Autonomous driving method, computer device, storage medium, program product and mobile platform

Abstract

An autonomous driving method, a computer device, a storage medium, a program product and a mobile platform. The method comprises: on the basis of environmental perception data, determining a topological path and a curb outer contour line within a perception range; generating a road centerline on the basis of the topological path and the curb outer contour line; and at least on the basis of the road centerline, generating a traveling trajectory for autonomous driving.

Description

Automatic driving method, computer device, storage medium, program product and mobile platform TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving, and in particular to an automatic driving method, a computer device, a storage medium, a program product and a mobile platform. BACKGROUND

[0002] Current mainstream automatic driving schemes mainly focus on structured roads, and driving safety and stability mainly rely on lane lines constructed by humans. For unstructured roads (for example, rural roads, road sections with chaotic lane lines), current perception can only provide road edges and non-road areas, and cannot generate rich road semantics. Most self-driving systems need user intervention to take over when encountering scenes with low road quality. Therefore, there is an urgent need for a scheme that can realize automatic driving on unstructured roads. SUMMARY

[0003] Embodiments of the present application provide an automatic driving method, a computer device, a storage medium, a program product and a mobile platform, which are used to at least solve one of the above technical problems.

[0004] In a first aspect, embodiments of the present application provide an automatic driving method, comprising:

[0005] determining a topological path and a road edge contour line within a perception range according to environment perception data;

[0006] generating a road center line according to the topological path and the road edge contour line;

[0007] generating a driving trajectory according to at least the road center line for automatic driving.

[0008] In some embodiments, the determining a topological path within a perception range according to environment perception data comprises: generating a Euclidean distance potential field map according to environment perception data; generating a V-Graph based on the Euclidean distance potential field map; and determining a topological path according to the V-Graph.

[0009] In some embodiments, the generating a Euclidean distance potential field map according to environment perception data comprises: generating a grid map of a current scene; determining an impassable area according to environment perception data, and filling corresponding grids on the grid map with a preset value; determining a passable area according to environment perception data, and filling distances to the nearest obstacles in corresponding grids on the grid map to obtain a Euclidean distance potential field map.

[0010] In some embodiments, determining the road edge contour line within the perception range according to the environment perception data comprises: obtaining road edge scatter data according to the environment perception data, and performing convex hull processing on the road edge scatter data to obtain a single polygon contour; and determining the road edge contour line according to the topological path and the polygon contour.

[0011] In some embodiments, performing convex hull processing on the road edge scatter data to obtain a single polygon contour comprises: performing convex hull clustering on the road edge scatter data; performing convex hull merging processing on the convex hull clustering result to obtain an initial polygon contour; and performing singular point filtering on the initial polygon contour to obtain the single polygon contour.

[0012] In some embodiments, generating the road center line according to the topological path and the road edge contour line comprises: dividing the topological path into a constrained valid segment and an unconstrained fuzzy segment according to attributes of obstacles on both sides of the topological path, wherein the attributes of the obstacles include road element obstacles and non-road region obstacles, the road element obstacles correspond to the constrained valid segment, and the non-road region obstacles correspond to the unconstrained fuzzy segment; and moving the road edge contour line corresponding to the constrained valid segment by a preset distance in a direction of the vehicle to generate the road center line.

[0013] In some embodiments, before determining the topological path and the road edge contour line within the perception range according to the environment perception data, the automatic driving method further comprises: generating the driving mode switching instruction in response to a driving mode switching operation of a user; or performing scene recognition according to the environment perception data, and generating the driving mode switching instruction when a scene recognition result meets a preset condition; and when the driving mode switching instruction is detected, controlling the vehicle to switch from a first driving mode to a second driving mode, wherein the first driving mode is a current driving mode of the vehicle, and the second driving mode is an unstructured road automatic driving mode.

[0014] In some embodiments, the automatic driving method further comprises: selecting the center line according to at least one of safety, comfort, and consistency; and performing trajectory planning according to the selected center line and obstacles on the center line for completing automatic driving on an unstructured road.

[0015] In a second aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method according to any one of the embodiments of the present application.

[0016] In a third aspect, the present application provides a computer readable storage medium having a computer program / instruction stored thereon, wherein the computer program / instruction is executed by a processor to implement the steps of the method according to any one of the embodiments of the present application.

[0017] In a fourth aspect, the present application provides a computer program product, comprising computer programs / instructions, wherein the computer programs / instructions, when executed by a processor, implement the steps of the method according to any one of the embodiments of the present application.

[0018] In a fifth aspect, the present application provides a mobile platform, wherein the mobile platform is installed with the computer device according to any one of the embodiments of the present application.

[0019] The present application determines a topology path and an along-outline line based on the environmental perception data of the vehicle, and further generates a road center line based on the topology path and the along-outline line for autonomous driving. The autonomous driving on an unstructured road is achieved without relying on the recognition of lane lines. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0021] FIG. 1 is a flowchart of an embodiment of the autonomous driving method of the present application;

[0022] FIG. 2 is a flowchart of another embodiment of the autonomous driving method of the present application;

[0023] FIG. 3 is an effect diagram of the Euclidean distance potential field calculated in an embodiment of the present application;

[0024] FIG. 4 is a schematic diagram of the generation of a Voronoi diagram in an embodiment of the present application;

[0025] FIG. 5 is a schematic diagram of sampling of a goal point candidate set in an embodiment of the present application;

[0026] FIG. 6 is a flowchart of another embodiment of the autonomous driving method of the present application;

[0027] FIG. 7 is a schematic diagram of the process of merging of convex hulls and extraction of an along-outline line of a single side in an embodiment of the present application;

[0028] FIG. 8 is a flowchart of another embodiment of the autonomous driving method of the present application;

[0029] FIG. 9 is a schematic diagram of local map cutting in the present application;

[0030] FIG. 10 is a flowchart of another embodiment of the autonomous driving method of the present application;

[0031] FIG. 11 is a schematic diagram of a rural mode state machine used in the autonomous driving method of the present application;

[0032] FIG. 12 is a flowchart of another embodiment of the automatic driving method of the present application;

[0033] FIG. 13 is a structural schematic diagram of an embodiment of the computer device of the present application. DETAILED DESCRIPTION

[0034] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions of the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0035] It should also be noted that, in this document, the terms "comprise", "contain", not only include those elements, but also include other elements not explicitly listed, or also include elements inherent to such a process, method, article or device. Without more limitations, the elements defined by the statement "comprise" do not exclude the presence of other identical elements in the process, method, article or device comprising the described elements.

[0036] Current mainstream automatic driving solutions are mostly limited to structured roads, and are still in an exploratory pre-research stage for unstructured roads. Most automatic driving systems need user intervention to take over when encountering scenes with low road quality. The working conditions of unstructured roads are complex, with high diversity, difficult data collection, poor annotation consistency, low reliability of lane line perception and reasoning, and cannot guarantee driving safety and consistency.

[0037] The present application proposes an automatic driving method which can be applied to unstructured road scenes. The unstructured road is relative to the structured road.

[0038] The structured road generally refers to highways, urban trunk roads and other well-structured roads. Such roads have clear road marking lines, the background environment of the road is relatively simple, and the geometric characteristics of the road are also relatively obvious. Therefore, the road detection problem for it can be simplified as lane line or road boundary detection.

[0039] Unstructured road generally refers to urban non-main road, rural street and other low-structured road. Such road has no lane line and clear road boundary, and is difficult to distinguish road area from non-road area due to the influence of shadow, water mark and the like. The variable road type, complex environment background, shadow, water mark and changing weather are the difficulties faced by unstructured road detection and the main research direction of current road recognition technology.

[0040] As shown in FIG. 1, it is a flowchart of an embodiment of the automatic driving method of the present application. The execution subject of the present application is a computer device, a server or a processor with computing function. The present application takes the application of the method in an automatic driving vehicle as an example for illustration. When the method is applied in an automatic driving vehicle, the execution subject of the method is an automatic driving vehicle control unit, and the present application includes the following steps:

[0041] S10, determining a topological path and a road edge contour line in a perception range according to environment perception data. Exemplarily, the environment perception data is obtained by a sensor carried by the vehicle. The sensor includes but is not limited to:

[0042] A binocular camera can be installed in front of the vehicle to capture front road images, and after computer vision processing, traffic elements (such as road edges, lane lines, center lines, guide strips, triangular islands and traffic lights, etc.) and static / dynamic obstacles (wherein the static obstacles are mainly temporary roadblocks such as cone barrels and stones; the dynamic obstacles include vehicles parked by the roadside or vehicles in motion, which are processed by computer vision as dynamic obstacles); a millimeter wave radar installed in front of the vehicle to detect the distance of static / dynamic obstacles; a fisheye camera installed on the side of the vehicle, which has similar functions to the binocular camera and lower precision.

[0043] The environment perception data includes but is not limited to at least one of the image data collected by the binocular camera, the detection data of the millimeter wave radar and the image data collected by the fisheye camera. According to the environment perception data collected by the one or more sensors carried by the vehicle, the topological path and the road edge contour line in the perception range are calculated.

[0044] S20, generating a road center line according to the topological path and the road edge contour line.

[0045] Exemplarily, the road center line can be generated according to the road edge contour lines on both sides of the topological path.

[0046] S30, generating a driving trajectory for automatic driving according to at least the road center line.

[0047] Exemplarily, the road center line is used as a driving track for automatic driving. Further, when an obstacle on the road center line is detected, a driving track for avoiding the obstacle is generated according to the position information of the obstacle and the road center line for automatic driving.

[0048] In the embodiment of the present application, the topological path and the road edge contour line are determined through the environmental perception data of the vehicle, and the road center line is generated according to the topological path and the road edge contour line for automatic driving. Without relying on the recognition of lane lines, automatic driving on unstructured roads is realized.

[0049] As shown in FIG. 2, it is a flowchart of another embodiment of the automatic driving method in the present application. In this embodiment, the topological path in the perception range is determined according to the environmental perception data, including:

[0050] S11, generating an Euclidean distance potential field map according to the environmental perception data.

[0051] Exemplarily, a grid map of the current scene is generated; the non-passable area is determined according to the environmental perception data, and the corresponding grid on the grid map is filled with a preset value; the passable area is determined according to the environmental perception data, and the distance to the nearest obstacle is filled in the corresponding grid on the grid map to obtain the Euclidean distance potential field map. The grid map is a grid-based map representation method, the map area is divided into uniform grid units, and each grid unit is assigned specific attribute information (for example, whether it is occupied by an obstacle, etc.). The grid map can be drawn using tools such as matlab, which is not limited in the present application. The preset value can be 0, indicating that the corresponding grid is a non-passable area; the non-passable area includes a border area and a non-road area, and the area outside the non-passable area is a passable area.

[0052] In this embodiment, various road elements (moving / static obstacles, etc.) are obtained through environmental perception data, combined with border and Nonroad areas, and the values of the corresponding grids in the grid map are filled (border / Nonroad grids represent the distance to the nearest obstacle as 0, so the distance of Border / Nonroad itself is also 0). Then, according to the EDF (Euclidean Distance Field), each grid is filled with the Euclidean distance to the nearest border and Nonroad obstacles by scanning columns first and then rows. The inside of the obstacle is all set to 0 (i.e., all grids on the grid map corresponding to the position of the obstacle are set to 0). The following formula 1 describes the algorithm for updating the Euclidean distance of the i-th point in the row direction after column scanning is completed. Wherein, j is the point on the same straight line as the i grid during row scanning. edf2 = min j [f row (C j )+||C j -C i || 2 ] Formula 1

[0053] Wherein, this formula 1 shows the EDF row-by-row scanning (column scanning has been simply completed according to the Euclidean distance) updates the nearest obstacle distance of a certain grid, wherein f row (C j ) represents the Euclidean distance of the j grid to the nearest obstacle in the same column (this distance is squared, and the calculation power problem caused by square root is omitted), and the coordinates of the grid in a row are represented as C 1， C2…C M , so C i represents the geometric position of the target grid (obstacle grid) to be solved, C j represents other grids in the same row as the target grid, f row (C j ) represents the nearest obstacle distance of the j grid, ||C j -C i || 2 represents the Euclidean distance from the i grid to the j grid. As shown in FIG. 3, it is an effect diagram of the Euclidean distance potential field diagram calculated by an embodiment of the present application. Wherein, the vehicle identifier is included, the A area represents an impassable area, the B area is a road edge area, and the area between the B areas is a passable area.

[0054] S12, generating a Voronoi diagram based on the Euclidean distance potential field diagram. The Voronoi diagram in the embodiment of the present application is shown in FIG. 4. The Voronoi diagram is composed of a continuous polygon composed of a vertical bisector connecting two adjacent points (for example, composed of nodes and edges), and has the following characteristics:

[0055] (1) The Voronoi diagram divides the plane into n polygonal domains, and each polygonal domain has only one generator;

[0056] (2) The distance from a point in each polygonal domain to the generator is shorter than the distance to other generators;

[0057] (3) The distance from a point on the boundary of a polygon to the two generators generating the boundary is equal;

[0058] (4) The boundary line of the adjacent Voronoi polygon is a subset of the original adjacent boundary line;

[0059] (5) The Voronoi diagram has at most 2*n-5 vertices and 3*n-6 edges;

[0060] (6) The generating element inside the polygon is the circumcenter of the triangle formed by the three points that form the three sides, and all of these circumcircles do not contain any vertex other than the three points inside (hollow circle property).

[0061] The present application does not limit the algorithm for specifically generating the Venn diagram.

[0062] Exemplarily, a sector area is constructed with the vehicle as the center (the radius of the sector area is determined according to the perceived range, for example, the longest distance is 200 m, and the central angle is 315°), and then the edge of the sector is sampled at equal intervals as the candidate points of the target area; then, for the candidate points, n representative points are calculated as the final goal points. The sampling method is shown in FIG. 5. The candidate set of goal points is sampled, and the sampling points are combined with the grid probability values of the grids in which the sampling points are located to select the sampling points that meet the set road threshold (the road threshold is a threshold for determining whether a point is a non-road, and the road threshold can be set according to experience. The higher the road threshold, the higher the probability that a point is a non-road. Generally, the road threshold is 0-10. Therefore, the set road threshold is that the grid probability value of a point is lower than the road threshold). The k-means algorithm is used for clustering of the sampling points that meet the set road threshold. The grid probability value can represent the probability of encountering an obstacle in the corresponding grid, and the grid probability value of each grid can be calculated through perception and analysis of the environment. The specific implementation algorithm is not limited in the present application. The sampling points are points sampled on the sector according to the central angle (for example, 5°). The candidate points are points whose Nonroad probability is less than 10. The goal points are selected from the candidate points. First, clustering is performed according to the geometric distance, and then a point that is farthest from the obstacle is selected as the final target point of the road in each cluster.

[0063] In some embodiments, a path search algorithm is used to select a point farthest from the obstacle in each cluster as the target point that the road finally needs to reach. In this regard, the path search algorithm generally uses a cost function + heuristic function to search a path from the current vehicle position to the goal point step by step. For each path point on the optimal path, the cost function is simply the cost paid to reach the current point from the starting point (including the Euclidean distance and the cost coefficient of each grid according to whether it contains an obstacle), and the heuristic function is the remaining cost from the current point to the goal point (also using the Euclidean distance, grid cost and angle cost). Finally, a path is obtained that minimizes the total cost of the cost function + heuristic function for all points on the path. The heuristic function of the k-means algorithm uses the Euclidean distance between candidate points as shown in formula 2 below. After clustering, the road exit area in the current scene is obtained. The road exit is defined as a passageway that can reach the boundary from the current vehicle position within the perception range. Therefore, it is necessary to find points on the perception edge (here, a fan-shaped region) that are below a certain threshold. A set of points will be found at the exit of each road (because the road width is generally not less than the accuracy of perception, there will not be only one point). A set of points cannot be used as an exit point to assist the search below. The role of the set of points is to assist the search of the exit of the road. Therefore, a clustering operation is performed, and only one representative point is selected for each road exit to simplify the problem and facilitate the search below.

[0064] The above is formula 2, where x represents the two-dimensional coordinates of a road candidate point in the world coordinate system, x i represents the two-dimensional coordinates of the i-th point in the point set, u i is the two-dimensional coordinates calculated by averaging all points in x, x i -u j represents the Euclidean distance from the i-th point to the average coordinates, C i represents the i-th set of points, D ij represents the distance from the i-th point in the point set to the center of the j-th point set.

[0065] Further, the node corresponding to the position of the vehicle is determined as the starting node of the Voronoi diagram, and the node corresponding to the exit point obtained in the above embodiment is determined as the terminal node of the Voronoi diagram, so as to obtain a directed Voronoi diagram, and the edges in the Voronoi diagram can be formed according to the passable path determined by the Euclidean distance potential field (for example, the path corresponding to the grid with a Euclidean distance greater than a set threshold in the Euclidean distance potential field, where the set threshold can be the minimum width of the vehicle, and the specific value is not limited in the present application).

[0066] S13, determining a topological path according to the Voronoi diagram.

[0067] Exemplarily, all possible topology paths in accordance with the direction of the ego vehicle are searched using depth-first search in the V-Graph path (i.e. the light-colored topology path I shown in the V-Graph of FIG. 4). In this process, the edges of the topology path are composed of points with Euclidean distance in accordance with a threshold (as mentioned in the foregoing embodiment, the Euclidean distance is greater than a set threshold, which will not be repeated here), and the geometric positions of the nearest obstacles on both sides of the topology path (e.g. the distance to the nearest obstacles on both sides) are recorded. Thus, when a topology path is obtained, a set of information recording the geometric positions of the nearest obstacles on both sides of the topology path is obtained, which can be used to construct the edges of the obstacles on both sides of the topology path.

[0068] Exemplarily, appropriate pruning is performed using path length and filtering repeated path rules in the process of determining the topology path, to remove redundant paths, and finally obtain the main trunk of the potential route and the edges formed by the associated obstacles on both sides of the main trunk, wherein the main trunk of the potential route is used as the topology path, and the edges formed by the associated obstacles on both sides of the main trunk define the boundaries of the main trunk, which is used for display in the V-Graph shown in FIG. 4.

[0069] In some embodiments, the path length refers to the Euclidean length on the edges of the V-Graph, and pruning using path length includes directly deleting paths with shorter length. Pruning using the filtering repeated path rule includes filtering repeated paths, i.e. when a repeated edge identifier (each edge is assigned a unique edge identifier in advance) is encountered during the depth-first search of the topology path, it is considered as a loop topology path, and the topology path is deleted.

[0070] As shown in FIG. 6, it is a flowchart of another embodiment of the automatic driving method in the present application. In this embodiment, the road edge contour line within the perception range is determined according to the environmental perception data, including:

[0071] S15, obtaining road edge scatter point data from the environmental perception data, and performing convex hull processing on the road edge scatter point data to obtain a single polygon contour.

[0072] Optionally, the following steps are included: performing convex hull clustering on the road edge scatter point data; performing convex hull merging processing on the convex hull clustering result to obtain an initial polygon contour; and performing singular point filtering on the road edge scatter point of the initial polygon contour to obtain a single polygon contour. Exemplarily, the border scatter point data (road edge scatter point data) obtained from the environmental perception data is first clustered into a plurality of convex hulls in a small range, and then the plurality of convex hulls are merged into an initial polygon contour.

[0073] Further, the inventor found that there are singular points in the initial polygon outer contour due to perception errors in the process of implementing the present application, so the obtained initial polygon outer contour is further filtered for singular points to obtain a usable polygon outer contour. Among them, the singular points are due to perception errors, and some misjudged border scatter points invade the lane, causing false braking and other problems, so it is necessary to exclude the interference of these misjudged points when obtaining the outer contour. Specifically, according to the characteristics that singular points will cause the convex burrs of the polygon outer contour, the road edge scatter points causing the convexity of the outer contour are determined and removed when the polygon outer contour is generated.

[0074] The above method of obtaining the polygon outer contour saves about 70% of the computing power compared with directly using alpha shape (an algorithm for extracting boundary points) to find the outer contour of the point set, and the outer contour can meet the optimization requirements in this scenario.

[0075] S16, determining a road edge contour line according to the topological path and the polygon outer contour.

[0076] As shown in FIG. 7, it is a convex hull merging and single-side road edge outer contour extraction process diagram in an embodiment of the present application. It includes the following process steps:

[0077] a, convex hull cutting: cutting the road edge scatter points (such as the solid circle points in FIG. 7) obtained according to the environmental perception data to obtain a plurality of convex hulls. As shown in FIG. 7, the road edge scatter points are clustered in a clustering manner, and divided into three rectangular regions;

[0078] b, calculating the center point: calculating the center point of the road edge scatter points contained in each region (i.e. the center of mass of the scatter points in each region corresponding to the convex hull, such as the filled circle points in FIG. 7);

[0079] c, merging polygon (polygon), connecting the center of mass: as shown in FIG. 7, connecting the centers of mass of adjacent convex hulls, and the connecting line of the two centers of mass intersects with two convex hull outer contour edges of the two convex hulls to form two intersection points;

[0080] d, merging polygon, arranging edge: arranging the convex hull outer contour lines of the two convex hulls, and then merging the two convex hulls when the distance between the two intersection points is less than a set threshold (for example, 0.5m) (for example, fusing the four vertices of the two convex hull outer contour edges to form the two dashed lines in FIG. 7, and deleting the two convex hull outer contour edges represented by the deletion lines in FIG. 7, and the two convex hulls are merged into one);

[0081] e, finding the leftmost and lowest point to string the entire contour: as shown in FIG. 7, finding the leftmost and lowest circle point to string all the scatter points in the three rectangular regions to obtain the entire contour;

[0082] f. deleting the line between the rectangular region contour line and the centroid to obtain an integral polygon outer contour;

[0083] g. obtaining a single-side outer contour using the selected line result: using the selected line result of the previous frame, determining a plurality of road edge scattered points on the side of the polygon outer contour close to the selected line result, and taking the line between the plurality of road edge scattered points as the single-side outer contour.

[0084] In the embodiment, the convex hull clustering method is mainly used to gather small piles of borders in a small range to obtain convex polygons (convex hulls), and then adjacent convex hulls are gradually merged (adjacent is defined as adjacent centroids, since the centroid outward line of the convex hull must pass through the convex polygon outer contour, so the line between the centroids of the two outer contours must pass through an edge of the outer contour of each, leaving an intersection point. If the Euclidean distance between the two intersection points is below a certain threshold (for example, 0.5 m), it is considered that the two convex polygons can be fused by the two outer contour edges, and the four vertices of the two line segments are fused to form two new edges, and the two polygons are merged into one. As shown in FIG. 7, the deleted lines are removed and two dashed lines are added), and the convex polygons formed by the convex hulls are gradually merged into one convex polygon according to the above steps, and then the outer contour of the road edge on the side closer to the selected line (such as the arrow line on the right side of the X line in FIG. 7) is found on the convex polygon using the selected line of the previous frame (such as the X line in FIG. 7). The selected line can be understood as a reference line selected in planning for autonomous driving, which is used to guide the vehicle to travel along the line). This method does not need to use the triangulation algorithm compared to the alpha shape, and only uses the convex hull clustering to quickly reduce the number of candidate point piles, saving computing power.

[0085] FIG. 8 shows a flowchart of another embodiment of the autonomous driving method of the present application. In this embodiment, the road center line is generated according to the topological path and the road edge outer contour line, including:

[0086] S21, according to the attributes of the obstacles on both sides of the topological path, the topological path is divided into a constrained effective segment and an unconstrained fuzzy segment; wherein the attributes of the obstacles include road element obstacles and non-road area obstacles, the road element obstacles correspond to the constrained effective segment, and the non-road area obstacles correspond to the unconstrained fuzzy segment.

[0087] Exemplarily, the nearest obstacles in the matching EDF potential field are sampled according to the searched topological path at a certain precision (for example, the precision is 2 m, that is, every 2 m along the topological path, the information of the nearest obstacles on the left and right is searched), and the local map is cut into a constrained effective segment and an unconstrained fuzzy segment according to the obstacle attribute. If there is a road edge, a lane line, a triangular island or other road element constraint on the left and right of a certain segment area, it can be considered that the segment is a constrained effective segment, the outer contour extraction scheme in the foregoing can be used to take out the single-side outer contour, and a half-plane constraint is provided for the generation of the center line (a certain point on the center line is not allowed to cross the half-plane of the single-side outer contour); if there is no road edge, lane line, triangular island or other road element constraint, only a Nonroad non-road constraint, that is, an unconstrained fuzzy segment, only curvature smoothing is performed on the optimization of the center line, and the interest area of each topological path is determined (for example, the road edge constrained effective segment, that is, the area in which the nearest obstacles on the left and right can be matched to the road element).

[0088] As shown in FIG. 9, it is a schematic diagram of local map cutting in the application. Wherein, the D area is a constrained effective segment, the E area is an unconstrained fuzzy segment, and the light-colored line segment L is a topological path after filtering processing. The interest area (that is, the constrained effective segment) is data associated with the road edge contour to determine the range of action of the road edge on the center line, the left and right attributes and the like, which are used for subsequent optimization calculation of the generated center line. The center line here is a reference line used to represent the lane semantics in decision planning, one reference line represents one lane in the decision planning, the trajectory of the decision planning follows the reference line, so the reference line itself guarantees smoothness, collision-free and provides lane semantics. Wherein, data association refers to: the EDF potential field mentioned in the foregoing embodiments can obtain the nearest obstacle (for example, the road element that can be encountered in the countryside) from a certain grid, so when the left and right boundaries of a certain point in the interest area are determined, the nearest obstacle from the current point, the two-dimensional coordinates and type of the next nearest obstacle on the opposite side can be directly obtained from the EDF potential field, which is called data association.

[0089] S22, moving the road edge outer contour line corresponding to the constrained effective segment to the vehicle direction by a preset distance to generate a road center line.

[0090] Exemplarily, the right road edge outer contour using data association is laterally translated to the position of the vehicle, that is, the right road edge outer contour is moved in the lateral direction (for example, the y axis) to the position of the vehicle, and the curve after moving a distance of half the lane width (for example, 3.75 / 2 m) is taken as the initial center line of the guess, and the points on the initial center line have an initial state (including the two-dimensional coordinates, direction, curvature and the like of the points).

[0091] Further, the initial center line of the guess is optimized. Exemplarily, points can be taken on the initial center line according to a preset step size (for example, 2m apart), and the initial state of the points is optimized. Wherein, the initial state angle and position on the initial center line are calculated as shown in formula 3. For the i+1 state of the i+1 point, the position of the i state of the i point is propagated in the direction of the i state, the two-dimensional coordinates, direction, curvature, etc. of the initial center line of the guess are optimized by gradient descent using LQR (Linear Quadratic Regulator), and finally the right center line is output. i+1 = X i + X iq *norm Formula 3, initial solution state propagation formula

[0092] Wherein, X is the state of a point on the optimized line (including the two-dimensional coordinates of the point, direction, curvature, etc.), X i+1 is the i+1 state, X i is the i state, X iq *norm represents advancing a distance of one precision (for example, the precision 2m mentioned in the foregoing embodiment) in the direction (sin(θ), cos(θ)) of the angle at the i position.

[0093] As shown in FIG. 10, it is a flowchart of another embodiment of the automatic driving method of the present application. In this embodiment, before determining the topological path and the road edge contour line within the perception range according to the environment perception data, the automatic driving method further comprises:

[0094] S40, generating the driving mode switching instruction in response to the user's driving mode switching operation; or, according to the environment perception data, performing scene recognition, and when the scene recognition result meets the preset condition, generating the driving mode switching instruction.

[0095] S50, when the driving mode switching instruction is detected, controlling the vehicle to switch from the first driving mode to the second driving mode; wherein, the first driving mode is the current driving mode of the vehicle, and the second driving mode is an unstructured road automatic driving mode. Wherein, the automatic driving method described in the foregoing embodiment is executed under the unstructured road automatic driving mode; the automatic driving method suitable for structured roads can be executed under the current driving mode of the vehicle (for example, structured road driving mode), and the specific implementation manner can refer to the related prior art, which is not limited by the present application.

[0096] Exemplarily, in actual use, through the embodiment, the user can manually trigger the generation of the driving mode switching instruction as needed, so as to complete the switching of the driving mode. For example, when the vehicle is about to reach a non-structured road scene such as a rural area, the user can trigger the driving mode switching instruction through a voice instruction or a gesture operation or a touch operation on the vehicle interaction screen, so as to switch the driving mode to the non-structured road automatic driving mode.

[0097] Further, scene recognition can also be performed through the perception data of the sensors carried by the vehicle. When the scene recognition result indicates that the current scene is a non-structured road scene, the driving mode switching instruction is automatically generated, so as to switch the driving mode to the non-structured road automatic driving mode. In the embodiment, the user does not need to manually switch, and the user's hands can be completely freed.

[0098] Exemplarily, the scene recognition includes: identifying the proportion of structured roads in the road according to the perception data, and determining that the current recognized scene is a non-structured road scene when the identification result indicates that the proportion of structured roads is lower than a set threshold (for example, 70%). Further, the scene recognition is performed through continuous multiple frames of data, and it is determined that the current recognized scene is a non-structured road scene when the identification results of the continuous multiple frames of data all indicate that the proportion of structured roads is lower than the set threshold (for example, 70%). The scene recognition and judgment through continuous multiple frames of data reduces the misjudgment caused by single frame error, and improves the reliability of the determination of the non-structured road scene.

[0099] In some embodiments, the scene recognition further includes: when it is determined according to the navigation positioning that the current road is not a highway, and it is identified according to the perception data that there is lane line missing or breaking in the road, it is determined that the current recognized scene is a non-structured road scene, so as to prevent false triggering in similar scenes such as highway intersections.

[0100] Further, in order to improve the accuracy and safety of the scene recognition according to the environmental perception data and the automatic triggering of the driving mode switching, the application introduces a rural mode state machine.

[0101] As shown in FIG. 11, a schematic diagram of a rural mode state machine used in the automatic driving method of the present application is shown. It includes an invalid state, an enable rural mode state, an activate rural mode state, and a close rural mode state; the enable rural mode state, the activate rural mode state, and the close rural mode state constitute a closed loop, and the invalid state is a state in which the rural mode state machine is not triggered at all. In the close rural mode state or the invalid state, the condition for entering the enable rural mode state is that the vehicle is not on a highway according to the navigation positioning display, and the road quality is low according to the perception data (for example, the proportion of structured roads in the current scene is lower than a set threshold); only when the state machine enters the enable rural mode state, and no intersection is detected and no normal route selection (normal route selection, i.e., no available route can be identified through structured road recognition) is detected, the activate rural mode state is switched to; in the activate rural mode state, if the road quality is high according to the perception or the intersection area is entered, the state machine is switched to the close rural mode state.

[0102] In this embodiment, the state jump of the rural mode state machine is triggered according to the scene recognition result, and the rural mode is switched to, so as to save the calculation power consumption in the normal mode, and the system mode recognition is added in the rural mode state machine (i.e., the scene is identified according to the navigation information and the lane line quality evaluation in the decision planning, only when the road quality is low according to the perception data and the current road is not a highway according to the navigation positioning, the mode is enabled), which prevents the misoperation in the similar scenes such as highway intersection.

[0103] In some embodiments, after the vehicle is switched from the first driving mode to the second driving mode (unstructured road automatic driving mode) according to the detected driving mode switching instruction, the automatic driving method of the present application further includes:

[0104] selecting the center line according to at least one of safety, comfort, and consistency;

[0105] performing trajectory planning according to the selected center line and the obstacles on the center line, for completing the automatic driving on the unstructured road.

[0106] Exemplarily, a centerline selection state machine is also provided in the present application to select the centerline. The centerline selection state machine selects the centerline after receiving an enabling flag (for example, included in a driving mode switching instruction, which can be obtained by parsing the instruction). The criteria for selecting the centerline are: focusing on evaluating safety and comfort, so as to select the centerline with the best consistency at present to keep the lane. Among them, the safety means that there is no collision in front of the centerline (not including various obstacles); the comfort means that the lateral and longitudinal accelerations of the centerline are within a preset range (for example, ±3.5 m / s^2); the consistency requires that, after sampling the points of the centerline finally determined in the foregoing embodiment and the initial centerline point by point, the corresponding points are kept within a certain threshold (0.5 m) in the lateral direction, to evaluate the similarity of the shapes of the two curves, and when more than 80% of the sampling points meet the above threshold, it is determined that the consistency of the two curves is optimal. Finally, the consistent centerline is selected based on at least one of the above three factors; further, the path planning is based on the selected centerline to plan the side deviation and detour trajectory for the static obstacles and stationary vehicles on the centerline, to complete the road driving on the rural road section. The present embodiment can realize the exploratory driving in the unstructured road scene, and the centerline trajectory selected and planned according to the three factors of safety, comfort and consistency also conforms to the natural habit of human driving on the side; in addition, by planning the trajectory for avoiding the static obstacles and stationary vehicles, the usability of the trajectory is ensured.

[0107] As shown in FIG. 12, it is a flowchart of another embodiment of the automatic driving method of the present application. In this embodiment, the automatic driving method mainly includes three parts: centerline estimation, state control and path planning. Among them,

[0108] The centerline estimation part includes: on the one hand, generating an EDF potential field according to the perception input (for example, the environmental perception data obtained by the vehicle sensor), obtaining a Voronoi diagram based on the EDF potential field, and then performing topological path search to obtain the required topological path; on the other hand, performing convex hull merging and clustered polygon processing on the road edge scatter points based on the perception input to obtain the road edge single-side outer contour; further, performing data association on the topological path and the road edge single-side outer contour, and generating and optimizing the centerline, to finally obtain the candidate centerline which can be used for automatic driving in the unstructured road scene.

[0109] The state control part includes: road quality evaluation and rural mode state machine. The state of the rural mode state machine is switched through the road quality evaluation result, and when the rural mode state is switched to the active state, an enabling flag is generated and sent to the centerline selection state machine.

[0110] In some embodiments, the switching of the autonomous driving mode can be achieved by the state control part. For example, the switching of the structured road driving mode and the unstructured road driving mode is achieved. Illustratively, the execution of the center line estimation part is controlled by the state control part to switch to the unstructured road driving mode. For example, when the state control part evaluates the road quality and switches to the active country mode state to generate the enable flag, the center line estimation part is controlled to execute the center line estimation step.

[0111] The path planning part includes a lane selection state machine and obstacle avoidance. When the lane selection state machine receives the enable flag from the state control part, the selection of the center line is performed (the specific selection method is described in the foregoing embodiments, which will not be described here). Then, the obstacle avoidance planning is performed according to the selected center line, and finally the planned trajectory is output for autonomous driving. This embodiment can realize the exploratory driving in the unstructured road scene. The selected planned center line trajectory according to the safety, comfort and consistency meets the natural habit of human side driving. In addition, the trajectory for avoiding static obstacles and static vehicles is planned to ensure the availability of the trajectory.

[0112] In some embodiments, the present application provides a computer device, including a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in any of the embodiments of the present application.

[0113] In some embodiments, the present application provides a computer readable storage medium, which stores a computer program / instruction, wherein the computer program / instruction is executed by a processor to implement the steps of the method described in any of the embodiments of the present application.

[0114] In some embodiments, the present application provides a computer program product, including a computer program / instruction, wherein the computer program / instruction is executed by a processor to implement the steps of the method described in any of the embodiments of the present application.

[0115] In some embodiments, the present application provides a mobile platform, wherein the mobile platform is installed with the computer device described in any of the embodiments of the present application. The mobile platform can be a vehicle, a robot, an unmanned aerial vehicle or other mobile device.

[0116] It should be noted that, for the foregoing method embodiments, for the purpose of simple description, they are all described as a series of actions, but those skilled in the art should know that the present application is not limited to the order of the actions described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application. In the above embodiments, the description of each embodiment is focused on, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0117] FIG. 13 is a schematic diagram of the hardware structure of a computer device for performing the automatic driving method according to another embodiment of the present application. As shown in FIG. 13, the device includes:

[0118] One or more processors 1310 and a memory 1320, of which one processor 1310 is taken as an example in FIG. 13.

[0119] The device for performing the automatic driving method can further include an input device 1330 and an output device 1340.

[0120] The processor 1310, the memory 1320, the input device 1330 and the output device 1340 can be connected through a bus or other means, of which the connection through the bus is taken as an example in FIG. 13.

[0121] The memory 1320 as a non-volatile computer readable storage medium can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions / modules corresponding to the automatic driving method in the embodiments of the present application. The processor 1310 performs various functional applications and data processing of the server by running the non-volatile software programs, instructions and modules stored in the memory 1320, that is, implements the automatic driving method of the above method embodiments.

[0122] The memory 1320 can include a program storage area and a data storage area, wherein the program storage area can store an operating system and at least one application required by a function; the data storage area can store data created according to the use of the automatic driving device, etc. In addition, the memory 1320 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 1320 can optionally include a memory remotely arranged with respect to the processor 1310, which can be connected to the automatic driving device through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0123] The input device 1330 can receive input digital or character information, and generate a signal corresponding to a user's setting or function control of the autonomous driving device. The output device 1340 can include a display device such as a display screen.

[0124] The one or more modules are stored in the memory 1320 and, when executed by the one or more processors 1310, perform the autonomous driving method of any of the above-described method embodiments.

[0125] The above-described product can perform the method provided by the embodiments of the present application, has the corresponding function modules and beneficial effects of performing the method. Technical details not described in detail in the embodiments can be referred to the method provided by the embodiments of the present application.

[0126] The computer device of the embodiments of the present application exists in various forms, including but not limited to:

[0127] (1) Mobile communication device: This kind of device is characterized by having mobile communication function, and taking providing voice and data communication as the main target. This kind of terminal includes: smart phone (such as iPhone), multimedia phone, functional phone, and low-end phone, etc.

[0128] (2) Ultra-mobile personal computer device: This kind of device belongs to the category of personal computer, has computing and processing functions, and generally also has the characteristics of mobile Internet. This kind of terminal includes: PDA, MID and UMPC device, etc., such as iPad.

[0129] (3) Portable entertainment device: This kind of device can display and play multimedia content. This kind of device includes: audio and video player (such as iPod), palm game console, electronic book, and smart toy and portable car navigation device.

[0130] (4) Server: A device that provides computing services. The components of a server include a processor, a hard disk, a memory, a system bus, etc. The server has a similar architecture to a general-purpose computer, but requires higher processing power, stability, reliability, security, scalability, and manageability due to the need to provide high-reliability services.

[0131] (5) Other electronic devices with data interaction function.

[0132] The device embodiments described above are only schematic, and the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e. they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiments of the present application.

[0133] Those skilled in the art can clearly understand the implementation of the various embodiments by means of software plus a general hardware platform from the above description of the embodiments, and of course, the various embodiments can also be implemented by hardware. Based on such an understanding, the above technical solutions, essentially or in other words, the part that contributes to the related art, can be embodied in the form of a software product. The computer software product can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a plurality of instructions to cause a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0134] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

1. An automatic driving method, comprising: determining a topology path and a road edge contour line within a perception range according to environment perception data; generating a road center line according to the topology path and the road edge contour line; generating a driving trajectory according to at least the road center line for automatic driving.

2. The method of claim 1, wherein, The determining of the topology path within the perception range according to the environment perception data comprises: generating a Euclidean distance potential field map according to the environment perception data; generating a Vornoi diagram based on the Euclidean distance potential field map; determining the topology path according to the Vornoi diagram.

3. The method of claim 2, wherein, The generating of the Euclidean distance potential field map according to the environment perception data comprises: generating a grid map of a current scene; determining an impassable region according to the environment perception data, and filling a corresponding grid on the grid map with a preset value; determining a passable region according to the environment perception data, and filling a distance to the nearest obstacle in a corresponding grid on the grid map to obtain the Euclidean distance potential field map.

4. The method of claim 1, wherein, The determining of the road edge contour line within the perception range according to the environment perception data comprises: obtaining road edge scattered point data according to the environment perception data, and performing convex hull processing on the road edge scattered point data to obtain a single polygon contour; determining the road edge contour line according to the topology path and the polygon contour.

5. The method of claim 4, wherein, The performing of the convex hull processing on the road edge scattered point data to obtain the single polygon contour comprises: performing convex hull clustering on the road edge scattered point data; performing convex hull merging processing on the convex hull clustering result to obtain an initial polygon contour; performing singular point filtering on the initial polygon contour to obtain the single polygon contour.

6. The method according to any one of claims 1-5, characterized in that, The generating of the road center line according to the topology path and the road edge contour line comprises: dividing the topology path into a constrained valid segment and an unconstrained fuzzy segment according to attributes of obstacles on both sides of the topology path; wherein the attributes of the obstacles include road element obstacles and non-road region obstacles, the road element obstacles correspond to the constrained valid segment, and the non-road region obstacles correspond to the unconstrained fuzzy segment; moving the road edge contour line corresponding to the constrained valid segment by a preset distance in a vehicle direction to generate the road center line.

7. The method according to any one of claims 1-5, characterized in that, Before the determining of the topology path and the road edge contour line within the perception range according to the environment perception data, the method further comprises: generating the driving mode switching instruction in response to a driving mode switching operation of a user; or, performing scene recognition according to the environment perception data, and generating the driving mode switching instruction when a scene recognition result meets a preset condition; when the driving mode switching instruction is detected, controlling the vehicle to switch from a first driving mode to a second driving mode; wherein the first driving mode is a current driving mode of the vehicle, and the second driving mode is an unstructured road automatic driving mode.

8. The method of claim 7, wherein, After the controlling of the vehicle to switch from the first driving mode to the second driving mode according to the detected driving mode switching instruction, the method further comprises: selecting a center line according to at least one of safety, comfort and consistency; performing trajectory planning according to the selected center line and obstacles on the center line for completing the automatic driving on the unstructured road.

9. A computer device comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program comprises instructions that, when executed by the processor, cause the processor to perform the method of any one of claims 1-8. The processor executes the computer program to implement the steps of the method of any one of claims 1-8.

10. A computer readable storage medium having stored thereon computer programs / instructions, characterized in that, The computer program / instructions, when executed by the processor, implement the steps of the method of any one of claims 1-8.

11. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instructions, when executed by the processor, implement the steps of the method of any one of claims 1-8.

12. A mobile platform, characterized by The computer device of claim 9 is installed.

Images