Automatic parking scoring method, device, equipment and storage medium

Abstract

The present invention relates to the field of autonomous driving and discloses an automatic parking scoring method, apparatus, device, and storage medium. The method comprises: obtaining scene data collected in an automatic parking scenario and marking the vehicle's dockable area in the scene data; inputting the scene data into a preset autonomous driving system to perform parking simulation, obtaining at least one simulated docking position output by the autonomous driving system; and evaluating the simulated docking position based on the scene data and the dockable area to obtain a scoring result for the autonomous driving system's automatic parking. This method, by evaluating the simulated docking position output by the autonomous driving system during the simulation process and comprehensively evaluating the simulated docking position, can more comprehensively evaluate the autonomous driving system's performance in a docking station scenario. This method can specifically help the autonomous driving system to rapidly iterate in a docking station scenario and optimize the autonomous driving system algorithm.

Description

Technical Field

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

[0002] With the continuous advancement of autonomous driving technology, there is significant room for improvement in automated parking due to complex road conditions and significant environmental factors. Picking up and dropping off passengers at parking stops is a crucial and complex scenario for commercial autonomous driving. Autonomous vehicles must not only select parking locations and plan their routes, but also ensure a reasonable final parking posture. Furthermore, parking stops are also one of the most frequent scenarios for autonomous vehicles to interact with other vehicles and pedestrians. Therefore, developing an autonomous driving system that can reliably and stably handle parking stop scenarios is a significant challenge. Currently, the primary method for evaluating autonomous driving systems for parking stop scenarios is through road testing. Key drawbacks include: long testing cycles to capture valuable scenarios; manual analysis is required to determine the appropriateness of the parking position for each stop; and, when comparing different versions of autonomous driving software through road testing, biased results can arise due to the varying scenarios encountered. Summary of the Invention

[0003] The main purpose of the present invention is to solve the existing technical problem of evaluating whether the choice of docking location of an autonomous driving vehicle in a docking station scenario is reasonable.

[0004] A first aspect of the present invention provides an automatic parking scoring method, the automatic parking scoring method comprising:

[0005] Obtaining scene data collected in an automatic parking scenario and marking the vehicle's dockable area in the scene data; inputting the scene data into a preset automatic driving system to perform parking simulation, and obtaining at least one simulated docking position output by the automatic system; and evaluating the simulated docking position based on the scene data and the dockable area to obtain a scoring result for automatic parking by the automatic driving system.

[0006] Optionally, in a first implementation manner of the first aspect of the present invention, the evaluation of the simulated docking position based on the scene data and the dockable area to obtain a scoring result of the automatic parking of the autonomous driving system includes: determining whether the simulated docking position falls within the dockable area; if it falls within the dockable area, calculating the optimal docking position of the dockable area based on the scene data, and calculating the docking score of the simulated docking position based on the optimal docking position; performing a regional scoring of the dockable area based on the docking score to obtain a regional score; performing an overall evaluation of the simulated docking position based on the regional score to obtain a scoring result of the automatic parking of the autonomous driving system; if it does not fall within the dockable area, setting the scoring result to zero.

[0007] Optionally, in a second implementation of the first aspect of the present invention, the scene data includes the site location and lane centerline of the site in the automatic parking scene, wherein the lane centerline is composed of multiple road points; the calculating the optimal parking position of the dockable area based on the scene data, and calculating the docking score of the simulated docking position based on the optimal parking position includes: calculating the direction of the dockable area based on the vertex coordinates of the dockable area and the road points on the lane centerline, and determining the vertex order on the dockable area based on the direction of the dockable area; determining the parking mode of the dockable area based on the site location and the road points on the lane centerline; calculating the optimal parking 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 calculating the docking score of the simulated docking position based on the optimal parking position and the scene data.

[0008] Optionally, in a third implementation of the first aspect of the present invention, the direction of the dockable area is calculated based on the vertex coordinates of the dockable area 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, including: calculating the area center of gravity of the dockable area based on the vertex coordinates; querying the road point on the road centerline that is closest to the area center of gravity, and taking the direction of the road point that is closest to the area center of gravity as the direction of the dockable area; connecting the area center of gravity and each vertex in the dockable area to obtain a first vector corresponding to each vertex; calculating the angle between the first vector and the direction of the dockable area, and sorting each vertex based on the size of the angle corresponding to each vertex to obtain the order of vertices on the dockable area.

[0009] Optionally, in a fourth implementation of the first aspect of the present invention, the scene data also includes a lane road edge position, a lane curb position, and a stationary obstacle position; the calculation of the docking score of the simulated docking position based on the optimal docking position and the scene data includes: calculating a first score of the simulated docking position based on the distance between the optimal docking position and the simulated docking position; calculating a second score and a third score of the simulated docking position based on the distance between the right side edge of the simulated vehicle corresponding to the simulated docking position and the lane road edge position and the lane curb position respectively; calculating a fourth score of the simulated docking position based on the distance between the stationary obstacle position of the stationary obstacle closest to the simulated docking position and the simulated docking position; obtaining the area level corresponding to the dockable area, and determining the area weight of the dockable area based on the area level; inputting the first score, the second score, the third score, the fourth score, and the area weight into a preset evaluation formula to obtain the docking score of the simulated docking position.

[0010] Optionally, in a fifth implementation manner of the first aspect of the present invention, the area scoring of the dockable area based on the docking score to obtain the area score includes: calculating the effective time of the dockable area, wherein the effective time is all time from the key frame to the end of the simulation, and the key frame is the moment when the dockable area is manually marked; recording the moment when the automatic driving system outputs the simulated docking area as a decision frame, and determining the decision frame with the highest docking score within the effective time; calculating the average value of the docking scores from the decision frame with the highest docking score to the last frame within the effective time; judging whether the docking score of the last frame is zero; if it is zero, dividing the average value by two to obtain the area score; if it is not zero, using the average value as the area score.

[0011] Optionally, in a sixth implementation of the first aspect of the present invention, the overall evaluation of the simulated parking positions based on the regional scores to obtain the scoring result of the automatic parking of the autonomous driving system includes: obtaining the regional scores corresponding to all simulated parking positions in the automatic parking scenario; and using the regional score with the highest score among all the regional scores as the scoring result of the automatic parking of the autonomous driving system.

[0012] A second aspect of the present invention provides an automatic parking scoring device, comprising: a position labeling module for acquiring scene data collected in an automatic parking scenario and labeling the vehicle's parking area in the scene data; a parking simulation module for inputting the scene data into a preset automatic driving system to perform parking simulation, thereby obtaining at least one simulated parking position output by the automatic system; and a scoring module for evaluating the simulated parking position based on the scene data and the parking area to obtain a scoring result for the automatic parking of the automatic driving system.

[0013] Optionally, in a first implementation manner of the second aspect of the present invention, the scoring module is specifically used for: an area judgment unit, used to judge whether the simulated docking position falls within the dockable area; a docking score calculation unit, used to calculate the optimal docking position of the dockable area based on the scene data if it falls within the dockable area, and calculate the docking score of the simulated docking position based on the optimal docking position; an area score calculation unit, used to perform an area score on the dockable area based on the docking score to obtain an area score; a scoring result summary unit, used to perform an overall evaluation of the simulated docking position based on the area score to obtain a scoring result of automatic parking of the autonomous driving system; a non-docking area unit, used to set the scoring result to zero if it does not fall within the dockable area.

[0014] Optionally, in a second implementation of the second aspect of the present invention, the docking score calculation unit is specifically used for: a vertex order determination subunit, used to calculate the direction of the dockable area based on the vertex coordinates of the dockable area 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 parking mode determination subunit, used to determine the parking mode of the dockable area based on the station position and the road points on the lane centerline; an optimal docking position calculation subunit, 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; a docking score calculation subunit, used to calculate the docking score of the simulated docking position based on the optimal docking position and the scene data.

[0015] Optionally, in a third implementation of the second aspect of the present invention, the vertex order determination subunit 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 road 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.

[0016] Optionally, in a fourth implementation of the second aspect of the present invention, the docking score calculation subunit is specifically used to: calculate a first score of the simulated docking position based on the distance between the optimal docking position and the simulated docking position; calculate a second score and a third score of the simulated docking position based on the distance between the right side edge of the simulated vehicle corresponding to the simulated docking position and the lane road edge position and the lane road curb position respectively; calculate a fourth score of the simulated docking position based on the distance between the stationary obstacle position of the stationary obstacle closest to the simulated docking position and the simulated docking position; obtain the area level corresponding to the dockable area, and determine the area weight of the dockable area based on the area level; input the first score, the second score, the third score, the fourth score and the area weight into a preset evaluation formula to obtain the docking score of the simulated docking position.

[0017] Optionally, in a fifth implementation of the second aspect of the present invention, the area score calculation unit is specifically used to: calculate the effective time of the dockable area, wherein the effective time is all time from the key frame to the end of the simulation, and the key frame is the moment when the dockable area is manually marked; record the moment when the automatic driving system outputs the simulated docking area as a decision frame, and determine the decision frame with the highest docking score within the effective time; calculate the average value of the docking scores from the decision frame with the highest docking score to the last frame within the effective time; determine whether the docking score of the last frame is zero; if it is zero, divide the average value by two to obtain the area score; if it is not zero, use the average value as the area score.

[0018] Optionally, in a sixth implementation of the second aspect of the present invention, the scoring result summary unit is specifically used to: obtain the regional scores corresponding to all simulated parking positions in the automatic parking scenario; and use the regional score with the highest score among all the regional scores as the scoring result of automatic parking of the automatic driving system.

[0019] A third aspect of the present invention provides an automatic parking scoring 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 circuit; the at least one processor invokes the instructions in the memory to cause the automatic parking scoring device to execute the steps of the automatic parking scoring method described above.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium having instructions stored therein, which, when executed on a computer, causes the computer to execute the steps of the above-mentioned automatic parking scoring method.

[0021] In the technical solution of the present invention, scene data collected in an automated parking scenario is acquired, and the vehicle's dockable areas are annotated in the scene data. The scene data is then input into a preset automated driving system for parking simulation, resulting in at least one simulated docking position output by the automated system. The simulated docking positions are then evaluated based on the scene data and the dockable areas, resulting in a scoring result for the automated parking performance of the automated driving system. This method, by evaluating the simulated docking positions output by the automated driving system during the simulation process and comprehensively evaluating the simulated docking positions, provides a more comprehensive assessment of the automated driving system's performance in a docking station scenario. This method can specifically assist the automated driving system in rapidly iterating in docking station scenarios and optimizing the automated driving system algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a schematic diagram of a first embodiment of an automatic parking scoring method according to an embodiment of the present invention;

[0023] Figure 2 2 is a schematic diagram of a second embodiment of the automatic parking scoring method according to an embodiment of the present invention;

[0024] Figure 3 Schematic diagram of a third embodiment of the automatic parking scoring method according to an embodiment of the present invention;

[0025] Figure 4 2 is a schematic diagram of a fourth embodiment of an automatic parking scoring method according to an embodiment of the present invention;

[0026] Figure 5 This is a first simulation diagram of the automatic parking scoring method according to an embodiment of the present invention;

[0027] Figure 6 A schematic diagram of an embodiment of an automatic parking scoring device according to an embodiment of the present invention;

[0028] Figure 7 2 is a schematic diagram of another embodiment of the automatic parking scoring device according to an embodiment of the present invention;

[0029] Figure 8 FIG. 1 is a schematic diagram of an embodiment of an automatic parking scoring device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In the technical solution of the present invention, scene data collected in an automated parking scenario is acquired, and the vehicle's dockable areas are annotated in the scene data. The scene data is then input into a preset automated driving system for parking simulation, resulting in at least one simulated docking position output by the automated system. The simulated docking positions are then evaluated based on the scene data and the dockable areas, resulting in a scoring result for the automated parking performance of the automated driving system. This method, by evaluating the simulated docking positions output by the automated driving system during the simulation process and comprehensively evaluating the simulated docking positions, provides a more comprehensive assessment of the automated driving system's performance in a docking station scenario. This method can specifically assist the automated driving system in rapidly iterating in docking station scenarios and optimizing the automated driving system algorithm.

[0031] The terms "first," "second," "third," "fourth," and the like (if any) in the description and claims of the present invention and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable 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 apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatus.

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

[0033] 101, acquiring scene data collected in an automatic parking scenario, and marking the parking area of ​​the vehicle in the scene data;

[0034] In this embodiment, scenario data for simulating automated parking is obtained and manually labeled to identify at least one parking area. Scenario data has data sources, and by classifying these sources, it's easier to select scenario data from different sources for simulation testing.

[0035] Specifically, scene data collection methods include at least the following three methods. First, by collecting complete stop requirements. Stop requirements are based on actual road conditions, client requirements, and policy influences. For example, tidal lanes, docking areas that can accommodate 1-3 vehicles at the same time, and docking areas adjacent to prohibited parking areas. Here, the data source can be divided into the worldsim category. Second, during actual road testing, drivers conducting road tests supplement the existing scene data with features and issues that have not been collected. The supplemented issues are used to supplement the original scene data, thereby obtaining scene data. Here, the data source can be divided into the incident category. Third, during normal operation of an autonomous vehicle, when the vehicle passes through a scene requiring automatic parking, the data collection module automatically records the position information collected by the vehicle's positioning module, the scene information and obstacle information collected by the perception module, and the decision information output by the planning and control module during the entire parking process. The auto cut bag category is obtained after integration.

[0036] 102, inputting the scene data into a preset autonomous driving system to perform parking simulation and obtain at least one simulated parking position;

[0037] In this embodiment, the scenario data is input into the autonomous driving system that needs to be scored for simulation, and the simulated docking position corresponding to the input scenario data is obtained from the output of the autonomous driving system.

[0038] Specifically, the preset autonomous driving system includes multiple different versions, using different kernels or decision algorithms. By scoring the simulated docking position output by at least one autonomous driving system that needs to be scored, the update direction of the autonomous driving system is determined, and the decision algorithm of the autonomous driving system is optimized more conveniently. The decision algorithm is used to determine the corresponding simulated docking position in the scene data.

[0039] 103. Evaluate the simulated parking position based on the scene data and the dockable area to obtain a scoring result of the automatic parking of the autonomous driving system.

[0040] In this embodiment, by evaluating the difference between the simulated parking position and the dockable area, the automatic parking level of the automatic driving system is objectively and systematically scored according to the degree of difference.

[0041] In practical applications, such as Figure 5As shown, the entire image is considered as a scene data, where the dotted box is the marked docking area; EGO is the simulated docking position output by the autonomous driving system; and there are bus stops, buses, lane lines, and roadside lines. In this scenario, the bus appears as an obstacle in the scene data. Therefore, the bus parking area cannot be designated as a docking area. After identifying the bus area, the obstacle position should be removed based on the proximity to the bus stop to output a suitable simulated docking position.

[0042] In this embodiment, scene data collected during an automated parking scenario is acquired and the vehicle's available parking areas are annotated in the scene data. The scene data is then input into a pre-set automated parking system for parking simulation, resulting in at least one simulated parking position output by the automated system. The simulated parking positions are then evaluated based on the scene data and the available parking areas, resulting in a scoring result for the automated parking system. This method evaluates the simulated parking positions output by the automated parking system during the simulation process. By comprehensively evaluating the simulated parking positions and providing a more comprehensive assessment of the automated parking system's performance in a docking station scenario, this method can help the automated parking system achieve rapid iteration in docking station scenarios and optimize the automated parking system algorithm.

[0043] See also Figure 2 The second embodiment of the automatic parking scoring method in the embodiment of the present invention includes:

[0044] 201, obtaining scene data collected in an automatic parking scenario, and marking the parking area of ​​the vehicle in the scene data;

[0045] 202, inputting the scene data into a preset autonomous driving system to perform parking simulation and obtain at least one simulated parking position;

[0046] 203, determining whether the simulated docking position falls within the dockable area;

[0047] In this embodiment, it is determined whether the simulated docking position output by the simulated automatic driving system falls within the marked dockable area.

[0048] Specifically, a coordinate system can be established by using scene data, wherein, since the simulated docking position and the dockable area are both based on the scene data, when establishing the coordinate system of the scene data, the simulated docking position and the dockable area have corresponding coordinates, and the judgment is made by determining whether the coordinates of the two overlap.

[0049] 204, if it falls within the dockable area, calculate the center of gravity of the dockable area based on the vertex coordinates;

[0050] In this embodiment, if the simulated docking position is determined to be within the dockable area, or if the coordinates of the simulated docking area overlap with the coordinates of the dockable area, the center of gravity of the dockable area is calculated using the vertex coordinates. The dockable area is generally marked as a rectangle or trapezoid, and the center of gravity of the dockable area is calculated by obtaining the vertex coordinates of the dockable area.

[0051] Specifically, the center of gravity of the dockable area can be calculated by connecting the coordinates of the diagonal vertices of the rectangular dockable area to obtain two intersecting diagonals. The intersection of the diagonals is the center of gravity of the dockable area. The center of gravity of dockable areas of other shapes is calculated in the same way and will not be further explained here.

[0052] 205, querying the road point on the road centerline closest to the center of gravity of the area, and taking the direction of the road point closest to the center of gravity of the area as the direction of the dockable area;

[0053] In this embodiment, two-way lanes are generally present on the road. By determining the vertical distances between the road center lines corresponding to the lanes in each direction and the center of gravity of the region, the perpendicular lines of the two lanes with respect to the center of gravity of the region are obtained respectively.

[0054] Specifically, by comparing the lengths of the two vertical lines, the lane corresponding to the shorter vertical line is selected, and the direction of the lane is determined as the direction of the simulated vehicle when it stops in the dockable area.

[0055] 206, connecting the center of gravity of the area and each vertex in the dockable area to obtain a first vector corresponding to each vertex;

[0056] In this embodiment, the vertex and the first vector corresponding to the vertex are obtained by connecting the center of gravity of the area and the vertex of the dockable area.

[0057] Specifically, the connection method is not limited to the starting and ending points between the region's centroid and the vertex. On the one hand, the first vector can start from the region's centroid and connect to the vertex; on the other hand, it can start from the vertex and connect to the region's centroid. However, it should be noted that during the entire automatic parking scoring process, only one connection method should be used, and it cannot be mixed.

[0058] 207 , calculating the angle between the first vector and the direction of the dockable area, and sorting the vertices based on the size of the angle corresponding to each vertex to obtain the order of the vertices on the dockable area;

[0059] In this embodiment, the angles between multiple first vectors and the centerline of the lanes to which the docking area belongs are calculated, and the vertices corresponding to the first vectors are sorted by the angles. The sorting can be from largest to smallest or from smallest to largest, and this embodiment does not impose any restrictions on the sorting method.

[0060] 208, determining a parking pattern for the available parking area based on the station location and the roadway point on the lane centerline;

[0061] In this embodiment, the relationship between the planned dockable area and the station is determined by determining the station location and the road point on the center line in the scene data, which is the parking mode.

[0062] Specifically, the parking mode can be divided into parking in the same lane: p_mode = same_lane_parking; and parking in different lanes: p_mode = side_lane_parking. This is used as a scoring criterion in subsequent judgments.

[0063] 209 , calculating an optimal parking position in the dockable area based on the parking pattern, the vertex order of the dockable area, and the road points on the lane centerline;

[0064] 210, calculating a docking score of the simulated docking position based on the optimal docking position and the scenario data;

[0065] In this embodiment, after determining the optimal parking position, the difference between the simulated parking position and the optimal parking position is determined through the scene data and the output simulated parking position. According to a preset scoring table, the corresponding scores between different differences are set in the scoring table, and the corresponding parking scores are obtained after summary calculation.

[0066] 211, performing regional scoring on the dockable area based on the docking score to obtain a regional score;

[0067] In this embodiment, in addition to calculating a parking score based on the difference between the simulated parking position and the optimal parking position, a regional score is also calculated in conjunction with the scene data. For example, the distance between the right edge of the vehicle and the road edge is calculated, where the road edge data is collected based on the scene data.

[0068] 212, perform an overall evaluation of the simulated parking positions based on the regional scores to obtain the scoring results of the automatic parking of the autonomous driving system;

[0069] In this embodiment, the simulated stop positions output by the autonomous driving system to be scored are evaluated overall based on their stop positions using the regional scores, and a final scoring result is obtained.

[0070] Specifically, the overall evaluation includes: score > 0.9, indicating that the stop point selection is very ideal and in line with human expectations; 0.5 < score < 0.9, indicating that the scenario test passed, but the stop point selection still has room for optimization; 0 < scenario score < = 0.5, indicating that the scenario test failed, the autonomous driving system selected a relatively ideal location, but ultimately did not stop in the candidate area; scenario score = 0, in this scenario, the decision of the autonomous driving system is completely inconsistent with expectations.

[0071] 213. If the vehicle does not fall within the dockable area, the score result is set to zero.

[0072] In this embodiment, if the simulated docking position output by the autonomous driving system fails to intersect with the dockable area, it is deemed that the output simulated docking position does not fall within the dockable area, and the autonomous driving system to be scored should be rated zero.

[0073] This embodiment, based on the previous embodiment, describes in detail the process of calculating the center of gravity of the dockable area based on the vertex coordinates; querying the road point on the road centerline that is closest to the center of gravity of the area, and using the direction of the road point closest to the center of gravity of the area as the direction of the dockable area; connecting the center of gravity of the area with each vertex in the dockable area to obtain a first vector corresponding to each vertex; calculating the angle between the first vector and the direction of the dockable area, and sorting the vertices based on the size of the angle corresponding to each vertex to obtain the order of vertices in the dockable area. Compared with traditional methods, this embodiment refines the method of determining the parking mode of the dockable area based on vertices and centerlines, and adds a parking mode evaluation dimension to the automatic parking scoring process.

[0074] See also Figure 3 and 5 The third embodiment of the automatic parking scoring method according to the present invention includes:

[0075] 301, acquiring scene data collected in an automatic parking scenario, and marking the parking area of ​​the vehicle in the scene data;

[0076] In this step, the available parking area can be directly specified by the user or calculated by an algorithm. In the case of the algorithm, the parking reference position executed by the user is obtained, and the platform search algorithm is used to search for a parking platform or a specified area within a certain range of the parking reference position, such as Figure 5 The dotted box in the figure shows the parking reference position, which is the platform sign area in the figure.

[0077] 302, inputting the scene data into a preset autonomous driving system to perform parking simulation and obtain at least one simulated parking position;

[0078] 303, determining whether the simulated docking position falls within the dockable area;

[0079] 304, if it falls within the dockable area, then based on the vertex coordinates of the dockable area and the road point on the lane centerline, calculate the direction of the dockable area, and based on the direction of the dockable area, determine the order of the vertices in the dockable area;

[0080] 305, determining a parking pattern for the available parking area based on the station location and the road points on the lane centerline;

[0081] 306 , calculating an optimal parking position in the dockable area based on the parking pattern, the vertex order of the dockable area, and the road points on the lane centerline;

[0082] Specifically, the optimal docking location is to select the one with the highest score from multiple algorithm output docking locations, and the score is achieved based on the distance between the station location and the algorithm output docking location, such as Figure 5 The EGO marked area in .

[0083] 307, calculating a first score of the simulated docking position based on the distance between the optimal docking position and the simulated docking position;

[0084] In this embodiment, by finding the optimal stop position in the candidate area, the longitudinal distance (distance along the lane direction) from the planned stop position to the optimal stop position is calculated, recorded as lon_dist; according to the formula lon_dist_score = 1–f_lon(lon_dist, p_mode), a first score is given.

[0085] 308 , calculating a second score and a third score of the simulated parking position based on the distances between the right edge of the simulated vehicle and the lane road edge and the lane curb, respectively, corresponding to the simulated parking position;

[0086] In this embodiment, the distance between the right edge of the vehicle and the edge of the road when the vehicle is at the planned stop position is calculated, which is recorded as lat_dist: the second score is given according to the formula lat_dist_score=1-f_lat(lat_dist,p_mode).

[0087] Specifically, the distance between the right edge of the vehicle and the curb when the vehicle is at the planned parking position is calculated, which is recorded as curb_dist; according to the formula curb_score=g_curb(curb_dist), the third score is given.

[0088] 309, calculating a fourth score of the simulated docking position based on the position of the stationary obstacle closest to the simulated docking position and the distance between the simulated docking position;

[0089] In this embodiment, the distance between the edge of the vehicle and the nearest stationary obstacle when the vehicle is at the planned parking position is calculated, which is recorded as obs_dist; and the fourth score is given according to the formula obs_score=g_obs(obs_dist).

[0090] 310, obtaining the area level corresponding to the dockable area, and determining the area weight of the dockable area based on the area level;

[0091] In this embodiment, window_weight represents the weight corresponding to the region level of the candidate region.

[0092] Specifically, for example, the weights for the three area levels of "Excellent," "Sub-Excellent," and "Good" are 1, 0.85, and 0.7, respectively. This weighting is merely an example of one aspect of the scoring method and can be adjusted to meet different needs. By adjusting the weighting parameters, the scoring results for each dimension can accurately reflect the expected demand for stops.

[0093] 311, inputting the first score, the second score, the third score, the fourth score, and the regional weight into a preset evaluation formula to obtain a docking score for the simulated docking location;

[0094] In this embodiment, the final stop score is given by the following formula: Frame_score = (lon_dist_score + lat_dist_score - 1) * window_weight + curb_score + obs_score - 2. By adjusting these parameters, the scoring results for each dimension accurately reflect the expected stop demand. Furthermore, the scoring criteria vary for different stop modes. This is reflected in the differences in the f_lon and f_lat parameters for different modes.

[0095] 312, performing regional scoring on the dockable area based on the docking score to obtain a regional score;

[0096] 313, based on the regional scores, an overall evaluation of the simulated parking positions is performed to obtain the scoring results of the automatic parking of the autonomous driving system;

[0097] 314. If the vehicle does not fall within the dockable area, the score result is set to zero.

[0098] Based on the previous embodiment, this embodiment describes in detail the process of calculating the first score of the simulated docking position based on the distance between the optimal docking position and the simulated docking position; calculating the second score and the third score of the simulated docking position based on the distance between the right edge of the simulated vehicle corresponding to the simulated docking position and the lane road edge position and the lane road curb position; calculating the fourth score of the simulated docking position based on the distance between the static obstacle position of the static obstacle closest to the simulated docking position and the simulated docking position; obtaining the area level corresponding to the dockable area and determining the area weight of the dockable area based on the area level; inputting the first score, the second score, the third score, the fourth score and the area weight into a preset evaluation formula to obtain the docking score of the simulated docking position. Compared with the traditional method, this embodiment refines the application process of the formula in the specific scoring method, and can also adjust the parameters therein so that the scoring results of each dimension can correctly reflect the expected demand for the docking station.

[0099] See also Figure 4 The fourth embodiment of the automatic parking scoring method according to the present invention includes:

[0100] 401, obtaining scene data collected in an automatic parking scenario, and marking the parking area of ​​the vehicle in the scene data;

[0101] 402 , inputting the scene data into a preset autonomous driving system to perform parking simulation and obtain at least one simulated parking position;

[0102] 403, determining whether the simulated docking position falls within the dockable area;

[0103] 404 , if the vehicle falls within the dockable area, then calculating an optimal docking position in the dockable area based on the scene data, and calculating a docking score for the simulated docking position based on the optimal docking position;

[0104] 405, calculate the effective time of the dockable area;

[0105] In this embodiment, the effective time in the process of calculating the docking score is defined as the total time from the key frame where the dockable area is located to the end of the simulation.

[0106] Specifically, if only one simulated stop position is output, the score of the last frame is calculated as zero.

[0107] 406 , recording the moment when the autonomous driving system outputs the simulated docking area as a decision frame, and determining the decision frame with the highest docking score within the effective time;

[0108] In this embodiment, if there is only one simulated docking position, the corresponding decision frame is the decision frame with the highest docking score within the effective time; if there are at least two simulated docking positions, the docking scores of each simulated docking position should be calculated separately and then the decision frame corresponding to the simulated docking position with the highest docking score should be determined.

[0109] 407, calculating the average of the docking scores from the decision frame with the highest docking score to the last frame within the effective time;

[0110] In this embodiment, the situation where there are at least two simulated docking positions is, for example, the automatic driving system outputs five simulated docking positions, where the docking score of the second simulated docking position is the highest, then the average of the docking scores corresponding to the second, third, fourth and fifth simulated docking positions is calculated.

[0111] 408, determining whether the docking score of the last frame is zero;

[0112] 409, if it is zero, divide the average by two to obtain the regional score;

[0113] 410, if not zero, the average value is used as the regional score;

[0114] 411, obtaining the area scores corresponding to all simulated parking positions in the automatic parking scenario;

[0115] 412, the highest regional score among all regional scores is used as the scoring result of the automatic parking of the autonomous driving system;

[0116] 413. If the vehicle does not fall within the dockable area, the score result is set to zero.

[0117] This embodiment, based on the previous embodiment, describes in detail the process of calculating the effective time of the dockable area, wherein the effective time is the total time from the key frame to the end of the simulation, and the key frame is the time when the dockable area is manually marked; the time when the autonomous driving system outputs the simulated docking area is recorded as the decision frame, and the decision frame with the highest docking score within the effective time is determined; the average of the docking scores from the decision frame with the highest docking score to the last frame within the effective time is calculated; and whether the docking score of the last frame is zero is determined; if it is zero, the average is divided by two to obtain the area score; if it is not zero, the average is used as the area score. By refining the area score calculation process, the scoring method for automatic parking is clarified.

[0118] The above describes the automatic parking scoring method in the embodiment of the present invention. The following describes the automatic parking scoring device in the embodiment of the present invention. Figure 6In one embodiment of the present invention, an automatic parking scoring device includes:

[0119] A location marking module 501 is used to obtain scene data collected in an automatic parking scenario and mark the parking area of ​​the vehicle in the scene data;

[0120] a parking simulation module 502 for inputting the scene data into a preset autonomous driving system to perform parking simulation and obtain at least one simulated parking position output by the autonomous system;

[0121] The scoring module 503 is configured to evaluate the simulated parking position based on the scenario data and the dockable area to obtain a scoring result of the automatic parking of the autonomous driving system.

[0122] In this embodiment, scene data collected during an automated parking scenario is acquired and the vehicle's available parking areas are annotated in the scene data. The scene data is then input into a pre-set automated parking system for parking simulation, resulting in at least one simulated parking position output by the automated system. The simulated parking positions are then evaluated based on the scene data and the available parking areas, resulting in a scoring result for the automated parking system. This method evaluates the simulated parking positions output by the automated parking system during the simulation process. By comprehensively evaluating the simulated parking positions and providing a more comprehensive assessment of the automated parking system's performance in a docking station scenario, this method can help the automated parking system achieve rapid iteration in docking station scenarios and optimize the automated parking system algorithm.

[0123] See also Figure 7 The second embodiment of the automatic parking scoring device in the embodiment of the present invention includes:

[0124] A location marking module 501 is used to obtain scene data collected in an automatic parking scenario and mark the parking area of ​​the vehicle in the scene data;

[0125] a parking simulation module 502 for inputting the scene data into a preset autonomous driving system to perform parking simulation and obtain at least one simulated parking position output by the autonomous system;

[0126] The scoring module 503 is configured to evaluate the simulated parking position based on the scenario data and the dockable area to obtain a scoring result of the automatic parking of the autonomous driving system.

[0127] In this embodiment, the scoring module 503 is specifically used to:

[0128] The area judgment unit 5031 is used to judge whether the simulated docking position falls within the dockable area; the docking score calculation unit 5032 is used to calculate the optimal docking position of the dockable area based on the scene data if it falls within the dockable area, and calculate the docking score of the simulated docking position based on the optimal docking position; the area score calculation unit 5033 is used to perform an area score on the dockable area based on the docking score to obtain an area score; the scoring result summary unit 5034 is used to perform an overall evaluation of the simulated docking position based on the area score to obtain a scoring result of automatic parking of the automatic driving system; the non-docking area unit 5035 is used to set the scoring result to zero if it does not fall within the dockable area.

[0129] In this embodiment, the docking score calculation unit 5032 is specifically configured to:

[0130] The vertex order determination subunit 50321 is used to calculate the direction of the dockable area based on the vertex coordinates of the dockable area and the road points on the center line of the lane, and determine the vertex order on the dockable area based on the direction of the dockable area; the parking mode determination subunit 50322 is used to determine the parking mode of the dockable area based on the station position and the road points on the center line of the lane; the optimal docking position calculation subunit 50323 is 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 center line of the lane; the docking score calculation subunit 50324 is used to calculate the docking score of the simulated docking position based on the optimal docking position and the scene data.

[0131] In this embodiment, the vertex order determination subunit 50321 is specifically used to:

[0132] The center of gravity of the dockable area is calculated based on the vertex coordinates; the road point on the road centerline that is closest to the center of gravity of the area is queried, and the direction of the road point closest to the center of gravity of the area is used as the direction of the dockable area; the center of gravity of the area 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 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.

[0133] In this embodiment, the docking score calculation subunit 50324 is specifically used to:

[0134] Based on the distance between the optimal parking position and the simulated parking position, calculate the first score of the simulated parking position; calculate the second score and the third score of the simulated parking position based on the distance between the right side edge of the simulated vehicle corresponding to the simulated parking position and the lane road edge position and the lane curb position respectively; calculate the fourth score of the simulated parking position based on the position of the stationary obstacle closest to the simulated parking position and the distance from the simulated parking position; obtain the area level corresponding to the dockable area, and determine the area weight of the dockable area based on the area level; input the first score, the second score, the third score, the fourth score and the area weight into a preset evaluation formula to obtain the docking score of the simulated parking position.

[0135] In this embodiment, the region score calculation unit 5033 is specifically configured to:

[0136] Calculate the effective time of the dockable area, wherein the effective time is all time from a key frame to the end of the simulation, and the key frame is the moment when the dockable area is manually marked; record the moment when the autonomous driving system outputs the simulated docking area as a decision frame, and determine the decision frame with the highest docking score within the effective time; calculate the average of the docking scores from the decision frame with the highest docking score to the last frame within the effective time; determine whether the docking score of the last frame is zero; if it is zero, divide the average by two to obtain the area score; if it is not zero, use the average as the area score.

[0137] In this embodiment, the scoring result aggregation unit 5034 is specifically used to:

[0138] Obtaining the regional scores corresponding to all simulated parking positions in the automatic parking scenario; and using the highest regional score among all regional scores as the scoring result of the automatic parking of the automatic driving system.

[0139] Based on the previous embodiment, this embodiment describes in detail the specific functions of each module and the unit structure of some modules. Through the above modules and the refinement of the specific functions of the original modules, the operation of the automatic parking scoring device is improved, its reliability during operation is enhanced, the actual logic between each step is clarified, and the practicality of the device is improved.

[0140] above Figure 6 and Figure 7 The automatic parking scoring device in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The automatic parking scoring device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0141] Figure 8FIG2 is a schematic diagram of the structure of an automatic parking scoring device provided in an embodiment of the present invention. The automatic parking scoring device 700 may vary significantly depending on configuration or performance. The device may include one or more central processing units (CPUs) 710 (e.g., one or more processors), memory 720, and one or more storage media 730 (e.g., one or more mass storage devices) storing applications 733 or data 732. The memory 720 and storage media 730 may be either transient or persistent storage. The program stored in the storage medium 730 may include one or more modules (not shown), each of which may include a series of instructions for operating the automatic parking scoring device 700. Furthermore, the processor 710 may be configured to communicate with the storage medium 730, allowing the automatic parking scoring device 700 to execute the series of instructions stored in the storage medium 730 to implement the steps of the automatic parking scoring method described above.

[0142] The automatic parking scoring device 700 may further include one or more power supplies 740, one or more wired or wireless network interfaces 750, one or more input and output interfaces 760, and / or one or more operating systems 731, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 8 The structure of the automatic parking scoring device shown does not constitute a limitation on the automatic parking scoring device provided in this application, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0143] 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 that, when executed on a computer, cause the computer to execute the steps of the automatic parking scoring method.

[0144] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0145] 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 and includes several instructions for enabling a computer device (which can be a personal computer, server, or 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), magnetic disk or optical disk, etc., various media that can store program code.

[0146] 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 above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can 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. An automatic parking scoring method, characterized in that: The automatic parking scoring method includes: Obtaining scene data collected in an automatic parking scenario, and marking the parking area of ​​the vehicle in the scene data; Inputting the scene data into a preset autonomous driving system to perform parking simulation, and obtaining at least one simulated parking position output by the autonomous driving system; evaluating the simulated parking position based on the scenario data and the dockable area to obtain a scoring result of the automatic parking of the autonomous driving system; The evaluating the simulated docking position based on the scenario data and the dockable area to obtain a scoring result of the automatic parking of the autonomous driving system includes: determining whether the simulated docking position falls within the dockable area; if it falls within the dockable area, calculating an optimal docking position for the dockable area based on the scenario data, and calculating a docking score for the simulated docking position based on the optimal docking position; performing a regional scoring on the dockable area based on the docking score to obtain a regional score; performing an overall evaluation of the simulated docking position based on the regional score to obtain a scoring result of the automatic parking of the autonomous driving system; and if it does not fall within the dockable area, setting the scoring result to zero; The scenario data includes a station location and a lane centerline of a station in an automatic parking scenario, wherein the lane centerline is composed of a plurality of road points; calculating the optimal parking position of the dockable area based on the scenario data, and calculating the parking score of the simulated parking position based on the optimal parking position includes: calculating a direction of the dockable area based on vertex coordinates of the dockable area and the road points on the lane centerline, and determining a vertex order in the dockable area based on the direction of the dockable area; determining a parking mode of the dockable area based on the station location and the road points on the lane centerline; calculating the optimal parking 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 calculating the parking score of the simulated parking position based on the optimal parking position and the scenario data; The method of calculating the direction of the dockable area based on the vertex coordinates of the dockable area 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 includes: calculating the area 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 area center of gravity, and taking the direction of the road point that is closest to the area center of gravity as the direction of the dockable area; connecting the area center of gravity and each vertex in the dockable area to obtain a first vector corresponding to each vertex; calculating the angle between the first vector and the direction of the dockable area, and sorting each vertex based on the size of the angle corresponding to each vertex to obtain the order of vertices on the dockable area.

2. The automatic parking scoring method according to claim 1, characterized in that: The scene data also includes the position of the lane road edge, the position of the lane curb and the position of the stationary obstacle; The calculating of the docking score of the simulated docking position based on the optimal docking position and the scenario data comprises: Calculating a first score of the simulated docking position based on a distance between the optimal docking position and the simulated docking position; Calculating a second score and a third score of the simulated parking position based on the distances between the right side edge of the simulated vehicle corresponding to the simulated parking position and the lane road edge and the lane curb, respectively; calculating a fourth score of the simulated docking position based on a distance between the simulated docking position and a stationary obstacle position of a stationary obstacle closest to the simulated docking position; Obtaining an area level corresponding to the dockable area, and determining an area weight of the dockable area based on the area level; The first score, the second score, the third score, the fourth score and the area weight are input into a preset evaluation formula to obtain the docking score of the simulated docking position.

3. The automatic parking scoring method according to claim 2, characterized in that: Scoring the dockable area based on the docking score to obtain the area score includes: Calculating the effective time of the dockable area, wherein the effective time is all time from a key frame to the end of the simulation, and the key frame is the moment when the dockable area is manually marked; Recording the moment when the automatic driving system outputs the simulated docking area as a decision frame, and determining the decision frame with the highest docking score within the effective time; Calculate the average of the stop scores from the decision frame with the highest stop score to the last frame within the effective time; Determine whether the docking score of the last frame is zero; If it is zero, the average value is divided by two to obtain the regional score; If it is not zero, the average value is used as the regional score.

4. The automatic parking scoring method according to any one of claims 2-3, characterized in that: The overall evaluation of the simulated parking position based on the regional score to obtain the automatic parking scoring result of the automatic driving system includes: Obtaining the area scores corresponding to all simulated parking positions in the automatic parking scenario; The highest regional score among all regional scores is used as the scoring result of the automatic parking of the automatic driving system.

5. An automatic parking scoring device, characterized in that: The automatic parking scoring device includes: A location marking module, configured to obtain scene data collected in an automatic parking scenario and mark the parking area of ​​the vehicle in the scene data; a parking simulation module, configured to input the scene data into a preset autonomous driving system to perform parking simulation and obtain at least one simulated parking position output by the autonomous driving system; a scoring module, configured to evaluate the simulated parking position based on the scenario data and the dockable area, and obtain a scoring result of the automatic parking of the autonomous driving system; The scoring module is specifically configured to: an area judgment unit for judging whether the simulated docking position falls within the dockable area; a docking score calculation unit for calculating, if the simulated docking position falls within the dockable area, an optimal docking position of the dockable area based on the scenario data, and calculating a docking score of the simulated docking position based on the optimal docking position; an area score calculation unit for performing an area score on the dockable area based on the docking score to obtain an area score; a scoring result summary unit for performing an overall evaluation of the simulated docking position based on the area score to obtain a scoring result of automatic parking by the autonomous driving system; and a non-docking area unit for setting the scoring result to zero if the simulated docking position does not fall within the dockable area. The scenario data includes a station location and a lane centerline of a station in an automatic parking scenario, wherein the lane centerline is composed of a plurality of road points. The parking score calculation unit is specifically configured to: a vertex order determination subunit, configured to calculate a direction of the dockable area based on vertex coordinates of the dockable area and the road points on the lane centerline, and determine a vertex order in the dockable area based on the direction of the dockable area; a parking mode determination subunit, configured to determine a parking mode of the dockable area based on the station location and the road points on the lane centerline; an optimal parking position calculation subunit, configured to calculate an optimal parking 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 a parking score calculation subunit, configured to calculate a parking score for the simulated parking position based on the optimal parking position and the scenario data. The vertex order determination subunit is specifically used to: calculate the 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 center of gravity of the area, and use the direction of the road point closest to the center of gravity of the area as the direction of the dockable area; connect the center of gravity of the area 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.

6. An automatic parking scoring device, characterized in that: The automatic parking scoring device includes: 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 to enable the automatic parking scoring device to perform each step of the automatic parking scoring method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, each step of the automatic parking scoring method according to any one of claims 1 to 4 is implemented.

Images