of the present invention, the direction of the dockable area is calculated based on the vertex coordinates and the road points on the lane centerline, and the order of vertices on the dockable area is determined based on the direction of the dockable area: the area center of gravity of the dockable area is calculated based on the vertex coordinates; the road point on the lane centerline closest to the area center of gravity is queried, and the direction of the road point closest to the area center of gravity is used as the direction of the dockable area; the area center of gravity and each vertex in the dockable area are connected to obtain a first vector corresponding to each vertex; the angle between the first vector and the direction of the dockable area is calculated, and each vertex is sorted based on the size of the angle corresponding to each vertex to obtain the order of vertices on the dockable area.

[0008] Optionally, in a fourth implementation manner of the first aspect of the present invention, the parking mode includes same-lane parking and non-same-lane parking; the parking mode includes the road points based on the site and the lane centerline, and determining the parking mode of the dockable area includes: based on the semantic map, querying the road point closest to the site; using a deep search algorithm to determine whether there is a connectable path between the road point closest to the site and the road point closest to the center of gravity of the area; if so, the parking mode is same-lane parking; if not, the parking mode is non-same-lane parking.

[0009] Optionally, in a fifth implementation manner of the first aspect of the present invention, the calculation of the optimal parking position in the dockable area based on the parking mode, the vertex sequence of the dockable area and the road points on the centerline of the lane includes: if the parking mode is non-lane parking, taking the preset half-width of the autonomous driving vehicle as the optimal position perpendicular to the lane; if the parking mode is same-lane parking, adding 0.3 meters to the half-width of the autonomous driving vehicle to obtain the optimal position perpendicular to the lane; determining the midpoint between the second vertex and the third vertex in the vertex sequence, calculating the distance value between the midpoint and the road point closest to the midpoint, and determining the optimal position along the lane of the dockable area based on the station and the distance value; determining the optimal parking position of the dockable area based on the optimal position perpendicular to the lane and the optimal position along the lane.

[0010] Optionally, in a sixth implementation manner of the first aspect of the present invention, the evaluation of the parking position of the autonomous driving vehicle based on the first annotation data and the second annotation data of the dockable area includes: determining the road point closest to the parking position based on the semantic map; calculating the longitudinal distance and lateral distance between the road point closest to the parking position and the optimal parking position; and evaluating the parking position based on the longitudinal distance and the lateral distance.

[0011] A second aspect of the present invention provides a parking assessment device, comprising: an acquisition module, for acquiring first annotation data, wherein the first annotation data is the vertex coordinates of a manually annotated dockable area; a labeling module, for calculating the position information of the dockable area based on the vertex coordinates and map information of a semantic map where the dockable area is located, and performing secondary labeling on the dockable area using the position information as second annotation data; and an assessment module, for assessing the parking position of the autonomous driving vehicle based on the first annotation data and the second annotation data of the dockable area when there is an autonomous driving vehicle parked in the dockable area.

[0012] Optionally, in a first implementation method of the second aspect of the present invention, the parking assessment device also includes a map module, and the map module is specifically used to: determine the site corresponding to the dockable area in the semantic map, wherein the site and the dockable area are in the same layer in the semantic map; and obtain map information of the layer in the semantic map.

[0013] Optionally, in a second implementation manner of the second aspect of the present invention, the map information includes a lane centerline, which is composed of multiple road points; the labeling module specifically includes: a sequence determination unit, used to calculate the direction of the dockable area based on the vertex coordinates and the road points on the lane centerline, and determine the vertex order on the dockable area based on the direction of the dockable area; a mode determination unit, used to determine the parking mode of the dockable area based on the station and the road points on the lane centerline; a position calculation unit, used to calculate the optimal docking position in the dockable area based on the parking mode, the vertex order of the dockable area, and the road points on the lane centerline, and use the direction of the dockable area and the optimal docking position as the position information; an information labeling unit, used to perform secondary labeling on the dockable area using the position information as second labeling data.

[0014] Optionally, in a third implementation of the second aspect of the present invention, the parking mode includes same-lane parking and non-same-lane parking; the order determination unit is specifically used to: calculate the area center of gravity of the dockable area based on the vertex coordinates; query the road point on the lane centerline that is closest to the area center of gravity, and take the direction of the road point closest to the area center of gravity as the direction of the dockable area; connect the area center of gravity and each vertex in the dockable area to obtain a first vector corresponding to each vertex; calculate the angle between the first vector and the direction of the dockable area, and sort the vertices based on the size of the angle corresponding to each vertex to obtain the vertex order on the dockable area.

[0015] Optionally, in a fourth implementation manner of the second aspect of the present invention, the parking mode includes same-lane parking and non-same-lane parking; the mode determination unit is specifically used to: based on the semantic map, query the road point closest to the site; use a deep search algorithm to determine whether there is a connectable path between the road point closest to the site and the road point closest to the center of gravity of the area; if so, the parking mode is same-lane parking; if not, the parking mode is non-same-lane parking.

[0016] Optionally, in a fifth implementation of the second aspect of the present invention, the position calculation unit is specifically used to: if the parking mode is non-lane parking, then the preset half-width of the autonomous driving vehicle is used as the optimal position of the perpendicular lane; if the parking mode is same-lane parking, then the half-width of the autonomous driving vehicle is added with 0.3 meters to obtain the optimal position of the perpendicular lane; determine the midpoint between the second vertex and the third vertex in the vertex sequence, calculate the distance value between the midpoint and the road point closest to the midpoint, and determine the optimal position along the lane of the parkable area based on the station and the distance value; determine the optimal parking position of the parkable area based on the optimal position of the perpendicular lane and the optimal position along the lane.

[0017] Optionally, in a sixth implementation manner of the second aspect of the present invention, the evaluation module is specifically used to: determine the road point closest to the stop position based on the semantic map; calculate the longitudinal distance and lateral distance between the road point closest to the stop position and the optimal stop position; and evaluate the stop position based on the longitudinal distance and the lateral distance.

[0018] A third aspect of the present invention provides a parking assessment device, comprising: a memory and at least one processor, wherein the memory stores instructions, and the memory and the at least one processor are interconnected via a line; the at least one processor calls the instructions in the memory so that the parking assessment device executes the steps of the above-mentioned parking assessment method.

[0019] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the above-mentioned parking assessment method.

[0020] In the technical solution of the present invention, by obtaining first annotation data, wherein the first annotation data is the vertex coordinates of the manually annotated dockable area; based on the vertex coordinates and the map information of the semantic map where the dockable area is located, the location information of the dockable area is calculated, and the location information is used as the second annotation data to perform secondary annotation on the dockable area; when there is an autonomous driving vehicle parked in the dockable area, the parking position of the autonomous driving vehicle is evaluated based on the first annotation data and the second annotation data of the dockable area. This method proposes to use semantic maps to assist in manual annotation in the docking station scenario. It only requires manual annotation of the rectangular outline of the allowable docking area. While ensuring that the annotation task is as lightweight, simple, and highly operable as possible, it can also ensure that the docking station evaluation system can obtain sufficient information to complete the evaluation of the automatic parking performance level of the autonomous driving system in the scenario. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of a first embodiment of a parking assessment method according to an embodiment of the present invention;

[0022] Figure 2 is a schematic diagram of a second embodiment of a parking assessment method according to an embodiment of the present invention;

[0023] Figure 3 A schematic diagram of an embodiment of a parking assessment device in an embodiment of the present invention;

[0024] Figure 4 It is a schematic diagram of another embodiment of the parking assessment device in an embodiment of the present invention;

[0025] Figure 5 A schematic diagram of an embodiment of a parking assessment device in an embodiment of the present invention;

[0026] Figure 6 A schematic diagram of an embodiment of a parking assessment device in an embodiment of the present invention;

[0027] Figure 7 A schematic diagram of a positional relationship between a station and a dockable area in an embodiment of the present invention;

[0028] Figure 8 Schematic diagram of a second positional relationship between a station and a dockable area in an embodiment of the present invention;

[0029] Fig. 9 This is a schematic diagram of the third positional relationship between a station and a dockable area in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The embodiments of the present application provide a parking assessment method, device, equipment and storage medium, which are used to solve the technical problem of low precision and accuracy of existing manual annotation parking assessment.

[0031] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1The first embodiment of the parking assessment method in the embodiment of the present invention includes:

[0033] 101. Acquire first annotation data, wherein the first annotation data is vertex coordinates of a manually-annotated dockable area;

[0034] It is understandable that the execution subject of the present invention may be a parking assessment device, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0035] In one embodiment of the present invention, the first annotation data are the vertex coordinates of a manually annotated dockable area, wherein the dockable area is an area artificially demarcated near a stop for parking autonomous driving vehicles in a scenario of picking up and dropping off passengers at a stop. Generally, the dockable area is a rectangular area. In addition, it can also be other polygonal figures, which is not limited in this embodiment.

[0036] In one embodiment of the present invention, taking the dockable area as a rectangle as an example, Figure 6 As shown, in order to simplify the manual labeling work, only the coordinates of the four vertices of the rectangular area need to be input, and there is no need to manually sort the coordinates of the four vertices.

[0037] 102. Based on the vertex coordinates and the map information of the semantic map where the dockable area is located, calculate the location information of the dockable area, and use the location information as the second annotation data to perform secondary annotation on the dockable area; in practical applications, the semantic map is used as a map for driving an autonomous vehicle, and contains richer semantic information, geometric relationships, and traffic rules and other information than ordinary maps (Raw Map). The process of establishing a semantic map is as follows: when a vehicle enters and exits a manually demarcated parking area, the parking area data is collected through sensors on the vehicle to form a high-precision semantic map. Among them, the parking lot data includes but is not limited to entrance gates, ramps, intersection lines between ramps and walls, parking spaces, arrows, lane lines, pillars, beams, fire hydrants, parking stations and other stable objects in the parking area. In this embodiment, the map information of the semantic map mainly used is to determine the center line of the lane based on the station corresponding to the dockable area.

[0038] In one embodiment of the present invention, based on the lane centerline of the semantic map, the order of the vertices on the dockable area can be determined, and then the direction of the dockable area in the semantic map can be determined. It can be understood that in actual applications, traffic roads usually include multiple lanes in parallel, and there may be a situation where two lanes are in opposite directions, so it is necessary to determine the direction of the lane in the semantic map.

[0039] In one embodiment of the present invention, the direction of the dockable area is determined to determine the parking mode of the dockable area, and the optimal parking position of the dockable area is determined based on the parking mode and the lane centerline information, and the direction of the dockable area and the optimal parking position are used as annotation data to annotate the dockable area. By annotating the dockable area, the dockable area corresponds to the calculated position information. When there are multiple dockable areas and the autonomous driving vehicle automatically parks in one of the dockable areas, the corresponding position information can be quickly used according to the annotation, thereby improving the efficiency of the automatic parking evaluation.

[0040] 103. When there is an autonomous driving vehicle parked in a dockable area, the parking position of the autonomous driving vehicle is evaluated based on the first labeled data and the second labeled data of the dockable area.

[0041] In one embodiment of the present invention, when an autonomous driving vehicle performs automatic parking, a parking area corresponding to the parking of the autonomous driving vehicle is determined, and position information marked on the determined parking area, such as the direction of the parking area, is obtained to determine whether the parking direction of the autonomous driving vehicle is correct, or the optimal parking position. The automatic parking is evaluated by determining the distance between the position of the autonomous driving vehicle after automatic parking and the optimal parking position. The further the position of the automatic parking deviates from the optimal parking position, the lower the evaluation score, and vice versa.

[0042] In this embodiment, by obtaining first annotation data, wherein the first annotation data is the vertex coordinates of the manually annotated dockable area; based on the vertex coordinates and the map information of the semantic map where the dockable area is located, the location information of the dockable area is calculated, and the location information is used as the second annotation data to re-annotate the dockable area; when there is an autonomous driving vehicle parked in the dockable area, the parking position of the autonomous driving vehicle is evaluated based on the first annotation data and the second annotation data of the dockable area. This method proposes to use semantic maps to assist in manual annotation in the docking station scenario. It only requires manual annotation of the rectangular outline of the allowable docking area. While ensuring that the annotation task is as lightweight, simple, and highly operable as possible, it can also ensure that the docking station evaluation system can obtain sufficient information to complete the evaluation of the automatic parking performance level of the autonomous driving system in the scenario.

[0043] See also Figure 2 , a second embodiment of the parking assessment method in the embodiment of the present invention includes:

[0044] 201. Acquire first annotation data, wherein the first annotation data is vertex coordinates of a manually-annotated dockable area;

[0045] 202. Determine a site corresponding to the dockable area in the semantic map, wherein the site and the dockable area are in the same layer in the semantic map;

[0046] It can be understood that the dockable area is used for the self-driving vehicle to automatically park and pick up customers at the docking station, so the dockable area corresponds to the station, and the station and the dockable area are in the semantic map.

[0047] 203. Obtain map information of a layer in a semantic map;

[0048] In one embodiment of the present invention, the map information includes a lane centerline, and the lane centerline is composed of a plurality of road points.

[0049] 204. Calculate the direction of the dockable area based on the vertex coordinates and the road point on the lane centerline, and determine the order of vertices on the dockable area based on the direction of the dockable area;

[0050] In one embodiment of the present invention, the direction of the dockable area is calculated based on the vertex coordinates and the road points on the lane centerline, and the order of vertices on the dockable area is determined based on the direction of the dockable area: the area center of gravity of the dockable area is calculated based on the vertex coordinates; the road point on the lane centerline closest to the area center of gravity is queried, and the direction of the road point closest to the area center of gravity is used as the direction of the dockable area; the area center of gravity and each vertex in the dockable area are connected to obtain a first vector corresponding to each vertex; the angle between the first vector and the direction of the dockable area is calculated, and each vertex is sorted based on the size of the angle corresponding to each vertex to obtain the order of vertices on the dockable area.

[0051] It can be understood that, taking the dockable area as a rectangle as an example, the formula for calculating the center of gravity of the rectangular area is: rect_gc = (p1+p2+p3+p4) / 4, where p1...p4 represents the coordinates of the four vertices of the rectangle, and the coordinate format is {x, y}. In addition, the dockable area can also be set with polygons of different shapes. The center of gravity of the area corresponding to polygons of different shapes is different, and the calculation formula is also different, which is not limited by the present invention. After calculating the center of gravity of the area, the semantic map is used to determine the road point on the center line of the lane that is closest to the road point. The road point on the center line of the lane contains semantic information, including the speed limit, direction, front and rear adjacent road points, road points on the center lines of the adjacent left and right lanes, and the distance from the point to the left and right boundary lines of the lane. The direction of the dockable area can be determined based on the direction of the road point.

[0052] It is understandable that in order to simplify the workload of manual labeling, only four vertices of the dockable area need to be labeled, and there is no requirement for the order of the labeled vertices, and the order of the vertices has no regularity.

[0053] It can be understood that in order to determine the optimal docking position of the dockable area later, it is necessary to determine the order of the four vertices of the rectangular area. This is achieved by connecting the center of gravity of the rectangular area and the four vertices of the dockable area to obtain four unit vectors, denoted as v i , get the direction vector in the direction of the dockable area, recorded as heading_vector, and calculate v i The angle between the heading_vector and the heading_vector is denoted as radian_diff, and the calculation formula is as follows: radian_diff = arccos(v i .dot(heading_vector)), determine the sign of the angle calculated above, and ensure that the vector heading_vector to the vector v i When it is clockwise, the sign is positive, and when it is counterclockwise, the sign is negative. Let the sign be sign, and we get sign = sign (heading_vector × v i ), and sort the vertices from large to small according to the result of radian_diff*sign.

[0054] 205. Determine the parking mode of the parking area based on the road points on the station and the lane centerline;

[0055] In one embodiment of the present invention, the parking mode includes same-lane parking and non-same-lane parking; the parking mode includes the road point based on the site and the lane centerline, and determining the parking mode of the dockable area includes: based on the semantic map, querying the road point closest to the site; using a deep search algorithm to determine whether there is a connectable path between the road point closest to the site and the road point closest to the center of gravity of the area; if so, the parking mode is same-lane parking; if not, the parking mode is non-same-lane parking.

[0056] It is understandable that the dockable area may not be on the same lane as the station. The identification method is as follows: query the road point closest to the station through the semantic map, and the road point closest to the center of gravity of the dockable area obtained when the direction of the dockable area was previously identified. The road point contains the information of the next road point in the forward direction. Use the depth search (DFS) algorithm to find a path that can be connected between the two road points. If there is such a path, it means that the two road points are in the same lane, otherwise, they are not in the same lane. For the special scenario of the stop station, the maximum search depth of DFS is limited to 200 meters.

[0057] It is understandable that in order to ensure the rationality of the manually marked docking area, when the calculated result is not in the same lane, the station road point is set to the road point of the adjacent left lane. Repeat the step of calculating whether there is a path that can be connected between the two road points. If all the left lanes of the station are traversed and no lane is found that is shared with the docking area, the manually marked area is determined to be unreasonable.

[0058] 206. Calculate an optimal parking position in the parking area based on the parking mode, the vertex sequence of the parking area, and the road points on the center line of the lane, and use the direction of the parking area and the optimal parking position as position information;

[0059] In one embodiment of the present invention, the calculation of the optimal parking position in the dockable area based on the parking mode, the vertex sequence of the dockable area and the road point on the center line of the lane includes: if the parking mode is non-lane parking, taking the preset half-width of the autonomous driving vehicle as the optimal position perpendicular to the lane; if the parking mode is same-lane parking, adding 0.3 meters to the half-width of the autonomous driving vehicle to obtain the optimal position perpendicular to the lane; determining the midpoint between the second vertex and the third vertex in the vertex sequence, calculating the distance value between the midpoint and the road point closest to the midpoint, and determining the optimal position along the lane of the dockable area based on the station and the distance value; determining the optimal parking position of the dockable area based on the optimal position perpendicular to the lane and the optimal position along the lane.

[0060] It can be understood that the best position of the dockable area is determined by calculating the best position in the direction perpendicular to the lane and the best position along the lane. The best position in the direction perpendicular to the lane is obtained by the distance to the edge of the current lane. The best position in the direction perpendicular to the lane is recorded as best_lat_dist. For the same-lane mode, best_lat_dist=0.3m+ego_half_width, where ego_half_width represents half the width of the autonomous driving vehicle. For the non-same-lane mode, best_lat_dist=ego_half_width. The best position along the lane direction is recorded as best_lon_dist, and the lane distance to the station is given by the driving distance from the optimal position to the station. The steps are as follows: a. Calculate the front midpoint of the standardized rectangular area, which is the second and third vertices in the rectangular area, find the road point front_middle_wp closest to the point, and calculate the lane distance fm_lon_dist to the station. b. Calculate the front midpoint of the standardized rectangular area, find the road point rear_middle_wp closest to the point, and calculate the lane distance rm_lon_dist to the station. Using the results in steps a and b, based on the positions of three different stations and the dockable area, as well as the geometric relationship of the relative positions, the lane distance from the optimal docking position to the station is given, where the positional relationship between the station and the dockable area is as follows: Figure 7 , Figure 8 and Fig. 9 .

[0061] It can be understood that the distance between two road points is obtained by using a depth search (DFS) algorithm to find a path connecting the two road points and calculating the length of the path, and the sign of the distance is specified, which is positive along the lane direction and negative against the lane direction.

[0062] 207. Use the position information as second annotation data to perform secondary annotation on the dockable area;

[0063] 208. Based on the semantic map, determine the road point closest to the stop location;

[0064] 209. Calculate the longitudinal distance and the lateral distance between the road point closest to the stop position and the optimal stop position;

[0065] 210. The docking position is evaluated based on the longitudinal distance and the lateral distance.

[0066] In one embodiment of the present invention, by finding the road point wp_s closest to the stop position and calculating the lane distance from wp_s to the optimal stop point best_wp, the longitudinal distance to the optimal stop position is obtained. For the lateral distance from the position to be evaluated to the optimal stop position, by calculating the distance lat_dist from the position to be evaluated to the right edge of the lane, the lateral distance to the optimal stop position = lat_dist-best_lat_dist is obtained.

[0067] Based on the previous embodiment, this embodiment describes in detail a process of calculating the location information of the dockable area based on the vertex coordinates and the map information of the semantic map where the dockable area is located, and re-labeling the dockable area using the location information as the second labeling data; when there is an autonomous driving vehicle parked in the dockable area, evaluating the parking position of the autonomous driving vehicle based on the first labeling data and the second labeling data of the dockable area. This method proposes to use semantic maps to assist manual labeling in the docking station scenario. Only the rectangular outline of the allowable docking area needs to be manually labeled. While ensuring that the labeling task is as lightweight, simple, and highly operational as possible, it can also ensure that the docking station evaluation system can obtain sufficient information to complete the evaluation of the automatic parking performance level of the autonomous driving system in the scenario.

[0068] The above describes the parking assessment method in the embodiment of the present invention. The following describes the parking assessment device in the embodiment of the present invention. Figure 3 In one embodiment of the present invention, a parking assessment device includes:

[0069] An acquisition module 301 acquires first annotation data, wherein the first annotation data is the vertex coordinates of the dockable area that are manually annotated;

[0070] A labeling module 302, configured to calculate the position information of the dockable area based on the vertex coordinates and the map information of the semantic map where the dockable area is located, and perform secondary labeling on the dockable area by using the position information as second labeling data;

[0071] The evaluation module 303 is used to evaluate the parking position of the autonomous driving vehicle based on the first labeled data and the second labeled data of the dockable area when there is an autonomous driving vehicle docked in the dockable area.

[0072] In an embodiment of the present invention, the parking assessment device runs the above parking assessment method, and the parking assessment device obtains first annotation data, wherein the first annotation data is the vertex coordinates of the manually annotated dockable area; based on the vertex coordinates and the map information of the semantic map where the dockable area is located, calculates the position information of the dockable area, and uses the position information as the second annotation data to perform secondary annotation on the dockable area; when there is an autonomous driving vehicle parked in the dockable area, the parking position of the autonomous driving vehicle is evaluated based on the first annotation data and the second annotation data of the dockable area. This method proposes to use semantic maps to assist in manual annotation in the docking station scenario, and only needs to manually annotate the rectangular outline of the allowable docking area. While ensuring that the annotation task is as lightweight, simple, and highly operable as possible, it can also ensure that the docking station evaluation system can obtain sufficient information to complete the evaluation of the automatic parking performance level of the autonomous driving system in the scenario.

[0073] See also Figure 4 The second embodiment of the parking assessment device in the embodiment of the present invention includes:

[0074] An acquisition module 301 acquires first annotation data, wherein the first annotation data is the vertex coordinates of the dockable area that are manually annotated;

[0075] A labeling module 302, configured to calculate the position information of the dockable area based on the vertex coordinates and the map information of the semantic map where the dockable area is located, and perform secondary labeling on the dockable area by using the position information as second labeling data;

[0076] The evaluation module 303 is used to evaluate the parking position of the autonomous driving vehicle based on the first labeled data and the second labeled data of the dockable area when there is an autonomous driving vehicle docked in the dockable area.

[0077] In this embodiment, the parking assessment device also includes a map module 304, and the map module 304 is specifically used to: determine the site corresponding to the dockable area in the semantic map, wherein the site and the dockable area are in the same layer in the semantic map; and obtain map information of the layer in the semantic map.

[0078] In this embodiment, the map information includes a lane centerline, which is composed of multiple road points; the annotation module 302 specifically includes: a sequence determination unit 3021, which is used to calculate the direction of the dockable area based on the vertex coordinates and the road points on the lane centerline, and determine the vertex order on the dockable area based on the direction of the dockable area; a mode determination unit 3022, which is used to determine the parking mode of the dockable area based on the station and the road points on the lane centerline; a position calculation unit 3023, which is used to calculate the best docking position in the dockable area based on the parking mode, the vertex order of the dockable area and the road points on the lane centerline, and use the direction of the dockable area and the best docking position as the position information; an information annotation unit 3024, which is used to perform secondary annotation on the dockable area using the position information as second annotation data.

[0079] In this embodiment, the parking modes include parking in the same lane and parking in different lanes; the order determination unit 3021 is specifically used to: calculate the area center of gravity of the dockable area based on the vertex coordinates; query the road point on the lane centerline that is closest to the area center of gravity, and take the direction of the road point closest to the area center of gravity as the direction of the dockable area; connect the area center of gravity and each vertex in the dockable area to obtain a first vector corresponding to each vertex; calculate the angle between the first vector and the direction of the dockable area, and sort the vertices based on the size of the angle corresponding to each vertex to obtain the vertex order on the dockable area.

[0080] In this embodiment, the parking mode includes same-lane parking and non-same-lane parking; the mode determination unit 3022 is specifically used to: based on the semantic map, query the road point closest to the site; use a deep search algorithm to determine whether there is a connectable path between the road point closest to the site and the road point closest to the center of gravity of the area; if so, the parking mode is same-lane parking; if not, the parking mode is non-same-lane parking.

[0081] In this embodiment, the position calculation unit 3023 is specifically used to: if the parking mode is non-lane parking, then the preset half-width of the autonomous driving vehicle is used as the optimal position of the perpendicular lane; if the parking mode is same-lane parking, then the half-width of the autonomous driving vehicle is added with 0.3 meters to obtain the optimal position of the perpendicular lane; determine the midpoint between the second vertex and the third vertex in the vertex sequence, calculate the distance value between the midpoint and the road point closest to the midpoint, and determine the optimal position along the lane of the parkable area based on the station and the distance value; determine the optimal parking position of the parkable area based on the optimal position of the perpendicular lane and the optimal position along the lane.

[0082] In this embodiment, the evaluation module 303 is specifically used to: determine the road point closest to the stop position based on the semantic map; calculate the longitudinal distance and lateral distance between the road point closest to the stop position and the optimal stop position; and evaluate the stop position based on the longitudinal distance and the lateral distance.

[0083] In this implementation, the specific functions of each module of the parking assessment device and the unit composition of some modules are described in detail. Through the modules and units of the device, semantic maps are used to assist manual labeling in the docking station scenario. It is only necessary to manually label the rectangular outline of the allowable docking area. While ensuring that the labeling task is as lightweight, simple, and easy to operate as possible, it can also ensure that the docking station evaluation system can obtain sufficient information to complete the evaluation of the automatic parking performance level of the autonomous driving system in the scenario.

[0084] above Figure 3 and Figure 4 The parking assessment device in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The parking assessment device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0085] Figure 5 is a schematic diagram of the structure of a parking assessment device provided by an embodiment of the present invention. The parking assessment device 500 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 510 (for example, one or more processors) and a memory 520, and one or more storage media 530 (for example, one or more mass storage devices) storing application programs 533 or data 532. Among them, the memory 520 and the storage medium 530 may be temporary storage or permanent storage. The program stored in the storage medium 530 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the parking assessment device 500. Furthermore, the processor 510 may be configured to communicate with the storage medium 530, and execute a series of instruction operations in the storage medium 530 on the parking assessment device 500 to implement the steps of the above-mentioned parking assessment method.

[0086] The parking assessment device 500 may also include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input and output interfaces 560, and / or one or more operating systems 531, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 5The parking assessment device structure shown does not constitute a limitation on the parking assessment device provided in the present application, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0087] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the parking assessment method.

[0088] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0089] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0090] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A parking assessment method, characterized in that: The parking assessment method comprises: Acquire first annotation data, wherein the first annotation data is vertex coordinates of the dockable area that are manually annotated; Calculating the direction of the dockable area based on the vertex coordinates and the road points on the lane centerline in the map information of the semantic map where the dockable area is located, and determining the order of vertices on the dockable area based on the direction of the dockable area; wherein the map information includes the lane centerline, and the lane centerline is composed of a plurality of road points; Determining a parking mode of the parking area based on the station and the road point on the center line of the lane; Calculating an optimal parking position in the dockable area based on the parking mode, the vertex sequence of the dockable area, and the road points on the lane centerline, and using the direction of the dockable area and the optimal parking position as position information; Using the position information as second annotation data to perform secondary annotation on the dockable area; When there is an autonomous driving vehicle parked in the dockable area, the parking position of the autonomous driving vehicle is evaluated based on the first labeled data and the second labeled data of the dockable area.

2. The parking assessment method according to claim 1, characterized in that: Before calculating the position information of the dockable area based on the vertex coordinates and the map information of the semantic map where the dockable area is located, the method further includes: Determine a site corresponding to the dockable area in the semantic map, wherein the site and the dockable area are in the same layer in the semantic map; Obtain map information of the layer in the semantic map.

3. The parking assessment method according to claim 1, characterized in that: Based on the vertex coordinates and the road point on the lane centerline, the direction of the dockable area is calculated, and based on the direction of the dockable area, the order of vertices on the dockable area is determined: Calculate the center of gravity of the dockable area based on the vertex coordinates; Querying the road point on the lane centerline that is closest to the center of gravity of the area, and taking the direction of the road point that is closest to the center of gravity of the area as the direction of the dockable area; Connecting the center of gravity of the region and each vertex in the dockable region to obtain a first vector corresponding to each vertex; The angle between the first vector and the direction of the dockable area is calculated, and the vertices are sorted based on the size of the angle corresponding to each vertex to obtain the order of vertices on the dockable area.

4. The parking assessment method according to claim 3, characterized in that: The parking modes include parking in the same lane and parking in different lanes; The parking mode includes determining the parking area based on the station and the road point on the lane center line, and the parking mode includes: Based on the semantic map, query the road point closest to the site; Using a deep search algorithm, determining whether there is a connectable path between the road point closest to the site and the road point closest to the center of gravity of the area; If so, the parking mode is parking in the same lane; If not present, the parking mode is non-lane parking.

5. The parking assessment method according to claim 4, characterized in that: The calculating the optimal parking position in the dockable area based on the parking mode, the vertex sequence of the dockable area and the road points on the lane centerline comprises: If the parking mode is non-lane parking, half the width of the preset autonomous driving vehicle is used as the optimal position of the perpendicular lane; If the parking mode is parking in the same lane, then half the width of the autonomous driving vehicle plus 0.3 meters is added to obtain the optimal position of the vertical lane; determining a midpoint between a second vertex and a third vertex in the vertex sequence, calculating a distance value between the midpoint and a road point closest to the midpoint, and determining an optimal position of the dockable area along a lane based on the station and the distance value; An optimal parking position of the dockable area is determined based on the optimal position perpendicular to the lane and the optimal position along the lane.

6. The parking assessment method according to any one of claims 1 to 5, characterized in that: The evaluating the parking position of the autonomous driving vehicle based on the first labeled data and the second labeled data of the dockable area includes: Based on the semantic map, determining the road point closest to the parking position; Calculating the longitudinal distance and the lateral distance between the road point closest to the stop position and the optimal stop position; The parking position is evaluated based on the longitudinal distance and the lateral distance.

7. A parking assessment device, characterized in that: The parking assessment device comprises: An acquisition module is used to acquire first annotation data, wherein the first annotation data is the vertex coordinates of the dockable area that are manually annotated; a labeling module, configured to calculate the direction of the dockable area based on the vertex coordinates and the road points on the lane centerline in the map information of the semantic map where the dockable area is located, and determine the order of vertices on the dockable area based on the direction of the dockable area; wherein the map information includes the lane centerline, and the lane centerline is composed of a plurality of road points; determine the parking mode of the dockable area based on the station and the road points on the lane centerline; calculate the best docking position in the dockable area based on the parking mode, the vertex order of the dockable area and the road points on the lane centerline, and use the direction of the dockable area and the best docking position as position information; and perform secondary labeling on the dockable area using the position information as second labeling data; An evaluation module is used to evaluate the parking position of the autonomous driving vehicle based on the first labeled data and the second labeled data of the dockable area when there is an autonomous driving vehicle docked in the dockable area.

8. A parking assessment device, characterized in that: The parking assessment device comprises: a memory and at least one processor, the memory stores instructions, and the memory and the at least one processor are interconnected via a line; The at least one processor calls the instructions in the memory to enable the parking assessment device to perform the steps of the parking assessment method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the parking assessment method according to any one of claims 1 to 6 are implemented.

Images