Reference line generation method, automatic driving method, device, equipment and vehicle

Abstract

The invention provides a reference line generation method, an automatic driving method and device, equipment and a vehicle, and relates to the field of artificial intelligence, in particular to the field of automatic driving. According to the specific implementation scheme, for at least one obstacle located in a target area in a road, the category of the obstacle is determined according to a first motion parameter of the obstacle in a historical time period, and the category of the obstacle represents the relative relation between the obstacle and a target vehicle; under the condition that the at least one obstacle comprises the target obstacle belonging to the target category, determining any candidate node in a plurality of candidate nodes determined for the position of the target vehicle, and determining the total cost value of the candidate nodes according to the historical driving track of the target obstacle and the second motion parameter of the candidate node; determining a target region reference line according to a target node determined by using the total cost values of the plurality of candidate nodes; wherein the target area reference line is used for planning a target driving track of the target vehicle.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to the field of autonomous driving. More specifically, the present disclosure provides a method for generating a road reference line, an autonomous driving method, a device for generating a road reference line, an autonomous driving device, an electronic device, a storage medium, a computer program product, and an autonomous driving vehicle. Background Art

[0002] In the field of autonomous driving, a reference line can be determined first, and then the driving path of the vehicle can be planned based on the reference line, so as to control the vehicle to drive along the driving path. Summary of the Invention

[0003] The present disclosure provides a method for generating a road reference line, an autonomous driving method, a device for generating a road reference line, an autonomous driving device, an electronic device, a storage medium, a computer program product, and an autonomous driving vehicle.

[0004] According to one aspect of the present disclosure, there is provided a method for generating a road reference line, including: for at least one obstacle located at a target area on a road, determining the category of the obstacle according to the first motion parameter of the obstacle in a historical period, where the category of the obstacle represents the relative relationship between the obstacle and the target vehicle; in the case that at least one obstacle includes a target obstacle belonging to the target category, for any candidate node determined according to the position of the target vehicle, determining the total cost value of the candidate node according to the historical driving trajectory of the target obstacle and the second motion parameter of the candidate node; determining a target area reference line according to the target node determined by using the total cost values of the multiple candidate nodes respectively; where the target area reference line is used to plan the target driving trajectory of the target vehicle.

[0005] According to another aspect of the present disclosure, there is provided an autonomous driving method, including: obtaining a target area reference line; determining a target driving trajectory of a target vehicle based on the target area reference line; controlling the target vehicle to drive autonomously based on the target driving trajectory; where the target area reference line is generated according to the above method.

[0006] According to another aspect of the present disclosure, there is provided an apparatus for generating an intersection reference line, including: a category determination module, an overall cost value determination module, and a reference line determination module. The category determination module is configured to determine the category of at least one obstacle located in a target area on a road according to the first motion parameters of the obstacle during a historical period, and the category of the obstacle represents the relative relationship between the obstacle and the target vehicle. The overall cost value determination module is configured to, when at least one obstacle includes a target obstacle belonging to a target category, determine the overall cost value of any candidate node among a plurality of candidate nodes determined according to the position of the target vehicle according to the historical driving trajectory of the target obstacle and the second motion parameters of the candidate node. The reference line determination module is configured to determine a target area reference line according to a target node determined by using the overall cost values of the plurality of candidate nodes respectively; wherein, the target area reference line is used to plan the target driving trajectory of the target vehicle.

[0007] According to another aspect of the present disclosure, there is provided an intersection automatic driving apparatus, including: an acquisition module, a determination module, and a control module. The acquisition module is configured to acquire a target area reference line, and the target area reference line is generated according to the above-mentioned apparatus. The determination module is configured to determine the target driving trajectory of the target vehicle based on the target area reference line. The control module is configured to control the automatic driving of the target vehicle based on the target driving trajectory.

[0008] According to another aspect of the present disclosure, there is provided an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method provided by the present disclosure.

[0009] According to another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute the method provided by the present disclosure.

[0010] According to another aspect of the present disclosure, there is provided a computer program product, including a computer program, and the computer program implements the method provided by the present disclosure when executed by a processor.

[0011] According to another aspect of the present disclosure, there is provided an autonomous vehicle, including the above-mentioned electronic device.

[0012] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. Description of the Drawings

[0013] The accompanying drawings are used to better understand the present solution and do not constitute a limitation to the present disclosure. Among them:

[0014] Figure 1 is a schematic diagram of the application scenario of the method for generating a road reference line and the autonomous driving method according to an embodiment of the present disclosure;

[0015] Figure 2 is a schematic flowchart of the method for generating a road reference line according to an embodiment of the present disclosure;

[0016] Figure 3 is a schematic principle diagram for determining the category of an obstacle according to an embodiment of the present disclosure;

[0017] Figure 4 is a schematic diagram of the positional relationship between a target vehicle and an obstacle according to an embodiment of the present disclosure;

[0018] Figure 5 is a schematic diagram of an intersection navigation line according to an embodiment of the present disclosure;

[0019] Figure 6 is a schematic diagram of the reference driving area for merging a side vehicle according to an embodiment of the present disclosure;

[0020] Figure 7 is a schematic principle diagram of the method for generating a road reference line according to an embodiment of the present disclosure;

[0021] Figure 8 is a schematic flowchart of the autonomous driving method according to an embodiment of the present disclosure;

[0022] Figure 9 is a schematic structural block diagram of the device for generating a road reference line according to an embodiment of the present disclosure;

[0023] Figure 10 is a schematic structural block diagram of the autonomous driving device according to an embodiment of the present disclosure; and

[0024] Figure 11 is a structural block diagram of an electronic device for implementing the method for generating a road reference line and the autonomous driving method according to an embodiment of the present disclosure. Detailed Embodiments

[0025] The following describes exemplary embodiments of the present disclosure in conjunction with the accompanying drawings. Various details of the embodiments of the present disclosure are included to assist in understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for clarity and conciseness, descriptions of well-known functions and structures are omitted below.

[0026] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision, and disclosure of the user's personal information comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0027] In the technical solution of the present disclosure, before obtaining or collecting the user's personal information, the user's authorization or consent is obtained.

[0028] In recent years, light maps have been increasingly widely used in the field of autonomous driving. Compared with traditional high-precision maps, light maps have the advantages of low production cost, convenient update, small data volume, and wide coverage. With the continuous progress of perception algorithms, especially the rapid development of visual perception technology, light maps have been able to meet the basic requirements of L2-level autonomous driving. More and more automobile enterprises have begun to adopt light map solutions. On the one hand, it can significantly reduce the map-related costs, and on the other hand, it can also more flexibly respond to changes in the road environment. Although there is still a gap between light maps and high-precision maps in terms of accuracy and reliability, their "lightweight" characteristics are very suitable for the current market demand and technological development trend of autonomous driving. It can be predicted that with the continuous improvement of perception technology, light maps will play an increasingly important role in the field of autonomous driving.

[0029] In light maps, the construction of reference lines is an important and difficult problem, especially in open space areas such as intersections: First, the geometric shapes of intersections in real roads vary significantly, ranging from regular crossroads to irregular multi-fork intersections, and this diversity makes it difficult to adopt a unified rule to describe and process intersection areas; Second, in actual scenarios, due to the existence of obstacles such as buildings, green belts, and other vehicles, as well as the perception distance limitation of the visual sensor itself, it is often impossible to obtain the complete information of the intersection area at one time, which brings great difficulties to perception-based intersection mapping; In addition, the constructed reference lines must conform to the driving habits of human drivers and match the actual traffic flow trajectories. This not only requires the reference lines to be smooth and reasonable in geometric shape, but also to reflect the characteristics of human driving, such as the driving path selection at different turning speeds and the interactive avoidance behaviors with other vehicles.

[0030] Currently, the methods for generating reference lines for open spaces such as intersections mainly include: rule-based reference line generation methods and artificial buried point-based reference line generation methods.

[0031] The rule-based reference line generation method mainly relies on preset geometric models and turning rules to generate reference lines. This type of method usually uses mathematical models such as Hermite curves, Dubins curves, Euler spirals, etc., and combines the entrance position and exit position of the intersection to perform curve fitting. In the implementation process, first, it is necessary to obtain the geometric feature information of the intersection, including the entrance position, exit position, road width, etc.; then, according to the desired turning type (left turn, straight ahead, or right turn), select an appropriate curve model to fit the reference line. To ensure the feasibility of the generated reference line, dynamic constraint conditions such as curvature constraints and speed constraints are usually introduced. However, the rule-based reference line generation method overly relies on preset geometric models and rules, cannot well accommodate intersection scenarios of different sizes and shapes, the generated reference lines are relatively idealized, and there is usually a certain deviation from the actual vehicle driving trajectory; moreover, this method lacks the ability to perceive and respond to the dynamic traffic environment and cannot make real-time adjustments according to traffic conditions such as the traffic flow within the current intersection.

[0032] The reference line generation method based on manual waypoint marking generates a complete reference line by pre-marking waypoints at several key positions in the intersection and then connecting these waypoints through a smoothing algorithm. This type of method usually marks a series of key waypoints within the intersection according to the road geometric features and typical manual driving trajectories, then constructs the original reference line based on these waypoints, and finally performs post-processing such as smoothing on the original reference line through an optimization algorithm to ensure that the generated reference line meets the requirements of continuity and smoothness. However, in addition to the high cost of manual annotation, the fixed waypoint configuration lacks flexibility and cannot be adjusted in a timely manner in the face of real-world changes (such as temporary construction).

[0033] The embodiments of the present disclosure aim to provide a method for generating a road reference line based on traffic flow, which can improve the ability to perceive and respond to the dynamic traffic environment. In the case of accidents, temporary construction, etc. on the road, this method can make real-time adjustments according to traffic conditions such as the traffic flow on the road, and reduce the collision risks such as squeezing and cutting in of the target vehicle with other vehicles within the target area.

[0034] The technical solutions provided by the embodiments of the present disclosure can be used to generate reference lines for various intersection scenarios such as straight ahead, left turn, and right turn of the target vehicle at intersections.

[0035] The following will elaborate in detail on the technical solutions provided by the present disclosure in combination with the accompanying drawings and specific embodiments.

[0036] Figure 1 It is a schematic diagram of the application scenarios of the method for generating a road reference line and the autonomous driving method according to the embodiments of the present disclosure.

[0037] It should be noted that Figure 1The figure shown is only an example of the system architecture to which the embodiments of the present disclosure can be applied, to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0038] As Figure 1 shown, the scenario 100 of this embodiment includes a server 110, a vehicle 120 traveling on a road, an obstacle 130, a communication base station 140, and a road traffic network. The road traffic network may include roads (such as roads 151, 152, 153), and intersections 154 formed by the intersection of roads. Figure 1 The arrows in indicate the driving directions of the roads.

[0039] For example, in this scenario 100, the vehicle 120 traveling on the road may be an autonomous vehicle. The obstacle 130 may be a pedestrian or other vehicle around the autonomous vehicle. For example, the vehicle 120 may go straight along the road 151 and pass through the intersection 154, and the obstacle 130 needs to travel from the road 153 to the intersection 154 and turn left into the road 152.

[0040] The vehicle 120 may be equipped with an autonomous driving system to detect the driving parameters of the obstacle 130, such as the current speed, acceleration, orientation information, etc. of the obstacle. And generate a reference line according to the driving parameters of the obstacle 130. The vehicle 120 may upload data to the background server through the communication base station 140. The server 110 may, for example, request data from the background server through a network to obtain the data uploaded by the vehicle 120.

[0041] It should be noted that the method for generating a road reference line and the autonomous driving method provided by the embodiments of the present disclosure may be executed by the vehicle 120. Correspondingly, the device for generating a road reference line and the autonomous driving device provided by the embodiments of the present disclosure may be provided in the vehicle 120. The method for generating a road reference line and the autonomous driving method provided by the embodiments of the present disclosure may also be executed by the server 110. Correspondingly, the device for generating a road reference line and the autonomous driving device provided by the embodiments of the present disclosure may also be provided in the server 110.

[0042] It should be understood that Figure 1 the number and type of the servers, roads, vehicles, obstacles, and communication base stations in are only illustrative. According to the implementation requirements, there may be any number and type of servers, roads, vehicles, obstacles, and communication base stations.

[0043] Figure 2 is a schematic flowchart of the method for generating a road reference line according to the embodiments of the present disclosure.

[0044] As Figure 2As shown, the method 200 for generating a road reference line may include operation S210 to operation S230.

[0045] In operation S210, for at least one obstacle at a target area in a road, according to the first motion parameters of the obstacle in a historical period, the category of the obstacle is determined, and the category of the obstacle characterizes the relative relationship between the obstacle and the target vehicle.

[0046] For example, the target area may be an open area lacking lane lines or with unclear lane lines, such as an intersection. The obstacle being at the target area may mean that the obstacle is located in the target area currently or in the past period of time.

[0047] For example, the obstacle may be other vehicles around the target vehicle. The first motion parameters may include, for example, the positions, directions, etc. of the obstacle at each moment in the historical period. The categories of the obstacle may include following targets, side vehicles, irrelevant vehicles, etc. Side vehicles may include same-direction side vehicles, reverse side vehicles, merging side vehicles, etc. The category of the obstacle characterizes the relative relationship between the obstacle and the target vehicle, and this relative relationship may include the relative position relationship and the relationship between the driving directions. The category of the obstacle can be determined based on the position of the target vehicle, the driving direction of the target vehicle, the position of the obstacle, and the driving direction of the obstacle.

[0048] In operation S220, in the case where at least one obstacle includes a target obstacle belonging to the target category, for any candidate node among the multiple candidate nodes determined based on the position of the target vehicle, according to the historical driving trajectory of the target obstacle and the second motion parameters of the candidate node, the total cost value of the candidate node is determined.

[0049] For example, the search for target nodes can be performed by the A* algorithm. The A* algorithm uses a multi-round iterative method for searching, and the target node of each round of iteration process is determined. In the first iteration round, multiple candidate nodes can be selected around the target vehicle, and then the target node is determined from them. In subsequent iteration rounds, multiple candidate nodes can be determined around the target node determined in the previous round, and then a target node is determined from them. It can be seen that the candidate nodes are directly or indirectly determined based on the position of the target vehicle. The target node is a node used to construct the reference line of the target area. It is necessary to determine whether to use the candidate node to construct the reference line of the target area according to the total cost value. If so, the candidate node is the target node; otherwise, the candidate node is not the target node. The second motion parameters of the candidate node may include position, direction, etc.

[0050] For example, the target categories may include following vehicle targets, adjacent vehicles, etc. For a target obstacle belonging to the target category, the total cost value can be determined based on the distance, angular difference, etc. between the candidate node and the historical driving trajectory of the target obstacle.

[0051] In operation S230, a target area reference line is determined based on the target node determined by using the total cost value of each of the multiple candidate nodes; wherein, the target area reference line is used to plan the target driving trajectory of the target vehicle.

[0052] For example, the candidate node with the minimum total cost value can be determined as the target node. After multiple rounds of iteration, multiple target nodes can be obtained, and the multiple target nodes can be fitted into a line, and this line is the target area reference line.

[0053] The technical solution provided by the embodiments of the present disclosure generates a road reference line based on the traffic flow of other obstacles on the road. This method classifies the obstacles observed in the intersection area (such as other vehicles around the target vehicle), and then for the target obstacles of the target category, uses the historical driving trajectory of the target obstacle and the second motion parameters such as the position and orientation angle of the node to calculate the total cost value of the node, and then selects the target node based on the total cost value.

[0054] It can be understood that the historical driving trajectory of the target obstacle can reflect the real traffic flow situation, and the real traffic flow situation changes dynamically with the traffic conditions on the road, rather than remaining fixed. For example, temporary construction, accidents, etc. on the road will all cause dynamic adjustment of the real traffic flow situation. Therefore, using the historical driving trajectory of the target obstacle to plan the reference line can improve the perception and response ability to the dynamic traffic environment. In the case of accidents, temporary construction, etc. on the road, this method can be adjusted in real time according to the traffic conditions such as the traffic flow on the road, and reduce the collision risks such as squeezing and cutting in between the target vehicle and other vehicles in the target area.

[0055] Figure 3 It is a schematic principle diagram for determining the obstacle category according to the embodiments of the present disclosure.

[0056] In this embodiment, the category of an obstacle can be determined based on the first motion parameter 301 of the obstacle within a historical period, in combination with the map static information 302 and the mapping perception information 303. The map static information 302 may include a navigation topology. The mapping perception information 303 includes, for example, a stop line, an entrance of an intersection, an exit of an intersection, etc. The driving direction of the target vehicle can be determined through the navigation topology, and information such as the position, driving direction, and driving trajectory of the obstacle can be determined through information such as the stop line, the entrance of the intersection, the exit of the intersection, and the first motion parameter of the obstacle, so as to determine the category of the obstacle. By classifying the obstacles based on the category, following targets 304, side vehicles 305, and other categories can be obtained. The side vehicles 305 may include same-direction side vehicles, reverse side vehicles, and merging side vehicles.

[0057] Figure 4 FIG. is a schematic diagram of the positional relationship between a target vehicle and an obstacle according to an embodiment of the present disclosure.

[0058] Next, in combination with Figure 4 , the category of the obstacle will be described.

[0059] In this embodiment, the categories of the obstacles may include following target categories, side vehicle categories, irrelevant categories, etc. The side vehicle categories may include same-direction side vehicle categories, reverse side vehicle categories, and merging side vehicle categories.

[0060] The following target 402 may be an obstacle that is in the same lane as the target vehicle 401, is located in front of the target vehicle 401, and has the same exit from the intersection as the target vehicle 401. The driving trajectory of the following target 402 is of great reference for the target trajectory planning of the target vehicle 401. The path characteristics of the following target 402 passing through the intersection can be tracked and learned in real time, so as to optimize the path selection and passing strategy of the target vehicle 401 and ensure safe and reasonable passing through complex intersections.

[0061] In addition, practical experience shows that sometimes there is at most one effective following target 402 (i.e., the vehicle in front of the target vehicle 401) in the intersection. Due to the influence of occlusion, the vehicle in front of the vehicle in front (i.e., the vehicle in front of the vehicle in front of the target vehicle 401) is usually difficult to detect, or even if detected, there will be problems with unstable detection results, which is difficult to meet the requirements of the continuity and reliability of target tracking for the construction of the intersection reference line. Therefore, considering the system robustness, only one following target 402 can be considered in the intersection scenario, that is: the vehicle that is in the same lane as the target vehicle 401, is located in front of the target vehicle 401, and has the closest longitudinal distance to the target vehicle 401.

[0062] The side vehicles may include the same - direction side vehicle 403, the reverse - direction side vehicle 405, and the merging - in side vehicle 404. These side vehicles are the interaction objects that the target vehicle 401 needs to focus on when driving at an intersection. Under the condition that the number of lanes permits, it is necessary to try to select a path and an exit with less interaction with such vehicles as much as possible to avoid unnecessary collision risks such as squeezing and cutting in.

[0063] The entrance of the same - direction side vehicle 403 entering the intersection is the same as the entrance of the target vehicle 401 entering the intersection, and the same - direction side vehicle 403 and the target vehicle 401 are in different lanes. Specifically, the same - direction side vehicle 403 and the target vehicle 401 enter the intersection from the same entrance but drive in different lanes. The same - direction side vehicle 403 may choose the same exit direction as the target vehicle 401, or may turn to other exits. Their movement trajectories within the intersection may intersect with those of the target vehicle 401.

[0064] The entrance of the reverse - direction side vehicle 405 entering the intersection is the same as the exit of the target vehicle 401 leaving the intersection, and the exit of the reverse - direction side vehicle 405 leaving the intersection is the same as the entrance of the target vehicle 401 entering the intersection. Specifically, the reverse - direction side vehicle 405 is a vehicle driving in the reverse direction from the exit lane and heading towards the entrance of the target vehicle 401. Although the driving path is quite different from that of the target vehicle 401, it may intersect with the target vehicle 401 when passing through the intersection.

[0065] The exit of the merging - in side vehicle 404 leaving the intersection is the same as the exit of the target vehicle 401 leaving the intersection. Specifically, the merging - in side vehicle 404 is a vehicle that enters the intersection from other entrances and finally heads towards the same exit as the target vehicle 401. The merging - in side vehicle 404 will form a confluence relationship with the target vehicle 401 when approaching the exit.

[0066] It should be noted that when the target vehicle 401 detects other leading vehicles in front of its own leading vehicle, there is a problem of unstable detection. However, the recognition of side vehicles by the target vehicle 401 is not restricted by this. Therefore, at the same moment, there can be multiple side vehicles around the target vehicle 401, and these side vehicles all correspond to independent historical driving trajectories. In the subsequent process of searching for the reference line, these side - vehicle trajectories can be comprehensively processed.

[0067] It can be understood that if the obstacle is neither the following - vehicle target 402 nor the side vehicle, it can be determined that the obstacle belongs to the irrelevant category. The obstacles belonging to the irrelevant category can refer to the obstacles that, although appearing in the target area, have no interaction with the target vehicle 401 or do not affect the construction of the reference line of the target vehicle 401, such as the left - turning vehicle or the crossing vehicle in the on - coming lane. For example, the route of the left - turning vehicle or the crossing vehicle in the on - coming lane intersects with the route of the autonomous vehicle, but this intersection mainly affects the yielding - and - overtaking strategy of the target vehicle 401 and does not affect the driving route of the target vehicle 401. Therefore, this type of obstacle can be ignored when constructing the reference line of the target area.

[0068] The categories of obstacles are introduced above, and the category can be determined by classifying the obstacles.

[0069] Figure 5 It is a schematic diagram of the intersection navigation line according to an embodiment of the present disclosure.

[0070] In this embodiment, the intersection navigation route can be constructed first, and the intersection navigation route can be used to determine the category of the obstacle.

[0071] Based on the horizontal and vertical coordinates and the orientation angle of the entrance and exit of the rough and low-precision intersection provided by the SD map (Standard Definition Map) or LD map (Lane-level Definition Map), an Euler spiral or other curve can be used to fit an intersection navigation line between the starting point and the ending point of the intersection. This original reference line of the intersection navigation line is used as the reference frame for calculating relevant projections in the Frenet coordinate system.

[0072] The Euler spiral is determined by formula (1), and a series of coordinate points of the final curve are obtained by solving the Fresnel integral by numerical methods.

[0073] Formula (1)

[0074] As Figure 5 shown, for a certain intersection 510, the lanes passing through the intersection can be determined based on the SD map, that is, the lane 521 extending in one direction and the lane 522 extending in the other direction. Since the accuracy of the SD map is low, the widths of the lane 521 and the lane 522 can be ignored. The intersection points 531 and 533 of the lane 521 and the stop line of the intersection 510 can be used as the entrance or exit of the intersection. The intersection points 532 and 534 of the lane 522 and the stop line of the intersection 510 can be used as the entrance or exit of the intersection. In this example, the target vehicle is expected to enter the intersection 510 from the entrance 533 and leave the intersection 510 from the exit 532. In this way, an intersection navigation route 540 can be determined. This intersection navigation route 540 passes through the entrance 533 and the exit 532 and is relatively smooth.

[0075] Next, the process of determining whether the obstacle is a following target will be described.

[0076] In this embodiment, an obstacle whose category belongs to the following target category is called a following target. When constructing the reference line of the target area, for a following target, it is expected that the reference line fits as closely as possible to the historical driving trajectory of the following target, so as to minimize the interaction risk with the surrounding vehicles and static obstacles in the intersection (such as fences, bridge piers, curbs, etc.).

[0077] According to the above definition of the following vehicle target, it can be known that the following vehicle target is in the same lane as the target vehicle and in front of the target vehicle. On this basis, it is necessary to determine whether the turning direction (for example, going straight, turning left or turning right) of the following vehicle target within the intersection is consistent with the following vehicle target. Therefore, the intersection navigation line mentioned above can be used as the Frenet reference system to judge whether the historical driving trajectory of the obstacle is within a certain range of this reference system. In this example, for an obstacle that is in the same lane as the target vehicle and in front of the target vehicle, when it is determined that the obstacle meets the following conditions, the obstacle is determined as the following vehicle target: at any moment within the historical period, the actual lateral distance between the position of the obstacle and the intersection navigation route is less than the target lateral distance, and the absolute value of the actual orientation angle of the obstacle is less than or equal to the target orientation angle.

[0078] For example, denote the current moment as , and the recording time of the trajectory points in the historical driving trajectory of the obstacle as . In the Frenet coordinate system, denote the entrance position of the intersection on the map as , and the exit position of the intersection as . For an obstacle that is in front of the target vehicle and in the same lane as the target vehicle, when its position and orientation angle at any moment satisfy formula (2), this obstacle is the following vehicle target. Formula (2) is as follows:

[0079] (2)

[0080] Among them, is the target lateral distance, is the target orientation angle. The position includes: the lateral distance from the guiding line of the SD map and the longitudinal distance . Among them, the point on the intersection navigation line with the shortest distance to the obstacle is called the reference point. The lateral distance can be the distance between the obstacle and the reference point, and the longitudinal distance can be the distance from the starting point to the reference point.

[0081] Exemplarily, the target lateral distance satisfies the following conditions: in the entrance area of the intersection, the target lateral distance is the predetermined minimum lateral distance; in the exit area of the intersection, the target lateral distance is the predetermined maximum lateral distance; in the middle area between the entrance area and the exit area of the intersection, the target lateral distance is positively correlated with the longitudinal distance, and the longitudinal distance is the distance that the obstacle moves along the intersection navigation route from the entrance of the intersection.

[0082] For example, the target lateral distance satisfies formula (3).

[0083] Formula (3)

[0084] Wherein, , , , , these are configurable parameters, is a predetermined minimum lateral distance, is a predetermined maximum lateral distance. It can be seen that the first segment in the above piecewise function corresponds to the entrance area, the second segment corresponds to the middle area, and the third segment represents the exit area.

[0085] Exemplarily, the target orientation angle satisfies the following conditions: in the entrance area of the intersection, the target orientation angle is a predetermined minimum angle; in the exit area of the intersection, the target lateral distance is a predetermined maximum angle; in the middle area between the entrance area and the exit area of the intersection, the target orientation angle is positively correlated with the longitudinal distance, and the longitudinal distance is the distance that the obstacle moves along the intersection navigation route from the entrance of the intersection.

[0086] For example, the target orientation angle satisfies Formula (4).

[0087] Formula (4)

[0088] Wherein, is a predetermined minimum angle, is a predetermined maximum angle, , , these are configurable parameters. It can be seen that the first segment in the above piecewise function corresponds to the entrance area, the second segment corresponds to the middle area, and the third segment represents the exit area.

[0089] The above describes the process of determining whether an obstacle is a following vehicle target.

[0090] After completing the screening and classification of the obstacles passing through the intersection within a period of time, the categories of the obstacles are obtained. If the obstacles include following vehicle targets, in the search and construction of the reference line, the reference line can be made to fit as closely as possible to the driving path of the following vehicle target.

[0091] Next, the process of determining whether an obstacle is a side vehicle is described.

[0092] In this embodiment, an obstacle whose category belongs to the side vehicle category is called a side vehicle, and the side vehicle can include a same-direction side vehicle, a reverse side vehicle, and an incoming side vehicle.

[0093] First, perform a preliminary screening. During the preliminary screening, if the obstacle is far from the lane where the target vehicle is to travel (for example, separated by several lanes), at this time, the influence of the obstacle on the target vehicle is small. Therefore, the influence of this obstacle on the target vehicle can be ignored, and thus the obstacle that satisfies the following formula (5) can be filtered out, and it is no longer determined whether this obstacle is a side vehicle. In other examples, the preliminary screening may not be performed, and only the subsequent re-screening process is carried out.

[0094] For example, denote the current moment as , and the recording time of the trajectory points in the historical driving trajectory of the obstacle is . Its position at any moment satisfies formula (5).

[0095] Formula (5)

[0096] For example, is a configurable parameter. In the Frenet coordinate system, when on the right side of the intersection navigation line, the lateral distance is negative, and when on the left side of the intersection navigation line, the lateral distance is positive. is the first predetermined lateral distance, is the second predetermined lateral distance.

[0097] ≤ indicates that the obstacle is on the right side of the intersection navigation line, and the distance between the obstacle and the intersection navigation line is less than or equal to , at this time, the distance between the obstacle and the intersection navigation line is small. indicates that the obstacle is on the left side of the intersection navigation line, and the distance between the obstacle and the intersection navigation line is less than or equal to , at this time, the distance between the obstacle and the intersection navigation line is small. If the distance between the obstacle and the intersection navigation line is greater than , it means that the obstacle is far from the lane where the target vehicle is to travel, for example, separated by several lanes. At this time, the influence of the obstacle on the target vehicle is small, so the influence of this obstacle on the target vehicle can be ignored.

[0098] After the initial screening, a re-screening can be performed. During this process, for the obstacles that have passed the initial screening, the obstacle can be determined as a side vehicle when it meets the following conditions: at any moment within a period in the historical period, the actual lateral distance between the position of the obstacle and the intersection navigation route is within the predetermined distance range, and the actual orientation angle of the obstacle is within the predetermined angle range. The actual orientation angle, for example, represents the angle between the direction of the candidate node and the direction of the reference point.

[0099] For example, one section of the historical driving trajectory of the adjacent vehicle is substantially parallel to the intersection navigation line at the entrance or exit of the target vehicle. The directions may be the same (such as an adjacent vehicle leading to the same entrance as the target vehicle, an adjacent vehicle merging into the same exit as the target vehicle, etc.) or opposite (such as a reverse adjacent vehicle, etc.).

[0100] Therefore, if there is an interval where the obstacle exists on the historical driving trajectory satisfying formula (6), the obstacle is determined as an adjacent vehicle.

[0101] Formula (6)

[0102] Among them, , , are configurable parameters. defines a predetermined distance range, defines a predetermined angle range.

[0103] In the above manner, it is possible to determine whether the obstacle is an adjacent vehicle. In the case where the obstacle is an adjacent vehicle, the subclass of the adjacent vehicle can also be determined according to the positional relationship between the entrance where the adjacent vehicle enters the intersection and the entrance where the target vehicle enters the intersection. This subclass is the same-direction adjacent vehicle, reverse adjacent vehicle, and merging adjacent vehicle. For example, the same-direction adjacent vehicle has the same entrance as the target vehicle when entering the intersection, the reverse adjacent vehicle has the opposite direction to the entrance of the target vehicle when entering the intersection, and the merging adjacent vehicle has a different entrance from the entrance of the target vehicle when entering the intersection and is not in the opposite direction.

[0104] The above describes the process of determining whether the obstacle is an adjacent vehicle. After completing the screening and classification of the obstacles passing through the intersection within a period of time, the categories of the obstacles are obtained. If the obstacles include adjacent vehicles, in the search and construction of the reference line, the reference line can minimize the overlap with the driving path of the adjacent vehicle to reduce the interaction risk in the intersection.

[0105] The above describes the process of determining the category of the obstacle. Next, taking the following vehicle target as an example, the process of determining the total cost value of the candidate node is described.

[0106] In this embodiment, the cost for determining the total cost value of the candidate node includes the following vehicle target cost value, and the following vehicle target cost value includes at least one of the distance cost value and the angle cost value. The following vehicle target cost is used to make the searched reference line maintain a high degree of spatial consistency with the historical driving trajectory of the following vehicle target, so as to enhance the naturalness and predictability of the driving trajectory of the target vehicle.

[0107] Next, taking an obstacle as the following - vehicle target as an example, the distance cost value in the following - vehicle target cost value is described. When at least one obstacle includes the following - vehicle target, the distance cost value can be determined according to the shortest distance from the position of the candidate node to the historical driving trajectory of the following - vehicle target.

[0108] For example, first, a Frenet reference system of the following - vehicle target is constructed. The historical driving trajectory of the following - vehicle target is usually a curve segment that is difficult to accurately depict by mathematical relations. In order to quickly and efficiently calculate the shortest distance and included - angle information from any point within the intersection to this curve segment, a Frenet coordinate system is constructed based on the historical driving trajectory of the following - vehicle target. For any position within the intersection, the shortest distance and relative angle can be calculated by projection on this Frenet coordinate system. It should be noted that the Frenet coordinate system in the above text is constructed based on the intersection navigation line, and the Frenet coordinate system here is constructed based on the historical driving trajectory of the following - vehicle target. Both belong to the Frenet coordinate system, but their construction bases are different, so the curves are generally different.

[0109] For example, the coordinates of the candidate node are projected into the Frenet coordinate system of the following - vehicle target to obtain the shortest distance from the candidate node to the historical driving trajectory of the following - vehicle target

[0110] The function of the distance cost can be formula (7).

[0111] Formula (7)

[0112] Among them, is the weight of the distance - cost function, , is an adjustable parameter for controlling the following - vehicle boundary.

[0113] When the candidate node is relatively close to the historical driving trajectory of the following - vehicle target ( ), at this time, the position of the candidate node is within a reasonable following - vehicle range, and the distance cost value is the first predetermined value. At this time, the penalty for the candidate node is small or there is no penalty. The above formula (7) is described with the first predetermined value being 0 as an example, and the first predetermined value can also be other values.

[0114] When the candidate node slightly deviates from the historical driving trajectory of the following - vehicle target ( When (), as the distance increases, the distance cost value increases accordingly. For example, in formula (7), the distance cost will increase in the form of a quadratic function to penalize candidate nodes that deviate from the historical driving trajectory of the car-following target. It should be noted that the purpose of using a quadratic function is: on the one hand, it can increase the distance cost value, and on the other hand, it can ensure that the distance cost value is a positive number. In other examples, the in formula (7) can be replaced with , where α can be any value greater than or equal to 1, and the distance cost value can be ensured to be a positive number by taking the absolute value. is replaced with , where α can be any value greater than or equal to 1, and the distance cost value can be ensured to be a positive number by taking the absolute value.

[0115] When the candidate node deviates significantly from the historical driving trajectory of the car-following target ( ), if the target vehicle moves to the position of this candidate node for car-following at this time, there will be overly intense lateral behavior. Therefore, the distance cost value is set to a second predetermined value to avoid overly intense lateral behavior. The above formula (7) is illustrated with the second predetermined value being 0, and the second predetermined value can also be other values. ), if the target vehicle moves to the position of this candidate node for car-following at this time, there will be overly intense lateral behavior. Therefore, the distance cost value is set to a second predetermined value to avoid overly intense lateral behavior. The above formula (7) is illustrated with the second predetermined value being 0, and the second predetermined value can also be other values.

[0116] The above takes an obstacle as the car-following target as an example to illustrate the distance cost value in the car-following target cost value.

[0117] Next, taking an obstacle as the car-following target as an example, the angle cost value in the car-following target cost value will be described.

[0118] For example, when at least one obstacle includes the car-following target, the angle cost value can be determined according to the included angle between the direction of the candidate node and the direction of the reference point.

[0119] The function of the angle cost can be formula (8).

[0120] Formula (8)

[0121] where is the unit direction vector corresponding to the candidate node. is the direction of the reference point, and this direction passes through the reference point and is tangent to the historical driving trajectory of the car-following target. It should be noted that the reference point refers to the projection point of the candidate node on the historical driving trajectory of the car-following target, and this reference point is usually the point on the historical driving trajectory of the car-following target whose distance from the candidate node meets a predetermined condition, and the predetermined condition can be the shortest distance.

[0122] In this embodiment, the angle cost value can be positively correlated with the size of the included angle. For example, if the included angle between the direction of the candidate node and the direction of the reference point is small, it indicates that the vehicle following is more reasonable, so the angle cost value is small. If the included angle between the direction of the candidate node and the direction of the reference point is large, it means that the target vehicle needs to adjust a large angle to move along the historical driving trajectory of the vehicle following target, and the vehicle following rationality is low, so the angle cost value is large.

[0123] Next, taking an obstacle as the vehicle following target as an example, the angle cost value in the vehicle following target cost value is described.

[0124] Next, taking an obstacle as a side vehicle as an example, the side vehicle traffic flow cost value is described. In the process of determining the side vehicle traffic flow cost value, the processed trajectory can be determined based on the historical driving trajectory of the side vehicle. The processed trajectory is mainly used to determine the reference driving area of the side vehicle, and then a grid map is constructed using the reference driving area, so as to determine the side vehicle traffic flow cost value using the grid map and realize the search for the target node.

[0125] Next, the process of determining the processed trajectory based on the historical driving trajectory of the side vehicle is introduced first.

[0126] In the case where the side vehicle is a reverse side vehicle belonging to the first subclass, the historical driving trajectory of the reverse side vehicle can be determined as the processed trajectory.

[0127] In the case where the side vehicle is a same - direction side vehicle belonging to the second subclass, the historical driving trajectory of the same - direction side vehicle can be offset in the direction away from the target vehicle to obtain the processed trajectory. The historical driving trajectory includes information such as the coordinates of the set center of the obstacle, the orientation angle, the left and right vehicle widths, etc. at each moment. The offset direction can be perpendicular to the driving direction of the obstacle, and the offset distance can be one vehicle width or other distances. For example, in some cases, such as when the same - direction side vehicle and the target vehicle need to leave the intersection from the same exit and the exit may have only one lane, at this time, the same - direction side vehicle and the target vehicle need to leave the exit from the same lane. If the processed trajectory of the same - direction side vehicle remains the same as the historical driving trajectory before processing, it will cause the reference driving area determined based on the same - direction side vehicle to occupy the lane that the target vehicle needs to drive on, further causing the passing times of the grids in this lane to increase, and the side vehicle traffic flow cost value of the candidate nodes in this lane to increase. Since the target vehicle needs to drive along this lane, therefore, the side vehicle traffic flow cost value of the candidate nodes in this lane needs to be small. For this reason, in this embodiment, the historical driving trajectory of the side vehicle is offset in the direction away from the target vehicle, that is, for the same - direction side vehicle on the left side of the target vehicle, the historical driving trajectory of the same - direction side vehicle is offset to the left; for the same - direction side vehicle on the right side of the target vehicle, the historical driving trajectory of the same - direction side vehicle is offset to the right. This can ensure that the candidate nodes in this lane have a small side vehicle traffic flow cost value and ensure the accuracy of searching for the target node.

[0128] In the case where the adjacent vehicle is an in-flowing adjacent vehicle belonging to the third subclass, the historical driving trajectory of the adjacent vehicle can be offset away from the target vehicle, and the offset distance can be one vehicle width or other distances, so as to obtain a processed trajectory. Similar to the in-lane adjacent vehicle, the in-flowing adjacent vehicle and the target vehicle sometimes need to leave the exit from the same lane. In this embodiment, by offsetting the historical driving trajectory, the candidate nodes in this lane can have a smaller adjacent vehicle traffic flow cost value, ensuring the accuracy of searching for the target node.

[0129] The above process of determining the processed trajectory based on the historical driving trajectory of the adjacent vehicle has been described.

[0130] Next, the process of obtaining the reference driving area by widening the processed trajectory will be described.

[0131] In the case where the adjacent vehicle is a reverse adjacent vehicle belonging to the first subclass, the processed trajectory of the adjacent vehicle is offset equidistantly to both sides to obtain the reference driving area. For example, the processed trajectory includes multiple trajectory points, and the trajectory points represent the geometric centers of obstacles. The left and right boundary points of the obstacle corresponding to this point can be calculated based on the trajectory points, the orientation angle, and the left and right vehicle widths. Connecting the boundary points on the left boundary can obtain the left boundary of the reference driving area corresponding to the obstacle. Similarly, connecting the boundary points on the right boundary can obtain the right boundary of the reference driving area corresponding to the obstacle. The width of the reference driving area can be the same as the vehicle width or slightly larger than the vehicle width.

[0132] In the case where the adjacent vehicle is an in-lane adjacent vehicle belonging to the second subclass, the processed trajectory of the in-lane adjacent vehicle is offset equidistantly to both sides to obtain the reference driving area. The process of widening the processed trajectory of the in-lane adjacent vehicle to the left and right can refer to the processing process of the reverse adjacent vehicle. The width of the reference driving area can be the same as the vehicle width or slightly larger than the vehicle width.

[0133] In the case where the adjacent vehicle is an in-flowing adjacent vehicle belonging to the third subclass, the processed trajectory of the in-flowing adjacent vehicle is offset by a first distance to the side close to the target vehicle, and the processed trajectory of the in-flowing adjacent vehicle is offset by a second distance to the side away from the target vehicle, where the first distance is less than the second distance. The first distance is, for example, equal to half of the vehicle width, and the second distance is, for example, equal to one vehicle width.

[0134] As Figure 6As shown, this example illustrates the case where the merging vehicle drives from left to right and the target vehicle turns right. For the merging vehicle, if the processed trajectory 602 of the merging vehicle is widened equidistantly to both sides, and each side is widened by half of the vehicle width, the area 603 can be obtained. It can be seen that when the candidate node is in the area 604, at this time, it crosses the lane space occupied by the merging vehicle, which will result in a relatively small cost of the merging vehicle flow, which will cause the target node to be selected from the area 604, and then cause the searched reference line 601 to cross the driving area 603 of the merging vehicle, affecting driving safety. Therefore, this embodiment widens the side of the merging vehicle away from the target vehicle additionally to obtain a reference driving area, and the reference driving area includes the area 603 and the area 604 in the figure. In this way, the cost of the merging vehicle flow of the candidate nodes in the area 604 can be increased, so as to avoid the situation where the reference line crosses the driving route of the merging vehicle.

[0135] Next, the process of constructing the grid map will be described.

[0136] In this embodiment, the historical driving trajectory of the merging vehicle can be processed into a processed trajectory according to at least one of the sub-class of the merging vehicle and the relative position relationship between the merging vehicle and the target vehicle. Then the historical driving trajectory of the merging vehicle is widened into a reference driving area. After that, for each grid, according to the positions of the reference driving areas of at least one merging vehicle respectively and the position of the grid, the passing times are determined.

[0137] For example, the corresponding continuous coordinates in the world coordinate system are converted into the corresponding discrete coordinates in the grid coordinate system with a certain precision for subsequent queries and calculations. Then traverse the grids in the grid map, and the ray casting algorithm can be used to judge whether the target point (such as the center point of the grid) in the grid is located in a certain reference driving area in turn. The number of times the target point falls within the reference driving area is the passing times, and the passing times are denoted as where and is the coordinate of the target point in the grid, and this coordinate can be used as the identifier of the grid.

[0138] The above describes the process of constructing the grid. In the grid map constructed in the above manner, the grid corresponds to the cumulative frequency of the obstacles around the target vehicle passing through the grid. Based on the established mapping relationship between the world coordinate system and the grid coordinate system, the traffic flow density at any position around the target vehicle can be quickly queried, providing an accurate quantitative expression of the traffic flow field and providing important environmental perception information for the subsequent reference line search based on traffic flow information.

[0139] Next, taking the obstacle as a merging vehicle as an example, the cost of the merging vehicle flow will be described.

[0140] In this embodiment, when at least one obstacle includes a sidecar, a grid map is obtained, and the grid map may include multiple grids, each of the multiple grids corresponds to a pass count, and the pass count represents the number of reference driving areas passing through the grid, and the reference driving area is determined based on the historical driving trajectory of the sidecar. Then, a target grid containing a candidate node among the multiple grids may be determined, and the sidecar traffic cost may be determined based on the pass count of the target grid. For example, the coordinates of the candidate node may be mapped to the grid map of the sidecar traffic, the target grid may be determined, and then the pass count of the target grid may be queried to obtain the number of sidecar traffic passing through the target node position.

[0141] For example, when the number of passes is less than or equal to a predetermined maximum value, the cost of the traffic flow by the side car is positively correlated with the number of passes. When the number of passes is greater than the predetermined maximum value, the cost of the traffic flow by the side car is a predetermined cost value. In this way, as the number of side cars passing the position of the candidate node increases, the cost of the traffic flow by the side car will increase accordingly, thereby avoiding selecting a target node that conflicts with the traffic flow by the side car in the search. The predetermined cost value can avoid the problem that the traffic flow by the side car cost value is too large when there are too many side cars passing the position of the candidate node, which affects the search results.

[0142] For example, the sidecar flow cost function is shown in formula (9).

[0143] Formula (9)

[0144] in, is the weight of the cost of the adjacent vehicle flow, is the number of paths to the target grid, is an adjustable parameter used to control The maximum value of .

[0145] In this embodiment, the cost value of the traffic flow of the adjacent vehicles will make it possible to avoid the interaction behaviors such as squeezing and cutting in with the adjacent vehicles during the search. In addition, as the number of adjacent vehicles passing the position of the candidate node increases, the cost value of the traffic flow of the adjacent vehicles increases super exponentially, thereby avoiding the selection of the target node that conflicts with the traffic flow of the adjacent vehicles during the search. The predetermined cost value can avoid the problem that the traffic flow cost value of the adjacent vehicles is too large when there are too many adjacent vehicles passing the position of the candidate node, which affects the search results.

[0146] Figure 7 is a schematic diagram of a method for generating a road reference line according to an embodiment of the present disclosure.

[0147] Next, taking the target area as an intersection as an example, the process of generating a road reference line is explained.

[0148] In this embodiment, the first motion parameters 701 of the obstacles at the intersection can be obtained first. For example, before the target vehicle enters the intersection at a certain distance, the first motion parameters 701 of other obstacles within a certain range of the target vehicle can be observed and recorded. The first motion parameters 701 may include position coordinates, orientation angles, timestamps, vehicle widths, etc. Through the traffic flow historical information accumulated over a period of time, the historical driving trajectories of each obstacle can be obtained.

[0149] The static map information 702 and the dynamic map perception information 703 can also be obtained. For example, navigation information can be obtained from the static map information 702 (such as SD map, etc.). Since the SD map is not a high-precision map and lacks information such as the number of lane lines, the number of lanes at the intersection entrance and exit, the position of the stop line, etc. can also be read from the upstream dynamic mapping information to establish a rough topological structure in all directions of the intersection for subsequent matching with the driving trajectories of social vehicles to judge the driving intentions of the corresponding vehicles.

[0150] Next, the first motion parameters 701 of the obstacles, the static map information 702, the map perception information 703, etc. can be combined to classify the obstacles. The categories of the obstacles may include following-vehicle targets 704, side vehicles 713, and irrelevant vehicle types.

[0151] Next, based on the screening and classification results of the obstacles, a Frenet coordinate system 706 of the following-vehicle target 704 can be constructed with the historical driving trajectory of the following-vehicle target 704 as the reference, and a grid map 705 can also be constructed using the historical driving trajectory of the side vehicle 713. In actual use, to further improve the calculation efficiency and reduce the system latency, the construction and initialization of the grid map 705 can be performed on the GPU (graphics processing unit).

[0152] Next, the A* algorithm can be used to determine the search for the target node 711. The A* algorithm can specifically be the ARA* (Anytime Repairing A*) algorithm. Compared with the traditional A* algorithm, this method can quickly find an initial solution, continuously improve the quality of the solution when there is more computing time, and can balance the advantages of search speed and path quality by adjusting the weight of the heuristic function, which is suitable for scenarios where the vehicle-side computing power and time of the autonomous driving system are limited. The ARA* algorithm uses a point at a certain distance in front of the target vehicle as the search starting point, and the goal is to search for a reference line 712 passing through the intersection along the intersection navigation line fitted by the Euler spiral mentioned above. The search process can include basic trajectory cost and heuristic cost, including the somatosensory cost, obstacle avoidance cost, traffic rule cost, etc. during the search process. Based on these costs, in this embodiment, for the following vehicle target 704, the distance cost value 708 and the angle cost value 709 can be determined according to the Frenet coordinate system 706, and the passing vehicle traffic flow cost value 707 can be determined according to the grid map 705. Then, various cost values can be processed by operations such as weighted summation to obtain the total cost value 710. And based on the total cost value 710, the target node 711 is selected.

[0153] After multiple rounds of iterative search, multiple target nodes 711 can be obtained. By fitting the multiple target nodes 711, the reference line 712 of the intersection can be obtained. By splicing the reference line 712 of the intersection with the reference lines of the structured roads before and after the intersection, a complete reference line for the autonomous driving vehicle to travel can be obtained.

[0154] In this embodiment, based on the traffic flow, the search for the reference line 712 is performed, and the generated reference line 712 is consistent with the real traffic flow behavior characteristics. For example, the reference line 712 is basically the same as the real traffic flow of the following vehicle target 704, or the reference line 712 avoids the real traffic flow of the surrounding passing vehicles 713. This solution has less dependence on the road structure, can construct a reasonable reference line 712 in intersections or other target areas without lane lines and with strong uncertainty of driving trajectories, and can be used for the construction of reference lines 712 in scenarios such as going straight through intersections, turning left at intersections, and turning right at intersections. At the same time, it also has the ability to migrate and generalize to non-intersection structured roads, and can efficiently and flexibly generate a reference line 712 for autonomous driving vehicles to travel in complex intersection scenarios with a large number of social vehicle flows.

[0155] Since this solution does not completely rely on rigid rules, it can quickly adapt to changes in static and dynamic environments within the intersection, and has stronger flexibility. It has less dependence on map elements, can support different accuracy levels of maps such as high-precision maps, lane-level accuracy maps, and standard accuracy maps, and has a wider range of applications.

[0156] Figure 8It is a schematic flowchart of an autonomous driving method according to an embodiment of the present disclosure.

[0157] The present disclosure also provides an autonomous driving method, which includes operations S810 to S830.

[0158] In operation S810, a reference line of a target area is obtained. For example, the reference line of the target area can be generated by using the method for generating a road reference line described above.

[0159] In operation S820, based on the reference line of the target area, a target driving trajectory of the target vehicle is determined.

[0160] In operation S830, based on the target driving trajectory, the target vehicle is controlled to drive autonomously.

[0161] Figure 9 It is a schematic structural block diagram of a device for generating a road reference line according to an embodiment of the present disclosure.

[0162] As Figure 9 shown, the device 900 for generating an intersection reference line may include a category determination module 910, a total cost value determination module 920, and a reference line determination module 930.

[0163] The category determination module 910 is configured to determine the category of at least one obstacle located in a target area on a road according to the first motion parameter of the obstacle in a historical period, and the category of the obstacle represents the relative relationship between the obstacle and the target vehicle.

[0164] The total cost value determination module 920 is configured to, when at least one obstacle includes a target obstacle belonging to a target category, determine the total cost value of any candidate node among a plurality of candidate nodes determined according to the position of the target vehicle, according to the historical driving trajectory of the target obstacle and the second motion parameter of the candidate node.

[0165] The reference line determination module 930 is configured to determine a reference line of the target area according to a target node determined by using the total cost value of each of the plurality of candidate nodes; wherein, the reference line of the target area is used to plan the target driving trajectory of the target vehicle.

[0166] According to another embodiment of the present disclosure, when the target area is an intersection, the total cost value determination module includes: a distance cost value determination sub-module, configured to, when at least one obstacle includes a following vehicle target, determine a distance cost value according to the shortest distance from the position of the candidate node to the historical driving trajectory of the following vehicle target; wherein, the following vehicle target represents: an obstacle whose exit from the intersection is the same as the exit of the target vehicle from the intersection, and is in the same lane as the target vehicle, and is in front of the target vehicle.

[0167] According to another embodiment of the present disclosure, the target area is an intersection, and the total cost value determination module includes: an angle cost value determination sub-module, configured to determine an angle cost value according to an included angle between a direction of a candidate node and a direction of a reference point when at least one obstacle includes a following vehicle target; wherein, the following vehicle target represents an obstacle that has the same exit from the intersection as the target vehicle, is in the same lane as the target vehicle, and is in front of the target vehicle; the reference point is a point on the historical driving trajectory of the following vehicle target whose distance from the candidate node meets a predetermined condition, and the direction of the reference point is tangential to the historical driving trajectory of the following vehicle target.

[0168] According to another embodiment of the present disclosure, the target area is an intersection, and the category determination module includes: a following vehicle target determination sub-module, configured to determine an obstacle as a following vehicle target when it is determined that the obstacle meets the following conditions: at any moment within a historical period, the actual lateral distance between the position of the obstacle and the intersection navigation route is less than a target lateral distance, and the absolute value of the actual orientation angle of the obstacle is less than or equal to a target orientation angle; wherein, the intersection navigation route is determined based on the entrance of the target vehicle into the intersection and the exit from the intersection.

[0169] According to another embodiment of the present disclosure, the target lateral distance meets the following conditions: in the entrance area of the intersection, the target lateral distance is a predetermined minimum lateral distance; in the exit area of the intersection, the target lateral distance is a predetermined maximum lateral distance; in the middle area between the entrance area and the exit area of the intersection, the target lateral distance is positively correlated with the longitudinal distance, and the longitudinal distance is the distance that the obstacle moves along the intersection navigation route from the entrance of the intersection.

[0170] According to another embodiment of the present disclosure, the target orientation angle meets the following conditions: in the entrance area of the intersection, the target orientation angle is a predetermined minimum angle; in the exit area of the intersection, the target lateral distance is a predetermined maximum angle; in the middle area between the entrance area and the exit area of the intersection, the target orientation angle is positively correlated with the longitudinal distance, and the longitudinal distance is the distance that the obstacle moves along the intersection navigation route from the entrance of the intersection.

[0171] According to another embodiment of the present disclosure, the target area is an intersection, and the total cost value determination module includes: a grid map acquisition sub-module, a target grid determination sub-module, and a passing vehicle traffic flow cost value determination sub-module. The grid map acquisition sub-module is configured to acquire a grid map when at least one obstacle includes a passing vehicle. The grid map includes a plurality of grids, and each grid in the plurality of grids corresponds to a passing frequency, where the passing frequency represents the number of reference driving areas passing through the grid, and the reference driving area is determined based on the historical driving trajectory of the passing vehicle. The target grid determination sub-module is configured to determine a target grid among the plurality of grids that contains a candidate node. The passing vehicle traffic flow cost value determination sub-module is configured to determine the passing vehicle traffic flow cost value according to the passing frequency of the target grid. The passing vehicle satisfies at least one of the following conditions: the passing vehicle and the target vehicle enter the intersection through the same entrance, and the passing vehicle and the target vehicle leave the intersection through the same exit.

[0172] According to another embodiment of the present disclosure, the number of passing vehicles is at least one; the grid map is constructed by the following modules: a processing module, a broadening module, and a frequency determination module. The processing module is configured to process the historical driving trajectory of the passing vehicle into a processed trajectory according to at least one of the subclass to which the passing vehicle belongs and the relative position relationship between the passing vehicle and the target vehicle. The broadening module is configured to broaden the historical driving trajectory of the passing vehicle into a reference driving area. The frequency determination module is configured to determine the passing frequency for each grid according to the positions of the reference driving areas of at least one passing vehicle and the position of the grid.

[0173] According to another embodiment of the present disclosure, the processing module includes: a first processing sub-module and a second processing sub-module. The first processing sub-module is configured to determine the historical driving trajectory of the reverse passing vehicle as the processed trajectory when the passing vehicle is a reverse passing vehicle belonging to the first subclass; wherein, the entrance where the reverse passing vehicle enters the intersection is the same as the exit where the target vehicle leaves the intersection, and the exit where the reverse passing vehicle leaves the intersection is the same as the entrance where the target vehicle enters the intersection. The second processing sub-module is configured to shift the historical driving trajectory of the passing vehicle in a direction away from the target vehicle to obtain a processed trajectory when the passing vehicle is a same-direction passing vehicle belonging to the second subclass or an incoming passing vehicle belonging to the third subclass; the entrance where the same-direction passing vehicle enters the intersection is the same as the entrance where the target vehicle enters the intersection, and the same-direction passing vehicle and the target vehicle are in different lanes; the exit where the incoming passing vehicle leaves the intersection is the same as the exit where the target vehicle leaves the intersection.

[0174] According to another embodiment of the present disclosure, the widening module includes: a first widening sub-module and a second widening sub-module. The first widening sub-module is configured to, when the side vehicle is a reverse side vehicle belonging to the first sub-category or a same-direction side vehicle belonging to the second sub-category, equally offset the processed trajectory of the side vehicle to both sides to obtain a reference driving area; wherein, the entrance of the reverse side vehicle entering the intersection is the same as the exit of the target vehicle leaving the intersection, and the exit of the reverse side vehicle leaving the intersection is the same as the entrance of the target vehicle entering the intersection; the entrance of the same-direction side vehicle entering the intersection is the same as the entrance of the target vehicle entering the intersection, and the same-direction side vehicle and the target vehicle are in different lanes. The second widening sub-module is configured to, when the side vehicle is a merging side vehicle belonging to the third sub-category, offset the processed trajectory of the merging side vehicle by a first distance towards the side close to the target vehicle, and offset the processed trajectory of the merging side vehicle by a second distance towards the side away from the target vehicle; wherein, the first distance is less than the second distance; the exit of the merging side vehicle leaving the intersection is the same as the exit of the target vehicle leaving the intersection.

[0175] According to another embodiment of the present disclosure, when the number of passing times is less than or equal to a predetermined maximum value, the side vehicle traffic flow cost value is positively correlated with the number of passing times; when the number of passing times is greater than the predetermined maximum value, the side vehicle traffic flow cost value is a predetermined cost value.

[0176] According to another embodiment of the present disclosure, the target area is an intersection, and the category determination module includes: a side vehicle determination sub-module, configured to determine an obstacle as a side vehicle when it is determined that the obstacle satisfies the following conditions: at any moment within a period in the historical period, the actual lateral distance between the position of the obstacle and the intersection navigation route is within a predetermined distance range, and the actual orientation angle of the obstacle is within a predetermined angle range; wherein, the intersection navigation route is determined based on the entrance of the target vehicle entering the intersection and the exit of the target vehicle leaving the intersection.

[0177] According to another embodiment of the present disclosure, it further includes: a sub-category determination module, configured to determine the sub-category of the side vehicle according to the positional relationship between the entrance of the side vehicle entering the intersection and the entrance of the target vehicle entering the intersection.

[0178] Figure 10 It is a schematic structural block diagram of an automatic driving device according to an embodiment of the present disclosure.

[0179] As Figure 10 shown, the intersection automatic driving device 1000 may include an acquisition module 1010, a determination module 1020, and a control module 1030.

[0180] The acquisition module 1010 is configured to acquire a target area reference line, and the target area reference line is generated according to the above-mentioned device.

[0181] The determination module 1020 is configured to determine the target driving trajectory of the target vehicle based on the target area reference line.

[0182] The control module 1030 is configured to control the target vehicle to drive autonomously based on the target driving trajectory.

[0183] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, including at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the above method.

[0184] According to an embodiment of the present disclosure, the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute the above method.

[0185] According to an embodiment of the present disclosure, the present disclosure also provides a computer program product, including a computer program which, when executed by a processor, implements the above method.

[0186] Figure 11 It is a structural block diagram of an electronic device for implementing the method for generating a road reference line and the method for autonomous driving according to the embodiments of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as, a personal digital processor, a cellular phone, a smart phone, a wearable device, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely exemplary and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0187] As Figure 11 shown, the device 1100 includes a computing unit 1101 which can execute various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1102 or a computer program loaded from a storage unit 1108 into a random access memory (RAM) 1103. In the RAM 1103, various programs and data required for the operation of the device 1100 can also be stored. The computing unit 1101, the ROM 1102, and the RAM 1103 are connected to each other through a bus 1104. An input / output (I / O) interface 1105 is also connected to the bus 1104.

[0188] Multiple components in device 1100 are connected to I / O interface 1105, including: an input unit 1106, such as a keyboard, a mouse, etc.; an output unit 1107, such as various types of displays, speakers, etc.; a storage unit 1108, such as a magnetic disk, an optical disc, etc.; and a communication unit 1109, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 1109 allows device 1100 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0189] The computing unit 1101 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 1101 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1101 executes the various methods and processes described above, such as any of the methods described above. For example, in some embodiments, any method can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as the storage unit 1108. In some embodiments, part or all of the computer program can be loaded and / or installed onto device 1100 via the ROM 1102 and / or the communication unit 1109. When the computer program is loaded into the RAM 1103 and executed by the computing unit 1101, one or more steps of any of the methods described above can be executed. Alternatively, in other embodiments, the computing unit 1101 can be configured to execute any method by any other suitable means (e.g., by means of firmware).

[0190] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special or general-purpose programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0191] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general purpose computer, a special purpose computer, or other programmable data processing device, such that the program codes, when executed by the processor or controller, cause the functions / operations specified in the flowchart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0192] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0193] In order to provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and the input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0194] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.

[0195] A computer system can include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The client-server relationship is generated by computer programs running on the respective computers and having a client-server relationship with each other.

[0196] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in this disclosure can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and no limitation is imposed herein.

[0197] The above specific embodiments do not constitute a limitation on the protection scope of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of this disclosure shall be included within the protection scope of this disclosure.

Claims

1. A method for generating a road reference line, comprising: For at least one obstacle located in a target area of ​​the road, determining a category of the obstacle according to a first motion parameter of the obstacle in a historical period, wherein the category of the obstacle represents a relative relationship between the obstacle and the target vehicle; In a case where the at least one obstacle includes a target obstacle belonging to a target category, for any candidate node among the plurality of candidate nodes determined by the position of the target vehicle, determining a total cost value of the candidate node according to a historical driving trajectory of the target obstacle and a second motion parameter of the candidate node; A target area reference line is determined according to a target node determined by using the total cost values ​​of each of the plurality of candidate nodes; wherein the target area reference line is used to plan a target driving trajectory of the target vehicle.

2. The method according to claim 1, wherein: The target area is an intersection, and when the at least one obstacle includes a target obstacle belonging to a target category, for any candidate node among the multiple candidate nodes determined for the position of the target vehicle, determining the total cost value of the candidate node according to the historical driving trajectory of the target obstacle and the second motion parameter of the candidate node includes: In the case where the at least one obstacle includes a following vehicle target, determining a distance cost value according to the shortest distance from the position of the candidate node to the historical driving trajectory of the following vehicle target; The following vehicle target represents an obstacle that has the same exit from the intersection as the exit from the target vehicle, is in the same lane as the target vehicle, and is located in front of the target vehicle.

3. The method according to claim 1, wherein: The target area is an intersection, and when the at least one obstacle includes a target obstacle belonging to a target category, for any candidate node among the multiple candidate nodes determined for the position of the target vehicle, determining the total cost value of the candidate node according to the historical driving trajectory of the target obstacle and the second motion parameter of the candidate node includes: In the case where the at least one obstacle includes a following vehicle target, determining an angle cost value according to an angle between a direction of the candidate node and a direction of a reference point; Among them, the following target is characterized by: an exit from the intersection is the same as the exit of the target vehicle from the intersection, and is in the same lane as the target vehicle, and is an obstacle located in front of the target vehicle; the reference point is the distance between the candidate node in the historical driving trajectory of the following target and meets the predetermined conditions, and the direction of the reference point is along the tangent of the historical driving trajectory of the following target.

4. The method according to any one of claims 1 to 3, wherein: The target area is a road intersection, and for at least one obstacle located in the target area on the road, determining the category of the obstacle according to a first motion parameter of the obstacle in a historical period includes: When it is determined that the obstacle meets the following conditions, the obstacle is determined to be a following vehicle target: At any time in the historical period, the actual lateral distance between the position of the obstacle and the intersection navigation route is less than the target lateral distance, and the absolute value of the actual orientation angle of the obstacle is less than or equal to the target orientation angle; The intersection navigation route is determined based on the entrance of the target vehicle into the intersection and the exit of the intersection.

5. The method according to claim 4, wherein: The target lateral distance meets the following conditions: At the entrance area of ​​the intersection, the target lateral distance is a predetermined minimum lateral distance; At the exit area of ​​the intersection, the target lateral distance is a predetermined maximum lateral distance; In the middle area between the entrance area and the exit area of ​​the intersection, the target lateral distance is positively correlated with the longitudinal distance, and the longitudinal distance is the distance that the entrance of the intersection of the obstacle moves along the navigation route of the intersection.

6. The method according to claim 4, wherein: The target heading angle satisfies the following conditions: At the entrance area of ​​the intersection, the target orientation angle is a predetermined minimum angle; At the exit area of ​​the intersection, the target lateral distance is a predetermined maximum angle; In a middle area between an entrance area and an exit area of ​​the intersection, the target heading angle is positively correlated with a longitudinal distance, where the longitudinal distance is a distance that the obstacle moves from the entrance of the intersection along the navigation route of the intersection.

7. The method according to claim 1, wherein: The target area is an intersection, and when the at least one obstacle includes a target obstacle belonging to a target category, for any candidate node among the multiple candidate nodes determined for the position of the target vehicle, determining the total cost value of the candidate node according to the historical driving trajectory of the target obstacle and the second motion parameter of the candidate node includes: In the case where the at least one obstacle includes a vehicle next to the vehicle, a grid map is obtained, the grid map includes a plurality of grids, each of the plurality of grids corresponds to a number of passes, the number of passes representing the number of reference driving areas passing through the grid, the reference driving area being determined based on the historical driving trajectory of the vehicle next to the vehicle; Determine a target grid including the candidate node among the multiple grids; Determining a traffic cost of a side vehicle according to the number of passes through the target grid; The side vehicle satisfies at least one of the following conditions: the side vehicle and the target vehicle enter the intersection through the same entrance, and the side vehicle and the target vehicle leave the intersection through the same exit.

8. The method according to claim 7, wherein: The number of the sidecars is at least one; and the grid map is constructed in the following manner: Processing the historical driving trajectory of the side vehicle into a processed trajectory according to at least one of a subclass to which the side vehicle belongs and a relative position relationship between the side vehicle and the target vehicle; widening the historical driving trajectory of the adjacent vehicle into the reference driving area; For each of the grids, the number of passes is determined according to the position of the reference driving area of ​​at least one of the adjacent vehicles and the position of the grid.

9. The method according to claim 8, wherein: The processing of the historical driving trajectory of the side vehicle into a processed trajectory according to at least one of the subclass to which the side vehicle belongs and the relative position relationship between the side vehicle and the target vehicle comprises: In the case where the bypass vehicle is an oncoming bypass vehicle belonging to the first subcategory, the historical driving trajectory of the oncoming bypass vehicle is determined as the processed trajectory; wherein the entrance of the oncoming bypass vehicle into the intersection is the same as the exit of the target vehicle from the intersection, and the exit of the oncoming bypass vehicle from the intersection is the same as the entrance of the target vehicle into the intersection; In the case that the bystander vehicle is a same-direction bystander vehicle belonging to the second subcategory or a merging bystander vehicle belonging to the third subcategory, the historical driving trajectory of the bystander vehicle is offset in a direction away from the target vehicle to obtain the processed trajectory; the entrance of the same-direction bystander vehicle into the intersection is the same as the entrance of the target vehicle into the intersection, and the same-direction bystander vehicle and the target vehicle are in different lanes; the exit of the merging bystander vehicle from the intersection is the same as the exit of the target vehicle from the intersection.

10. The method according to claim 8, wherein: The step of widening the historical driving trajectory of the adjacent vehicle into the reference driving area comprises: In the case where the side car is an oncoming side car belonging to the first subcategory or a same-direction side car belonging to the second subcategory, the processed trajectory of the side car is shifted to both sides by equal distances to obtain the reference driving area; wherein the entrance of the oncoming side car into the intersection is the same as the exit of the target vehicle from the intersection, and the exit of the oncoming side car from the intersection is the same as the entrance of the target vehicle into the intersection; the entrance of the same-direction side car into the intersection is the same as the entrance of the target vehicle into the intersection, and the same-direction side car and the target vehicle are in different lanes; In the case that the merging vehicle is a merging vehicle belonging to the third subcategory, the processed trajectory of the merging vehicle is offset by a first distance to a side close to the target vehicle, and the processed trajectory of the merging vehicle is offset by a second distance to another side away from the target vehicle; wherein the first distance is smaller than the second distance; and an exit of the merging vehicle from the intersection is the same as an exit of the target vehicle from the intersection.

11. The method according to claim 7, wherein: When the number of passes is less than or equal to a predetermined maximum value, the sidecar traffic cost is positively correlated with the number of passes; When the number of passes is greater than the predetermined maximum value, the side vehicle traffic cost value is a predetermined cost value.

12. The method according to any one of claims 7 to 11, wherein: The target area is a road intersection, and for at least one obstacle located in the target area on the road, determining the category of the obstacle according to a first motion parameter of the obstacle in a historical period includes: When it is determined that the obstacle meets the following conditions, the obstacle is determined to be a side vehicle: At any time in a period of the historical period, the actual lateral distance between the position of the obstacle and the intersection navigation route is within a predetermined distance range, and the actual orientation angle of the obstacle is within a predetermined angle range; The intersection navigation route is determined based on the entrance of the target vehicle into the intersection and the exit of the intersection.

13. The method according to claim 12, further comprising: The subclass of the adjacent vehicle is determined according to a positional relationship between an entrance through which the adjacent vehicle enters the intersection and an entrance through which the target vehicle enters the intersection.

14. An automatic driving method, comprising: Get the target area reference line; Determining a target driving trajectory of a target vehicle based on the target area reference line; Based on the target driving trajectory, controlling the target vehicle to automatically drive; Wherein, the target area reference line is generated according to the method according to any one of claims 1 to 13.

15. A device for generating a reference line of an intersection, comprising: a category determination module, for determining, for at least one obstacle located in a target area on a road, a category of the obstacle according to a first motion parameter of the obstacle in a historical period, wherein the category of the obstacle represents a relative relationship between the obstacle and a target vehicle; a total cost value determination module, configured to determine, for any candidate node among the plurality of candidate nodes determined by the position of the target vehicle, a total cost value of the candidate node according to a historical driving trajectory of the target obstacle and a second motion parameter of the candidate node, when the at least one obstacle includes a target obstacle belonging to a target category; The reference line determination module is used to determine a target area reference line according to a target node determined by using the total cost values ​​of each of the multiple candidate nodes; wherein the target area reference line is used to plan a target driving trajectory of the target vehicle.

16. An automatic driving device at an intersection, comprising: An acquisition module, used to acquire a reference line of a target area; A determination module, used to determine a target driving trajectory of a target vehicle based on a target area reference line; A control module, used for controlling the automatic driving of the target vehicle based on the target driving trajectory; Wherein, the target area reference line is generated according to the method of claim 15.

17. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 14.

18. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 14.

19. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 14.

20. An autonomous driving vehicle comprising the electronic device of claim 17.

Images