Vehicle pose information optimization method and device, medium and product

Abstract

The invention provides a vehicle pose information optimization method and device, a medium and a product, and the method comprises the steps: determining the predicted pose information of a vehicle according to the point cloud data of at least one sensing point in the driving environment of the vehicle; lane association operation is executed for each sensing point, a target lane line corresponding to each sensing point is obtained, and the lane association operation comprises the steps that the target lane line corresponding to the sensing point is determined from the associated lane lines according to the projection position of the sensing point on each associated lane line and the position distribution characteristics of lane line points on each associated lane line, the associated lane lines are determined based on the sensing points; and according to the distance between each sensing point and the target lane line corresponding to each sensing point, optimizing the predicted pose information to obtain target pose information of the vehicle. According to the embodiment of the invention, the accuracy of vehicle positioning can be improved.

Description

Technical Field

[0001] The present application relates to the field of vehicle technology, and in particular to a method, device, medium and product for optimizing vehicle posture information. Background Art

[0002] In the current L2+ autonomous driving technology, the semantic positioning algorithm is the key technology to achieve precise vehicle positioning. Although it has been widely used in scenes such as highways and urban expressways, it still faces the challenge of insufficient vehicle positioning accuracy. The semantic positioning algorithm of the related technology mainly relies on the registration of high-precision vector maps and perception point clouds. However, in practical applications, traditional registration methods have certain limitations when processing environmental information, and cannot comprehensively and efficiently extract and utilize effective information in the environment, thus affecting the accuracy of vehicle positioning. Therefore, the accuracy of vehicle positioning in the related technology is low. Summary of the invention

[0003] The present application provides a method, device, medium and product for optimizing vehicle posture information, which can improve the accuracy of vehicle positioning.

[0004] In a first aspect, an embodiment of the present application provides a method for optimizing posture information of a vehicle, the method comprising:

[0005] Determining predicted position information of the vehicle based on point cloud data of at least one sensing point in a driving environment of the vehicle;

[0006] Performing lane association operations for each sensing point to obtain a target lane line corresponding to each sensing point. The lane association operation includes: determining a target lane line corresponding to the sensing point from the associated lane lines according to a projection position of the sensing point on each associated lane line and a position distribution feature of lane line points on each associated lane line. The associated lane line is determined based on the sensing point.

[0007] According to the distance between each sensing point and the target lane line corresponding to each sensing point, the predicted posture information is optimized to obtain the target posture information of the vehicle.

[0008] In a second aspect, the present application provides a vehicle posture information optimization device, the device comprising:

[0009] A determination module, configured to determine predicted position information of the vehicle based on point cloud data of at least one sensing point in a driving environment of the vehicle;

[0010] an association module, configured to perform a lane association operation on each sensing point to obtain a target lane line corresponding to each sensing point, wherein the lane association operation includes: determining a target lane line corresponding to the sensing point from the associated lane lines according to a projection position of the sensing point on each associated lane line and a position distribution feature of lane line points on each associated lane line, wherein the associated lane line is determined based on the sensing point;

[0011] The optimization module is used to optimize the predicted posture information according to the distance between each sensing point and the target lane line corresponding to each sensing point, so as to obtain the target posture information of the vehicle.

[0012] In a third aspect, an embodiment of the present application provides an electronic device, the electronic device comprising: a processor and a memory storing computer program instructions;

[0013] When the processor executes the computer program instructions, it implements the vehicle posture information optimization method as in any one of the embodiments in the first aspect.

[0014] In a fourth aspect, an embodiment of the present application provides a computer storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, a method for optimizing vehicle posture information as in any one of the embodiments in the first aspect is implemented.

[0015] In a fifth aspect, an embodiment of the present application provides a computer program product. When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes a method for optimizing vehicle posture information as in any one of the embodiments in the first aspect above.

[0016] In a method, device, medium and product for optimizing the posture information of a vehicle provided in an embodiment of the present application, the predicted posture information is first determined based on the point cloud data of the sensing point in the vehicle driving environment. Then, a lane association operation is performed for each sensing point, and the target lane line corresponding to each sensing point is determined by analyzing the projection position of the sensing point on multiple associated lane lines and the position distribution characteristics between the lane line points on these associated lane lines. This association operation makes full use of the key environmental element of the lane line, and effectively makes up for the defect of insufficient utilization of environmental information by traditional methods. Finally, the predicted posture information is optimized based on the distance between each sensing point and the corresponding target lane line, so as to obtain the target posture information. Therefore, by introducing the lane association operation and the posture information optimization strategy based on the distance between the sensing point and the target lane line, a deep understanding and efficient utilization of the vehicle driving environment is achieved, thereby effectively improving the accuracy of vehicle positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0018] Figure 1 It is a flow chart of a method for optimizing vehicle posture information provided in an embodiment of the present application;

[0019] Figure 2 This is one of the principle schematic diagrams of the vehicle posture information optimization method provided in the embodiment of the present application;

[0020] Figure 3 This is the second principle schematic diagram of the vehicle posture information optimization method provided in the embodiment of the present application;

[0021] Figure 4 is a schematic diagram of an initial map provided in an embodiment of the present application;

[0022] Figure 5 This is the third principle schematic diagram of the vehicle posture information optimization method provided in the embodiment of the present application;

[0023] Figure 6 It is a structural schematic diagram of a vehicle posture information optimization device provided in an embodiment of the present application;

[0024] Figure 7 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0026] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0027] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprising a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0028] In L2+ autonomous driving technology, semantic positioning algorithms, as a key technology for achieving precise vehicle navigation, have been widely used in scenarios such as highways and urban expressways, but their performance is still limited by several technical challenges. The semantic positioning algorithm is highly dependent on high-precision vector maps, which must be acquired in advance through a complex map collection, construction, and mapping process. In the real-time positioning stage, the system needs to accurately align the high-precision map (covering key elements such as lane lines and curbs) loaded from the cloud or offline with the lane lines and curbs point cloud captured by the on-board camera.

[0029] There are three main semantic positioning technology solutions in related technologies: one is a particle filtering-based method, which randomly generates multiple particles within the predicted posture range, rotates and translates the perception point cloud represented by each particle, and evaluates its matching degree to find the best matching posture; the second is an optimization-based method, which constructs a residual function between the perception point cloud and the lane line point cloud, and uses the least squares optimization algorithm to solve the posture with the smallest residual; the third is an algorithm based on brute force matching, which divides the posture according to a certain grid size within the predicted posture range, and rotates and translates the perception point cloud for each divided posture, and also finds the best posture through matching degree evaluation.

[0030] However, these related technical solutions face significant difficulties in the registration process: although the particle filter-based method can find an approximate optimal solution through random sampling and traversal evaluation, it is computationally intensive and inefficient; the optimization-based method solves the optimal pose by constructing a residual function and applying a least squares optimization algorithm, but the construction and optimization process of the residual function in a complex environment is complex and time-consuming; the algorithm based on brute force matching is simple and direct, but it also faces the problem of low computational efficiency, especially in large-scale grid division and pose evaluation. These technical problems not only affect the real-time and accuracy of the semantic positioning algorithm, but also limit the application and development of autonomous driving technology in a wider range of scenarios.

[0031] In order to solve the problems existing in the related art, the embodiments of the present application provide a method, device, medium and product for optimizing the posture information of a vehicle.

[0032] The embodiments of the present application provide a method, device, medium and product for optimizing the posture information of a vehicle. The following first introduces the method for optimizing the posture information of a vehicle provided in the embodiments of the present application. Figure 1 As shown, the method specifically comprises the following steps:

[0033] S100, determining predicted position information of the vehicle based on point cloud data of at least one sensing point in a driving environment of the vehicle.

[0034] Optionally, in the embodiments of the present application, point cloud data refers to three-dimensional point cloud information about the vehicle's driving environment collected in real time by on-board sensors (such as lidar, cameras, etc.). These point cloud data are composed of a large number of points, each of which contains its position information in three-dimensional space (such as X, Y, Z coordinates), and some may also contain other attributes such as color and reflection intensity. Point cloud data is an important basis for autonomous driving systems to understand the surrounding environment, make decisions and plan.

[0035] A perception point is a single point extracted from point cloud data, which represents a specific location in the vehicle's driving environment. These perception points represent the surface positions of specific objects (such as trees, buildings, vehicles, pedestrians, lane lines, etc.) in the vehicle's surroundings, and provide three-dimensional information about the vehicle's surroundings.

[0036] Predicted posture information refers to the position and posture of the vehicle in three-dimensional space, including the vehicle's position coordinates (such as X, Y, Z coordinates) and the vehicle's orientation (such as yaw angle, pitch angle, roll angle). Predicted posture information is based on the vehicle's historical motion data, current sensor data, and possible path planning information, and is initially predicted by the algorithm to obtain the possible position and posture of the vehicle.

[0037] Optionally, in a feasible implementation of the present application, when the autonomous driving vehicle is driving on the road, the various sensors it carries (such as laser radar, camera, etc.) will continuously collect three-dimensional point cloud data of the surrounding environment, and these point cloud data contain a large amount of information about the position, shape and attributes of objects around the vehicle. These point cloud data are acquired in real time, and they provide the vehicle with accurate perception of the surrounding environment.

[0038] Subsequently, these point cloud data are preprocessed to extract key information useful for vehicle positioning. Preprocessing specifically includes filtering out noise points, removing redundant data, and converting point cloud data into a format suitable for subsequent processing. Subsequently, algorithms (such as Simultaneous Localization and Mapping (SLAM) algorithms, deep learning models, etc.) are used to further analyze and interpret these preprocessed point cloud data.

[0039] In this process, the algorithm will try to infer the vehicle's current possible position and posture based on the spatial distribution, shape characteristics, and movement patterns of the point cloud data. Specifically, it involves operations such as registration, feature extraction, and matching of point cloud data. Registration refers to aligning point cloud data acquired at different times or from different perspectives to the same coordinate system for subsequent analysis and comparison. Feature extraction is to extract information with significant differentiation from point cloud data, such as edges, corners, planes, etc. Matching refers to comparing these features with features pre-stored in the map to determine the exact position of the vehicle relative to the map.

[0040] Based on the above information, the algorithm will calculate the predicted position and posture information of the vehicle at the current moment, including position coordinates and posture information. This information is the basis for lane matching, position and posture optimization and other operations in subsequent steps.

[0041] It should be noted that since point cloud data may have problems such as noise, incompleteness or occlusion, the accuracy of the predicted pose information may be affected to a certain extent in practical applications. In order to improve the robustness and accuracy of positioning, the predicted pose information needs to be corrected and optimized.

[0042] S200, performing lane association operations for each perception point respectively to obtain a target lane line corresponding to each perception point, the lane association operation comprising: determining a target lane line corresponding to the perception point from the associated lane lines according to a projection position of the perception point on each associated lane line and a position distribution feature of lane line points on each associated lane line, the associated lane line being determined based on the perception point.

[0043] Optionally, in an embodiment of the present application, the associated lane line is determined based on the perception point. The associated lane line can be a set of lane lines related to the perception point currently being processed in the vehicle driving environment. Specifically, it can be all lane lines around the vehicle that are associated with the perception point. These lane lines can be the left and right boundary lane lines of the lane where the vehicle is located, or the lane lines of the adjacent lane, etc. The associated lane line provides a basis for the subsequent determination of the target lane line specifically corresponding to the perception point.

[0044] Lane points refer to a series of discrete points on the associated lane line, which define the shape and position of the associated lane line. Lane points can be points on the lane boundary or on the lane centerline, depending on the representation of lane data and the needs of subsequent processing.

[0045] Optionally, the position distribution feature mainly describes the relative position between the projection position of the sensing point and the lane line point on the associated lane line, for example, whether the projection position of the sensing point is within the range defined by a certain section of lane line points.

[0046] Optionally, the target lane line is a specific lane line selected from a set of associated lane lines determined based on the sensing point when performing the lane association operation. The target lane line is the lane line that best represents the actual lane position of the sensing point after comprehensively considering the projection position of the sensing point on each associated lane line and the position distribution characteristics between lane line points on each associated lane line.

[0047] Optionally, in a feasible implementation of the present application, each perception point is processed one by one: the perception point is projected onto each determined associated lane line to obtain the corresponding projection position of the perception point on each associated lane line. Then, the specific positional relationship between this projection position and the lane line point on the associated lane line is analyzed. For example, check whether the projection position of the perception point falls between two adjacent lane line points on the associated lane line. If the projection position happens to be in such an interval, it means that from the perspective of the spatial position relationship, the perception point fits closely with this associated lane line, and the associated lane line is the target lane line corresponding to this perception point.

[0048] For another example, we can also see how far the projection position is from the nearest lane line point. If the distance between the projection position and a lane line point on an associated lane line meets a preset distance threshold range, and the distribution of other surrounding lane line points does not show any contradiction, then we can also determine that this associated lane line is the target lane line corresponding to the perception point.

[0049] By comparing and analyzing each perception point with the associated lane lines in this way, the target lane line corresponding to each perception point is determined, thereby providing the necessary basis for optimizing the predicted posture information of the vehicle according to the distance between the perception point and the target lane line in the subsequent S300, thereby ultimately achieving the purpose of improving the vehicle positioning accuracy.

[0050] S300, optimizing the predicted posture information according to the distance between each sensing point and the target lane line corresponding to each sensing point to obtain the target posture information of the vehicle.

[0051] Optionally, in an embodiment of the present application, the predicted posture information is obtained through preliminary calculation and has certain deviations; while the target posture information is obtained after optimization processing and is more accurate and reliable.

[0052] Optionally, in a feasible implementation of the present application, after the target lane line corresponding to each sensing point is determined, the distance from each sensing point to its target lane line must first be determined. For distance calculation, a spatial geometry algorithm can be used to regard the sensing point as a point in space, and the target lane line as a spatial curve or straight line segment composed of a series of ordered points, and the distance value is calculated using the distance calculation principle from a point to a straight line (or curve) or line segment.

[0053] Next, these distance information are used as key basis to optimize the predicted posture information of the vehicle to obtain the target posture information. For example, an optimization model can be constructed to set the distance between the sensing point and the target lane line as the error term in the optimization target. For example, if the sensing point is far away from the target lane line, it means that the current predicted posture may have a large deviation. By adjusting the position coordinates (such as x, y, z coordinates) and attitude angles (such as pitch angle, yaw angle, roll angle) and other parameters in the predicted posture information, the sum of the distances from all sensing points to the target lane line is minimized. This process can be achieved with the help of optimization algorithms such as gradient descent algorithms. Through continuous iterative calculations, the posture parameters are updated in the direction that reduces the sum of the distances until the preset convergence conditions are reached, such as the change in the sum of the distances is less than a certain threshold or the number of iterations reaches the upper limit, thereby obtaining the target posture information.

[0054] In addition, probability-based methods such as Bayesian estimation can also be used. The distance from the sensing point to the target lane line is used as the observation data, and the predicted posture information is used as the variable to be estimated. Based on the prior posture probability distribution and the likelihood function obtained based on the distance observation, the posterior probability distribution is calculated using the Bayesian formula to obtain the optimized posture information estimate. In this process, as more distance information between sensing points and the target lane line is incorporated, the estimation of posture information will become more and more accurate.

[0055] Filtering algorithms, such as extended Kalman filtering, can also be used. The distance between the sensing point and the target lane line is used as the measurement value, and the predicted posture information is used as the state quantity. Using the prediction and update steps of the Kalman filter, the posture information is predicted according to the vehicle's motion model in the prediction step, and the predicted posture information is corrected in combination with the sensing point distance measurement value in the update step. Through continuous cycle prediction and update, the error of the predicted posture information is effectively reduced, and finally the optimized vehicle target posture information is obtained.

[0056] In other embodiments, the curvature information of the lane line can also be introduced as an additional optimization constraint. Specifically, the sensing point can be projected onto a curved line segment that takes into account the curvature of the lane line instead of the straight line of the target lane line, so that the actual distance from the sensing point to the lane line can be calculated more accurately. Since the lane line in the real environment often presents a certain curvature rather than a simple straight line, projecting the sensing point onto a curved line segment can be closer to the actual situation and reduce the error caused by the straight line projection.

[0057] Specifically, we can first use map data and real-time perception point cloud data to accurately estimate the curvature of the target lane line and build a curve model of the target lane line. Then, we project each perception point onto this curve model and calculate the shortest distance from the projected point to the curve. This distance will serve as an important indicator in the optimization process.

[0058] During the optimization process, an iterative method is used to continuously adjust the vehicle's posture parameters until an optimal solution is found, which minimizes the sum (or weighted sum) of the distances from all sensing points to the target lane curve. This optimal solution is the target posture information, which can not only more accurately reflect the vehicle's position and posture at the current moment, but also provide a more reliable foundation for subsequent autonomous driving and navigation tasks.

[0059] In a method for optimizing the posture information of a vehicle provided in an embodiment of the present application, the predicted posture information is first determined based on the point cloud data of the sensing point in the vehicle driving environment. Then, a lane association operation is performed for each sensing point, and the target lane line corresponding to each sensing point is determined by analyzing the projection position of the sensing point on multiple associated lane lines and the position distribution characteristics between the lane line points on these associated lane lines. This association operation makes full use of the lane line as a key environmental element, and effectively makes up for the defect of insufficient utilization of environmental information by traditional methods. Finally, the predicted posture information is optimized based on the distance between each sensing point and the corresponding target lane line, so as to obtain the target posture information. Therefore, by introducing the lane association operation and the posture information optimization strategy based on the distance between the sensing point and the target lane line, a deep understanding and efficient utilization of the vehicle driving environment is achieved, thereby effectively improving the accuracy of vehicle positioning.

[0060] In one embodiment, determining the target lane line corresponding to the perception point from the associated lane lines according to the projection position of the perception point on each associated lane line and the position distribution characteristics of the lane line points on each associated lane line includes:

[0061] Determine at least two lane line points on each associated lane line;

[0062] The projection position of the sensing point on each associated lane line is determined as the reference point of each associated lane line;

[0063] Based on the position distribution relationship between at least two lane line points of each associated lane line and the reference point, a target lane line corresponding to the perception point is determined from the associated lane lines.

[0064] Optionally, in a specific implementation of the present application, the associated lane line includes a plurality of lane line points. In order to obtain at least two lane line points therefrom, an efficient spatial search algorithm may be used, such as a data structure based on a spatial index, such as a K-dimensional tree (kdtree), but not limited to the algorithm):

[0065] First, the perception point is converted to a unified world coordinate system based on the current vehicle predicted pose information, and this perception point is used as the source of the search. Then, all lane line points contained in the associated lane line are used as the target set.

[0066] Using the selected spatial search algorithm, search among the many lane line points corresponding to the associated lane line, with the goal of finding several lane line points that are close to the sensing point in space. For example, set a distance threshold and search for at least two lane line points that are closest to the sensing point within this threshold. The lane line points that were not found are relatively far away from the sensing point. In the preliminary analysis of determining the target lane line for the sensing point, they are considered to have relatively weak relevance to the judgment of the lane position of the sensing point, so they are not considered as the focus of consideration for the time being, but they are still part of the associated lane line as a whole, and can be included in further comprehensive judgments if the analysis requires it.

[0067] After the two lane line points on the associated lane line are determined by the above search method, the projection position of the sensing point on the associated lane line is determined as the reference point. Next, the focus is on determining whether the projection reference point of the sensing point is located between the two lane line points to determine the target lane line. The specific operations are as follows:

[0068] The judgment is made by comparing the relative position relationship of the projected reference point on the line segment formed by the two lane line points. A line segment is constructed with the two lane line points as endpoints, and the coordinate values ​​of the two lane line points and the reference point in the corresponding coordinate system are obtained.

[0069] Subsequently, the vector method or coordinate comparison method can be used to determine whether the projected reference point is on the line segment formed by the two lane line points. For example, using the vector method, the relationship between the vector dot product and the vector modulus can be used to determine whether the projected reference point is on the line segment formed by the two endpoints.

[0070] If the projection reference point is not on the line segment formed by these two lane line points, it means that the spatial position of the perception point relative to this associated lane line is not as expected, and the associated lane line is not the target lane line corresponding to the perception point. At this time, it can be excluded and other associated lane lines can continue to be judged and screened in the same way.

[0071] When more than two lane line points on the associated lane line are determined by the above search method, the projection position of the sensing point on the associated lane line is also determined as the reference point, and then the following analysis is performed on multiple lane line points:

[0072] Calculate the distance from the reference point to each lane line point, and count the lane line points closest to the reference point (for example, take the 3 or 5 closest lane line points, the specific number can be set according to factors such as actual positioning accuracy requirements). Observe the relative position relationship between these relatively close lane line points. If they are relatively concentrated, forming a local gathering area, and the reference point is located in this area or is close to this area (the distance threshold can also be set for judgment), it means that the sensing point is closely related to the associated lane line part represented by these relatively close lane line points in space, and the associated lane line is the target lane line.

[0073] Alternatively, you can also use a curve fitting algorithm (such as the least squares method) to fit these two or more lane line points into a curve, treat the reference point as a point outside the curve, and calculate the shortest distance from the reference point to the fitted curve. If this shortest distance is within the preset minimum value range, it means that the sensing point is very close to the shape fitted by the associated lane line in space, and then combine the overall trend of the lane line point fitting curve with the current driving direction and posture changes of the vehicle (such as steering, acceleration and deceleration, etc.) to make a comprehensive judgment. If the curve trend is consistent with the vehicle's driving trend, for example, the vehicle is turning, and the fitting curve also shows a corresponding change in the curvature of the curve, then the associated lane line is the target lane line corresponding to the sensing point.

[0074] In these optional embodiments, determining at least two lane line points on the associated lane line can focus on key position information and narrow the scope of analysis. Setting the projection position of the perception point as a reference point facilitates intuitive comparison of the spatial relationship with the lane line points on the associated lane line. Determining the target lane line based on the position distribution relationship between the two can improve the accuracy of lane matching. Especially in complex road conditions, it can effectively eliminate interference, allowing the vehicle to more accurately know its position in the lane, providing a reliable basis for subsequent optimization of the vehicle's posture information, thereby enhancing the accuracy of autonomous driving vehicle positioning and the safety and stability of driving.

[0075] In one embodiment, the determining the target lane line corresponding to the perception point from the associated lane lines based on the position distribution relationship between at least two lane line points of each associated lane line and the reference point includes:

[0076] For each associated lane line, executing: if the reference point of the associated lane line is located between at least two lane line points of the associated lane line, determining the associated lane line as a lane line to be selected;

[0077] Determine the target lane line corresponding to the sensing point from the determined lane lines to be selected.

[0078] Optionally, in a specific implementation of the present application, when two lane line points are determined on the associated lane line, it can be determined whether the reference point is on the line segment formed by the two lane line points based on the coordinate values ​​of the two lane line points and the reference point in the corresponding coordinate system. If the reference point is on the line segment formed by the two lane line points, the associated lane line can be determined as the lane line to be selected.

[0079] When there are more than two lane line points on the associated lane line, such as Figure 2 As shown in the figure, assuming that five lane line points on the associated lane line are found, a triangle consisting of the sensing point and the two lane line points closest to the sensing point among the five points is then constructed, and it is checked whether the triangle is an acute triangle. If the triangle consisting of the two lane line points and the sensing point meets the conditions of an acute triangle, then it can be considered that the reference point of the sensing point is located between at least two lane line points of the associated lane line, and therefore the associated lane line can be determined as the candidate lane line corresponding to the sensing point.

[0080] After obtaining all the candidate lane lines, the target lane line corresponding to the perception point is determined from the determined candidate lane lines. Other factors can be comprehensively considered, such as the degree of matching between each candidate lane line and the vehicle's driving direction, the type of lane line (such as main lane, auxiliary lane, etc.) and the environmental characteristics around the lane line, the distance between the lane line and the current position of the vehicle, etc., and finally the target lane line is accurately screened out, providing an accurate basis for the subsequent optimization of vehicle posture information.

[0081] In these optional embodiments, the lane line to be selected is determined by judging whether the reference point is located between the lane line points, which can accurately screen out the lane line that has a strong correlation with the spatial position of the perception point and eliminate interference. Secondly, the target lane line can be further determined from the lane lines to be selected, and more practical factors can be combined to make the selected target lane line more consistent with the actual lane of the vehicle. Overall, the accuracy of lane matching is improved, providing a reliable basis for the subsequent optimization of vehicle posture information, helping to enhance the accuracy of autonomous driving vehicle positioning and ensure driving safety and stability.

[0082] In one embodiment, determining the target lane line corresponding to the perception point from the determined lane lines to be selected includes:

[0083] Determine the distance from the sensing point to each lane line to be selected;

[0084] A target lane line corresponding to the sensing point is determined from the candidate lane lines according to the distance.

[0085] Optionally, in a specific implementation of the present application, first, the distance from the perception point to each lane line to be selected is determined. The perception point can be regarded as a point in space. For each lane line to be selected, a suitable distance measurement method is selected according to the actual application scenario and accuracy requirements, such as Euclidean distance or Manhattan distance. If the Euclidean distance is selected, when the lane line to be selected is presented as a series of coordinate points, the distance formula between two points is used to calculate the distance between the perception point and each point on the lane line, and then the minimum value is taken as the distance from the perception point to the lane line to be selected; if the Manhattan distance is used, the sum of the absolute values ​​of the differences in each coordinate dimension is calculated for measurement.

[0086] Next, the target lane line corresponding to the sensing point is determined from the candidate lane lines based on the calculated distance. The distances from the sensing point to each candidate lane line are compared, and the candidate lane line with the smallest distance is selected and determined as the target lane line corresponding to the sensing point. In this way, the target lane line can be further accurately locked on the basis of the selected candidate lane lines, providing an accurate basis for the subsequent optimization of vehicle posture information.

[0087] In these optional embodiments, the distance from the sensing point to each lane line to be selected is first determined, and the degree of spatial correlation between the sensing point and the lane line to be selected can be evaluated in a quantitative manner. The target lane line is further determined based on the distance, and the target lane line that is closest to the sensing point and has the most consistent position relationship can be accurately selected. Such precise matching effectively reduces lane misjudgment and improves the accuracy and reliability of lane positioning.

[0088] In one embodiment, the step of optimizing the predicted posture information according to the distance between each sensing point and the target lane line corresponding to each sensing point to obtain the target posture information of the vehicle includes:

[0089] Establishing an optimization function according to a distance parameter between each of the perception points and a target line segment corresponding to each of the perception points, wherein the target line segment is composed of lane line points on the target lane line corresponding to the perception point;

[0090] The predicted position and posture information is optimized based on the optimization target of the optimization function to obtain the target position and posture information of the vehicle.

[0091] Optionally, the target line segment may be formed by two lane line points on the target lane line that have a specific positional relationship with the corresponding sensing point. For example, two lane line points closest to the corresponding sensing point are selected to form the target line segment.

[0092] Specifically, we first use the sensing point as a reference, and find the two lane line points closest to the sensing point among the many lane line points contained in the corresponding target lane line through the corresponding distance calculation method (such as Euclidean distance calculation method). The line segment formed by these two lane line points is used as the target line segment, which is then used to calculate the distance parameter between the sensing point and the target line segment, and then used to construct optimization functions and other operations.

[0093] For example, lane line points are selected in a certain order, and two adjacent lane line points that form a specific geometric relationship with the perception point (such as the acute triangle mentioned above, which is a geometric shape that can effectively constrain the position of the perception point) are selected as elements that constitute the target line segment. This ensures that the determined target line segment can more accurately reflect the spatial position relationship between the perception point and the target lane line, laying the foundation for accurately optimizing the vehicle's predicted posture information, so that the final vehicle target posture information is more in line with actual driving conditions.

[0094] Subsequently, the residual distance between each sensing point and the target line segment is calculated. Next, an optimization function is constructed based on these residual distances, and the optimization objective of the optimization function is to minimize the sum of the residual distances from all sensing points to their respective corresponding target line segments. This optimization function can be an error term or a cost function, which comprehensively considers the position errors of all sensing points to find an optimal vehicle posture.

[0095] To solve this optimization problem, you can use optimization libraries such as G2O, Ceres, or GTSAM. These libraries provide a wealth of optimization algorithms and tools that can efficiently handle a variety of complex optimization problems. The optimization function and related parameters are input into these libraries, and the optimization goal is to minimize the sum of residual distances. Then, the optimization library will use internal algorithms and iterative processes to continuously adjust the predicted posture information of the vehicle until an optimal solution, namely the target posture information, is found. At this optimal posture, the sum of the residual distances from all sensing points to their corresponding lane line segments reaches the minimum or reaches the preset threshold, thereby achieving accurate optimization of the vehicle posture information.

[0096] In the embodiment of the present application, an optimization method of the distance from the sensing point to the target line segment is adopted, which has more obvious advantages than the optimization method from point to line. Through the schematic diagram of the acute triangle formed by the sensing point and the lane line point, it can be easily concluded that the point to line segment method has a certain constraint effect on the longitudinal direction, that is, the two points of the acute triangle surround the sensing point. Therefore, this advantage is particularly prominent when the vehicle is driving in a scene with a large turn.

[0097] like Figure 3 As shown in the figure, the large arc here represents the lane line during a large turn. In the point-to-line optimization method, there is a significant difference between the distance from the sensing point to the extended line of the tangent of the two points on the left and the distance from the sensing point to the point of the acute triangle segment formed by these two points and another lane line point. This difference mainly comes from the change in curvature of the lane line during a large turn. In the case of a large turn, the lane line presents a large arc, and the point-to-line optimization method fails to fully consider this curvature change, resulting in a large deviation between the calculated distance and the actual distance.

[0098] In contrast, the point-to-line segment optimization method can more accurately reflect the actual distance from the sensing point to the lane line. This is because this method takes into account the line segment characteristics of the lane line, that is, the lane line can be approximately regarded as a straight line segment in a small segment. When making a large turn, although the lane line is in an arc shape as a whole, it can still be regarded as a straight line segment in a small local segment. Therefore, the point-to-line segment optimization method can more accurately calculate the distance from the sensing point to the lane line, thereby obtaining a more accurate vehicle posture.

[0099] In summary, by optimizing the distance from the perception point to the target line segment of the target lane line, the distance from the perception point to the lane line can be calculated more accurately in large turning scenarios, thereby obtaining a more accurate vehicle posture, thereby improving the accuracy and reliability of the autonomous driving system.

[0100] In these optional embodiments, by calculating the distance between the sensing point and the target line segment on the corresponding target lane line and constructing an optimization function, the accuracy of the vehicle posture information is effectively improved. This method not only takes into account the spatial relationship between the sensing point and the lane line, but also iteratively adjusts the predicted posture through the optimization algorithm until the optimal solution is found, thereby significantly improving the accuracy and reliability of vehicle positioning, providing more accurate basic information for autonomous driving decision-making and navigation, and helping to achieve safer and more efficient autonomous driving.

[0101] In one embodiment, the associated lane line is a lane line in a target map; the target map is determined based on the perception point from a loaded map corresponding to the driving environment, and the geographical area covered by the target map is smaller than that of the loaded map.

[0102] Optionally, the loaded map is a complete map corresponding to the vehicle driving environment, which includes all geographic information of the vehicle driving area, such as various roads, buildings, transportation facilities, etc. It is a relatively comprehensive map resource, which provides basic data for the subsequent generation of the target map.

[0103] The target map is a specific map determined from the loaded map corresponding to the vehicle's driving environment. It is determined based on the perception point and is generated to more accurately focus on the area around the vehicle related to the perception point. Its geographical area is smaller than the loaded map in order to reduce unnecessary information interference and focus on analyzing the situation near the vehicle.

[0104] The associated lane line is the lane line of the lane in the target map. This means that the associated lane line is a component of the lane boundary line in the local area of ​​the target map. When determining the target lane line corresponding to the sensing point, select the associated lane lines contained in the target map. The target map provides a range limit for the associated lane line.

[0105] In one embodiment, the predicted posture information includes a predicted position and a predicted heading;

[0106] Before performing lane association operations on each sensing point to obtain a target lane line corresponding to each sensing point, the method further includes:

[0107] According to the predicted position, acquiring an initial map within the perception range of the vehicle from the loaded map;

[0108] According to the predicted heading, the oncoming lane in the initial map is deleted to obtain a target map, wherein the driving direction of the oncoming lane is opposite to the current driving direction of the vehicle.

[0109] Optionally, in the embodiment of the present application, the initial map refers to high-precision map data within the vehicle's perception range. The target map is a map obtained by filtering and narrowing the initial map according to the predicted position and predicted heading of the vehicle. The target map is more focused on the area where the vehicle may travel in the future, which helps to reduce the computational burden of the autonomous driving system and improve the efficiency and accuracy of decision-making and planning.

[0110] Optionally, in a specific implementation of the present application, as Figure 4As shown, according to the predicted current vehicle posture P_pre (predicted position) and H_pre (predicted heading), only the length of the vehicle's forward L_front perception range (for example, if the perception range is 100 meters, then the map of the forward 100-meter range is obtained) and the short-distance vehicle rearward L_back distance (determined by the length of the historical perception lane line mosaic frame) are obtained, and the ranges W_left and W_right of 10 meters on the left and right of the vehicle are obtained (because the highway generally does not exceed 5 lanes, and the width of each lane is generally 3.75 meters, so the width of 10 meters on the left and right, i.e. 20 meters, is sufficient).

[0111] Specifically, if Figure 4 As shown, mathematical tools such as trigonometric functions can be used to calculate the coordinates of the four key points of the vehicle, namely, the coordinates of the left front point P_front_left, the right front point P_front_right, the left rear point P_back_left, and the right rear point P_back_right of the current position, which are as follows:

[0112] Left front point P_front_left (where the subscripts e and n represent the east and north components of the point respectively):

[0113] P_front_left_e=L_front*sin(H_pre)-W_left*cos(H_pre)+P_pre_e

[0114] P_front_left_n=L_front*cos(H_pre)+W_left*sin(H_pre)+P_pre_n

[0115] Right front point P_front_right (where the subscripts e and n represent the east and north components of the point respectively):

[0116] P_front_right_e=L_front*sin(H_pre)+W_left*cos(H_pre)+P_pre_e

[0117] P_front_right_n=L_front*cos(H_pre)-W_left*sin(H_pre)+P_pre_n

[0118] The left back point P_back_left (where the subscripts e and n represent the east and north components of the point respectively):

[0119] P_back_left_e=L_back*sin(H_pre)-W_left*cos(H_pre)+P_pre_e

[0120] P_back_left_n=L_back*cos(H_pre)+W_left*sin(H_pre)+P_pre_n

[0121] The right back point P_back_right (where the subscripts e and n represent the east and north components of the point respectively):

[0122] P_back_right_e=L_back*sin(H_pre)+W_left*cos(H_pre)+P_pre_e

[0123] P_back_right_n=L_back*cos(H_pre)-W_left*sin(H_pre)+P_pre_n

[0124] The range of the rectangular frame enclosed by the four points in front, behind, left and right obtained by the above calculation is the range of the initial map that needs to be screened.

[0125] After obtaining the initial map, it is necessary to filter the lanes in the initial map according to the predicted heading to further narrow and generate the target map. Specifically, all lane lines contained in the initial map are first traversed. In this process, effective mathematical tools such as heading angles or direction vectors are used to measure the relationship between each lane line and the predicted heading. For example, by calculating the angle between them, the direction of the lane line can be expressed by a direction vector based on the coordinate system and vector operation knowledge, and then combined with the vector corresponding to the predicted heading of the vehicle, the specific angle value is calculated using the vector angle calculation formula. The size of this value intuitively reflects the degree of deviation between the lane line and the predicted heading; or a certain algorithm can be used to calculate the similarity between the two, reflecting the closeness of the relationship between the two from another perspective.

[0126] Next, set reasonable thresholds, including angle thresholds and similarity thresholds (specific values ​​can be determined based on actual map accuracy, vehicle positioning requirements and other factors). When judging a lane line, if the angle between it and the predicted heading is less than the set angle threshold, it means that the direction of this lane line is close to the predicted heading of the vehicle, and they are lane lines in the same direction; or if the similarity between it and the predicted heading is higher than the set similarity threshold, it also indicates that the lane line meets the expectations of the vehicle's driving direction. Such lane lines will be retained and become part of the target map.

[0127] On the contrary, if the angle between a lane line and the predicted heading is greater than the angle threshold, or the similarity is lower than the similarity threshold, it means that this lane line is likely to be an oncoming lane with a driving direction opposite to the current driving direction of the vehicle. In this case, it will be deleted from the initial map.

[0128] After such a comprehensive and detailed filtering and screening process, the numerous lane lines in the initial map will be effectively reduced, and only those lane lines that are highly consistent with the predicted heading will be left. These retained lane lines together constitute the target map, which enables the target map to focus on the relevant lane information in the expected direction of travel of the vehicle, providing a more targeted data basis for subsequent operations such as determining associated lane lines and optimizing vehicle posture.

[0129] In these optional embodiments, by first acquiring the initial map within the perception range according to the predicted position, and then filtering the lanes according to the predicted heading, the amount of map data processing can be significantly reduced, while ensuring a high degree of consistency between the target map and the vehicle's future driving path. This not only improves the response speed and accuracy of the autonomous driving system, but also effectively reduces the consumption of computing resources, providing a strong guarantee for the safe and efficient driving of the vehicle.

[0130] In one embodiment, the step of deleting the oncoming lane in the initial map according to the predicted heading to obtain the target map includes:

[0131] Each lane in the initial map is used as an initial lane and processed as follows: multiple reference points corresponding to the initial lane are determined; multiple candidate point groups are determined from the multiple reference points; based on the reference distance between each candidate point group and the predicted position, a target point group corresponding to the initial lane is determined from the multiple candidate point groups; the reference point is a point on the center line of the initial lane, and the candidate point group is at least two adjacent reference points from the multiple reference points;

[0132] Determining an opposite lane from each of the initial lanes according to the target point groups corresponding to each of the initial lanes;

[0133] The oncoming lane is deleted from the initial map to obtain the target map.

[0134] Alternatively, the reference point may be a point on the initial lane centerline. The lane centerline is a line connecting the center positions of the lane in a geometric sense. These points are arranged continuously in a certain order, which can well reflect the shape and direction of the lane.

[0135] A candidate point group is a combination selected from multiple reference points. Specifically, it is composed of at least two adjacent reference points from multiple reference points. You can think of them as a small continuous set of points intercepted from the sequence of reference points. For example, if the reference points are points numbered in sequence along the center line of the lane, then the candidate point group may be a group of two or three adjacent points, etc. These groups are the basis for the subsequent determination of the target point group. The point group that best meets the requirements is found by comparing different candidate point groups.

[0136] The reference distance is an indicator used to measure the relationship between the candidate point group and the predicted position of the vehicle. The reference distance is determined based on the sub-distances between each reference point in the candidate point group and the predicted position, and comprehensively considers the sub-distances from multiple reference points (at least two) in the candidate point group to the predicted position. For example, it can be the sum, average, or result obtained by some weighted calculation of these sub-distances. By comparing the reference distances of different candidate point groups, it is possible to more accurately determine which candidate point group (i.e., target point group) has the closest relationship with the predicted position of the vehicle.

[0137] The target point group is the most representative point group for each initial lane, selected from all candidate point groups based on the reference distance to the predicted position. This point group can better reflect the positional relationship between the vehicle and the initial lane, and plays a key role in subsequent operations such as determining whether the lane is an oncoming lane. The target point group can be used to calculate information such as the lane heading, thereby determining whether the lane is consistent with the vehicle's driving direction.

[0138] Optionally, in a specific implementation of the present application, as Figure 5 As shown, first, for each initial lane within the initial map range, all reference points arranged in a preset order on the initial lane are traversed, and two consecutive adjacent reference points among these reference points are combined into multiple groups of candidate point groups. For any group of candidate point groups (such as Figure 5 As shown, it contains two consecutive lane centerline points P1 and P2), calculates the norm distance between the predicted position P_pre of the vehicle and the two reference points in the candidate point group and obtains the reference distance corresponding to the candidate point group, that is, |P_pre-P1|+|P_pre-P2|, where "|……|" represents the norm (or Euclidean distance) calculation between the coordinates of the two points. This process is equivalent to measuring the sum of the vertical projection distances from the predicted position of the vehicle to the line segments connecting each two adjacent reference points on the lane in two-dimensional or three-dimensional space, which is used as an indicator to measure the proximity of the vehicle to the lane.

[0139] After calculating the reference distances of all candidate point groups, we then compare the reference distances of all candidate point groups and find the candidate point group with the smallest reference distance. This candidate point group is the target point group corresponding to the initial lane to be determined. Because this target point group is closest to the predicted position of the vehicle, it can relatively accurately reflect the position of the lane where the vehicle may currently be located. It also provides a key data basis for further judging the lane conditions and determining the oncoming lane. For example, we can calculate the lane heading and other related information based on the reference points in the target point group to analyze the relationship between the lane and the vehicle's driving direction.

[0140] It should be noted that if the candidate point group includes more than two reference points, the norm distance between the predicted vehicle position and each reference point in the candidate point group can be calculated first, and then the average of these distance values ​​can be calculated as the reference distance. This method reflects the proximity between the predicted vehicle position and the candidate point group from the perspective of overall average through averaging, which to a certain extent avoids the significant impact of abnormal distances of individual reference points on the judgment, and can more smoothly measure the distance relationship between the two, thereby helping to screen out a suitable target point group.

[0141] After obtaining the target point groups corresponding to each initial lane, the oncoming lane is determined based on these target point groups. The lane heading of each initial lane is determined by corresponding mathematical calculations (such as using straight line equations or vector operations) using the coordinate information of each point in the target point group. The lane heading reflects the tangent direction of the center line of the lane, that is, the direction of the lane. Afterwards, the lane heading of each initial lane is compared with the predicted heading of the vehicle, and the heading difference between them is calculated. By setting a reasonable heading difference threshold (the threshold can be determined comprehensively based on factors such as actual map accuracy and vehicle driving characteristics), when the difference between the lane heading of an initial lane and the predicted heading exceeds this threshold, the initial lane is determined to be an oncoming lane.

[0142] Finally, the determined oncoming lanes are deleted from the initial map. After such screening and filtering, only the lanes that match the predicted heading of the vehicle remain in the initial map. These retained lanes together constitute the target map, which makes it more consistent with the subsequent driving path of the vehicle, reduces unnecessary data interference, and provides a more accurate and streamlined data basis for related operations such as vehicle posture information optimization. At the same time, it improves the efficiency of map data processing to a certain extent and reduces the consumption of computing resources.

[0143] In these optional embodiments, by determining the reference points, candidate point groups, and target point groups corresponding to the initial lane, the positional relationship between the vehicle and each lane can be accurately measured, making subsequent judgments more reliable. Next, the oncoming lane is determined based on the target point group, and the lane that does not match the vehicle's driving direction can be accurately screened out to avoid interference from irrelevant lane data. Finally, the oncoming lane is deleted to obtain the target map, which not only makes the map data more streamlined and consistent with the vehicle's subsequent driving path, but also improves map processing efficiency, reduces computing resource consumption, and provides an accurate and targeted data basis for operations such as vehicle posture information optimization, helping vehicles to drive more safely and efficiently.

[0144] In one embodiment, determining the oncoming lane from each of the initial lanes according to the target point groups corresponding to each of the initial lanes includes:

[0145] For each initial lane, respectively: determining the lane heading of the initial lane according to the coordinate difference of the target point group corresponding to the initial lane, wherein the coordinate difference is the horizontal coordinate difference and the vertical coordinate difference between each reference point in the target point group in the two-dimensional coordinate system;

[0146] The oncoming lane is determined based on the lane headings and predicted headings of each initial lane.

[0147] Optionally, in a specific implementation of the present application, for each initial lane, the lane heading is first determined based on its corresponding target point group. Specifically, the horizontal coordinate difference and the vertical coordinate difference between each reference point in the target point group in the two-dimensional coordinate system, that is, the coordinate difference, are focused on.

[0148] For example, Figure 5 As shown, the target point group includes two reference points P1 and P2. First, their difference in the ordinate (corresponding to the y-axis of the map) is calculated, such as the ordinate difference between P2 and P1 (P2_y-P1_y). This difference reflects the change of the lane centerline in the vertical direction; then, their difference in the abscissa (corresponding to the x-axis of the map) is calculated, such as (P2_x-P1_x). This difference reflects the change of the lane centerline in the horizontal direction.

[0149] Then, the lane heading is determined by using the difference between the two coordinate directions. The calculation method can be to take the arc tangent of the ratio between the difference between the ordinate and the abscissa, that is, the lane heading H_p1p2 = arctan((P2_y-P1_y) / (P2_x-P1_x)). It should be noted here that the order of subtraction cannot be reversed, because the reference points in the map are arranged in a preset order, P2 is after P1, which ensures that the calculated lane heading is consistent with the actual direction of the lane.

[0150] After obtaining the lane heading of each initial lane, the next step is to compare it with the predicted heading of the vehicle to determine the opposite lane. The specific operation is to calculate the difference between the lane heading H of each initial lane and the predicted heading H_pre of the vehicle (delta_H=fabs(H_p1p2-H_pre)), where fabs means taking the absolute value to measure the degree of deviation between the two. Then, determine whether this difference is greater than a preset threshold (this threshold can be set comprehensively based on factors such as actual map accuracy and vehicle driving scenarios, such as setting it to a suitable angle value). If delta_H is greater than the threshold, it means that the direction of this initial lane is opposite to the current driving direction of the vehicle, and the initial lane can be determined to be an opposite lane. It can be deleted from the initial map later, thereby completing the screening of the initial map and obtaining a target map that better fits the vehicle's driving direction.

[0151] In other embodiments, when the target point group includes more than two reference points, the lane heading of the initial lane can be calculated by a segmented calculation method. Specifically, first select two adjacent reference points (such as P1 and P2) in the target point group, and calculate the heading angle corresponding to this small segment in the same way as the lane heading is determined by the previous two reference points, which is recorded as H1; then select another two adjacent reference points (such as P2 and P3) to calculate the heading angle H2, and so on, to obtain multiple local heading angle values. Then, these local angle values ​​are comprehensively considered, and a comprehensive angle value that can represent the lane heading corresponding to the entire target point group is obtained by averaging, weighted averaging (different weights are assigned according to factors such as the relative importance of each segment in the entire lane), etc., so as to reflect the overall direction of the lane, facilitate subsequent comparison with the predicted vehicle heading, and accurately determine whether it is an oncoming lane.

[0152] In these optional embodiments, the lane heading is determined based on the coordinate difference of the target point group, which can accurately quantify the actual direction of the lane and present the lane direction characteristics in a digitized form. Then, it is compared with the predicted heading to determine the opposite lane, so that the judgment has a reliable basis and the lane opposite to the vehicle's travel direction can be accurately screened out. In this way, irrelevant opposite lane information can be effectively eliminated, map data can be streamlined, and subsequent map-based operations can be more targeted and improve computing efficiency.

[0153] Figure 6 A schematic diagram of the structure of a vehicle posture information optimization device provided in another embodiment of the present application is shown. For the sake of convenience of explanation, only the parts related to the embodiment of the present application are shown.

[0154] Reference Figure 6 , the vehicle posture information optimization device may include:

[0155] A determination module 601 is used to determine predicted position information of the vehicle based on point cloud data of at least one sensing point in the driving environment of the vehicle;

[0156] An association module 602 is used to perform a lane association operation on each perception point to obtain a target lane line corresponding to each perception point, wherein the lane association operation includes: determining a target lane line corresponding to the perception point from the associated lane lines according to a projection position of the perception point on each associated lane line and a position distribution feature of lane line points on each associated lane line, wherein the associated lane line is determined based on the perception point;

[0157] The optimization module 603 is used to optimize the predicted posture information according to the distance between each sensing point and the target lane line corresponding to each sensing point, so as to obtain the target posture information of the vehicle.

[0158] In one embodiment, the association module 602 may include:

[0159] A first determination submodule, used to determine at least two lane line points on each associated lane line;

[0160] A second determination submodule is used to determine the projection position of the perception point on each associated lane line as the reference point of each associated lane line;

[0161] The third determination submodule is used to determine the target lane line corresponding to the perception point from the associated lane lines based on the position distribution relationship between at least two lane line points of each associated lane line and the reference point.

[0162] In one embodiment, the third determining submodule may include:

[0163] A first determining unit is configured to execute, for each associated lane line: if a reference point of the associated lane line is located between at least two lane line points of the associated lane line, determining the associated lane line as a lane line to be selected;

[0164] The second determining unit is used to determine the target lane line corresponding to the perception point from the determined lane lines to be selected.

[0165] In one embodiment, the second determining unit may include:

[0166] A first determination subunit is used to determine the distance from the sensing point to each lane line to be selected;

[0167] The second determining subunit is used to determine the target lane line corresponding to the sensing point from the lane lines to be selected according to the distance.

[0168] In one embodiment, the optimization module 603 may include:

[0169] A construction submodule, for establishing an optimization function according to a distance parameter between each of the perception points and a target line segment corresponding to each of the perception points, wherein the target line segment is composed of lane line points on the target lane line corresponding to the perception point;

[0170] The optimization submodule is used to optimize the predicted posture information based on the optimization target of the optimization function to obtain the target posture information of the vehicle.

[0171] In one embodiment, the associated lane line is a lane line in a target map; the target map is determined based on the perception point from a loaded map corresponding to the driving environment, and the geographical area covered by the target map is smaller than that of the loaded map.

[0172] In one embodiment, the predicted posture information includes a predicted position and a predicted heading; the posture information optimization device of the vehicle may further include:

[0173] an acquisition module, configured to acquire an initial map within a sensing range of the vehicle from the loaded map according to the predicted position;

[0174] The deletion module is used to delete the opposite lane in the initial map according to the predicted heading to obtain a target map, wherein the driving direction of the opposite lane is opposite to the current driving direction of the vehicle.

[0175] In one embodiment, the deletion module may include:

[0176] The fourth determination submodule is used to process each lane in the initial map as an initial lane as follows: determine multiple reference points corresponding to the initial lane; determine multiple candidate point groups from the multiple reference points; determine a target point group corresponding to the initial lane from the multiple candidate point groups based on a reference distance between each candidate point group and a predicted position; the reference point is a point on the center line of the initial lane, and the candidate point group is at least two adjacent reference points from the multiple reference points;

[0177] A fifth determination submodule, configured to determine an oncoming lane from each of the initial lanes according to a target point group corresponding to each of the initial lanes;

[0178] The deletion submodule is used to delete the oncoming lane from the initial map to obtain the target map.

[0179] In one embodiment, the fifth determining submodule may include:

[0180] A third determination unit is configured to respectively perform, for each initial lane: determining a lane heading of the initial lane according to a coordinate difference of a target point group corresponding to the initial lane, wherein the coordinate difference is a horizontal coordinate difference and a vertical coordinate difference between reference points in the target point group in a two-dimensional coordinate system;

[0181] The fourth determining unit is used to determine the opposite lane according to the lane headings and the predicted headings of the initial lanes.

[0182] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present application, and are devices corresponding to the above-mentioned methods. All implementation methods in the above-mentioned method embodiment are applicable to the embodiments of the device. Its specific functions and technical effects brought about can be found in the method embodiment part, and will not be repeated here.

[0183] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0184] Figure 7 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application is shown.

[0185] The device may include a processor 701 and a memory 702 storing program instructions.

[0186] When the processor 701 executes the program, the steps in any of the above method embodiments are implemented.

[0187] Exemplarily, the program may be divided into one or more modules / units, one or more modules / units are stored in the memory 702 and executed by the processor 701 to complete the present application. One or more modules / units may be a series of program instruction segments capable of completing a specific function, and the instruction segments are used to describe the execution process of the program in the device.

[0188] Specifically, the processor 701 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0189] The memory 702 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 702 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 702 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 702 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 702 is a non-volatile solid-state memory.

[0190] The memory may include read-only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical or other physical / tangible memory storage devices. Thus, typically, the memory includes one or more tangible (non-transitory) readable storage media (e.g., memory devices) encoded with software including computer executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present disclosure.

[0191] The processor 701 implements any one of the methods in the above embodiments by reading and executing program instructions stored in the memory 702 .

[0192] In one example, the electronic device may further include a communication interface 703 and a bus 710. The processor 701, the memory 702, and the communication interface 703 are connected via the bus 710 and communicate with each other.

[0193] The communication interface 703 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0194] Bus 710 includes hardware, software or both, and the parts of online data flow billing equipment are coupled to each other. For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front-end bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 710 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the present application considers any suitable bus or interconnection.

[0195] In addition, in combination with the method in the above embodiment, the embodiment of the present application can provide a storage medium for implementation. The storage medium stores program instructions; when the program instructions are executed by a processor, any one of the methods in the above embodiment is implemented.

[0196] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0197] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0198] An embodiment of the present application provides a computer program product, which is stored in a storage medium. The program product is executed by at least one processor to implement the various processes of the above-mentioned method embodiment and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0199] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0200] The functional modules shown in the above block diagram can be implemented as hardware, software, firmware or their combination. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), suitable firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. Programs or code segments can be stored in machine-readable media, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable media" can include any medium that can store or transmit information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. Code segments can be downloaded via computer grids such as the Internet, intranets, etc.

[0201] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0202] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device (system) and program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0203] The above are only specific implementation methods of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.

Claims

1. A method for optimizing vehicle posture information, characterized in that: The method comprises: Determining predicted position information of the vehicle based on point cloud data of at least one sensing point in a driving environment of the vehicle; Performing a lane association operation for each sensing point to obtain a target lane line corresponding to each sensing point, wherein the lane association operation includes: determining a target lane line corresponding to the sensing point from the associated lane lines according to a projection position of the sensing point on each associated lane line and a position distribution feature of lane line points on each associated lane line, wherein the associated lane line is determined based on the sensing point; The predicted posture information is optimized according to the distance between each sensing point and the target lane line corresponding to each sensing point to obtain the target posture information of the vehicle.

2. The vehicle posture information optimization method according to claim 1, characterized in that: The determining, from the associated lane lines, a target lane line corresponding to the perception point according to the projection position of the perception point on each associated lane line and the position distribution characteristics of the lane line points on each associated lane line, comprises: Determine at least two lane line points on each associated lane line; The projection position of the sensing point on each associated lane line is determined as the reference point of each associated lane line; Based on the position distribution relationship between at least two lane line points of each associated lane line and a reference point, a target lane line corresponding to the perception point is determined from the associated lane lines.

3. The vehicle posture information optimization method according to claim 2, characterized in that: The determining, from the associated lane lines, a target lane line corresponding to the perception point based on a position distribution relationship between at least two lane line points of each associated lane line and a reference point, comprises: For each associated lane line, executing: if the reference point of the associated lane line is located between at least two lane line points of the associated lane line, determining the associated lane line as a lane line to be selected; Determine the target lane line corresponding to the sensing point from the determined lane lines to be selected.

4. The vehicle posture information optimization method according to claim 3, characterized in that: The step of determining a target lane line corresponding to the sensing point from the determined lane lines to be selected includes: Determine the distance from the sensing point to each lane line to be selected; A target lane line corresponding to the sensing point is determined from the candidate lane lines according to the distance.

5. The vehicle posture information optimization method according to claim 1, characterized in that: The step of optimizing the predicted posture information according to the distance between each sensing point and the target lane line corresponding to each sensing point to obtain the target posture information of the vehicle includes: Establishing an optimization function according to a distance parameter between each of the perception points and a target line segment corresponding to each of the perception points, wherein the target line segment is composed of lane line points on the target lane line corresponding to the perception point; The predicted position and posture information is optimized based on the optimization target of the optimization function to obtain the target position and posture information of the vehicle.

6. The vehicle posture information optimization method according to claim 1, characterized in that: The associated lane line is a lane line in a target map; the target map is determined based on the perception point from a loaded map corresponding to the driving environment, and a geographical area covered by the target map is smaller than that of the loaded map.

7. The vehicle posture information optimization method according to claim 6, characterized in that: The predicted position information includes a predicted position and a predicted heading; Before performing lane association operations on each sensing point to obtain a target lane line corresponding to each sensing point, the method further includes: According to the predicted position, acquiring an initial map within the perception range of the vehicle from the loaded map; According to the predicted heading, the oncoming lane in the initial map is deleted to obtain a target map, wherein the driving direction of the oncoming lane is opposite to the current driving direction of the vehicle.

8. The method for optimizing vehicle posture information according to claim 7, characterized in that: The step of deleting the opposite lane in the initial map according to the predicted heading to obtain a target map includes: Each lane in the initial map is used as an initial lane and processed as follows: multiple reference points corresponding to the initial lane are determined; multiple candidate point groups are determined from the multiple reference points; based on the reference distance between each candidate point group and the predicted position, a target point group corresponding to the initial lane is determined from the multiple candidate point groups; the reference point is a point on the center line of the initial lane, and the candidate point group is at least two adjacent reference points from the multiple reference points; Determining an opposite lane from each of the initial lanes according to the target point groups corresponding to each of the initial lanes; The oncoming lane is deleted from the initial map to obtain the target map.

9. The vehicle posture information optimization method according to claim 8, characterized in that: The determining of the oncoming lane from each of the initial lanes according to the target point groups corresponding to each of the initial lanes includes: For each initial lane, respectively: determining the lane heading of the initial lane according to the coordinate difference of the target point group corresponding to the initial lane, wherein the coordinate difference is the horizontal coordinate difference and the vertical coordinate difference between each reference point in the target point group in the two-dimensional coordinate system; The oncoming lane is determined based on the lane headings and predicted headings of each initial lane.

10. An electronic device, characterized in that: The device comprises: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the vehicle posture information optimization method as described in any one of claims 1-9 is implemented.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by the processor, the vehicle posture information optimization method as described in any one of claims 1-9 is implemented.

12. A computer program product, characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the vehicle posture information optimization method as described in any one of claims 1-9.

Images