Driving track determination method, computer device, and program product

Abstract

A driving track determination method, a computer device, and a program product. The method comprises: on the basis of historical driving tracks corresponding to a target road, determining a plurality of historical track clusters having a topological relationship, wherein each historical track cluster comprises at least one historical driving track, and the topological relationship is used for indicating whether the plurality of historical track clusters have a sequential relationship (S101); for each historical track cluster, on the basis of first pose information of track points of each historical driving track in each historical track cluster, and the topological relationship between said historical track cluster and the other historical track clusters, determining second pose information of aggregation points corresponding to said historical track cluster, wherein pose information of the last aggregation point in said historical track cluster is the same as pose information of the first aggregation point of a subsequent historical track cluster corresponding thereto (S102); and on the basis of the second pose information of aggregation points respectively corresponding to the plurality of historical track clusters, determining reference track information corresponding to the target road, wherein the reference track information is used for controlling a vehicle to travel on the target road (S103).

Description

A driving track determination method, computer device and program product

[0001] The present disclosure claims priority to a Chinese patent application No. 2024107329653, filed on June 6, 2024, and entitled "A driving track determination method, computer device and program product", the entire content of which is incorporated herein by reference. TECHNICAL FIELD

[0002] The present disclosure relates to the technical field of computer, in particular to a driving track determination method, computer device and program product. BACKGROUND

[0003] In order to solve the high cost problem brought by high-precision maps, lightweight maps have gradually been favored by major automobile manufacturers. Although the construction cost of lightweight maps is relatively low, the requirements for the perception ability and positioning ability of autonomous driving technology are also relatively high. Real-time perception data at the vehicle end, driving behavior experience track and other information are important data sources and information priors for constructing lightweight maps. Among them, the driving experience layer derived from multiple historical driving tracks can not only provide experience tracks for driving roads, but also provide more rich driving experience data such as recommended speed on curved roads or bumpy road sections, multi-lane lane selection probability, and pedestrian easy-to-cross road information, which are not possessed by traditional maps.

[0004] However, how to obtain a relatively reliable reference track for assisting driving based on multiple historical driving tracks is a problem worthy of study.

[0005] SUMMARY

[0006] The present disclosure at least provides a driving track determination method, device, computer device and program product.

[0007] In a first aspect, the present disclosure provides a driving track determination method, comprising:

[0008] Based on each historical driving track corresponding to a target road, a plurality of historical track clusters having a topological relationship are determined; at least one historical driving track is included in the historical track cluster; the topological relationship is used to indicate whether the plurality of historical track clusters have a predecessor-successor relationship;

[0009] For each historical track cluster, the second pose information of each aggregation point corresponding to the historical track cluster is determined according to the first pose information of the track points of each historical driving track in the historical track cluster and the topological relationship between the historical track cluster and other historical track clusters; wherein the last aggregation point of the historical track cluster has the same pose information as the first aggregation point of the successor historical track cluster corresponding thereto;

[0010] determine reference trajectory information corresponding to the target road according to the second pose information of each aggregation point corresponding to the plurality of historical trajectory clusters; the reference trajectory information is used to control the vehicle to travel on the target road.

[0011] In this embodiment, according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster and the topological relationship between the historical trajectory clusters, not only the aggregation of the multiple historical driving trajectories in the historical trajectory cluster can be realized, but also the trajectory common point problem between the historical trajectory clusters with the predecessor and successor topological relationship can be fully considered, so that the reference trajectories corresponding to the historical trajectory clusters with the predecessor and successor topological relationship can be connected by the common point (i.e. the last aggregation point corresponding to the predecessor historical trajectory cluster and the first aggregation point corresponding to the successor historical trajectory cluster), thereby improving the continuity and rationality of the determined reference trajectory information. By using the reference trajectory information, a reasonable and effective experience layer in the lightweight map can be constructed, and then the lightweight map is used to assist the automatic driving system to make accurate and reasonable driving behaviors.

[0012] In a possible implementation, the second pose information of each aggregation point corresponding to the historical trajectory cluster is determined according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster and the topological relationship between the historical trajectory cluster and other historical trajectory clusters, including:

[0013] determining the second pose information of the first aggregation point of the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster;

[0014] taking the first aggregation point as a current target aggregation point and taking the second pose information of the first aggregation point as a current target pose information;

[0015] determining the aggregation direction corresponding to the target aggregation point according to the target pose information, and determining the first coordinate information of a seed point corresponding to a next aggregation point according to the aggregation direction and a target step length;

[0016] determining a plurality of neighbor points of the seed point from the trajectory points of each historical driving trajectory according to the first pose information of the trajectory points of each historical driving trajectory and the first coordinate information, and determining the second pose information of the next aggregation point according to the first pose information of the plurality of neighbor points;

[0017] determining the second pose information of the last aggregation point corresponding to the historical trajectory cluster, and returning to perform the step of determining the target aggregation point corresponding to the target pose information until the second pose information of the last aggregation point corresponding to the historical trajectory cluster is obtained; wherein the plurality of neighbor points of the seed point corresponding to the last aggregation point include head points of each historical driving trajectory in a subsequent historical trajectory cluster of the historical trajectory cluster; the head point refers to a trajectory point within a preset search radius of a starting trajectory point of the historical driving trajectory.

[0018] In this embodiment, the seed point corresponding to the aggregation point and the plurality of neighbor points of the seed point are used to determine the aggregation point, so that the pose information of the plurality of trajectory points within a point range in each historical driving trajectory can be fully combined, thereby improving the accuracy and rationality of the determined aggregation point. The aggregation process of the entire historical trajectory cluster is completed based on the recursive search and aggregation of the seed point, so that the rationality and accuracy of the experience trajectory corresponding to the historical trajectory cluster obtained by aggregation are also greatly improved. Moreover, the neighbor points used when aggregating the last aggregation point of the historical trajectory cluster include the head points of each historical driving trajectory in the subsequent historical trajectory cluster, so that the last aggregation point of the previous historical trajectory cluster in the aggregation not only combines the trajectory characteristics of the historical trajectory cluster, but also fits the trajectory characteristics of the subsequent historical trajectory cluster, thereby improving the rationality of the last aggregation point obtained by aggregation.

[0019] In a possible implementation, the plurality of neighbor points of the seed point are determined from the trajectory points of each historical driving trajectory according to the first pose information of the trajectory points of each historical driving trajectory and the first coordinate information, including:

[0020] constructing an R-tree corresponding to the historical trajectory cluster according to the second coordinate information indicated by each first pose information and a preset search radius;

[0021] determining a matching point of the seed point according to the first coordinate information and the second coordinate information;

[0022] determining the plurality of neighbor points of the seed point from the R-tree by using the matching point.

[0023] In this embodiment, the R-tree corresponding to the historical trajectory cluster is constructed, so that the position relationship between the trajectory points in the historical trajectory cluster can be reflected by using the R-tree. Furthermore, when searching for the neighbor points, the neighbor points corresponding to the seed point can be quickly determined by using the R-tree, thereby improving the search efficiency of the neighbor points.

[0024] In a possible implementation, the plurality of neighbor points of the seed point are determined from the R-tree by using the matching point, including:

[0025] determine, from the R-tree, a plurality of candidate neighboring points corresponding to the seed point according to the matching points;

[0026] determine a first projection distance of each of the candidate neighboring points on a plane with the aggregation direction as a normal vector;

[0027] take the candidate neighboring point corresponding to the first projection distance satisfying the preset distribution as a neighboring point of the seed point.

[0028] In this embodiment, the first projection distance is used to screen the candidate neighboring points, so that outliers can be screened from the plurality of candidate neighboring points. Then, the outliers are deleted, so that the aggregation and effectiveness of the remaining neighboring points are improved, and a reasonable aggregation point is aggregated.

[0029] In a possible implementation, the second pose information of the first aggregation point of the historical trajectory cluster is determined according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster, and includes:

[0030] determining a first end point set and a last end point set corresponding to the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory;

[0031] in a case where the topological relationship indicates that the historical trajectory cluster has a plurality of predecessor historical trajectory clusters, adding the first pose information of each first end point in the first end point set of the historical trajectory cluster and the first pose information of each last end point in the last end point set corresponding to each of the predecessor historical trajectory clusters, performing an average operation on the added pose information to obtain average pose information;

[0032] taking the average pose information as the second pose information of the first aggregation point of the historical trajectory cluster.

[0033] In this embodiment, each first end point in the first end point set can reflect the starting position of the historical driving trajectory in the historical trajectory cluster. Each last end point in the last end point set can reflect the ending position of the historical driving trajectory in the historical trajectory cluster. For the historical trajectory cluster having a plurality of predecessor historical trajectory clusters, the pose information of the last end point corresponding to each predecessor historical trajectory cluster and the pose information of the first end point of the historical trajectory cluster are averaged, so that the obtained average pose information can reflect the pose information of the common point between the plurality of predecessor historical trajectory clusters and the historical trajectory cluster. By taking the average pose information as the second pose information of the first aggregation point of the historical trajectory cluster, the effective connection between the plurality of predecessor historical trajectory clusters and the historical trajectory cluster can be realized.

[0034] In a possible implementation, the first pose information of the trajectory points of each historical driving trajectory is used to determine the first end point set and the last end point set corresponding to the historical trajectory cluster, including:

[0035] According to the second coordinate information indicated by each first pose information and a preset search radius, an R-tree corresponding to the historical trajectory cluster is constructed.

[0036] For any historical driving trajectory in the historical trajectory cluster, the first neighbor point of the starting trajectory point of the historical driving trajectory and the second neighbor point of the ending trajectory point of the historical driving trajectory are determined according to the first pose information of the trajectory points in the historical driving trajectory and the R-tree.

[0037] The first end point set is determined according to the first neighbor points corresponding to each historical driving trajectory, and the last end point set is determined according to the second neighbor points corresponding to each historical driving trajectory.

[0038] In this implementation, each first neighbor point corresponding to the starting trajectory point of a historical driving trajectory is used as a first end point in the first end point set, and each second neighbor point corresponding to the ending trajectory point of a historical driving trajectory is used as a last end point in the last end point set. When the first end points and the last end points are used for subsequent aggregation, the influence of the division error of the historical driving trajectories on the aggregation result can be reduced.

[0039] In a possible implementation, the second pose information of the first aggregation point of the historical trajectory cluster is determined according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster, including:

[0040] In a case where the topology relationship indicates that the predecessor historical trajectory cluster of the historical trajectory cluster is one, if the second pose information of the last aggregation point corresponding to the predecessor historical trajectory cluster has been aggregated, the second pose information is used as the second pose information of the first aggregation point of the historical trajectory cluster.

[0041] If the second pose information of the last aggregation point corresponding to the predecessor historical trajectory cluster of the historical trajectory cluster has not been aggregated, or the historical trajectory cluster does not have a predecessor historical trajectory cluster, the second pose information of the first aggregation point of the historical trajectory cluster is determined according to the first pose information of the starting trajectory point of at least one historical driving trajectory in the historical trajectory cluster.

[0042] In the embodiment, in the case that the historical trajectory cluster has only one preceding historical trajectory cluster, if the preceding historical trajectory cluster has been aggregated, the continuity of the reference trajectory obtained when the historical trajectory cluster is aggregated can be improved by taking the last aggregation point corresponding to the preceding historical trajectory cluster as the first aggregation point of the historical trajectory cluster. If the last aggregation point corresponding to the preceding historical trajectory cluster has not been aggregated, the aggregation of the historical trajectory cluster can be ensured to be completed in time by aggregating the starting trajectory point of the historical driving trajectory, thereby improving the aggregation efficiency.

[0043] In a possible implementation, after the second pose information of the first aggregation point of the historical trajectory cluster is determined according to the first pose information of the starting trajectory point of at least one historical driving trajectory in the historical trajectory cluster, the method further includes:

[0044] After the second pose information of the last aggregation point corresponding to the preceding historical trajectory cluster is aggregated, it is determined whether the second pose information matches the second pose information of the first aggregation point of the historical trajectory cluster. If the second pose information does not match the second pose information of the first aggregation point of the historical trajectory cluster, the second pose information of the first aggregation point of the historical trajectory cluster is updated by using the second pose information of the last aggregation point corresponding to the preceding historical trajectory cluster.

[0045] In the embodiment, if the first aggregation point used when the historical trajectory cluster is aggregated is the starting trajectory point of a historical driving trajectory, after the preceding historical trajectory cluster is aggregated, the continuity between the reference trajectory corresponding to the preceding historical trajectory cluster and the reference trajectory corresponding to the historical trajectory cluster can be improved by updating the first aggregation point of the historical trajectory cluster by using the last aggregation point corresponding to the preceding historical trajectory cluster.

[0046] In a possible implementation, after the step of returning to execute the step of determining the aggregation direction corresponding to the target aggregation point according to the target pose information is performed until the second pose information of the last aggregation point corresponding to the historical trajectory cluster is obtained, the method further includes:

[0047] If the historical trajectory cluster is one of a plurality of preceding historical trajectory clusters corresponding to a succeeding historical trajectory cluster, in the case that the second pose information of the last aggregation point of the historical trajectory cluster is inconsistent with the mean pose information corresponding to the historical trajectory cluster, the second pose information of the last aggregation point is updated by using the mean pose information.

[0048] In this embodiment, if one historical trajectory cluster belongs to one of the plurality of preceding historical trajectory clusters corresponding to the historical trajectory cluster, the mean pose information corresponding to each preceding historical trajectory cluster is aggregated in advance. Then, when the second pose information of the last aggregation point corresponding to the historical trajectory cluster is different from the mean pose information, the second pose information of the last aggregation point is updated using the mean pose information, so that the updated last aggregation point and the succeeding historical trajectory cluster have a common point, thereby improving the continuity of the aggregated reference trajectory information.

[0049] In a possible implementation, the updating the second pose information of the last aggregation point using the mean pose information comprises:

[0050] determining a target vector according to the second pose information of the preceding aggregation point of the last aggregation point and the mean pose information;

[0051] determining a second projection distance of the last aggregation point in the direction indicated by the target vector;

[0052] in a case where the second projection distance is negative, replacing the second pose information of the last aggregation point with the mean pose information;

[0053] in a case where the second projection distance is positive, determining a projection point at the second projection distance; updating the second pose information of the last aggregation point to the mean pose information, and adding the projection point as an aggregation point between the preceding aggregation point and the last aggregation point.

[0054] In this embodiment, in a case where the second projection distance is negative, it indicates that the distance between the last aggregation point and the preceding aggregation point is relatively close, and thus the distance between the updated last aggregation point and the preceding aggregation point is reasonable by directly replacing the second pose information of the last aggregation point with the mean pose information. In a case where the second projection distance is positive, it indicates that the distance between the last aggregation point and the preceding aggregation point is relatively far, and thus the density and continuity between the aggregation points corresponding to the historical trajectory cluster are improved by adding the projection point as an aggregation point and replacing the second pose information of the last aggregation point with the mean pose information.

[0055] In a second aspect, the embodiments of the present disclosure further provide a driving trajectory determination device, comprising:

[0056] a trajectory division module, configured to determine a plurality of historical trajectory clusters having a topological relationship based on a plurality of historical driving trajectories corresponding to a target road; the historical trajectory cluster includes at least one historical driving trajectory; the topological relationship is used to indicate whether the plurality of historical trajectory clusters have a preceding and succeeding relationship;

[0057] a clustering module configured to determine, for each of the historical trajectory clusters, second pose information of each of the aggregated points corresponding to the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster and the topological relationship between the historical trajectory cluster and other historical trajectory clusters, wherein the last aggregated point of the historical trajectory cluster has the same pose information as the first aggregated point of the subsequent historical trajectory cluster corresponding thereto;

[0058] a planning module configured to determine reference trajectory information corresponding to the target road according to the second pose information of each of the aggregated points corresponding to the plurality of historical trajectory clusters respectively, wherein the reference trajectory information is used to control the vehicle to travel on the target road.

[0059] In a third aspect, an optional implementation of the present disclosure further provides a computer device, a processor and a memory, the memory stores machine readable instructions executable by the processor, and the processor is configured to execute the machine readable instructions stored in the memory, and the machine readable instructions are executed by the processor to execute the steps of the first aspect or any possible implementation manner of the first aspect.

[0060] In a fourth aspect, an optional implementation of the present disclosure further provides a computer program product, which, when executed, performs the steps of the first aspect or any possible implementation manner of the first aspect.

[0061] For the effects of the driving trajectory determination apparatus, the computer device and the computer program product, refer to the description of the driving trajectory determination method, which will not be repeated here.

[0062] In order to make the above objectives, characteristics and advantages of the present disclosure more apparent, the following will specifically describe a preferred embodiment in conjunction with the accompanying drawings, and make a detailed description as follows. BRIEF DESCRIPTION OF DRAWINGS

[0063] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following will briefly introduce the drawings needed to be used in the embodiments, the drawings herein are incorporated into the description and form a part of the description, which show the embodiments consistent with the present disclosure, and are used to explain the technical solutions of the present disclosure together with the description. It should be understood that the following drawings only show some embodiments of the present disclosure, and therefore should not be considered as a limitation to the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor.

[0064] FIG. 1 shows a flowchart of a driving trajectory determination method provided by an embodiment of the present disclosure;

[0065] FIG. 2 shows a schematic diagram of clustering historical trajectories according to an embodiment of the present disclosure;

[0066] FIG. 3 shows a schematic diagram of target reference trajectories obtained by clustering historical trajectories corresponding to a target road according to an embodiment of the present disclosure; wherein FIG. 3-1 shows target reference trajectories corresponding to each historical trajectory cluster, and FIG. 3-2 shows target reference trajectories corresponding to each historical trajectory cluster;

[0067] FIG. 4 shows a schematic diagram of trajectory merging points, trajectory separating points, optimized merging points, and optimized separating points according to an embodiment of the present disclosure;

[0068] FIG. 5 shows a schematic diagram of a driving trajectory determination apparatus according to an embodiment of the present disclosure;

[0069] FIG. 6 shows a schematic diagram of a computer device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0070] In order to make the objects, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be described clearly and completely below with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments. The components of the embodiments of the present disclosure described and shown herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure is not intended to limit the scope of the claimed present disclosure, but only represents selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present disclosure.

[0071] In addition, the terms "first", "second", and the like in the specification and claims of the embodiments of the present disclosure and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein.

[0072] As referred to herein, "a plurality" or "a number of means two or more. "And / or", describing the association relationship between the associated objects, means that there can be three relationships, for example, A and / or B, which can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. The character " / " generally represents that the front and rear associated objects are in an "or" relationship.

[0073] It is found through research that the accuracy of lightweight map construction affects the safety of autonomous driving. The construction accuracy of the experience layer in the lightweight map for reflecting the experience trajectory on the road affects the rationality and accuracy of driving decisions when autonomous driving on the road. Generally, the experience trajectory can be determined according to various historical driving trajectories on the road. The traditional determination method of the experience trajectory can include the following methods: ①, the time-related driving experience trajectory extraction method, which obtains the experience trajectory of the driver passing through the intersection in different time periods by griding the road network in the space-time coordinate system. ②, the point-to-point driving experience trajectory extraction method, which generates the best trajectory from point to point in the road network by counting the frequency of trajectories in the road network. ③, the method of extracting experience trajectory by using historical driving trajectories of high-income taxi drivers, which weights the income of the driver, and the higher the income of the driver, the greater the weight of the driver in the generation of the driving experience trajectory, thereby obtaining the best experience trajectory at the intersection. However, the above methods all lack attention to the correlation between the historical driving trajectories, resulting in poor accuracy and rationality of the experience trajectory, which is difficult to adapt to the lightweight map as the experience layer, affecting the construction accuracy of the experience layer.

[0074] Based on the above research, the present disclosure provides a driving trajectory determination method, device, computer equipment and program product. According to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster and the topological relationship between the historical trajectory clusters, not only the aggregation of multiple historical driving trajectories in the historical trajectory cluster can be realized, but also the trajectory common point problem between the historical trajectory clusters with the predecessor and successor topological relationship can be fully considered, so that the reference trajectories corresponding to the historical trajectory clusters with the predecessor and successor topological relationship aggregated respectively can be connected by the common point (i.e. the last aggregated point corresponding to the predecessor historical trajectory cluster and the first aggregated point corresponding to the successor historical trajectory cluster), thereby improving the continuity and rationality of the determined reference trajectory information. Using the reference trajectory information, a reasonable and effective experience layer in the lightweight map can be constructed, and then the lightweight map is used to assist the autonomous driving system to make accurate and reasonable driving behavior.

[0075] The defects of the above scheme are the result of the inventors' practice and careful research, therefore, the discovery process of the above problems and the solutions proposed by the present disclosure to solve the above problems should be the contribution of the inventors to the present disclosure in the process of the present disclosure.

[0076] It should be noted that similar reference numerals and letters refer to similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0077] It can be understood that, before using the technical solutions disclosed in the embodiments of the present disclosure, the type of personal information involved in the present disclosure, the use range, the use scenario, etc. should be informed to the user and the authorization of the user should be obtained through appropriate means according to relevant laws and regulations.

[0078] It should be noted that the specific terms mentioned in the embodiments of the present disclosure include:

[0079] An R-tree is a spatial index data structure used to organize and manage data composed of two-dimensional or higher-dimensional regions, such as points in a multidimensional space, irregular multidimensional shapes, etc. The basic idea of the R-tree is to divide the spatial data according to the multidimensional space, each node corresponds to a region and a disk page, the disk page of the non-leaf node stores the region range of all its child nodes, and the region of all child nodes of the non-leaf node falls within its region range; the disk page of the leaf node stores the circumscribed rectangle of all spatial objects within its region range.

[0080] In order to facilitate the understanding of the present embodiment, first, a driving track determination method disclosed by the present embodiment is described in detail. The execution subject of the driving track determination method provided by the present embodiment is generally a terminal device or other processing device with certain computing power, wherein the terminal device can be a user equipment (UE), a mobile device, a user terminal, a terminal, a personal digital assistant device (PDA), a handheld device, a computer device, etc. In some possible implementation manners, the driving track determination method can be realized by a processor calling computer readable instructions stored in a memory.

[0081] The driving track determination method provided by the present embodiment is described below taking a computer device as an example.

[0082] As shown in FIG. 1, a flowchart of a driving track determination method provided by the present embodiment can include the following steps:

[0083] S101: Based on each historical driving track corresponding to a target road, a plurality of historical track clusters having a topological relationship are determined; at least one historical driving track is included in the historical track cluster; the topological relationship is used to indicate whether the plurality of historical track clusters have a preceding and following relationship.

[0084] Here, the target road can be any road for vehicle driving, and the historical driving trajectory can be a segment of a complete driving trajectory generated by any vehicle on the target road. The complete driving trajectory can be a continuous driving trajectory of the vehicle. For any complete driving trajectory generated by a vehicle on the target road, the complete driving trajectory can be composed of historical driving trajectories corresponding to the vehicle on the target road. The historical driving trajectories corresponding to the target road can be obtained by dividing at least one complete driving trajectory generated by at least one vehicle.

[0085] For example, the complete driving trajectory generated by the vehicle 1 on the target road can include the trajectory A and the trajectory B, the complete driving trajectory generated by the vehicle 2 on the target road can include the trajectory C, the trajectory D and the trajectory E, and the complete driving trajectory generated by the vehicle 3 on the target road can include the trajectory F. For the trajectory A, the historical driving trajectories A-1, A-2 and A-3 corresponding to the target road can be obtained by division; for the trajectory B, the historical driving trajectories B-1 and B-2 corresponding to the target road can be obtained by division; for the trajectory C, the historical driving trajectories C-1, C-2 and C-3 corresponding to the target road can be obtained by division; for the trajectory D, the historical driving trajectories D-1, D-2 and D-3 corresponding to the target road can be obtained by division; for the trajectory E, the historical driving trajectories E-1 and E-2 corresponding to the target road can be obtained by division; and for the trajectory F, the historical driving trajectories F-1 and F-2 corresponding to the target road can be obtained by division.

[0086] The historical driving trajectories in the historical trajectory cluster can include at least one historical driving trajectory with the same driving intention and associated lane, where the associated lane can be the same lane or adjacent lane. The historical driving trajectories included in one historical trajectory cluster belong to different complete driving trajectories respectively. For example, the historical driving trajectories A-1, B-1, C-1 and D-1 can be included in the historical trajectory cluster 1, and the historical driving trajectories E-1 and F-1 can be included in the historical trajectory cluster 2. The number of historical driving trajectories included in different historical trajectory clusters can be the same or different, which is not limited in the embodiments of the present application.

[0087] The plurality of historical trajectory clusters can have a topological relationship, and the topological relationship is used to indicate whether the historical trajectory clusters have a predecessor-successor relationship. When the two historical trajectory clusters have a topological relationship, it can be indicated that the two historical trajectory clusters have a predecessor relationship and a successor relationship. The predecessor relationship can indicate a predecessor historical trajectory cluster of a historical trajectory cluster, and each historical driving trajectory in the predecessor historical trajectory cluster can be considered as a predecessor trajectory of each historical driving trajectory in the historical trajectory cluster. The lane corresponding to each historical driving trajectory in the predecessor historical trajectory cluster can be considered as a predecessor lane corresponding to each historical driving trajectory in the historical trajectory cluster. The successor relationship is used to indicate a successor historical trajectory cluster of a historical trajectory cluster, and each historical driving trajectory in the successor historical trajectory cluster can be considered as a successor trajectory of each historical driving trajectory in the historical trajectory cluster. The lane corresponding to each historical driving trajectory in the successor historical trajectory cluster can be considered as a successor lane corresponding to each historical driving trajectory in the historical trajectory cluster.

[0088] A group of associated lanes (i.e., one lane or a group of adjacent lanes) can correspond to a historical trajectory cluster, and each historical driving trajectory in the historical trajectory cluster can be a trajectory generated on the lane corresponding to the historical trajectory cluster. The topological relationship between the historical trajectory clusters can be determined according to the lane-level matching of the trajectory points and the road, that is, can be determined according to the topological relationship between the lanes corresponding to the historical trajectory clusters. For example, if the lane Lane A is the predecessor of the lanes Lane B and Lane C , then the successor historical trajectory cluster of the historical trajectory cluster A associated with the lane Lane A is the historical trajectory cluster B associated with the lane Lane B and the historical trajectory cluster C associated with the lane Lane C .

[0089] One historical trajectory cluster can have a topological relationship with at least one other historical trajectory cluster, for example, one historical trajectory cluster can have one or more predecessor historical trajectory clusters, and one historical trajectory cluster can have one or more successor historical trajectory clusters. The plurality of historical trajectory clusters can form a historical trajectory group, and each historical driving trajectory corresponding to the historical trajectory group can be combined into each complete driving trajectory corresponding to the target road.

[0090] As shown in FIG. 2, a diagram of dividing a historical trajectory cluster is provided in the embodiment of the present application, and seven complete driving trajectories (i.e., trajectories 1-7) are included in FIG. 2. Among them, the historical trajectory cluster obtained by division includes g1-g6, and g1-g6 form a historical trajectory group. Different historical trajectory clusters are identified by different dashed boxes. Among them, the g1 trajectory cluster includes seven historical driving trajectories, the successor trajectory cluster of g1 includes g2 and g4, that is, the predecessor trajectory cluster of g2 and g4 is g1, the g2 trajectory cluster includes one historical driving trajectory corresponding to each of trajectories 3-7, and the g4 trajectory cluster includes one historical driving trajectory corresponding to trajectory 1 and one historical driving trajectory corresponding to trajectory 2. The successor trajectory cluster of g4 includes g5, that is, the predecessor trajectory cluster of g5 is g4, and the g5 trajectory cluster includes one historical driving trajectory corresponding to each of trajectories 1 and 2. The successor trajectory cluster of g2 includes g3 and g6, that is, the predecessor trajectory cluster of g3 and g6 is g2, the g3 trajectory cluster includes one historical driving trajectory corresponding to each of trajectories 6 and 7, and the g6 trajectory cluster includes one historical driving trajectory corresponding to each of trajectories 3-5.

[0091] It can be understood that if each historical driving trajectory included in a historical trajectory cluster is the last trajectory corresponding to the target road of the complete driving trajectory, the historical trajectory cluster can have no successor historical trajectory cluster. Similarly, if each historical driving trajectory included in a historical trajectory cluster is the starting trajectory corresponding to the target road of the complete driving trajectory, the historical trajectory cluster can have no predecessor historical trajectory cluster.

[0092] In specific implementation, one historical trajectory cluster can correspond to a unique cluster identifier (i.e., cluster id), and the topological relationship can indicate the cluster id of the predecessor historical trajectory cluster corresponding to the historical trajectory cluster and / or the cluster id of the successor historical trajectory cluster corresponding to the historical trajectory cluster. Meanwhile, each historical driving trajectory in the historical trajectory cluster can also have a trajectory identifier (i.e., trajectory id). The predecessor historical trajectory cluster of a historical trajectory cluster is also the predecessor historical trajectory cluster of the historical driving trajectory in the historical trajectory cluster. The successor historical trajectory cluster of a historical trajectory cluster is also the successor historical trajectory cluster of the historical driving trajectory in the historical trajectory cluster. The topological relationship can also indicate the cluster id of the predecessor historical trajectory cluster corresponding to the trajectory id and / or the cluster id of the successor historical trajectory cluster corresponding to the trajectory id. Optionally, if a historical driving trajectory has no predecessor and successor driving trajectory, the trajectory id of the historical driving trajectory can be set to a default value.

[0093] Each trajectory point in the historical driving trajectory can have first pose information, which is the ego-vehicle pose information of the pose point collected in the driving process of the historical driving trajectory. The first pose information can specifically include coordinate information of the trajectory point, orientation information of the trajectory point, speed information corresponding to the trajectory point, positioning state information, etc. The coordinate information of the trajectory point can include coordinate information of the trajectory point in the longitude-latitude-height coordinate system and / or coordinate information of the trajectory point in the East-North-Up coordinate system (ENU).

[0094] In specific implementation, each complete driving trajectory corresponding to the target road can be divided to obtain each historical driving trajectory. Then, based on the driving intention of each historical driving trajectory, the topological relationship between the trajectory points in the historical driving trajectory and the road, a plurality of historical trajectory clusters and the topology between the historical trajectory clusters are determined. The plurality of historical trajectory clusters form a historical trajectory group G={group i , i=1, 2, …, M}, where the value of M is determined according to the number of the divided historical trajectory clusters. group i , i=1, 2, …, M} indicates the i-th historical trajectory cluster, which is composed of a plurality of historical driving trajectories with the same driving intention and associated lanes, and can be described as group i ={traj j , j=0, 1, 2, N}, where the value of N is determined according to the number of historical driving trajectories included in the historical trajectory cluster, and traj j indicates the j+1-th historical driving trajectory in group i .

[0095] Alternatively, a topological grouping algorithm in the prior art can be directly used to obtain a plurality of historical trajectory clusters with topological relationships based on each historical driving trajectory corresponding to the target road.

[0096] S102: For each historical trajectory cluster, second pose information of each aggregation point corresponding to the historical trajectory cluster is determined according to first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster and the topological relationship between the historical trajectory cluster and other historical trajectory clusters; wherein the last aggregation point of the historical trajectory cluster has the same pose information as the first aggregation point of the successor historical trajectory cluster corresponding thereto.

[0097] Here, the aggregation point is a trajectory point aggregated for the historical trajectory cluster, which is usually generated based on the existing trajectory points included in each historical driving trajectory in the historical trajectory cluster. Each aggregation point corresponding to a historical trajectory cluster is connected, which can form a reference trajectory corresponding to the historical trajectory cluster. The first pose information is the pose information of the trajectory point in the historical driving trajectory, and the second pose information is the pose information of the aggregation point.

[0098] The first aggregation point is the first trajectory point on the reference trajectory corresponding to the historical trajectory cluster, and the last aggregation point is the last trajectory point on the reference trajectory corresponding to the historical trajectory cluster.

[0099] In specific implementation, each historical trajectory cluster can be traversed. For the currently traversed historical trajectory cluster, the predecessor historical trajectory cluster and / or the successor historical trajectory cluster of the historical trajectory cluster can be determined according to the topological relationship between the historical trajectory cluster and other historical trajectory clusters. Then, the first pose information of the trajectory points on each historical driving trajectory included in the historical trajectory cluster is used to perform trajectory fitting on the historical driving trajectories, to obtain the second pose information of each aggregation point corresponding to the historical trajectory cluster. Then, for the first aggregation point of the historical trajectory cluster, if the historical trajectory cluster has a predecessor historical trajectory cluster, the first aggregation point can be updated by using the last aggregation point corresponding to the predecessor historical trajectory cluster, to ensure the continuity between the historical trajectory cluster and the predecessor historical trajectory cluster. That is, the pose information of the first aggregation point of the historical trajectory cluster is consistent with the pose information of the last aggregation point of the predecessor historical trajectory cluster corresponding to the historical trajectory cluster. For the last aggregation point of the historical trajectory cluster, if the historical trajectory cluster has a successor historical trajectory cluster, the last aggregation point can be updated by using the first aggregation point corresponding to the successor historical trajectory cluster, to ensure the continuity between the historical trajectory cluster and the successor historical trajectory cluster. That is, the pose information of the last aggregation point of the historical trajectory cluster is consistent with the pose information of the first aggregation point of the successor historical trajectory cluster corresponding to the historical trajectory cluster.

[0100] In this way, after the traversal of each historical trajectory cluster is completed, the second pose information of each aggregation point corresponding to each historical trajectory cluster can also be obtained. Moreover, since the topological relationship between the historical trajectory clusters is used when the aggregation points are determined, the trajectory common point problem of the two adjacent historical trajectory clusters can be fully considered when the trajectory aggregation is performed, and each aggregation point with continuity can be obtained.

[0101] S103: Determine reference trajectory information corresponding to the target road according to the second pose information of each aggregation point corresponding to each historical trajectory cluster; the reference trajectory information is used to control the vehicle to travel on the target road.

[0102] Here, the reference trajectory information can represent optimal driving state information planned on the target road, where the optimal driving state information can include a reference trajectory planned on the target road and information such as a recommended vehicle speed, a recommended lane selection, etc. at each road position. The reference trajectory information can be matched with an experience layer of a lightweight map, so as to be used to indicate information such as an experience trajectory and an experience speed in the experience layer. The reference trajectory information can be used to control the driving behavior of the vehicle when the vehicle performs automatic driving on the target road.

[0103] In actual implementation, according to the topological relationship between the historical trajectory clusters, the at least one target reference trajectory corresponding to the target road can be obtained by sequentially connecting the aggregated points corresponding to each historical trajectory cluster according to the second pose information. Alternatively, the reference trajectory corresponding to each historical trajectory cluster can be obtained by aggregating the aggregated points corresponding to each historical trajectory cluster according to the second pose information of each aggregated point. Then, the topological relationship between the historical trajectory clusters is integrated into the topological relationship of the reference trajectory corresponding to the historical trajectory cluster, and each reference trajectory is connected and smoothed according to the topological relationship of the reference trajectory, to obtain the at least one target reference trajectory corresponding to the target road. Meanwhile, according to the orientation information, speed information, etc. in the second pose information, the recommended speed, recommended orientation, etc. corresponding to each aggregated point can be determined. According to the lane corresponding to the aggregated point in each reference trajectory, the recommended lane, etc. can be determined. In this way, the reference trajectory information corresponding to the target road can be obtained.

[0104] In this way, according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster and the topological relationship between the historical trajectory clusters, not only the aggregation of the multiple historical driving trajectories in the historical trajectory cluster can be realized, but also the trajectory common point problem between the historical trajectory clusters with the topological relationship of predecessor and successor can be fully considered, so that the reference trajectories corresponding to the historical trajectory clusters with the topological relationship of predecessor and successor can be connected by the common point (i.e. the last aggregated point corresponding to the predecessor historical trajectory cluster and the first aggregated point corresponding to the successor historical trajectory cluster), thereby improving the continuity and rationality of the determined reference trajectory information. Moreover, when the multiple historical driving trajectories in the historical trajectory cluster are aggregated, the second pose information of each aggregated point is relied on. Since the pose information can not only reflect the position of the aggregated point, but also represent the speed, orientation, positioning state, etc. of the aggregated point, the reference trajectory information obtained by using the second pose information of each aggregated point can represent the experience trajectory and experience speed, etc. corresponding to the target road. In this way, the aggregated reference trajectory information can be adapted to the lightweight map to construct an accurate and reasonable experience layer, thereby assisting the automatic driving system to make accurate and reasonable driving behaviors.

[0105] In an embodiment, S102 can be implemented according to the following steps:

[0106] S102-1: According to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster, the second pose information of the first aggregated point of the historical trajectory cluster is determined.

[0107] For example, for any historical trajectory cluster, when determining the first aggregation point corresponding to the historical trajectory cluster, the first pose information of the starting trajectory point corresponding to each historical driving trajectory can be determined according to the first pose information of the trajectory points on each historical driving trajectory, and then the second pose information of the first aggregation point of the historical trajectory cluster can be obtained by performing an averaging operation on the first pose information of each starting trajectory point.

[0108] S102-2: The first aggregation point is taken as the current target aggregation point, and the second pose information of the first aggregation point is taken as the current target pose information.

[0109] S102-3: The aggregation direction corresponding to the target aggregation point is determined according to the target pose information, and the first coordinate information of the seed point corresponding to the next aggregation point is determined according to the aggregation direction and the target step length.

[0110] Here, the target step length is a step length that is pre-set according to experience, and the step length can affect the distance between adjacent aggregation points obtained by aggregation. The seed point is a point used when determining the next aggregation point, and the first coordinate information can be the coordinate information of the seed point in the longitude-latitude-height coordinate system and / or the coordinate information of the seed point in the ENU.

[0111] The pose information can include orientation information of the point, and the orientation information can specifically include a translation part and a rotation part corresponding to the pose of the point. In specific implementation, the rotation part in the pose can be determined according to the target pose information. The x-axis coordinate in the ego vehicle coordinate system is determined according to the rotation part, and then a vector of the x-axis coordinate in the world coordinate system is determined, and the direction represented by the vector is taken as the aggregation direction. That is, the aggregation direction can be calculated according to the rotation part in the target pose information, and can be represented by the vector of the x-axis of the ego vehicle coordinate system in the world coordinate system at this moment. Then, the first coordinate information of the seed point corresponding to the next aggregation point of the target aggregation point can be obtained by advancing the target step length from the first target aggregation point according to the aggregation direction.

[0112] S102-4: A plurality of neighbor points of the seed point are determined from the trajectory points of each historical driving trajectory according to the first pose information and the first coordinate information of the trajectory points of each historical driving trajectory, and the second pose information of the next aggregation point is determined according to the first pose information of the plurality of neighbor points.

[0113] Here, the neighbor points of the seed point are the trajectory points adjacent to the seed point included in each historical driving trajectory. The neighbor points of the seed point are located within a preset search radius of the seed point.

[0114] In specific implementation, the second coordinate information of each trajectory point can be determined according to the first pose information of the trajectory points of each historical driving trajectory. Then, the distance between each trajectory point and the seed point can be determined according to the first coordinate information and the second coordinate information. Then, each neighbor point of the seed point can be selected from each trajectory point according to the distance corresponding to each trajectory point and the preset distance. Then, the first pose information of each neighbor point can be processed by mean operation, and the obtained pose information can be taken as the second pose information of the next aggregation point.

[0115] S102-5: taking the next aggregation point as a new target aggregation point, and taking the second pose information of the next aggregation point as new target pose information, and returning to perform the step of determining the aggregation direction corresponding to the target aggregation point according to the target pose information until the second pose information of the last aggregation point corresponding to the historical trajectory cluster is obtained; wherein the plurality of neighbor points of the seed point corresponding to the last aggregation point include the first end points of each historical driving trajectory in the subsequent historical trajectory cluster of the historical trajectory cluster; the first end point refers to a trajectory point located within a preset search radius of the starting trajectory point of the historical driving trajectory.

[0116] Optionally, the first end point can directly refer to each trajectory point located within a preset search radius of the starting trajectory point of the historical driving trajectory, or refer to a trajectory point located within a preset search radius of the starting trajectory point of the historical driving trajectory and satisfying a preset condition, wherein the preset condition can be that the projection distance on a plane with the aggregation direction as the normal vector satisfies a preset distribution.

[0117] In specific implementation, after the second pose information of the next aggregation point corresponding to the first aggregation point of the historical trajectory cluster (the next aggregation point is the second aggregation point) is determined according to the steps of S102-2 to S102-4, the next aggregation point of the second aggregation point can be searched. Therefore, the newly obtained second aggregation point can be taken as a new target aggregation point, and the second pose information of the newly obtained second aggregation point can be taken as new target pose information. Then, S102-3 and S102-5 are executed to obtain the second pose information corresponding to the third aggregation point, the fourth aggregation point, …, and the last aggregation point of the historical trajectory cluster, respectively. In this way, the second pose information of each aggregation point corresponding to the historical trajectory cluster can be accurately determined in sequence by using the seed point through recursive search.

[0118] In the embodiment, the second pose information of the second aggregation point is determined according to the second pose information of the first aggregation point, and the second pose information of the third aggregation point is determined according to the second pose information of the second aggregation point, and so on, to obtain all the aggregation points of a historical trajectory cluster. That is, for any aggregation point of a historical trajectory cluster except the first aggregation point, the second pose information of the aggregation point is determined according to the second pose information of the last aggregation point of the aggregation point.

[0119] It should be noted that after the penultimate aggregation point corresponding to the historical trajectory cluster is obtained by using the above steps, the second pose information of the penultimate aggregation point can be used to determine the aggregation direction, and the first coordinate information of the seed point corresponding to the last aggregation point (for the convenience of description, the seed point corresponding to the last aggregation point is referred to as the last seed point hereinafter) can be determined according to the aggregation direction and the target step. Then, in order to improve the continuity of the last aggregation point of the historical trajectory cluster and the subsequent historical trajectory cluster of the historical trajectory cluster, the first pose information of each trajectory point in the historical trajectory cluster can be used to determine each third neighbor point of the last seed point from each trajectory point. At the same time, each subsequent historical trajectory cluster of the historical trajectory cluster can be determined, and then the first end point of each historical driving trajectory in each subsequent historical trajectory cluster can be determined as the fourth neighbor point of the last seed point. Finally, each third neighbor point and each fourth neighbor point can be used as the neighbor point of the last seed point, and then the second pose information of the last aggregation point can be determined according to the pose information of the neighbor points. That is, the neighbor points of the last seed point include the trajectory points in the historical trajectory cluster and the trajectory points in the subsequent historical trajectory cluster.

[0120] The first end point can be each neighbor point of the starting trajectory point of the historical driving trajectory. Specifically, for any historical trajectory cluster, the starting trajectory point corresponding to each historical driving trajectory in the historical trajectory cluster can be determined, and then each neighbor point of the starting trajectory point can be determined according to the pose information of the starting trajectory point and the pose information of the other trajectory points of each historical driving trajectory in the historical trajectory cluster.

[0121] If the next seed point of a seed point does not exist in any neighbor point in the R-tree corresponding to the historical trajectory cluster, the seed point can be determined as the last seed point corresponding to the historical trajectory cluster, and the aggregation point corresponding to the seed point can be determined as the last aggregation point to be aggregated corresponding to the historical trajectory cluster.

[0122] In an embodiment, for S102-1, the following steps can be implemented:

[0123] S102-1-1: Determine the first end point set and the last end point set corresponding to the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory.

[0124] Here, one historical trajectory cluster corresponds to one start point set and one end point set, the start point set stores respective start points corresponding to the historical trajectory cluster, and the end point set stores respective end points corresponding to the historical trajectory cluster. For example, the start point set can be represented by Map {id, start_poses}, where id is the cluster id of the historical trajectory cluster, and start_poses includes respective start points; and the end point set can be represented by Map {id, end_poses}, where id is the cluster id of the historical trajectory cluster, and end_poses includes respective end points. The start point can be a neighbor of the start trajectory point of the historical driving trajectory, and the end point can be a neighbor of the end trajectory point of the historical driving trajectory.

[0125] In specific implementation, for any historical trajectory cluster, for each historical driving trajectory in the historical trajectory cluster, the start point and the end point corresponding to the historical driving trajectory can be determined according to the first pose information of respective pose points in the historical driving trajectory, by searching according to the distance between the trajectory points. After obtaining the start point and the end point corresponding to each historical driving trajectory in the historical trajectory cluster, the respective start points can be saved in the start point set Map {id, start_poses} corresponding to the historical trajectory cluster after deduplication, and the respective end points can be saved in the end point set Map {id, end_poses} corresponding to the historical trajectory cluster after deduplication.

[0126] In one embodiment, S102-1-1 can be implemented according to the following steps:

[0127] S102-1-1-1: Construct an R-tree corresponding to the historical trajectory cluster according to the second coordinate information indicated by the respective first pose information and the preset search radius.

[0128] Here, one historical trajectory cluster corresponds to one R-tree, and the preset search radius is the shortest distance from the rectangular edge of the circumscribed rectangle corresponding to each trajectory point in the R-tree to the trajectory point. The second coordinate information can be the ENU coordinate in the first pose information.

[0129] In specific implementation, for any historical trajectory cluster, the R-tree corresponding to the historical trajectory cluster can be constructed according to the ENU coordinate in the first pose information of all trajectory points in the historical trajectory cluster. Since the search range of the R-tree cannot be represented by coordinate information alone, a rectangular frame within the preset search radius of the trajectory point is also needed to be set as the search range of the trajectory point when constructing the R-tree.

[0130] S102-1-1-2: For any historical driving trajectory in the historical trajectory cluster, first neighbor points of a starting trajectory point of the historical driving trajectory and second neighbor points of an ending trajectory point of the historical driving trajectory are determined according to the first pose information of each trajectory point in the historical driving trajectory and the R-tree.

[0131] Here, the first neighbor points are each neighbor point corresponding to the starting trajectory point, and the second neighbor points are each neighbor point corresponding to the ending trajectory point.

[0132] In specific implementation, for any historical driving trajectory in the historical trajectory cluster, the starting trajectory point and the ending trajectory point corresponding to the historical driving trajectory can be determined according to the first pose information of each trajectory point in the historical driving trajectory. Then, according to the first pose information of the starting trajectory point, a trajectory point matching the starting trajectory point can be searched from the R-tree, and each trajectory point in the rectangular frame corresponding to the trajectory point is taken as the first neighbor point of the starting trajectory point. The trajectory point matching the starting trajectory point can be a point consistent with the starting trajectory point. Similarly, according to the first pose information of the ending trajectory point, a trajectory point matching the ending trajectory point can be searched from the R-tree, and each trajectory point in the rectangular frame corresponding to the trajectory point is taken as the second neighbor point of the ending trajectory point.

[0133] S102-1-1-3: A first end point set is determined according to the first neighbor points corresponding to each historical driving trajectory, and an ending point set is determined according to the second neighbor points corresponding to each historical driving trajectory.

[0134] In specific implementation, after obtaining the first neighbor points corresponding to each historical driving trajectory in the historical trajectory cluster, a first end point set corresponding to the historical trajectory cluster can be obtained after deduplication of each first neighbor point. Similarly, after obtaining the second neighbor points corresponding to each historical driving trajectory in the historical trajectory cluster, an ending point set corresponding to the historical trajectory cluster can be obtained after deduplication of each second neighbor point.

[0135] S102-1-2: In a case where the topological relationship indicates that the historical trajectory cluster has a plurality of predecessor historical trajectory clusters, the first pose information of each first end point in the first end point set of the historical trajectory cluster and the first pose information of each ending point in the ending point set corresponding to each predecessor historical trajectory cluster are added, and a mean value operation is performed on the added pose information to obtain mean value pose information.

[0136] In a specific implementation, if it is determined according to the topological relationship that the historical trajectory cluster has multiple preceding historical trajectory clusters, in order to improve the continuity between the first aggregated point and each preceding historical trajectory cluster, the first pose information of each first end point in the first end point set of the historical trajectory cluster and the first pose information of each last end point in the last end point set corresponding to each preceding historical trajectory cluster can be determined first. Then, the first pose information of each first end point and the first pose information of each last end point are added, and the result obtained by the addition is subjected to a mean value operation to obtain a mean pose information. In the mean pose information, the coordinate information is obtained by taking the mean value of the coordinate information in each first pose information, and the orientation information, the speed information, etc. in the mean pose information are also obtained by taking the mean value of the corresponding information in each first pose information.

[0137] S102-1-3: Taking the mean pose information as the second pose information of the first aggregated point of the historical trajectory cluster.

[0138] In a specific implementation, the mean pose information can be taken as the second pose information of the first aggregated point of the historical trajectory cluster, and the mean pose information can also be taken as the second pose information of the last aggregated point of each preceding historical trajectory cluster. In this way, the first aggregated point of the historical trajectory cluster and the last aggregated point of each preceding historical trajectory cluster can be ensured to be consistent, thereby ensuring the continuity of the reference trajectory obtained by aggregation.

[0139] Optionally, after obtaining each historical trajectory cluster, a historical trajectory cluster having multiple preceding historical trajectory clusters can be determined according to the topological relationship. Then, for such a historical trajectory cluster, the first aggregated point of the historical trajectory cluster and the last aggregated point of each preceding historical trajectory cluster of the historical trajectory cluster can be aggregated in advance according to the steps of S102-1-2 and S102-1-3 described above. That is, the first aggregated points of multiple pairs of trajectory clusters (i.e., multiple preceding historical trajectory clusters corresponding to one succeeding historical trajectory cluster) are aggregated in advance. Then, when aggregating each historical trajectory cluster, if the historical trajectory cluster corresponds to multiple preceding historical trajectory clusters, the first aggregated point aggregated in advance can be directly taken as the first aggregated point of the historical trajectory cluster.

[0140] In another embodiment, for S102-1 described above, if the historical trajectory cluster has only one preceding historical trajectory cluster or has no preceding historical trajectory cluster, the first aggregated point can also be determined according to the following steps:

[0141] In the case where the topological relationship indicates that the preceding historical trajectory cluster of the historical trajectory cluster is one, if the second pose information of the last aggregated point corresponding to the preceding historical trajectory cluster has been aggregated, the second pose information is taken as the second pose information of the first aggregated point of the historical trajectory cluster.

[0142] In a specific implementation, if the topological relationship of the historical trajectory cluster currently in need of aggregation indicates that there is a predecessor historical trajectory cluster of the historical trajectory cluster, it can be determined whether the last aggregation point corresponding to the predecessor historical trajectory cluster is completed, that is, whether the predecessor historical trajectory cluster has been aggregated. If yes, the second pose information of the last aggregation point of the predecessor historical trajectory cluster that has been aggregated can be taken as the second pose information of the first aggregation point of the historical trajectory cluster. In this way, the historical trajectory cluster and the predecessor historical trajectory cluster can have a common point.

[0143] On the contrary, if the last aggregation point corresponding to the predecessor historical trajectory cluster is not completed, that is, the predecessor historical trajectory cluster is not aggregated, in order to ensure the aggregation efficiency of the historical trajectory cluster, the second pose information of the first aggregation point of the historical trajectory cluster can be determined according to the first pose information of the starting trajectory point of at least one historical driving trajectory in the historical trajectory cluster. For example, the first pose information of the starting trajectory point of any historical driving trajectory in the historical trajectory cluster can be taken as the second pose information of the first aggregation point of the historical trajectory cluster. Alternatively, the first pose information of the starting trajectory point of each historical driving trajectory in the historical trajectory cluster can be averaged to obtain the second pose information of the first aggregation point of the historical trajectory cluster.

[0144] In addition, if it is determined according to the topological relationship of the historical trajectory cluster that there is no predecessor historical trajectory cluster of the historical trajectory cluster, the second pose information of the first aggregation point of the historical trajectory cluster can also be determined according to the first pose information of the starting trajectory point of at least one historical driving trajectory in the historical trajectory cluster. For example, any historical driving trajectory can be selected from the historical driving trajectories corresponding to the historical trajectory cluster, and the first pose information of the starting trajectory point of the selected historical driving trajectory can be taken as the second pose information of the first aggregation point of the historical trajectory cluster. Alternatively, the first pose information of the starting trajectory point corresponding to each historical driving trajectory in the historical trajectory cluster can be averaged, and the obtained pose information can be taken as the second pose information of the first aggregation point of the historical trajectory cluster.

[0145] In an embodiment, if the historical trajectory cluster has a preceding historical trajectory cluster, and the first pose information of the first trajectory point of at least one historical driving trajectory in the historical trajectory cluster is used to determine the first aggregated point of the historical trajectory cluster, in order to improve the continuity between the reference trajectory of the preceding historical trajectory cluster and the reference trajectory of the historical trajectory cluster, after the second pose information of the last aggregated point of the preceding historical trajectory cluster is determined, it is further determined whether the second pose information of the last aggregated point matches the second pose information of the first aggregated point of the historical trajectory cluster. That is, after the last aggregated point of the preceding historical trajectory cluster is determined, it is determined whether the pose information of the first aggregated point of the historical trajectory cluster matches the pose information of the last aggregated point.

[0146] If the two pose information match, that is, the pose information of the two aggregated points is consistent, the first aggregated point does not need to be updated. If the two pose information do not match, the second pose information of the last aggregated point of the preceding historical trajectory cluster is used to update the second pose information of the first aggregated point of the historical trajectory cluster. For example, the second pose information of the last aggregated point of the preceding historical trajectory cluster is directly used as the second pose information of the first aggregated point of the historical trajectory cluster, so that the reference trajectories corresponding to the two historical trajectory clusters match at a point.

[0147] In an embodiment, the step of determining the plurality of neighboring points of the seed point in S102-4 can be implemented according to the following steps:

[0148] S102-4-1: Constructing an R-tree corresponding to the historical trajectory cluster according to the second coordinate information indicated by the first pose information and the preset search radius.

[0149] For the specific implementation process of this step, please refer to S102-1-1-1 described above, which will not be repeated here.

[0150] S102-4-2: Determining the matching point of the seed point according to the first coordinate information and the second coordinate information.

[0151] Here, in the case where there is a trajectory point in the R-tree with coordinate information consistent with the seed point, the matching point can be the trajectory point in the R-tree with coordinate information consistent with the seed point. In the case where there is no trajectory point in the R-tree with coordinate information consistent with the seed point, the matching point can be the point with coordinate information closest to the first coordinate information of the seed point, or a new node can be added in the R-tree according to the first coordinate information, and the new node is used as the matching point.

[0152] In implementation, whether there is a trajectory point with the same coordinate information as the seed point can be determined according to the first coordinate information of the seed point and the second coordinate information of each trajectory point in the R-tree. If yes, the point is taken as the matching point of the seed point. If no, the distance between each trajectory point and the seed point can be determined, and the trajectory point with the shortest distance is taken as the matching point of the seed point. Or, in the case where there is no trajectory point with the same coordinate information as the seed point, the node added in the R-tree using the first coordinate information is taken as the matching point.

[0153] S102-4-3: determining a plurality of neighbor points of the seed point from the R-tree using the matching point.

[0154] In implementation, the rectangular frame corresponding to the matching point in the R-tree can be determined, each trajectory point located in the rectangular frame is retrieved from the R-tree, and the neighbor points of the seed point are determined according to the trajectory points. For example, each trajectory point located in the rectangular frame corresponding to the matching point can be taken as the neighbor point of the seed point.

[0155] In the formula, if the matching point is the added node, the rectangular frame corresponding to the added node in the R-tree can be constructed using the preset radius and the second coordinate information indicated by each first pose information when the added node is added. Then, the neighbor points of the seed point are determined according to each trajectory point located in the rectangular frame.

[0156] In an embodiment, S102-4-3 can be implemented according to the following steps:

[0157] S102-4-3-1: determining a plurality of candidate neighbor points corresponding to the seed point from the R-tree using the matching point.

[0158] For example, each trajectory point located in the rectangular frame corresponding to the matching point can be taken as the candidate neighbor point.

[0159] S102-4-3-2: determining the first projection distance of each candidate neighbor point on the plane with the aggregation direction as the normal vector.

[0160] In implementation, after each candidate neighbor point is determined, the plane can be constructed with the aggregation direction used when the seed point is determined as the normal vector, and then the projection distance of each candidate neighbor point on the plane is determined according to the first pose information of each candidate neighbor point, and the projection distance is taken as the first projection distance of the candidate neighbor point.

[0161] S102-4-3-3: taking the candidate neighbor point with the corresponding first projection distance satisfying the preset distribution as the neighbor point of the seed point.

[0162] In a specific implementation, the preset distribution can be a Gaussian distribution. After obtaining the first projection distance of each candidate neighboring point, it can be determined whether the first projection distance exceeds the 3σ range of the Gaussian distribution. If not, the candidate neighboring point is determined to be an outlier and is removed. If it does not exceed the 3σ range, the candidate neighboring point is regarded as an effective neighboring point corresponding to the seed point.

[0163] In an embodiment, after obtaining the second pose information of the last aggregated point corresponding to the historical trajectory cluster, the following steps can be used to determine whether the last aggregated point needs to be updated:

[0164] Step A: If the historical trajectory cluster is one of the multiple predecessor historical trajectory clusters corresponding to a successor historical trajectory cluster, and the second pose information of the last aggregated point of the historical trajectory cluster is inconsistent with the mean pose information corresponding to the historical trajectory cluster, the mean pose information is used to update the second pose information of the last aggregated point.

[0165] Here, for any historical trajectory cluster that needs to be aggregated at present, if the historical trajectory cluster has a successor historical trajectory cluster, and the successor historical trajectory cluster has other predecessor historical trajectory clusters in addition to the historical trajectory cluster, the mean pose information obtained by early aggregation corresponding to the historical trajectory cluster can also be obtained.

[0166] That is, if a historical trajectory cluster is one of the multiple predecessor historical trajectory clusters of a certain historical trajectory cluster, when performing S102-1-2 and S102-1-3, the mean pose information corresponding to the historical trajectory cluster will be determined, and the mean pose information will be used as the second pose information of the last aggregated point corresponding to the historical trajectory cluster.

[0167] Then, after determining the second pose information of the last aggregated point (described as the last aggregated point obtained by recursive search) corresponding to the historical trajectory cluster according to S102-5, it can be determined whether the second pose information is consistent with the mean pose information corresponding to the historical trajectory cluster. If it is consistent, the second pose information of the historical trajectory cluster does not need to be updated. If it is not consistent, the mean pose information can be used to update the second pose information, so that the updated last aggregated point corresponding to the historical trajectory cluster can be consistent with the first aggregated point of the subsequent historical trajectory cluster. For example, the second pose information of the last aggregated point obtained by recursive search can be directly updated to the mean pose information corresponding to the historical trajectory cluster, thereby updating the last aggregated point corresponding to the historical trajectory cluster.

[0168] In an embodiment, the step of updating the second pose information in step A can be implemented according to the following steps:

[0169] S1: determining a target vector according to second pose information of a previous aggregation point of the last aggregation point and mean pose information.

[0170] In implementation, a previous aggregation point of the last aggregation point obtained by the recursive search can be determined, and then a target vector can be calculated according to second pose information of the previous aggregation point and mean pose information corresponding to the historical trajectory cluster.

[0171] S2: determining a second projection distance of the last aggregation point in a direction indicated by the target vector.

[0172] In implementation, the target vector exists in a direction as a vector, and after the target vector is determined, a projection distance of the last aggregation point obtained by the recursive search in the direction indicated by the target vector can be determined according to second pose information of the last aggregation point obtained by the recursive search, and the projection distance can be taken as the second projection distance corresponding to the last aggregation point obtained by the recursive search.

[0173] S3: in a case where the second projection distance is negative, replacing the second pose information of the last aggregation point with the mean pose information; or in a case where the second projection distance is positive, determining a projection point at the second projection distance; updating the second pose information of the last aggregation point to the mean pose information, and taking the projection point as a new aggregation point added between the previous aggregation point and the last aggregation point.

[0174] In implementation, after the second projection distance is obtained, in a case where the second projection distance is negative, the second pose information of the last aggregation point obtained by the recursive search can be directly updated to the mean pose information corresponding to the historical trajectory cluster, that is, the last aggregation point obtained by the recursive search is discarded, and an aggregation point corresponding to the mean pose information is taken as the last aggregation point of the historical trajectory cluster. In this way, the last aggregation point obtained by the recursive search can be updated to the aggregation point corresponding to the mean pose information.

[0175] On the contrary, in a case where the second projection distance is positive, pose information of a projection point corresponding to the second projection distance can be determined according to the second pose information of the last aggregation point obtained by the recursive search, then the last aggregation point obtained by the recursive search can be discarded, the aggregation point corresponding to the mean pose information is taken as the last aggregation point of the historical trajectory cluster, and the projection point is added between the previous aggregation point and the updated last aggregation point as a new supplementary aggregation point.

[0176] For example, the last aggregation point corresponding to the recursive search of the historical trajectory cluster is denoted as aggregation point A, the aggregation point indicated by the mean pose information corresponding to the historical trajectory cluster is denoted as aggregation point B, the point at the second projection distance corresponding to the aggregation point A is denoted as point C, and the previous aggregation point of the aggregation point A is denoted as aggregation point D. If the second projection distance is negative, the aggregation point A is discarded and the aggregation point B is taken as the last aggregation point corresponding to the historical trajectory cluster. If the second projection distance is positive, the aggregation point A is discarded and the aggregation point B is taken as the last aggregation point corresponding to the historical trajectory cluster, and the point C is taken as a newly added aggregation point between the aggregation point B and the aggregation point D.

[0177] In an embodiment, for S103 described above, the following steps can be implemented:

[0178] S103-1: determining initial reference trajectories corresponding to each historical trajectory cluster and topological relations between the initial reference trajectories according to the second pose information of each aggregation point corresponding to each historical trajectory cluster and the topological relations between the historical trajectory clusters.

[0179] In implementation, for each historical trajectory cluster, the initial reference trajectory corresponding to the historical trajectory cluster can be determined according to the second pose information of each aggregation point corresponding to the historical trajectory cluster. Meanwhile, the topological relation between two historical trajectory clusters can be taken as the topological relation between the initial reference trajectories corresponding to the two historical trajectory clusters.

[0180] For example, if the historical trajectory cluster 1 is a predecessor historical trajectory cluster of the historical trajectory clusters 2 and 3, the initial reference trajectory 1 corresponding to the historical trajectory cluster 1 is also a predecessor reference trajectory of the initial reference trajectory 2 corresponding to the historical trajectory cluster 2 and the initial reference trajectory 3 corresponding to the historical trajectory cluster 3.

[0181] S103-2: determining reference trajectory information corresponding to the target road according to the initial reference trajectories and the topological relations between the initial reference trajectories.

[0182] In implementation, the initial reference trajectories can be connected and trajectory smoothing processing can be performed according to the topological relations between the initial reference trajectories, so as to obtain each target reference trajectory corresponding to the target road. Meanwhile, the planning speed, planning orientation and other information of each aggregation point on the target reference trajectory can be determined according to the second pose information of each aggregation point on the target reference trajectory. In this way, the reference trajectory information corresponding to the target road can be determined according to each target reference trajectory and the planning speed, planning orientation and other information of each aggregation point on each target reference trajectory.

[0183] Thus, according to the second pose information of each aggregation point corresponding to the historical trajectory cluster, the initial reference trajectory corresponding to the historical trajectory cluster can be fitted. Through the topological relationship between the historical trajectory clusters, the topological relationship between the initial reference trajectories can be obtained. By using the topological relationship and the initial reference trajectory, reasonable reference trajectory information corresponding to the target road can be obtained.

[0184] For example, for a historical trajectory group G = {group i , i = 1, 2, …, M}, by using the driving trajectory determination method provided in the above embodiments, the target reference trajectory T = {traj i , i = 1, 2, …, M} can be aggregated.

[0185] As shown in FIG. 3, it is a schematic diagram of the target reference trajectory obtained by aggregating each historical driving trajectory corresponding to the target road according to an embodiment of the present application. In FIG. 3, 3-1 is a schematic diagram of each historical driving trajectory, wherein each point on the historical driving trajectory is a trajectory point on the historical driving trajectory. In FIG. 3-1, the target reference trajectory corresponding to each historical trajectory cluster is shown, but in order to more clearly show the target reference trajectory corresponding to each historical trajectory cluster, FIG. 3-2 separately shows the target reference trajectory corresponding to each historical trajectory cluster, and a line in FIG. 3-2 is a target reference trajectory. In FIG. 3-1 and FIG. 3-2, since the picture is processed in grayscale, the color is also shown in grayscale.

[0186] In an embodiment, for S103-2, the following steps can also be implemented:

[0187] S103-2-1: According to the initial reference trajectory corresponding to each historical trajectory cluster, a trajectory separation point and a trajectory merging point are screened out; wherein the trajectory separation point is the last aggregation point on the initial reference trajectory in the case that the initial reference trajectory corresponds to multiple subsequent reference trajectories; and the trajectory merging point is the first aggregation point on the initial reference trajectory in the case that the initial reference trajectory corresponds to multiple previous reference trajectories.

[0188] In specific implementation, according to the initial reference trajectory corresponding to each historical trajectory cluster and the topological relationship between the initial reference trajectories, each initial reference trajectory with a topological relationship can be connected to obtain each candidate reference trajectory corresponding to the target road. Then, according to the candidate reference trajectory and the topological relationship between the candidate reference trajectories, each trajectory separation point and each trajectory merging point are screened out from the candidate reference trajectories.

[0189] S103-2-2: determining an optimized separation point corresponding to the trajectory separation point according to a distance change trend between the historical trajectory cluster and the trajectory separation point, and determining an optimized merging point corresponding to the trajectory merging point according to a distance change trend between the historical trajectory cluster and the trajectory merging point.

[0190] Here, the distance change trend is used to represent a change trend of the distance between each historical driving trajectory corresponding to the historical trajectory cluster and the trajectory separation point, and a change trend of the distance between the trajectory merging point.

[0191] In specific implementation, a first distance between the trajectory separation point and each historical driving trajectory can be determined, a distance change trend between the historical trajectory cluster and the trajectory separation point is determined according to the first distance, and second pose information of the trajectory separation point is optimized according to the distance change trend to obtain the optimized separation point. Similarly, a second distance between the trajectory merging point and each historical driving trajectory can be determined, a distance change trend between the historical trajectory cluster and the trajectory merging point is determined according to the second distance, and second pose information of the trajectory merging point is optimized according to the distance change trend to obtain the optimized merging point.

[0192] It can be understood that there can be trajectory separation points and / or trajectory merging points that do not need to be optimized, and such points do not have corresponding optimized points, or the corresponding optimized points are the points themselves.

[0193] As shown in FIG. 4, it is a schematic diagram of a trajectory merging point, a trajectory separation point, an optimized merging point and an optimized separation point provided by an embodiment of the present application, wherein each white trajectory is a historical driving trajectory, a trajectory point corresponding to a square indicated by a solid line is a trajectory merging point or a trajectory separation point before optimization, and a trajectory point corresponding to a square indicated by a dashed line with a horizontal line in the middle is an optimized merging point or an optimized separation point obtained after optimization. In FIG. 4, one trajectory merging point that does not need to be optimized is shown.

[0194] S103-2-3: re-dividing each historical driving trajectory by using the optimized separation point and the optimized merging point to obtain a plurality of new historical trajectory clusters having a topological relationship.

[0195] In specific implementation, after obtaining each optimized separation point and each optimized merging point, each historical driving trajectory can be re-divided by using the optimized separation point and each optimized merging point and a topological relationship between the historical driving trajectory to obtain a plurality of new historical trajectory clusters having a topological relationship.

[0196] S103-2-4: Return to execute the step of determining the second pose information of each aggregation point corresponding to the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster, and the topological relationship between the historical trajectory cluster and other historical trajectory clusters, until the number of cycles reaches the preset number, to obtain each target reference trajectory and the corresponding target topological relationship.

[0197] For example, after obtaining a plurality of new historical trajectory clusters, the above S102 and S103 can be executed again until the number of cycles reaches the preset number, or the distance between the optimization merging point and the trajectory merging point determined in the current cycle is less than the preset distance, and the distance between the optimization separation point and the trajectory separation point determined in the current cycle is less than the preset distance. The latest obtained each candidate reference trajectory is taken as the target reference trajectory, and the topological relationship of each candidate reference trajectory is taken as the target topological relationship of each target reference trajectory.

[0198] S103-2-5: Determine the reference trajectory information corresponding to the target road according to each target reference trajectory and the target topological relationship.

[0199] For the specific implementation process of this step, please refer to the implementation process of S103-2 described above, which will not be repeated here.

[0200] In this way, the initial reference trajectory can be used to re-divide the historical trajectory cluster, and then the new historical trajectory cluster obtained by re-dividing can be re-aggregated, which can improve the rationality of the initial reference trajectory obtained by re-aggregation compared with the initial reference trajectory obtained by the last aggregation. In this way, through repeated aggregation, the rationality and accuracy of the target reference trajectory obtained can be improved.

[0201] To facilitate understanding of the driving trajectory determination method provided by the embodiments of the present disclosure, a specific embodiment is described below:

[0202] For the historical trajectory group G = {group i , i = 1, 2, …, M}, group i = {traj j , j = 0, 1, 2, N}:

[0203] Step 1, traverse each historical trajectory cluster in the historical trajectory group G, for each historical trajectory cluster, the R-tree corresponding to the historical trajectory cluster can be constructed according to the second coordinate information of the first pose indication of the trajectory points of each historical driving trajectory and the preset search radius. Then for any historical driving trajectory in the historical trajectory cluster, the first neighbor point of the starting trajectory point of the historical driving trajectory and the second neighbor point of the ending trajectory point of the historical driving trajectory are determined according to the first pose information of each trajectory point in the historical driving trajectory and the R-tree; the first end point set Map {id, start_poses} corresponding to the historical trajectory cluster is determined according to the first neighbor point of each historical driving trajectory, and the second end point set Map {id, end_poses} corresponding to the historical trajectory cluster is determined according to the second neighbor point of each historical driving trajectory.

[0204] Step 2, aggregate the first aggregation point of the multi-to-one trajectory cluster in advance: traverse the historical trajectory group G, for the historical trajectory cluster X with multiple predecessor historical trajectory clusters, add the first pose information of each first end point in the first end point set of the historical trajectory cluster X and the first pose information of each second end point in the second end point set corresponding to each predecessor historical trajectory cluster, and take the average of the added pose information to obtain the average pose information. The average pose information is taken as the second pose information of the first aggregation point of the historical trajectory cluster X.

[0205] Step 3, for any historical trajectory cluster, the first aggregation point of the historical trajectory cluster is determined. In the case that the historical trajectory cluster has multiple predecessor historical trajectory clusters, the first aggregation point is the point corresponding to the average pose information aggregated in step 2.

[0206] In the case that the historical trajectory cluster has one predecessor historical trajectory cluster, if the second pose information of the last aggregation point corresponding to the predecessor historical trajectory cluster has been aggregated, the second pose information is taken as the second pose information of the first aggregation point of the historical trajectory cluster.

[0207] If the second pose information of the last aggregation point corresponding to the predecessor historical trajectory cluster of the historical trajectory cluster has not been aggregated, the first pose information of the starting trajectory point of any historical driving trajectory in the historical trajectory cluster is taken as the second pose information of the first aggregation point of the historical trajectory cluster. And after the last aggregation point corresponding to the predecessor historical trajectory cluster is aggregated, if the last aggregation point corresponding to the predecessor historical trajectory cluster is inconsistent with the last aggregation point recursively searched corresponding to the historical trajectory cluster, the last aggregation point corresponding to the historical trajectory cluster is updated to the last aggregation point corresponding to the predecessor historical trajectory cluster.

[0208] If the historical trajectory cluster does not exist a preceding historical trajectory cluster, the first pose information of the start trajectory point of any historical driving trajectory in the historical trajectory cluster is taken as the second pose information of the first aggregation point of the historical trajectory cluster.

[0209] Step 4, taking the first aggregation point as a target aggregation point, and taking the second pose information of the first aggregation point as target pose information; determining an aggregation direction according to the first pose information of the target aggregation point, and determining the first coordinate information of the seed point corresponding to the next aggregation point according to the aggregation direction and a target step length, searching for a plurality of candidate neighbor points corresponding to the seed point by using the R-tree corresponding to the historical trajectory cluster, and determining the first projection distance of each candidate neighbor point on a plane with the aggregation direction as a normal vector. The candidate neighbor point corresponding to the first projection distance satisfying a preset distribution is taken as a neighbor point of the seed point.

[0210] Step 5, performing an averaging operation on the first pose information of the plurality of neighbor points to obtain the second pose information of the next aggregation point.

[0211] Step 6, taking the next aggregation point as a new target aggregation point, and taking the second pose information of the next aggregation point as new target pose information, and returning to execute step 4 until the last seed point corresponding to the last aggregation point of the historical trajectory cluster is obtained, determining each third neighbor point of the last seed point by using the R-tree corresponding to the historical trajectory cluster, and taking each first end point in the first end point set of each subsequent historical trajectory cluster of the historical trajectory cluster as a fourth neighbor point of the last seed point. Adding the first pose information of each third neighbor point and the first pose information of each fourth neighbor point, and performing an averaging operation on the added pose information to obtain the second pose information of the last aggregation point. If the historical trajectory cluster is one of the plurality of preceding historical trajectory clusters corresponding to a subsequent historical trajectory cluster, in the case that the second pose information of the last aggregation point obtained by the recursive search of the historical trajectory cluster is inconsistent with the mean pose information corresponding to the historical trajectory cluster, a target vector is determined according to the second pose information of the preceding aggregation point of the last aggregation point obtained by the recursive search and the mean pose information. The second projection distance of the last aggregation point obtained by the recursive search in the direction indicated by the target vector is determined. In the case that the second projection distance is negative, the second pose information of the last aggregation point obtained by the recursive search is replaced by the mean pose information; or in the case that the second projection distance is positive, a projection point at the second projection distance is determined; the second pose information of the last aggregation point obtained by the recursive search is updated to the mean pose information, and the projection point is taken as an aggregation point newly added between the preceding aggregation point and the last aggregation point.

[0212] Step 7, determining initial reference trajectories corresponding to each historical trajectory cluster and topological relations between the initial reference trajectories according to second position information of each aggregation point corresponding to each historical trajectory cluster and the topological relations between the historical trajectory clusters.

[0213] Step 8, screening out trajectory separation points and trajectory merging points according to the initial reference trajectories and the topological relations corresponding to each historical trajectory cluster; determining optimized separation points corresponding to the trajectory separation points according to distance variation trends between the historical trajectory clusters and the trajectory separation points, and determining optimized merging points corresponding to the trajectory merging points according to distance variation trends between the historical trajectory clusters and the trajectory merging points; re-dividing each historical driving trajectory by using the optimized separation points and the optimized merging points to obtain a plurality of new historical trajectory clusters having topological relations; returning to execute steps 1-7 until a preset number of cycles is reached to obtain target reference trajectories and corresponding target topological relations.

[0214] Step 9, determining reference trajectory information corresponding to a target road according to the target reference trajectories and the corresponding target topological relations.

[0215] For the specific implementation process of steps 1-9, refer to the above embodiments, which will not be repeated here.

[0216] Those skilled in the art can understand that in the above method of the specific implementation, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the implementation process, and the specific execution order of each step should be determined by its function and possible internal logic.

[0217] Based on the same inventive concept, the disclosure embodiments also provide a driving trajectory determination device corresponding to the driving trajectory determination method. Since the principle of solving problems by the device in the disclosure embodiments is similar to the above-mentioned driving trajectory determination method of the disclosure embodiments, the implementation of the device can be referred to the implementation of the method, and the repeated parts will not be repeated.

[0218] As shown in FIG. 5, a schematic diagram of a driving trajectory determination device provided by the disclosure embodiments includes:

[0219] The trajectory division module 501 is configured to determine a plurality of historical trajectory clusters having topological relations based on each historical driving trajectory corresponding to a target road; the historical trajectory cluster includes at least one historical driving trajectory; and the topological relation is used to indicate whether the plurality of historical trajectory clusters have a front-back succession relation.

[0220] The aggregation module 502 is configured to determine, for each of the historical trajectory clusters, second position information of each aggregation point corresponding to the historical trajectory cluster according to first position information of the trajectory points of each historical driving trajectory in the historical trajectory cluster and topological relations between the historical trajectory cluster and other historical trajectory clusters; wherein the last aggregation point of the historical trajectory cluster has the same position information as the first aggregation point of a subsequent historical trajectory cluster corresponding to the historical trajectory cluster.

[0221] The planning module 503 is configured to determine reference trajectory information corresponding to the target road according to the second position information of each aggregation point corresponding to each of the historical trajectory clusters; the reference trajectory information is used to control the vehicle to travel on the target road.

[0222] In a possible implementation, when determining the second position information of each aggregation point corresponding to the historical trajectory cluster according to the first position information of the trajectory points of each historical driving trajectory in the historical trajectory cluster and the topological relations between the historical trajectory cluster and other historical trajectory clusters, the aggregation module 502 is configured to:

[0223] determine the second position information of the first aggregation point of the historical trajectory cluster according to the first position information of the trajectory points of each historical driving trajectory in the historical trajectory cluster;

[0224] take the first aggregation point as a current target aggregation point and take the second position information of the first aggregation point as current target position information;

[0225] determine an aggregation direction corresponding to the target aggregation point according to the target position information, and determine first coordinate information of a seed point corresponding to a next aggregation point according to the aggregation direction and a target step length;

[0226] determine a plurality of neighbor points of the seed point from the trajectory points of each historical driving trajectory according to the first position information of the trajectory points of each historical driving trajectory and the first coordinate information, and determine the second position information of the next aggregation point according to the first position information of the plurality of neighbor points;

[0227] take the next aggregation point as a new target aggregation point and take the second position information of the next aggregation point as new target position information, and return to perform the step of determining the aggregation direction corresponding to the target aggregation point according to the target position information until the second position information of a last aggregation point corresponding to the historical trajectory cluster is obtained; wherein the plurality of neighbor points of the seed point corresponding to the last aggregation point include first end points of each historical driving trajectory in a subsequent historical trajectory cluster of the historical trajectory cluster; the first end point refers to a trajectory point located within a preset search radius of a starting trajectory point of the historical driving trajectory.

[0228] In a possible implementation, the aggregation module 502, when determining the plurality of neighbor points of the seed point from the trajectory points of each historical driving trajectory according to the first pose information of the trajectory points of each historical driving trajectory and the first coordinate information, is configured to:

[0229] construct an R-tree corresponding to the historical trajectory cluster according to the second coordinate information indicated by each first pose information and a preset search radius;

[0230] determine the matching point of the seed point according to the first coordinate information and the second coordinate information;

[0231] determine the plurality of neighbor points of the seed point from the R-tree by using the matching point.

[0232] In a possible implementation, the aggregation module 502, when determining the plurality of neighbor points of the seed point from the R-tree by using the matching point, is configured to:

[0233] determine the plurality of candidate neighbor points corresponding to the seed point from the R-tree by using the matching point;

[0234] determine a first projection distance of each candidate neighbor point on a plane with the aggregation direction as a normal vector;

[0235] take the candidate neighbor point corresponding to the first projection distance satisfying a preset distribution as the neighbor point of the seed point.

[0236] In a possible implementation, the aggregation module 502, when determining the second pose information of the first aggregation point of the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster, is configured to:

[0237] determine a head point set and a tail point set corresponding to the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory;

[0238] in a case where the topological relationship indicates that the historical trajectory cluster has a plurality of predecessor historical trajectory clusters, add the first pose information of each head point in the head point set of the historical trajectory cluster and the first pose information of each tail point in the tail point set corresponding to each predecessor historical trajectory cluster, respectively, and perform an average operation on the added pose information to obtain mean pose information;

[0239] take the mean pose information as the second pose information of the first aggregation point of the historical trajectory cluster.

[0240] In a possible implementation, the aggregation module 502, when determining the first-end point set and the last-end point set corresponding to the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory, is configured to:

[0241] construct an R-tree corresponding to the historical trajectory cluster according to the second coordinate information indicated by each first pose information and a preset search radius;

[0242] for any historical driving trajectory in the historical trajectory cluster, determine the first neighbor point of the starting trajectory point of the historical driving trajectory and the second neighbor point of the ending trajectory point of the historical driving trajectory according to the first pose information of each trajectory point in the historical driving trajectory and the R-tree;

[0243] determine the first-end point set according to the first neighbor points corresponding to each historical driving trajectory, and determine the last-end point set according to the second neighbor points corresponding to each historical driving trajectory.

[0244] In a possible implementation, the aggregation module 502, when determining the second pose information of the first aggregation point of the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster, is configured to:

[0245] if the predecessor historical trajectory cluster of the historical trajectory cluster is one, and the second pose information of the last aggregation point corresponding to the predecessor historical trajectory cluster has been aggregated, the second pose information is taken as the second pose information of the first aggregation point of the historical trajectory cluster;

[0246] if the second pose information of the last aggregation point corresponding to the predecessor historical trajectory cluster of the historical trajectory cluster has not been aggregated, or the historical trajectory cluster does not have a predecessor historical trajectory cluster, the second pose information of the first aggregation point of the historical trajectory cluster is determined according to the first pose information of the starting trajectory point of at least one historical driving trajectory in the historical trajectory cluster.

[0247] In a possible implementation, after the aggregation module 502 determines the second pose information of the first aggregation point of the historical trajectory cluster according to the first pose information of the starting trajectory point of at least one historical driving trajectory in the historical trajectory cluster, the aggregation module 502 is further configured to:

[0248] after the second pose information of the last aggregation point corresponding to the predecessor historical trajectory cluster is aggregated, it is determined whether the second pose information matches the second pose information of the first aggregation point of the historical trajectory cluster; if not, the second pose information of the first aggregation point of the historical trajectory cluster is updated by using the second pose information of the last aggregation point corresponding to the predecessor historical trajectory cluster.

[0249] In a possible implementation, the aggregation module 502, after returning to perform the step of determining the aggregation direction corresponding to the target aggregation point according to the target pose information until the second pose information of the last aggregation point corresponding to the historical trajectory cluster is obtained, is further configured to:

[0250] If the historical trajectory cluster is one of the plurality of preceding historical trajectory clusters corresponding to a subsequent historical trajectory cluster, and the second pose information of the last aggregation point of the historical trajectory cluster is inconsistent with the mean pose information corresponding to the historical trajectory cluster, the mean pose information is used to update the second pose information of the last aggregation point.

[0251] In a possible implementation, the aggregation module 502, when updating the second pose information of the last aggregation point by using the mean pose information, is configured to:

[0252] determine a target vector according to the second pose information of a preceding aggregation point of the last aggregation point and the mean pose information;

[0253] determine a second projection distance of the last aggregation point in the direction indicated by the target vector;

[0254] if the second projection distance is negative, replace the second pose information of the last aggregation point by using the mean pose information;

[0255] if the second projection distance is positive, determine a projection point at the second projection distance; update the second pose information of the last aggregation point to the mean pose information, and add the projection point as an aggregation point between the preceding aggregation point and the last aggregation point.

[0256] The description of the processing procedure of each module in the apparatus and the interaction procedure between the modules can refer to the related description in the method embodiments, and will not be repeated here.

[0257] Based on the same technical concept, the embodiments of the present application further provide a computer device. Referring to FIG. 6, a structural schematic diagram of a computer device provided by an embodiment of the present application includes:

[0258] The processor 601, the memory 602 and the bus 603. The memory 602 stores machine readable instructions executable by the processor 601, and the processor 601 is configured to execute the machine readable instructions stored in the memory 602. When the machine readable instructions are executed by the processor 601, the processor 601 performs the following steps: S101: determining a plurality of historical trajectory clusters having a topological relationship based on a plurality of historical driving trajectories corresponding to a target road; at least one historical driving trajectory is included in the historical trajectory cluster; the topological relationship is used to indicate whether the plurality of historical trajectory clusters have a predecessor-successor relationship; S102: for each historical trajectory cluster, determining second pose information of each aggregation point corresponding to the historical trajectory cluster according to first pose information of trajectory points of each historical driving trajectory in the historical trajectory cluster and the topological relationship between the historical trajectory cluster and other historical trajectory clusters; wherein the last aggregation point of the historical trajectory cluster has the same pose information as the first aggregation point of the successor historical trajectory cluster corresponding to the historical trajectory cluster; and S103: determining reference trajectory information corresponding to the target road according to the second pose information of each aggregation point corresponding to the plurality of historical trajectory clusters; the reference trajectory information is used to control a vehicle to drive on the target road.

[0259] The memory 602 includes an internal memory 6021 and an external memory 6022. The internal memory 6021 is also referred to as an internal storage, and is used to temporarily store operation data in the processor 601 and exchange data with the external memory 6022 such as a hard disk. The processor 601 exchanges data with the external memory 6022 through the internal memory 6021. When the computer device is running, the processor 601 communicates with the memory 602 through the bus 603, so that the processor 601 executes the instructions mentioned in the above method embodiments.

[0260] The disclosure also provides a computer program product. When the computer program product is executed by a processor, the steps of the driving trajectory determination method described in the above method embodiments are performed. The program product can be a volatile or non-volatile computer program product.

[0261] The computer program product of the driving trajectory determination method provided by the disclosure includes a computer readable storage medium storing program codes. The program codes include instructions for performing the steps of the driving trajectory determination method described in the above method embodiments. For details, refer to the above method embodiments, which will not be described here.

[0262] The computer program product can be implemented by hardware, software or a combination thereof. In one optional embodiment, the computer program product is embodied as a computer storage medium. In another optional embodiment, the computer program product is embodied as a software product, such as a software development kit (SDK) and the like.

[0263] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the apparatus described above can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here. In several embodiments provided in the present disclosure, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are only schematic. For example, the division of the units is only a logical function division, and there can be another division in actual implementation. For example, a plurality of units or components can be combined or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some communication interfaces, devices or units, and can be electrical, mechanical or other forms.

[0264] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e. can be located in one place or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0265] In addition, each functional unit in each embodiment of the present disclosure can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit.

[0266] If the functions are realized in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer readable storage medium executable by a processor. Based on this understanding, the technical solutions of the present disclosure or the part of the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present disclosure. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various program code storage media.

[0267] If the technical solution of the present application involves personal information, the product applying the technical solution of the present application has been explicitly informed of the personal information processing rules before processing the personal information, and has obtained the personal independent consent. If the technical solution of the present application involves sensitive personal information, the product applying the technical solution of the present application has obtained the personal independent consent before processing the sensitive personal information, and at the same time meets the requirement of "explicit consent". For example, at the personal information collection device such as camera, a clear and prominent mark is set to inform that it has entered the personal information collection range and will collect personal information. If the individual voluntarily enters the collection range, it is considered to agree to collect personal information. Or on the device for processing personal information, through the pop-up information or by uploading personal information by the individual, the individual's authorization is obtained under the condition of using obvious mark / information to inform the individual of the personal information processing rules. The personal information processing rules can include personal information processor, personal information processing purpose, processing method, personal information type, etc.

[0268] Finally, it should be noted that: the above-described embodiments are only specific embodiments of the present disclosure, used to illustrate the technical solutions of the present disclosure, and are not limitations thereof. The protection scope of the present disclosure is not limited thereto. Although the present disclosure has been described in detail with reference to the foregoing embodiments, it should be understood by those skilled in the art that any skilled person familiar with the technical field can modify or easily think of changes to the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features thereof within the technical range disclosed by the present disclosure. The modification, change or replacement does not make the corresponding technical solution deviate from the spirit and scope of the technical solution of the embodiments of the present disclosure, and should be covered within the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be subject to the protection scope of the claims.

Claims

1. A driving trajectory determination method characterized by, The method comprises the following steps: determining a plurality of historical trajectory clusters with topological relations based on a plurality of historical driving trajectories corresponding to a target road; at least one historical driving trajectory is included in the historical trajectory cluster; the topological relation is used to indicate whether there is a predecessor-successor relationship between the plurality of historical trajectory clusters; for each historical trajectory cluster, determining second pose information of each aggregation point corresponding to the historical trajectory cluster according to first pose information of trajectory points of each historical driving trajectory in the historical trajectory cluster and the topological relationship between the historical trajectory cluster and other historical trajectory clusters; wherein the last aggregation point of the historical trajectory cluster has the same pose information as the first aggregation point of the successor historical trajectory cluster corresponding thereto; determining reference trajectory information corresponding to the target road according to the second pose information of each aggregation point corresponding to the plurality of historical trajectory clusters; the reference trajectory information is used to control the vehicle to drive on the target road.

2. The method of claim 1, wherein, determining the second pose information of each aggregation point corresponding to the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster and the topological relationship between the historical trajectory cluster and other historical trajectory clusters, comprising: determining the second pose information of the first aggregation point of the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster; taking the first aggregation point as the current target aggregation point, and taking the second pose information of the first aggregation point as the current target pose information; determining the aggregation direction corresponding to the target aggregation point according to the target pose information, and determining the first coordinate information of the seed point corresponding to the next aggregation point according to the aggregation direction and the target step length; determining a plurality of neighbor points of the seed point from the trajectory points of each historical driving trajectory according to the first pose information of the trajectory points of each historical driving trajectory and the first coordinate information, and determining the second pose information of the next aggregation point according to the first pose information of the plurality of neighbor points; taking the next aggregation point as a new target aggregation point, and taking the second pose information of the next aggregation point as a new target pose information, and returning to execute the step of determining the aggregation direction corresponding to the target aggregation point according to the target pose information until the second pose information of the last aggregation point corresponding to the historical trajectory cluster is obtained; wherein the plurality of neighbor points of the seed point corresponding to the last aggregation point include the first end points of each historical driving trajectory in the successor historical trajectory cluster of the historical trajectory cluster; the first end point refers to a trajectory point within a preset search radius of the starting trajectory point of the historical driving trajectory.

3. The method of claim 2, wherein, determining a plurality of neighbor points of the seed point from the trajectory points of each historical driving trajectory according to the first pose information of the trajectory points of each historical driving trajectory and the first coordinate information, comprising: constructing an R-tree corresponding to the historical trajectory cluster according to the second coordinate information indicated by each first pose information and a preset search radius; determining a matching point of the seed point according to the first coordinate information and the second coordinate information; determining a plurality of neighbor points of the seed point from the R-tree by using the matching point.

4. The method of claim 3, wherein, The determining, from the R-tree, the plurality of neighbor points of the seed point by using the matching point comprises: The determining, from the R-tree, a plurality of candidate neighbor points corresponding to the seed point by using the matching point; The determining of a first projection distance of each of the candidate neighbor points on a plane with the aggregation direction as a normal vector; The candidate neighbor point corresponding to the first projection distance satisfying a preset distribution is taken as a neighbor point of the seed point.

5. The method of claim 2, wherein, The determining of the second pose information of the first aggregation point of the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises:

6. The method of claim 5, wherein, The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises:

7. The method of claim 5, wherein, The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises:

8. The method of claim 7, wherein, The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in the historical trajectory cluster according to the first pose information of the trajectory points of each historical driving trajectory in the historical trajectory cluster comprises: The determining of a first pose information of each of the trajectory points of each historical driving trajectory in After the second pose information of the last aggregation point corresponding to the preceding historical trajectory cluster is obtained, it is determined whether the second pose information matches the second pose information of the first aggregation point of the historical trajectory cluster; if not, the second pose information of the first aggregation point of the historical trajectory cluster is updated by using the second pose information of the last aggregation point corresponding to the preceding historical trajectory cluster.

9. The method of claim 5, wherein, After the step of returning to perform the step of determining the aggregation direction corresponding to the target aggregation point according to the target pose information until the second pose information of the last aggregation point corresponding to the historical trajectory cluster is obtained, the method further comprises: If the historical trajectory cluster is one of a plurality of preceding historical trajectory clusters corresponding to a succeeding historical trajectory cluster, and the second pose information of the last aggregation point of the historical trajectory cluster is inconsistent with the mean pose information corresponding to the historical trajectory cluster, the second pose information of the last aggregation point is updated by using the mean pose information.

10. The method of claim 9, wherein, The step of updating the second pose information of the last aggregation point by using the mean pose information comprises: determining a target vector according to the second pose information of a preceding aggregation point of the last aggregation point and the mean pose information; determining a second projection distance of the last aggregation point in a direction indicated by the target vector; in a case where the second projection distance is negative, replacing the second pose information of the last aggregation point by using the mean pose information; in a case where the second projection distance is positive, determining a projection point at the second projection distance; updating the second pose information of the last aggregation point to the mean pose information, and taking the projection point as an aggregation point newly added between the preceding aggregation point and the last aggregation point. comprises:

11. A computer device, comprising: a processor and a memory, the memory storing machine readable instructions executable by the processor, the processor being configured to execute the machine readable instructions stored in the memory, and the machine readable instructions, when executed by the processor, causing the processor to perform the steps of the driving trajectory determination method according to any one of claims 1 to 10. The computer program product, when executed by a computer device, causes the computer device to perform the steps of the driving trajectory determination method according to any one of claims 1 to 10.

12. A computer program product, characterised in that, ​

Images