Road structure identification method, device, vehicle and storage medium

Abstract

The present application relates to a road structure recognition method, device, vehicle and storage medium. The method includes: obtaining a vehicle driving trajectory collected on a set route and multiple road boundaries within the vehicle's perception range; respectively determining the relative road widths of the vehicle driving trajectory and multiple road boundaries; using the vehicle driving trajectory, multiple road boundaries and relative road widths to construct an autonomous driving map of the set route; according to the autonomous driving map, identifying the road structure of the set route; wherein the road structure includes at least one of a diverging road and a merging road. The solution provided in the present application can solve the problem of identifying diverging and merging roads of a set route without the support of a high-precision map.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a road structure recognition method, device, vehicle and storage medium. Background Art

[0002] High-precision maps provide strong support for the field of autonomous driving with their centimeter-level accuracy, which plays an important role in the decision-making and planning algorithms of autonomous driving. Therefore, high-precision maps are an important facility in the field of autonomous driving.

[0003] In the related technologies, the road structure of the merge-divergence scenarios is relatively complex, and high-precision maps can well identify complex road structures such as merge-divergence scenarios. However, problems such as limited coverage, low timeliness, and high construction costs of high-precision maps have gradually emerged. Therefore, it is urgent to solve the problem of identifying merge-divergence roads without the support of high-precision maps. Summary of the invention

[0004] In order to solve or partially solve the problems existing in the related art, the present application provides a road structure identification method, device, vehicle and storage medium, which can solve the problem of identifying the diverging and merging roads of a set route without the support of high-precision maps.

[0005] The first aspect of the present application provides a road structure recognition method, comprising:

[0006] Obtain the vehicle driving trajectory collected on the set route and multiple road boundaries within the vehicle's perception range;

[0007] respectively determining the relative road widths of the vehicle driving trajectory and the boundaries of the plurality of road sections;

[0008] constructing an autonomous driving map of the set route using the vehicle driving trajectory, the multiple road boundaries and the relative road width;

[0009] According to the autonomous driving map, a road structure of the set route is identified; wherein the road structure includes at least one of a diverging road and a merging road.

[0010] A second aspect of the present application provides a road structure recognition device, comprising:

[0011] A route information acquisition module is set to acquire the vehicle driving trajectory collected on the set route and multiple road boundaries within the vehicle's sensing range;

[0012] A relative road width determination module, used to respectively determine the relative road widths of the vehicle's driving trajectory and the boundaries of the multiple road segments;

[0013] An autonomous driving map construction module, configured to construct an autonomous driving map of the set route by using the vehicle driving trajectory, the multiple road boundaries and the relative road width;

[0014] A road structure recognition module is used to identify the road structure of the set route based on the autonomous driving map; the road structure includes at least one of a diverging road and a merging road.

[0015] A third aspect of the present application provides a vehicle, comprising:

[0016] Processor; and

[0017] The memory stores executable codes thereon, and when the executable codes are executed by the processor, the processor is caused to execute the method as described above.

[0018] A fourth aspect of the present application provides a computer-readable storage medium having executable code stored thereon. When the executable code is executed by a processor of a vehicle, the processor is caused to execute the method as described above.

[0019] The technical solution provided by this application may have the following beneficial effects:

[0020] The solution provided by the present application obtains the vehicle driving trajectory collected on the set route and the multiple road boundaries within the vehicle's perception range; determines the relative road widths of the vehicle driving trajectory and the multiple road boundaries respectively; uses the vehicle driving trajectory, the multiple road boundaries and the relative road widths to construct an autonomous driving map of the set route; and identifies the road structure of the set route according to the autonomous driving map; wherein the road structure includes at least one of a diverging road and a merging road. The present application collects the vehicle driving trajectory and the multiple road boundaries within the perception range on the set route so as to obtain the relative road width of the set route, and then uses the vehicle driving trajectory, the multiple road boundaries and the relative road widths to construct an autonomous driving map of the set route. In this way, even without the support of a high-precision map, the constructed autonomous driving map can be used to identify the diverging road or merging road of the set route.

[0021] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and other objects, features and advantages of the present application will become more apparent by describing in more detail the exemplary embodiments of the present application in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the present application.

[0023] Figure 1is a flowchart of a road structure identification method shown in an embodiment of the present application;

[0024] Figure 2 is another flowchart of the road structure identification method shown in an embodiment of the present application;

[0025] Figure 3 is a schematic diagram of a learning map shown in an embodiment of the present application;

[0026] Figure 4 is a schematic diagram of a rightward diversion scenario shown in an embodiment of the present application;

[0027] Figure 5 is a schematic diagram of a leftward diversion scenario shown in an embodiment of the present application;

[0028] Figure 6 is a schematic diagram of a left merging scenario shown in an embodiment of the present application;

[0029] Figure 7 is a schematic diagram of a rightward merging scenario shown in an embodiment of the present application;

[0030] Figure 8 is a schematic diagram of the structure of a road structure identification device shown in an embodiment of the present application;

[0031] Fig. 9 It is a schematic diagram of the structure of a vehicle shown in an embodiment of the present application. DETAILED DESCRIPTION

[0032] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0033] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0034] It should be understood that although the terms "first", "second", "third", etc. may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0035] In the related technologies, the road structure of the merge-divergence scenarios is relatively complex, and high-precision maps can well identify complex road structures such as merge-divergence scenarios. However, problems such as limited coverage, low timeliness, and high construction costs of high-precision maps have gradually emerged. Therefore, it is urgent to solve the problem of identifying merge-divergence roads without the support of high-precision maps.

[0036] In response to the above problems, an embodiment of the present application provides a road structure recognition method, which collects the vehicle driving trajectory and multiple road boundaries within the perception range on the set route so as to obtain the relative road width of the set route, and then uses the vehicle driving trajectory, multiple road boundaries and relative road width to construct an autonomous driving map of the set route. In this way, even without the support of high-precision maps, the constructed autonomous driving map can be used to identify the diverging roads or merging roads of the set route.

[0037] The technical solution of the embodiments of the present application is described in detail below with reference to the accompanying drawings.

[0038] Figure 1 It is a flowchart of the road structure identification method shown in an embodiment of the present application.

[0039] See also Figure 1 , the road structure recognition method of the present application comprises:

[0040] S110, obtaining a vehicle driving trajectory collected on a set route and multiple road boundaries within a vehicle sensing range.

[0041] Among them, the executor of this road structure recognition can be an automatic driving system mounted on a vehicle. The embodiment of the present application can be divided into a learning process, a construction process and an identification process. During the learning process, the user can set the departure point and the destination, and the route from the departure point to the destination can be used as the set route. Among them, there may be multiple routes from the departure point to the destination, and the automatic driving system can select one of the fixed routes commonly used by the user as the set route. For example, assuming that the route from the departure point A to the destination B includes route 1, route 2 and route 3, and the user often takes route 2, then route 2 can be used as the set route from the departure point A to the destination B.

[0042] It should be noted that if the departure place and the destination are swapped, the corresponding set route will also be different due to the change in the vehicle's driving direction. For example, taking the route between location A and location B as an example, it is divided into an outbound route and a return route. Assuming that the outbound route refers to the route from location A to location B, the return route refers to the route from location B to location A. If the route from location A to location B is to drive on the right, that is, the outbound route is the route on the right road, then the route from location B to location A is to drive on the left, that is, the return route is the route on the left road. Since the road structure, road condition, road width, etc. of the left road and the right road are different, the outbound route and the return route need to be set as different routes.

[0043] Since the set route is not supported by high-precision maps, the automatic driving mode of a fixed route can be turned on during the learning process. In this mode, the fixed route is used as the set route, and the automatic driving system can control the vehicle to automatically drive along the set route based on the perception data of various sensors. During the process of the vehicle automatically driving along the set route, the vehicle's driving trajectory can be collected and recorded by the data acquisition system, and the road boundaries can be collected and recorded by the vehicle's radar, camera and other sensors. In addition, in order to improve the quality of learning and driving safety, or if the set route is not a fixed route, the embodiment of the present application can also allow the user to drive the vehicle along the set route. During the process of the vehicle driving along the set route, the vehicle's driving trajectory can be collected and recorded by the data acquisition system, and the road boundaries can be collected and recorded by the vehicle's radar, camera and other sensors.

[0044] Due to the limited perception range of vehicle sensors, the road boundaries captured by vehicle sensors are discontinuous curves. For example, if there is an intersection in front of the vehicle and the intersection is very wide, the vehicle sensors cannot capture the road boundaries outside the perception range and can only capture the road boundaries within the perception range. Therefore, the road boundaries within the vehicle's perception range are segmented.

[0045] S120, respectively determining the relative road widths of the vehicle driving trajectory and the boundaries of the multiple road sections.

[0046] After the autonomous driving system obtains the vehicle driving trajectory from the data acquisition system and obtains multiple road boundaries from various vehicle sensors, it enters the construction process. During the construction process, the autonomous driving system can first determine the relative road width between the vehicle driving trajectory and each road boundary, wherein the road boundary can include the left road boundary and the right road boundary, the left road boundary refers to the road boundary collected by the vehicle sensor on the left side of the vehicle driving trajectory, and the right road boundary refers to the road boundary collected by the vehicle sensor on the right side of the vehicle driving trajectory, so the relative road width can include the left road width and the right road width, the left road width refers to the relative road width between the vehicle driving trajectory and the left road boundary, and the right road width refers to the relative road width between the vehicle driving trajectory and the right road boundary.

[0047] S130, constructing an autonomous driving map of a set route using a vehicle driving trajectory, multiple road boundaries, and relative road widths.

[0048] By collecting the vehicle driving trajectory on the set route and the multiple road boundaries within the perception range, the relative road width of the set route can be obtained. Then, the vehicle driving trajectory, the multiple road boundaries and the relative road width can be used to construct an autonomous driving map of the set route. For example, the vehicle driving trajectory, the left road boundary, the right road boundary, the left road width and the right road width can be used to construct an autonomous driving map of the set route, thereby solving the problem of constructing an autonomous driving map for a set route without the support of a high-precision map.

[0049] S140, identifying a road structure of a set route based on the autonomous driving map; wherein the road structure includes at least one of a diverging road and a merging road.

[0050] After constructing the autonomous driving map of the set route, the recognition process begins. During the recognition process, since the autonomous driving map contains prior information of the set route, the prior information may include at least one of road boundary information, relative road width information, and selected road information, the autonomous driving system can identify the road structure of the set route based on the road boundary information and relative road width information, such as identifying complex road structures such as diverging roads or merging roads of the set route, thereby solving the problem of identifying diverging and merging roads of the set route without the support of high-precision maps.

[0051] In addition, the automatic driving system can determine the passable road area of ​​the vehicle according to the road information selected for passage, so that the vehicle can be directly controlled to drive automatically in the passable road area, thereby avoiding the situation of taking the wrong road due to driving according to the navigation information. For example, at a fork in the road, the road that the vehicle has traveled during the learning process indicates the road that the vehicle will take in the future. For automatic driving with only navigation information, it is still necessary to select the road to be taken at the fork in the road. For example, assuming that the vehicle encounters forks A and B in front, if the vehicle selects the road at fork A during the learning process, the road at fork A can be used as the vehicle's driving trajectory, and an automatic driving map is constructed based on the vehicle's driving trajectory. The automatic driving map carries the road information selected for passage. When the vehicle encounters forks A and B again, the automatic driving system can determine that the passable road area of ​​the vehicle is the road at fork A according to the road information selected for passage. Therefore, the vehicle can be directly controlled to drive automatically in the road at fork A, so that there is no need to select which fork road again. For automatic driving with only navigation information, it is still necessary to select the road to be taken at forks A and B, and it is easy to take the wrong road when driving according to the navigation information.

[0052] It can be seen that even without the support of high-precision maps, the embodiments of the present application can identify complex road structures such as diverging roads or merging roads of the set route through the constructed autonomous driving map, and drive according to the autonomous driving map to avoid taking the wrong road.

[0053] It should be noted that the road structure recognition method provided in the embodiment of the present application has a wider scope of application and can be applied in areas not covered by high-precision maps as well as in areas covered by high-precision maps.

[0054] As can be seen from this example, the solution provided by this application obtains the vehicle driving trajectory collected on the set route and the multiple road boundaries within the vehicle's perception range; determines the relative road widths of the vehicle driving trajectory and the multiple road boundaries respectively; uses the vehicle driving trajectory, the multiple road boundaries and the relative road width to construct an autonomous driving map of the set route; and identifies the road structure of the set route according to the autonomous driving map; wherein the road structure includes at least one of a diverging road and a merging road. This application collects the vehicle driving trajectory and the multiple road boundaries within the perception range on the set route so as to obtain the relative road width of the set route, and then uses the vehicle driving trajectory, the multiple road boundaries and the relative road width to construct an autonomous driving map of the set route. In this way, even without the support of a high-precision map, the constructed autonomous driving map can be used to identify the diverging road or merging road of the set route.

[0055] Figure 2 It is another flowchart of the road structure identification method shown in an embodiment of the present application.

[0056] See also Figure 2 , the road structure recognition method of the present application comprises:

[0057] S210, obtaining a vehicle driving trajectory collected on a set route and multiple road boundaries within a vehicle sensing range.

[0058] During the learning process, the departure point and destination can be set, and the route from the departure point to the destination can be used as the set route. When the vehicle is driving along the set route, the vehicle's driving trajectory can be collected and recorded by the data acquisition system, and the road boundaries can be collected and recorded by the vehicle's radar, camera and other sensors. Therefore, the autonomous driving system can obtain the vehicle's driving trajectory from the data acquisition system and the road boundaries from various vehicle sensors.

[0059] Due to the limited sensing range of the vehicle sensor, the road boundary captured by the vehicle sensor is a discontinuous curve, so the road boundary within the vehicle's sensing range is segmented. Figure 3 As shown, the road boundary may include a left road boundary and a right road boundary, and both the left road boundary and the right road boundary are discontinuous curves. Therefore, the road boundary obtained by the autonomous driving system from various vehicle sensors may include multiple sections of left road boundaries and multiple sections of right road boundaries within the vehicle's perception range.

[0060] Among them, the types of road boundaries may include at least one of surrounding ground markings that define the passage area (such as guide lines, double yellow solid lines, etc.), physical dividing facilities (such as fences, water barriers, etc.), and boundary outlines of buildings.

[0061] S220, taking the vehicle driving trajectory as a learning trajectory, and extracting a plurality of learning trajectory points from the learning trajectory according to a preset interval distance.

[0062] During the construction process, the vehicle driving trajectory collected during the learning process can be used as the learning trajectory. The embodiment of the present application can record the relative position relationship between the road boundary and the learning trajectory. The relative position relationship is used to characterize that the road boundary and the learning trajectory are in the same coordinate system, so as to construct a learning map of the set route based on multiple road boundaries and learning trajectories in the same coordinate system. Figure 3 That is, the learning map. Since the road boundaries of the learning map are still discontinuous curves, the passable road area of ​​the set route cannot be obtained based on the learning map, nor can the road structure of the set route be accurately identified. Therefore, the embodiment of the present application can further construct a more detailed autonomous driving map based on the information of the learning map.

[0063] The embodiment of the present application can extract multiple learning trajectory points from the learning trajectory according to a preset interval, wherein the learning trajectory points can include the starting point and the end point of the learning trajectory. Figure 3 As shown, learning trajectory points O, A, B and C are extracted from the learning trajectory at a fixed preset interval distance, where learning trajectory point O can be the starting point of the learning trajectory. The smaller the preset interval distance, the more detailed the learning trajectory information, so the accuracy of the autonomous driving map constructed based on the learning trajectory information is also higher, but the corresponding calculation amount and storage space are larger, so the preset interval distance is generally not more than 2.0m, preferably, the preset interval distance can be 0.5m to 1.0m.

[0064] It should be noted that, in general, the learning trajectory can be a continuous curve or a discontinuous curve. The discontinuous curve may be artificially produced according to different learning trajectory schemes. When the learning trajectory is a discontinuous curve, there is no need to extract learning trajectory points in the interrupted area. It is only necessary to extract learning trajectory points at the preset interval distance on each curve. The vehicle can drive in the discontinuous area through the free area driving mode (a driving mode of autonomous driving).

[0065] It should be noted that the embodiments of the present application may use other similar navigation line types or line types with direction indications as learning trajectories, or may use discrete points or line segments as learning trajectories. When the learning trajectory is a discrete point, the learning trajectory points may be selected by interpolation, or the discrete points may be directly used as learning trajectory points. Regardless of which form of learning trajectory is used, the embodiments of the present application require the vehicle to travel along the set route once in order to obtain the road boundaries of the set route collected during the driving process.

[0066] S230: Determine the relative road width between each learning trajectory point and the corresponding road boundary.

[0067] The autonomous driving system may first determine the relative road width between each learning trajectory point and the corresponding road boundary. Specifically, the autonomous driving system may first determine the left road width between each learning trajectory point and the corresponding left road boundary, and determine the right road width between each learning trajectory point and the corresponding right road boundary.

[0068] In one embodiment, determining the relative road width between each learning trajectory point and the corresponding road boundary includes:

[0069] Determine the tangent direction of the learning trajectory at each learning trajectory point; use the direction perpendicular to the tangent direction as the emission direction; start from each learning trajectory point, emit rays along the corresponding emission direction until the ray encounters the corresponding road boundary, and terminate the emission to obtain the length of the ray; determine the length of the ray as the relative road width between each learning trajectory point and the corresponding road boundary.

[0070] The autonomous driving system may first determine the tangent direction of the learning trajectory at each learning trajectory point. The tangent direction at each learning trajectory point includes an upward tangent direction and a downward tangent direction. The upward tangent direction indicates that the tangent direction points to the next learning trajectory point, and the downward tangent direction indicates that the tangent direction points to the previous learning trajectory point. The embodiment of the present application may select one type of tangent direction, such as Figure 3 As shown, the embodiment of the present application can select the upward tangent direction at each learning trajectory point. For example, the upward tangent direction b at the learning trajectory point B indicates that the tangent direction b points to the next learning trajectory point C.

[0071] After determining the tangent direction at each learning trajectory point, the direction perpendicular to the tangent direction can be used as the emission direction. The emission direction indicates the direction in which the learning trajectory point emits rays toward the corresponding road boundary. Since the road boundary includes the left road boundary and the right road boundary, the emission direction can include the left emission direction and the right emission direction. The direction perpendicular to the tangent direction and located on the left side of the learning trajectory can be used as the left emission direction, and the direction perpendicular to the tangent direction and located on the right side of the learning trajectory can be used as the right emission direction. Specifically, the target angle θ of each learning trajectory point can be calculated. The target angle θ can refer to the angle between the tangent direction and the reference direction, such as Figure 3 As shown, the target angle θ of the learning trajectory point B is the angle between the tangent direction b and the reference direction. The reference direction can be selected arbitrarily, and the reference directions of all learning trajectory points remain consistent. After calculating the target angle θ of each learning trajectory point, the target angle θ of each learning trajectory point can be added to the preset angle to obtain the left emission angle corresponding to each learning trajectory point, and the target angle θ of each learning trajectory point can be subtracted from the preset angle to obtain the right emission angle corresponding to each learning trajectory point.

[0072] The preset angle can be set to π / 2, so the left emission angle can be (θ+π / 2), and the right emission angle can be (θ-π / 2). (θ+π / 2) and (θ-π / 2) are both expressed in radians, so π / 2 corresponds to an angle of 90 degrees. Figure 3 As shown, the direction of the left emission angle (θ+π / 2) may be determined as the left emission direction, and the direction of the right emission angle (θ-π / 2) may be determined as the right emission direction.

[0073] The ray corresponding to each learning trajectory point can include a left ray and a right ray, such as Figure 3 As shown, each learning trajectory point can be used as the starting point of the left ray, that is, starting from each learning trajectory point, the left ray is emitted along the corresponding left emission direction, that is, the left ray is emitted in the direction of the left emission angle (θ+π / 2), until the left ray encounters the corresponding left road boundary and the emission is terminated, and the length of the left ray is obtained. And, each learning trajectory point can be used as the starting point of the right ray, that is, starting from each learning trajectory point, the right ray is emitted along the corresponding right emission direction, that is, the right ray is emitted in the direction of the right emission angle (θ-π / 2), until the right ray encounters the corresponding right road boundary and the emission is terminated, and the length of the right ray is obtained.

[0074] The length of the ray can refer to the length from the end point of the ray to the starting point of the ray (learning trajectory point), so the length of the ray can be determined as the relative road width of each learning trajectory point and the corresponding road boundary. It can be seen that each learning trajectory point corresponds to two length values. For example, when the left ray is emitted in the direction of the left emission angle (θ+π / 2), the length of the left ray can be used as the left road width, which is recorded as The right ray is emitted in the direction of the right emission angle (θ-π / 2). The length of the right ray can be used as the width of the right road, which is recorded as

[0075] It should be noted that the width of the road on the left is In the figure, l represents the left side (left); the right side represents the road width In , r represents the right side (right). Where n represents the nth learning trajectory point.

[0076] In one embodiment, the method may further include:

[0077] When the ray does not encounter the corresponding road boundary and the length of the ray is greater than or equal to the preset length, the preset length is used as the length of the ray.

[0078] Since the road boundary is a discontinuous curve, there may be a situation where the ray does not encounter the corresponding road boundary. For this, the embodiment of the present application can pre-set a preset length, which can be any value. Preferably, the preset length can be 25.0m. When the ray does not encounter the corresponding road boundary, and the length of the ray is greater than or equal to the preset length of 25.0m, the preset length of 25.0m can be used as the length of the ray, so that the length of the ray 25.0m is determined as the relative road width between the current learning trajectory point and the corresponding virtual road boundary. Specifically, when the left ray does not encounter the corresponding left road boundary, and the length of the left ray is greater than or equal to the preset length of 25.0m, the preset length of 25.0m can be used as the length of the left ray, so that the length of the left side 25.0m is determined as the left road width between the current learning trajectory point and the corresponding left virtual road boundary; and / or, when the right ray does not encounter the corresponding right road boundary, and the length of the right ray is greater than or equal to the preset length of 25.0m, the preset length of 25.0m can be used as the length of the right ray, so that the length of the right side 25.0m is determined as the right road width between the current learning trajectory point and the corresponding right virtual road boundary.

[0079] S240, performing filtering processing on each relative road width to obtain a target relative road width corresponding to each learning trajectory point.

[0080] The relative road width determined based on the learning trajectory points and road boundaries is usable to a certain extent, but it is not smooth enough, and the vehicle may sway left and right when driving. Therefore, the embodiment of the present application can filter the relative road width corresponding to each learning trajectory point to obtain the target relative road width corresponding to each learning trajectory point after filtering. It can be seen that compared with the relative road width that has not been filtered, the target relative road width can be smoother.

[0081] Relative road width includes left road width and right road width Therefore, the width of each left road can be Perform filtering to obtain the target left road width corresponding to each learning trajectory point And the width of each right road Perform filtering to obtain the target right road width corresponding to each learning trajectory point

[0082] In one implementation, filtering each relative road width to obtain a target relative road width corresponding to each learning trajectory point may include:

[0083] Clustering is performed on each relative road width to obtain multiple first clusters; each first cluster is traversed, and if the relative road width of the current first cluster is greater than the relative road width of the adjacent first cluster, the relative road width of the current first cluster is adjusted to the relative road width of the adjacent first cluster, until the traversal is completed, and the adjusted target relative road width corresponding to each learning trajectory point is obtained.

[0084] The embodiment of the present application can perform clustering processing on each relative road width to obtain multiple first clusters. The first cluster may include a left first cluster and a right first cluster. The left first cluster may be obtained by performing clustering processing on each left road width, and the right first cluster may be obtained by performing clustering processing on each right road width.

[0085] In one example, for the clustering process of the left road width, each left road width can be arranged in the order of each learning trajectory point, so as to obtain the left road width sequence Where N represents the number of learning trajectory points. Where the order of each learning trajectory point can be determined according to the distance s between each learning trajectory point and the starting point of the learning trajectory, where s can be defined as Among them, d i represents the Euclidean distance from the i-th learning trajectory point to the i-1-th learning trajectory point, so the s value corresponding to the current learning trajectory point can be the sum of the Euclidean distances of all adjacent learning trajectory points before the learning trajectory point. For example, Figure 3 As shown, the starting point of the learning trajectory is the learning trajectory point O, then the s value corresponding to the learning trajectory point O is 0, the s value corresponding to the learning trajectory point A is d1, the s value corresponding to the learning trajectory point B is d1+d2, and the s value corresponding to the learning trajectory point C is d1+d2+d3, where d1 represents the Euclidean distance from the learning trajectory point A to the learning trajectory point O, d2 represents the Euclidean distance from the learning trajectory point B to the learning trajectory point A, and d3 represents the Euclidean distance from the learning trajectory point C to the learning trajectory point B. After that, the difference between the widths of two adjacent left roads can be calculated. If the difference is less than or equal to the clustering threshold, the two adjacent left road widths can be clustered into the same category, that is, the two adjacent left road widths can be divided into the same left first cluster, wherein the clustering threshold can be any value, for example, the clustering threshold can be set to 3.0m. For example, Figure 3 As shown, assuming that the left road width of the learning trajectory point O is The road width on the left side of the learning trajectory point A The difference is ≤3m, and the width of the road on the left side of the learning trajectory point A The road width on the left side of the learning trajectory point B The difference is ≤3m, and the width of the left road of the learning trajectory point B The road width on the left side of the learning trajectory point C If the difference is greater than 3m, the left road width of the learning trajectory point O can be The left road width of learning trajectory point A and the left road width of the learning trajectory point B Divided into the same left first cluster.

[0086] According to the above clustering method, the first cluster sequence on the left can be obtained in, represents the Mth left first cluster, M represents the number of first clusters (left first cluster / right first cluster), and m represents the mth learning trajectory point. Contains m+1 consecutive left road width sequences, and the number of left road width sequences contained in each left first cluster is not necessarily the same. For example, assuming Then we can cluster and get three left first clusters: Thus, we can get the first cluster sequence on the left:

[0087] Similarly, we can cluster the first cluster sequence on the right The clustering process of the right road width is the same as the clustering process of the left road width described above, and will not be described in detail here.

[0088] Get the first cluster sequence on the left Each left first cluster can be traversed. If the left road width of the current left first cluster is greater than the left road width of the adjacent left first cluster, the left road width of the current left first cluster can be adjusted to the left road width of the adjacent left first cluster until the traversal is completed, and the adjusted target left road width sequence can be obtained. Target left road width sequence Contains the width of the road on the left side of the target corresponding to each learning trajectory point.

[0089] Specifically, traverse each left first cluster, if all left road widths in the current left first cluster are greater than the last left road width in the previous left first cluster by more than the second threshold, and are greater than the first left road width in the next left first cluster by more than the second threshold, it can be explained that the current road features are: narrow-wide-narrow, so all left road widths in the current left first cluster can be adjusted to the last left road width in the previous left first cluster, or all left road widths in the current left first cluster can be adjusted to the first left road width in the next left first cluster, where the second threshold can be set to 3.0m. Exemplarily, continuing with the above example, three left first clusters are obtained by clustering: When traversing to the first cluster on the left When the first cluster on the left is Can be respectively connected to the left first cluster Compare. Among them, the first cluster on the left The first cluster on the left side of the previous The comparison process is: Among them, the first cluster on the left The first cluster on the left The comparison process is: It can be seen that the first cluster on the left All left road widths in Both are larger than the first cluster on the left The last left road width in Greater than 3.0m, and the first cluster on the left All left road widths in The first cluster on the left is The first left road width in If the current cluster is greater than 3.0m, the first cluster on the left can be All left road widths in All are adjusted to the first cluster on the left side of the previous The last left road width in Thus, the adjusted target left road width is obtained: Or you can put the first cluster on the left All left road widths in All are adjusted to the first cluster on the left The first left road width in Thus, the adjusted target left road width is obtained:

[0090] Similarly, the adjusted target right road width sequence can be obtained Target right road width sequence Contains the target right road width corresponding to each learning trajectory point. The adjustment process of the target right road width is the same as the adjustment process of the target left road width described above, and will not be repeated here.

[0091] It can be seen that in the case where the road characteristics are "narrow-wide-narrow", by unifying the road width, on the premise of ensuring that the vehicle does not collide with the road boundary during the automatic driving process, it is possible to prevent the vehicle from straying into the oncoming lane. In addition, other road characteristics will not be rashly processed, otherwise problems such as collisions may occur. For example, if the road characteristics are "wide-narrow-wide", this situation does not require unifying the road width, because if the narrow part is changed to wide, it will cause the vehicle to possibly collide with the road boundary when driving automatically in the passable road area.

[0092] It should be noted that capital letters are used to represent fixed values. Therefore, M represents the number of the first cluster, N represents the number of learning trajectory points, and generally, M < N. Lowercase letters are used to represent index values. Therefore, n represents the nth learning trajectory point, and m represents the mth learning trajectory point. For example represents the left road width corresponding to the nth learning trajectory point, represents the left road widths corresponding to N learning trajectory points.

[0093] S250, adopt the corresponding target relative road width to calculate the passable road boundary points corresponding to each learning trajectory point; wherein, the passable road boundary points include left passable road boundary points and right passable road boundary points.

[0094] After obtaining the target relative road width corresponding to each learning trajectory point, the corresponding target relative road width can be adopted to calculate the passable road boundary points corresponding to each learning trajectory point. Among them, the passable road boundary points can include left passable road boundary points and right passable road boundary points. The left passable road boundary points can be calculated based on the target left road width, and the right passable road boundary points can be calculated based on the target right road width.

[0095] Specifically, each learning trajectory point can be translated by a distance of the target left road width in the corresponding left emission direction, that is, each learning trajectory point can be translated by a distance of the target left road width in the direction of the left emission angle (θ + π / 2), so as to obtain the left passable road boundary points corresponding to each learning trajectory point. Exemplarily, as Figure 3 shown, assuming that the target left road width corresponding to the learning trajectory point B is 4m, then the learning trajectory point B can be translated by 4m in the direction of the left emission angle (θ + π / 2), so as to obtain the left passable road boundary point corresponding to the learning trajectory point B.

[0096] Similarly, each learning trajectory point can be translated in the corresponding right emission direction by the width of the target right road, that is, each learning trajectory point can be translated in the direction of the right emission angle (θ-π / 2) by the width of the target right road, so that the right passable road boundary point corresponding to each learning trajectory point can be obtained.

[0097] S260, connecting a plurality of left-side passable road boundary points in sequence to obtain a left boundary of the passable road, and connecting a plurality of right-side passable road boundary points in sequence to obtain a right boundary of the passable road.

[0098] A plurality of left-side passable road boundary points can be connected in sequence according to the order of each learning trajectory point, so as to obtain the left boundary of the passable road. For example, the left-side passable road boundary points can be connected into a curve from small to large according to the s value of the corresponding learning trajectory point, and this curve is the left boundary of the passable road.

[0099] Similarly, multiple right-side passable road boundary points can be connected in sequence according to the order of each learning trajectory point, so as to obtain the right boundary of the passable road. For example, the right-side passable road boundary points can be connected into another curve according to the s value of the corresponding learning trajectory point from small to large, and this other curve is the right boundary of the passable road.

[0100] S270, using the left boundary of the passable road and the right boundary of the passable road to construct a passable road area of ​​the set route, and output the passable road area as an autonomous driving map of the set route.

[0101] After obtaining the left boundary of the passable road and the right boundary of the passable road, the left boundary of the passable road and the right boundary of the passable road can be used to construct the passable road area of ​​the set route. For example, the area between the left boundary of the passable road and the right boundary of the passable road can be used as the passable road area of ​​the set route. The passable road area refers to the automatic driving area where the vehicle can stably pass, that is, the automatic driving system can control the vehicle to stably drive automatically in the passable road area, so the passable road area can be output as an automatic driving map of the set route.

[0102] S280, identifying a road structure of the passable road area according to a target relative road width change of the passable road area and surrounding ground markings; wherein the road structure includes at least one of a diverging road and a merging road.

[0103] During the identification process, since the passable road area has prior information of the set route, the prior information may include at least one of road boundary information, relative road width information, and selected road information. The road boundary information may include surrounding ground markings of the passable road area, and the relative road width information may include target relative road width changes of the passable road area. Therefore, the automatic driving system can identify the road structure of the passable road area based on the surrounding ground markings of the passable road area and the target relative road width changes, such as identifying complex road structures such as diverging roads or merging roads in the passable road area, thereby solving the problem of identifying diverging and merging roads of a set route without the support of high-precision maps.

[0104] It should be noted that the road structure recognition method provided in the embodiment of the present application has a wider scope of application and can be applied in areas not covered by high-precision maps as well as in areas covered by high-precision maps.

[0105] In one embodiment, the surrounding ground markings include guide lines; identifying the road structure of the passable road area according to the target relative road width change of the passable road area and the surrounding ground markings, including:

[0106] Clustering is performed on each target relative road width to obtain multiple second clusters; the target relative road width difference between two adjacent second clusters is calculated, and the target relative road width change between two adjacent second clusters is determined, and whether the two adjacent second clusters correspond to guide lines; if the target relative road width difference is greater than a first threshold, and the target relative road width changes from large to small, and there is a corresponding guide line, then the road structure of the passable road area at the two adjacent second clusters is identified as a diversion road; or, if the target relative road width difference is greater than the first threshold, and the target relative road width changes from small to large, and there is a corresponding guide line, then the road structure of the passable road area at the two adjacent second clusters is identified as a merging road.

[0107] Since the relative road width is filtered to obtain the target relative road width, it is necessary to cluster the filtered target relative road width again according to the above clustering processing method, so as to obtain multiple second clusters, wherein the second cluster may include a left second cluster and a right second cluster. The left second cluster may be obtained by clustering the left road width of each target, and the right second cluster may be obtained by clustering the right road width of each target. Specifically, according to the above clustering of the left road width sequence The clustering processing method of each left road width in the target left road width sequence after filtering is The left road width of each target in is clustered again, so that the left second cluster sequence after clustering can be obtained. Wherein, K represents the number of the second cluster (second cluster on the left / second cluster on the right), K≤M. For example, assuming the sequence of the first cluster on the left: in, Filter each left road width to obtain the target left road width after filtering: By clustering the width of the road on the left side of each target again, the second cluster sequence on the left side can be obtained: in,

[0108] Similarly, we can obtain the second cluster sequence on the right after clustering again

[0109] The surrounding ground markings may include guide lines, which may be traffic markings used to indicate vehicle divergence and merging. Therefore, when obtaining the second cluster sequence on the left side, Then, the target left road width difference between two adjacent left second clusters can be calculated. And determine the change in the target left road width of the two adjacent left second clusters, and determine whether the two adjacent left second clusters correspond to the guide line. And, after obtaining the right second cluster sequence Then, the difference in the target right road width between two adjacent right second clusters can be calculated. And determine the change in the target right road width of two adjacent right second clusters, and judge whether the two adjacent right second clusters correspond to the guide line.

[0110] If the target left road width difference If the width of the target left road changes from large to small, it can be shown that the type of the diverging and merging scene is a diverging road; if the width of the target left road changes from small to large, it can be shown that the type of the diverging and merging scene is a merging road.

[0111] Similarly, if the target right road width difference If the width of the target right road changes from large to small, it can be said that the type of the diverging and merging scene is a diverging road; if the width of the target right road changes from small to large, it can be said that the type of the diverging and merging scene is a merging road.

[0112] It should be noted that the first threshold can be adjusted according to specific needs. For example, the first threshold can be set to 3.5m, which is not limited in the embodiment of the present application.

[0113] In one embodiment, calculating the target relative road width difference between two adjacent second clusters and determining the target relative road width change between two adjacent second clusters may include:

[0114] Among two adjacent second clusters, the target relative road width at the last position in the previous second cluster is selected, and the target relative road width at the first position in the next second cluster is selected; the absolute value of the difference between the target relative road width at the last position and the target relative road width at the first position is calculated as the target relative road width difference between the two adjacent second clusters; and, the width change from the target relative road width at the last position to the target relative road width at the first position is determined as the target relative road width change between the two adjacent second clusters.

[0115] Among them, a target left road width can be selected from each of the two adjacent left second clusters as their representative. Preferably, the two selected target left road widths can have an adjacent relationship, and the change of the target left road width of the two adjacent left second clusters can be accurately reflected through the two adjacent target left road widths. Specifically, in the two adjacent left second clusters, the target left road width at the last position in the previous left second cluster and the target left road width at the first position in the next third cluster have an adjacent relationship, so the target left road width at the last position can be selected from the previous left second cluster, and the target left road width at the first position can be selected from the next third cluster. Exemplarily, assuming that the two adjacent left second clusters are in, Then we can start from the second cluster on the left of the previous Select the left road width of the target at the end And the second cluster from the left Select the first target left road width The width of the left road between two adjacent targets and Can be used to characterize the two adjacent left second clusters The target left road width changes.

[0116] Similarly, in two adjacent right second clusters, the target right road width at the last position in the previous right second cluster is selected, and the target right road width at the first position in the next right second cluster is selected. Select the right road width of the target at the end And the second cluster from the right Select the first target right road width Width of two adjacent target right roads and Can be used to characterize two adjacent right second clusters The target right road width changes.

[0117] When selecting the road width on the left side of the target at the end Target left road width at first position After that, the width of the left road at the end can be calculated. The road width to the left of the first target The absolute value of the difference between the two adjacent left second clusters is used as the target left road width difference between the two adjacent left second clusters. Right now And, determine the road width to the left of the target at the end Road width to the left of the first target ’s width changes so that their width changes are taken as the target left road width changes of the two adjacent left second clusters.

[0118] Similarly, when selecting the right road width of the target at the end Target right road width at first position After that, the width of the target right road at the end can be calculated The road width to the right of the first target The absolute value of the difference between the two adjacent right second clusters is used as the target right road width difference between the two adjacent right second clusters. Right now And, determine the road width to the right of the target at the end Road width to the right of the first target ’s width changes so that their width changes are taken as the target right road width changes of the two adjacent right second clusters.

[0119] In one embodiment, after identifying the road structure of the traversable road area, the method may further include:

[0120] According to the position of the guide line, the diverging direction of the diverging road is identified, or the merging direction of the merging road is identified.

[0121] Specifically, after identifying that the road structure of the passable road area at two adjacent left second clusters is a diverging road, the diverging direction of the diverging roads at two adjacent left second clusters can be identified according to the position of the guide line. Alternatively, after identifying that the road structure of the passable road area at two adjacent right second clusters is a diverging road, the diverging direction of the diverging roads at two adjacent right second clusters can be identified according to the position of the guide line. Alternatively, after identifying that the road structure of the passable road area at two adjacent left second clusters is a merging road, the merging direction of the merging roads at two adjacent left second clusters can be identified according to the position of the guide line. Alternatively, after identifying that the road structure of the passable road area at two adjacent right second clusters is a merging road, the merging direction of the merging roads at two adjacent right second clusters can be identified according to the position of the guide line.

[0122] In one embodiment, the target relative road width includes a target left road width and a target right road width; and identifying the diverging direction of the diverging road or the merging direction of the merging road according to the position of the guide line may include:

[0123] If the guide line is within the preset range of the width of the left road of the target located in the first position, the divergence direction of the divergence road is identified as the right divergence direction; or, if the guide line is within the preset range of the width of the right road of the target located in the first position, the divergence direction of the divergence road is identified as the left divergence direction; or, if the guide line is within the preset range of the width of the left road of the target located in the last position, the merging direction of the merging road is identified as the left merging direction; or, if the guide line is within the preset range of the width of the right road of the target located in the last position, the merging direction of the merging road is identified as the right merging direction.

[0124] In one example, if the guide line is at the first target left road width If the preset range is within the range of , it can be identified that the divergence direction of the diverging roads at the two adjacent left second clusters is the right divergence direction, wherein the preset range can be dynamically adjusted according to the specific mapping method, for example, the preset range can be set to 3.0m. In short, if Figure 4 As shown, if Right now And the width of the road on the left side of the target changes from large to small, that is, And at a smaller target left road width If there is a guide line within 3.0m of the target, the width of the road on the left side is identified. and There is a scene where the flow is diverted to the right.

[0125] In another example, if the guide line is at the first target right road width If the pre-set range is within , it can be identified that the diverging direction of the diverging roads at the two adjacent right second clusters is the left diverging direction. Figure 5 As shown, if Right now And the width of the road on the right side of the target changes from large to small, that is, And at a smaller target right road width If there is a guide line within 3.0m of the vicinity, the width of the road on the right side of the target is identified. and There is a scene of left diversion.

[0126] In another example, if the guide line is at the target left road width at the end If the merging direction of the two adjacent left second clusters is within the preset range, it can be identified that the merging direction of the merging roads is the left merging direction. Figure 6 As shown, if Right now And the width of the road on the left side of the target changes from small to large, that is And at a smaller target left road width If there is a guide line within 3.0m of the target, the width of the road on the left side is identified. and There is a scene where the flow merges to the left.

[0127] In another example, if the guide line is located at the end of the target right road width If the preset range is within , it can be identified that the merging direction of the diverging roads at the adjacent right second clusters is the right merging direction. Figure 7 As shown, if Right now And the width of the target right road changes from small to large, that is And at a smaller target right road width If there is a guide line within 3.0m of the vicinity, the width of the road on the right side of the target is identified. and There is a scene where the flow merges to the right.

[0128] It can be seen from this example that the solution provided by the present application, when the set route is not supported by a high-precision map, the embodiment of the present application collects the vehicle's driving trajectory and multiple road boundaries within the perception range on the set route so as to obtain the relative road width of the set route, and then converts the relative road width into a passable road area that is more suitable for autonomous driving through a filtering method, thereby avoiding the vehicle from swaying left and right when driving. Subsequently, through the target relative road width change in the passable road area and the surrounding ground markings, the divergence and merging scenes can be accurately identified, and specific divergence and merging directions can be given, thereby providing more detailed road structure information for autonomous driving, so that the autonomous driving system can provide a basis for autonomous driving links such as vehicle path planning, vehicle lane change timing, and vehicle speed planning in the scene.

[0129] Corresponding to the aforementioned application function implementation method embodiment, the present application also provides a road structure recognition device, a vehicle, a computer-readable storage medium and corresponding embodiments.

[0130] Figure 8 It is a schematic diagram of the structure of a road structure identification device shown in an embodiment of the present application.

[0131] See also Figure 8 , the present application provides a road structure recognition device, which may include:

[0132] A set route information acquisition module 810 is used to acquire the vehicle driving trajectory collected on the set route and multiple road boundaries within the vehicle sensing range;

[0133] A relative road width determination module 820, for respectively determining the relative road widths of the vehicle's driving trajectory and the boundaries of multiple road segments;

[0134] An autonomous driving map construction module 830, for constructing an autonomous driving map of a set route using a vehicle driving trajectory, multiple road boundaries, and relative road widths;

[0135] The road structure recognition module 840 is used to identify the road structure of the set route according to the autonomous driving map; the road structure includes at least one of a diverging road and a merging road.

[0136] In one embodiment, the relative road width determination module 820 may include:

[0137] A learning trajectory point extraction submodule is used to use the vehicle driving trajectory as a learning trajectory and extract multiple learning trajectory points from the learning trajectory according to a preset interval distance;

[0138] The relative road width determination submodule is used to determine the relative road width between each learning trajectory point and the corresponding road boundary.

[0139] In one embodiment, the relative road width determination submodule may include:

[0140] A tangent direction determination unit, used to determine the tangent direction of the learning trajectory at each learning trajectory point;

[0141] An emission direction determination unit, used for taking a direction perpendicular to the tangent direction as the emission direction;

[0142] A first length obtaining unit is used to respectively start from each learning trajectory point, emit a ray along the corresponding emission direction, and terminate the emission when the ray encounters the corresponding road boundary, so as to obtain the length of the ray;

[0143] The relative road width determination unit is used to determine the length of the ray as the relative road width between each learning trajectory point and the corresponding road boundary.

[0144] In one embodiment, the relative road width determination submodule may further include:

[0145] The second length obtaining unit is configured to use the preset length as the length of the ray when the ray does not encounter the corresponding road boundary and the length of the ray is greater than or equal to the preset length.

[0146] In one embodiment, the autonomous driving map construction module 830 may include:

[0147] The filtering processing submodule is used to filter each relative road width to obtain the target relative road width corresponding to each learning trajectory point;

[0148] The passable road boundary point calculation submodule is used to calculate the passable road boundary point corresponding to each learning trajectory point by using the corresponding target relative road width; wherein the passable road boundary point includes the left passable road boundary point and the right passable road boundary point;

[0149] The passable road boundary point connection submodule is used to sequentially connect multiple left passable road boundary points to obtain the left boundary of the passable road, and sequentially connect multiple right passable road boundary points to obtain the right boundary of the passable road;

[0150] The passable road area construction submodule is used to construct the passable road area of ​​the set route using the passable road left boundary and the passable road right boundary, and output the passable road area as an autonomous driving map of the set route.

[0151] In one embodiment, the filtering processing submodule may include:

[0152] A first clustering processing unit, used for performing clustering processing on each relative road width to obtain a plurality of first clusters;

[0153] The relative road width adjustment unit is used to traverse each first cluster. If the relative road width of the current first cluster is greater than the relative road width of the adjacent first cluster, the relative road width of the current first cluster is adjusted to the relative road width of the adjacent first cluster until the traversal is completed, and the adjusted target relative road width corresponding to each learning trajectory point is obtained.

[0154] In one embodiment, the road structure identification module 840 may include:

[0155] The road structure recognition submodule is used to identify the road structure of the passable road area according to the target relative road width change and the surrounding ground markings in the passable road area.

[0156] In one embodiment, the surrounding ground markings include guide lines; the road structure identification submodule may include:

[0157] A second clustering processing unit is used to perform clustering processing on the relative road width of each target to obtain a plurality of second clusters;

[0158] A second cluster information processing unit, used to calculate the target relative road width difference between two adjacent second clusters, determine the target relative road width change between two adjacent second clusters, and determine whether two adjacent second clusters correspond to a guide line;

[0159] A diversion road identification unit is used to identify that the road structure of the passable road area at two adjacent second clusters is a diversion road if the target relative road width difference is greater than a first threshold value, and the target relative road width changes from large to small, and there is a corresponding guide line; or

[0160] The merging road identification unit is used to identify that the road structure of the passable road area at two adjacent second clusters is a merging road if the target relative road width difference is greater than a first threshold, and the target relative road width changes from small to large, and there is a corresponding guide line.

[0161] In one embodiment, the second cluster information processing unit may include:

[0162] The adjacent two target relative road width selection subunit is used to select the target relative road width located at the last position in the previous second cluster and the target relative road width located at the first position in the next second cluster from among the two adjacent second clusters;

[0163] a target relative road width difference calculation subunit, used to calculate the absolute value of the difference between the target relative road width at the last position and the target relative road width at the first position, as the target relative road width difference between two adjacent second clusters; and

[0164] The target relative road width change determination subunit is used to determine the width change from the target relative road width at the last position to the target relative road width at the first position as the target relative road width change of two adjacent second clusters.

[0165] In one embodiment, after identifying the road structure of the traversable road area, the device may further include:

[0166] The diverging and merging direction recognition module is used to identify the diverging direction of the diverging road or the merging direction of the merging road according to the position of the guide line.

[0167] In one embodiment, the target relative road width includes a target left road width and a target right road width; and the diverging and merging direction recognition module may include:

[0168] If the guide line is within the preset range of the width of the left road of the first target, it is identified that the diversion direction of the diversion road is the right diversion direction; or,

[0169] If the guide line is within the preset range of the width of the target right road located at the first position, it is identified that the diversion direction of the diversion road is the left diversion direction; or,

[0170] If the guide line is within a preset range of the width of the target left road at the last position, it is identified that the merging direction of the merging road is a left merging direction; or,

[0171] If the guide line is within a preset range of the width of the target right road located at the last position, it is identified that the merging direction of the merging road is the right merging direction.

[0172] As can be seen from this example, the solution provided by this application obtains the vehicle driving trajectory collected on the set route and the multiple road boundaries within the vehicle's perception range; determines the relative road widths of the vehicle driving trajectory and the multiple road boundaries respectively; uses the vehicle driving trajectory, the multiple road boundaries and the relative road width to construct an autonomous driving map of the set route; and identifies the road structure of the set route according to the autonomous driving map; wherein the road structure includes at least one of a diverging road and a merging road. This application collects the vehicle driving trajectory and the multiple road boundaries within the perception range on the set route so as to obtain the relative road width of the set route, and then uses the vehicle driving trajectory, the multiple road boundaries and the relative road width to construct an autonomous driving map of the set route. In this way, even without the support of a high-precision map, the constructed autonomous driving map can be used to identify the diverging road or merging road of the set route.

[0173] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated again here.

[0174] Fig. 9 It is a schematic diagram of the structure of a vehicle shown in an embodiment of the present application.

[0175] See also Fig. 9 , vehicle 900 includes a memory 910 and a processor 920 .

[0176] The processor 920 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0177] The memory 910 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. Among them, ROM can store static data or instructions required by the processor 920 or other modules of the computer. The permanent storage device may be a readable and writable storage device. The permanent storage device may be a non-volatile storage device that does not lose the stored instructions and data even after the computer is powered off. In some embodiments, the permanent storage device uses a large-capacity storage device (such as a magnetic or optical disk, flash memory) as a permanent storage device. In some other embodiments, the permanent storage device may be a removable storage device (such as a floppy disk, optical drive). The system memory may be a readable and writable storage device or a volatile readable and writable storage device, such as a dynamic random access memory. The system memory may store some or all instructions and data required by the processor at run time. In addition, the memory 910 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (such as DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, the memory 910 may include a readable and / or writable removable storage device, such as a laser disc (CD), a read-only digital versatile disc (such as a DVD-ROM, a double-layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (such as an SD card, a mini SD card, a Micro-SD card, etc.), a magnetic floppy disk, etc. The computer-readable storage medium does not include carrier waves and transient electronic signals transmitted wirelessly or wired.

[0178] The memory 910 stores executable codes, and when the executable codes are processed by the processor 920 , the processor 920 can execute part or all of the methods described above.

[0179] In addition, the method according to the present application may also be implemented as a computer program or a computer program product, which includes computer program code instructions for executing some or all of the steps in the above method of the present application.

[0180] Alternatively, the present application can also be implemented as a computer-readable storage medium (or non-temporary machine-readable storage medium or machine-readable storage medium) on which executable code (or computer program or computer instruction code) is stored. When the executable code (or computer program or computer instruction code) is executed by the vehicle's processor, the processor executes part or all of the steps of the above-mentioned method according to the present application.

[0181] The embodiments of the present application have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A road structure recognition method, characterized in that: include: Obtain the vehicle driving trajectory collected on the set route and multiple road boundaries within the vehicle's perception range; respectively determining the relative road widths of the vehicle driving trajectory and the boundaries of the plurality of road sections; constructing an autonomous driving map of the set route using the vehicle driving trajectory, the multiple road boundaries and the relative road width; identifying a road structure of the set route according to the autonomous driving map; wherein the road structure includes at least one of a diverging road and a merging road; The step of identifying the road structure of the set route according to the autonomous driving map includes: Clustering is performed on each target relative road width to obtain multiple second clusters; the target relative road width difference between two adjacent second clusters is calculated, and the target relative road width change between the two adjacent second clusters is determined, and whether the two adjacent second clusters correspond to guide lines; if the target relative road width difference is greater than a first threshold value, and the target relative road width changes from large to small, and there is a corresponding guide line, then the road structure of the passable road area at the two adjacent second clusters is identified as a diversion road; or, if the target relative road width difference is greater than a first threshold value, and the target relative road width changes from small to large, and there is a corresponding guide line, then the road structure of the passable road area at the two adjacent second clusters is identified as a merging road.

2. The method according to claim 1, characterized in that The respectively determining the relative road widths of the vehicle driving trajectory and the boundaries of the multiple road sections includes: Using the vehicle driving trajectory as a learning trajectory, and extracting a plurality of learning trajectory points from the learning trajectory according to a preset interval; Determine the relative road width of each learning trajectory point and the corresponding road boundary.

3. The method according to claim 2, characterized in that The determining of the relative road width between each learning trajectory point and the corresponding road boundary includes: Determining the tangent direction of the learning trajectory at each learning trajectory point; Taking the direction perpendicular to the tangent direction as the emission direction; Starting from each of the learning trajectory points, a ray is emitted along the corresponding emission direction until the ray encounters a corresponding road boundary and the emission is terminated, thereby obtaining the length of the ray; The length of the ray is determined as the relative road width between each learning trajectory point and the corresponding road boundary.

4. The method according to claim 3, characterized in that The method further comprises: When the ray does not encounter the corresponding road boundary and the length of the ray is greater than or equal to a preset length, the preset length is used as the length of the ray.

5. The method according to claim 2, characterized in that: The step of constructing the autonomous driving map of the set route by using the vehicle driving trajectory, the multiple road boundaries and the relative road width includes: Filtering each relative road width to obtain a target relative road width corresponding to each learning trajectory point; The corresponding target relative road width is used to calculate the passable road boundary points corresponding to each of the learning trajectory points; wherein the passable road boundary points include left passable road boundary points and right passable road boundary points; Connecting a plurality of left-side passable road boundary points in sequence to obtain a left passable road boundary, and connecting a plurality of right-side passable road boundary points in sequence to obtain a right passable road boundary; The drivable road left boundary and the drivable road right boundary are used to construct a drivable road area of ​​the set route, and the drivable road area is output as an automatic driving map of the set route.

6. The method according to claim 5, characterized in that The filtering process is performed on each relative road width to obtain the target relative road width corresponding to each learning trajectory point, including: Perform clustering on each relative road width to obtain multiple first clusters; Traverse each first cluster, and if the relative road width of the current first cluster is greater than the relative road width of the adjacent first cluster, adjust the relative road width of the current first cluster to the relative road width of the adjacent first cluster, until the traversal is completed, and obtain the adjusted target relative road width corresponding to each learning trajectory point.

7. The method according to claim 1, characterized in that The calculating the target relative road width difference between two adjacent second clusters and determining the target relative road width change between the two adjacent second clusters includes: In two adjacent second clusters, the target relative road width at the last position in the first second cluster is selected, and the target relative road width at the first position in the second cluster is selected; The absolute value of the difference between the target relative road width at the last position and the target relative road width at the first position is calculated as the target relative road width difference between the two adjacent second clusters; and the width change from the target relative road width at the last position to the target relative road width at the first position is determined as the target relative road width change between the two adjacent second clusters.

8. The method according to claim 7, characterized in that After identifying the road structure of the set route, the method further includes: According to the position of the guide line, the diverging direction of the diverging road is identified, or the merging direction of the merging road is identified.

9. The method according to claim 8, characterized in that The target relative road width includes a target left road width and a target right road width; and the identifying the diverging direction of the diverging road or the identifying the merging direction of the merging road according to the position of the guide line includes: If the guide line is within the preset range of the width of the left side road of the first target, it is identified that the diversion direction of the diversion road is a right diversion direction; or, If the guide line is within the preset range of the width of the target right road located in the first position, it is identified that the diversion direction of the diversion road is a left diversion direction; or, If the guide line is within the preset range of the width of the target left road at the last position, it is identified that the merging direction of the merging road is a left merging direction; or, If the guide line is within a preset range of the width of the target right road at the last position, it is identified that the merging direction of the merging road is a right merging direction.

10. A road structure recognition device, characterized in that: include: A route information acquisition module is set to acquire the vehicle driving trajectory collected on the set route and multiple road boundaries within the vehicle's sensing range; A relative road width determination module, used to respectively determine the relative road widths of the vehicle's driving trajectory and the boundaries of the multiple road segments; An autonomous driving map construction module, configured to construct an autonomous driving map of the set route by using the vehicle driving trajectory, the multiple road boundaries and the relative road width; A road structure recognition module, configured to recognize a road structure of the set route according to the autonomous driving map; the road structure includes at least one of a diverging road and a merging road; The road structure recognition module comprises: The road structure recognition submodule is used to perform clustering processing on the relative road width of each target to obtain multiple second clusters; Calculate the target relative road width difference between two adjacent second clusters, determine the target relative road width change between the two adjacent second clusters, and judge whether the two adjacent second clusters correspond to guide lines; if the target relative road width difference is greater than a first threshold value, and the target relative road width changes from large to small, and corresponds to a guide line, then identify that the road structure of the passable road area at the two adjacent second clusters is a diversion road; or, if the target relative road width difference is greater than a first threshold value, and the target relative road width changes from small to large, and corresponds to a guide line, then identify that the road structure of the passable road area at the two adjacent second clusters is a merging road.

11. A vehicle, characterized in that: include: processor; as well as A memory having executable codes stored thereon, which, when executed by the processor, causes the processor to execute the method according to any one of claims 1 to 9.

12. A computer-readable storage medium having executable codes stored thereon, which, when executed by a processor of a vehicle, causes the processor to execute the method according to any one of claims 1 to 9.

Images