Vehicle decision planning method, device, apparatus, and media

Abstract

The present invention provides a vehicle decision planning method, device, apparatus, and medium. The method comprises the steps of: generating a basic coordinate system; generating guidelines in the basic coordinate system to determine the approximate future driving trajectory of the vehicle; performing obstacle decision-making under the constraints of the guidelines; and generating a drivable area based on the obstacle decision-making. The method enables integrated decision-making for grid-type and convex-hull-type obstacles, simplifies the obstacle decision-making flow for mixed-type obstacles, accelerates the obstacle decision-making process, and allows the decision planning module to perform obstacle decision-making conveniently and quickly.

Description

Technology Field

[0001] Cross-reference regarding related applications

[0002] The present application claims priority to a Chinese patent application filed with the Chinese Intellectual Property Office on August 25, 2021, with application number 202110984268.3 and invention title “Method, apparatus, device and medium for decision-making for obstacle avoidance” and to a Chinese patent application filed with the Chinese Intellectual Property Office on September 17, 2021, with application number 202111095293.2 and invention title “Method, apparatus, device and medium for generating a drivable area of ​​a vehicle”, the entire contents of said applications are incorporated by reference into the present application.

[0003] The present invention relates to the field of autonomous driving technology, and in particular to a vehicle decision planning method, device, apparatus, and medium. Background Technology

[0004] With the advancement of vehicle intelligence technology, automatic control technology for autonomous vehicles is gradually emerging as a hotspot in the field of vehicle research. To ensure that vehicles do not collide with obstacles, autonomous driving systems must plan a smooth, safe, and passable route.

[0005] Generally, the sensing module of an autonomous driving system outputs two types of obstacles: one is a convex hull obstacle containing rich semantic information, and the other is a grid obstacle that does not contain semantic information. For convex hull obstacles, the decision planning module can conveniently perform obstacle decisions; however, for grid obstacles that are highly discrete and numerous, it is difficult for the decision planning module to conveniently and rapidly perform obstacle decisions, making it difficult for the decision planning module to perform obstacle decisions regarding mixed-type obstacles. The problem to be solved

[0006] To solve the above technical problem or to solve the above technical problem at least partially, the present invention provides a vehicle decision planning method, device, apparatus, and medium. means of solving the problem

[0007] An embodiment of the present invention provides a vehicle decision planning method, said method, said method,

[0008] Step of creating a basic coordinate system;

[0009] A step of generating guidelines in the above basic coordinate system to determine the approximate future driving trajectory of the vehicle;

[0010] A step of performing obstacle decision-making under the constraints of the above guidelines; and

[0011] Includes the step of generating a drivable area based on obstacle decision-making.

[0012] In some embodiments, the step of performing obstacle decision-making is,

[0013] A step of obtaining road information, first grid obstacle information of a first grid obstacle and first convex hull obstacle information of a first convex hull obstacle;

[0014] A step of obtaining a second grid obstacle by preprocessing the first grid obstacle based on the above road information and the above first grid obstacle information;

[0015] A step of converting the above-mentioned second grid obstacle into a second convex hull obstacle; and

[0016] The method includes the step of making an avoidance decision regarding the target convex hull obstacle based on the target convex hull obstacle information of the target convex hull obstacle, wherein the target convex hull obstacle includes the first convex hull obstacle and / or the second convex hull obstacle.

[0017] In some embodiments, the step of generating a drivable area based on obstacle decision-making is,

[0018] Step of acquiring environmental sensing information - the environmental sensing information includes at least two of lane information, obstacle information, and ego vehicle information, and the obstacle information includes static obstacle information and / or dynamic obstacle information - ;

[0019] A step of determining lane decision semantic information for each lane based on the above environment sensing information - the lane decision semantic information includes passage time costs and safety costs -; and

[0020] It includes the step of generating a drivable area based on the lane decision semantic information above.

[0021] An embodiment of the present invention provides a vehicle decision planning device, said device, said device,

[0022] Basic coordinate system generator for generating a basic coordinate system;

[0023] A guideline generator for generating a guideline in the above basic coordinate system to determine the approximate future driving trajectory of a vehicle;

[0024] An obstacle decision maker for performing obstacle decision-making under the constraints of the above guidelines; and

[0025] Includes a driving space generator for generating a driving area based on obstacle decision-making.

[0026] In some embodiments, the obstacle decision-maker,

[0027] Information acquisition module for acquiring road information, first grid obstacle information of a first grid obstacle, and first convex hull obstacle information of a first convex hull obstacle;

[0028] A preprocessing module for obtaining a second grid obstacle by preprocessing the first grid obstacle based on the above road information and the above first grid obstacle information - the number of the second grid obstacles is smaller than the number of the first grid obstacles - ;

[0029] A type conversion module for converting the above-mentioned second grid obstacle into a second convex hull obstacle; and

[0030] An avoidance decision module for making an avoidance decision regarding a target convex hull obstacle based on target convex hull obstacle information of the target convex hull obstacle, wherein the target convex hull obstacle includes the first convex hull obstacle and / or the second convex hull obstacle.

[0031] In some embodiments, the driving space generator is,

[0032] A sensing information acquisition module for acquiring environmental sensing information - the environmental sensing information includes at least two of lane information, obstacle information, and ego vehicle information, and the obstacle information includes static obstacle information and / or dynamic obstacle information - ;

[0033] A lane decision semantic information determination module for determining lane decision semantic information for each lane based on the above environment sensing information - the lane decision semantic information includes passage time cost and safety cost -; and

[0034] It includes a drivable area generation module for generating a drivable area based on the above lane decision semantic information.

[0035] An embodiment of the present invention provides an electronic device, said electronic device, said electronic device,

[0036] It includes memory and one or more processors;

[0037] The above memory is connected to the above one or more processors in communication, and the memory stores instructions that can be executed by the above one or more processors, and by executing the instructions by the above one or more processors, a vehicle decision planning method provided by any embodiment of the present invention is performed by the electronic device.

[0038] An embodiment of the present invention provides a computer-readable storage medium in which a computer-executable instruction is stored, and a vehicle decision planning method provided by any embodiment of the present invention is performed by executing the computer-executable instruction by a computing device. Effects of the invention

[0039] The technical solution provided by the embodiment of the present invention has the following advantages over the prior art.

[0040] First, after preprocessing the first grid obstacle to obtain the second grid obstacle, the second grid obstacle is converted into a second convex hull obstacle—that is, a grid-type obstacle is converted into a convex hull-type obstacle—thereby realizing integrated decision-making for both grid-type and convex hull-type obstacles (i.e., mixed-type obstacles). This simplifies the obstacle decision-making flow for mixed-type obstacles, accelerates the obstacle decision-making process, and enables the decision planning module to conveniently and quickly perform obstacle decisions.

[0041] Second, lane decision semantic information for each lane is determined according to environmental sensing information, and the lane decision semantic information is converted into constraint boundaries of a drivable area. By simultaneously considering drivability and safety, a drivable area with high drivability and safety can be rapidly generated, and by accelerating the generation of a driving trajectory, rapid avoidance of obstacles can be realized. At the same time, both the passing time cost and the safety cost can represent the passing cost of dynamic obstacles, and the passing time cost can represent the passing cost of static obstacles. Therefore, the technical solution of the present invention can simultaneously realize a passing plan for dynamic obstacles and static obstacles by generating a drivable area based on the passing time cost and the safety cost, and is applicable to the processing of obstacles in a dynamic environment. Brief explanation of the drawing

[0042] The attached drawings are incorporated into the specification and constitute part of the specification, and are used to illustrate embodiments according to the present invention and to explain the principles of the present invention together with the specification. In order to explain the embodiments of the present invention or the technical methods of the prior art more clearly, the attached drawings used in the description of the embodiments or background art are briefly described below, and a person skilled in the art can obtain other attached drawings based on these attached drawings without undergoing creative labor. FIG. 1 is a flowchart of a vehicle decision planning method provided by an embodiment of the present invention. FIG. 2 is a block diagram of a functional module of a decision planning module provided by an embodiment of the present invention. FIG. 3 is a schematic diagram of an application scenario of a vehicle decision planning method provided by an embodiment of the present invention. FIG. 4 is a partial flowchart of a vehicle decision planning method provided by an embodiment of the present invention. FIG. 5 is a block diagram of a functional module of an obstacle decision maker provided by an embodiment of the present invention. FIG. 6 is a schematic diagram of a road bounding box provided by an embodiment of the present invention. FIG. 7 is a diagram of a scenario for avoidance decision-making provided by an embodiment of the present invention. FIG. 8 is a partial flowchart of another vehicle decision planning method provided by an embodiment of the present invention. FIG. 9 is a diagram of a scenario corresponding to the lane passage time cost provided by an embodiment of the present invention. FIG. 10 is an ST diagram of lane safety judgment provided by an embodiment of the present invention. FIG. 11 is a schematic diagram showing the discretized boundaries of a drivable area provided by an embodiment of the present invention. FIG. 12 is a diagram of a scenario corresponding to the lane width passing cost provided by an embodiment of the present invention. FIG. 13 is a schematic diagram of updating a drivable area based on preset traffic rules provided by an embodiment of the present invention. FIG. 14 is a schematic diagram of updating the drivable range of a vehicle based on kinematic and dynamic constraints provided by an embodiment of the present invention. FIG. 15 is a schematic diagram of updating a drivable area based on obstacle semantic information and a preset stable area provided by an embodiment of the present invention. FIG. 16 is a schematic diagram of creating a frenet bounding box provided by an embodiment of the present invention. FIG. 17 is a schematic diagram of updating a drivable area based on static obstacle information, obstacle decision semantic information, and a ray tracing algorithm provided by an embodiment of the present invention. FIG. 18 is a block diagram of a functional module of an obstacle decision maker of a vehicle decision planning device provided by an embodiment of the present invention. FIG. 19 is a block diagram of a functional module of a driving space generator of a vehicle decision planning device provided by an embodiment of the present invention. FIG. 20 is a schematic diagram of the structure of an electronic device of a preferred embodiment of the present invention. Specific details for implementing the invention

[0043] To better understand the above-mentioned objectives, features, and advantages of the present invention, technical methods of the present invention are further described below. Embodiments and features of the present invention may be combined with one another, provided they do not conflict.

[0044] In the following description, many specific details have been provided to fully understand the invention, but the invention may be implemented in ways different from those described herein; and the embodiments presented herein are some embodiments of the invention and not all embodiments.

[0045] FIG. 1 is a flowchart of a vehicle decision planning method provided by an embodiment of the present invention. The method is applied to obstacle decision-making and the generation of a drivable area of ​​an autonomous vehicle. As illustrated in FIG. 1, the method comprises the following steps.

[0046] In step (S110), a basic coordinate system is created.

[0047] In step (S120), a guideline is generated in the basic coordinate system to determine the approximate future driving trajectory of the vehicle.

[0048] In step (S130), obstacle decision-making is performed under the constraints of the guidelines.

[0049] In step (S140), a drivable area is created based on obstacle decision-making.

[0050] In this embodiment, a base coordinate system is created, and subsequent guidelines, obstacle decision data, and a drivable area are all generated under the base coordinate system, thereby providing a reference for positioning the vehicle and the obstacle. The base coordinate system may be a Frenet coordinate system. After creating the guidelines, obstacles are avoided by performing obstacle decision on obstacles on the approximate driving trajectory according to the approximate driving trajectory indicated by the guidelines. Next, the drivable area of ​​the vehicle is determined by deciding whether the vehicle will pass to the left of the obstacle, pass to the right of the obstacle, or follow the obstacle according to the obstacle decision.

[0051] FIG. 2 is a block diagram of a functional module of a decision planning module. As shown in FIG. 2, the decision planning module (1) may include a constraint generation unit (11), a trajectory generation unit (12), and a trajectory smoothing unit (13).

[0052] In some embodiments, the constraint generation unit (11) may include a base coordinate system generator (111), a guideline generator (112), an obstacle decisioner (113), and a driving space generator (114). Here, the base coordinate system generator (111) is used to generate a base coordinate system, such as a Frenet coordinate system; the guideline generator (112) generates guidelines to determine the approximate future driving trajectory of the vehicle; the obstacle decisioner (113) is used to perform obstacle decisions; and the driving space generator is used to generate a driving area based on the obstacle decisions. In some embodiments, the trajectory generation unit (12) is used to generate a driving trajectory of the autonomous vehicle based on the driving area; and the trajectory smoothing unit (13) is used to smooth the driving trajectory.

[0053] In some embodiments, the obstacle decision-maker (113) specifically obtains road information, first grid obstacle information of a first grid obstacle, and first convex hull obstacle information of a first convex hull obstacle; based on the road information and the first grid obstacle information, preprocesses the first grid obstacle to obtain a second grid obstacle; converts the second grid obstacle into a second convex hull obstacle; and based on the target convex hull obstacle information of a target convex hull obstacle, makes an avoidance decision for the target convex hull obstacle.

[0054] In some embodiments, the driving space generator (114) specifically obtains environment sensing information; determines lane decision semantic information for each lane based on the environment sensing information, wherein the lane decision semantic information includes a passing time cost and a safety cost; and generates a driving area based on the lane decision semantic information.

[0055] Based on the above technical method, an embodiment of the present invention provides a vehicle decision planning method, which is applied to an autonomous vehicle to make decisions regarding static obstacles and / or dynamic obstacles, such as grid obstacles and convex hull obstacles in a road environment. FIG. 3 illustrates an application scenario of the vehicle decision planning method. Referring to FIG. 3, there are convex hull obstacles (200) (including static and dynamic obstacles) and grid obstacles (300) in front of an autonomous vehicle (100). The autonomous vehicle (100) can realize an integrated decision regarding convex hull obstacles (200) and grid obstacles (300) by acquiring obstacle information regarding the convex hull obstacles (200) and grid obstacles (300) and converting the type of the grid obstacles (300) into a convex hull type. The above method can be applied to an autonomous vehicle, specifically to a decision planning module in an autonomous vehicle autonomous driving system. Based on the vehicle decision planning method provided by the embodiment of the present invention, integrated decision-making of mixed-type obstacles is realized, thereby enabling obstacle decision-making to be performed conveniently and quickly.

[0056] Based on the above technical plan, FIG. 4 is a partial flowchart of a vehicle decision planning method provided by an embodiment of the present invention. As shown in FIG. 4, the step of performing an obstacle decision (or, a decision method for obstacle avoidance) includes the following steps.

[0057] In step (S210), road information, first grid obstacle information of the first grid obstacle and first convex hull obstacle information of the first convex hull obstacle are obtained.

[0058] In an embodiment of the present invention, the grid obstacle is a grid-type obstacle, and the convex hull obstacle is a convex hull-type obstacle.

[0059] In some embodiments, road information may be acquired by a high-precision map or a vehicle-mounted camera, and the road information may include road boundary information and road curvature information, etc. At the same time, obstacle information may be acquired through a vehicle detection module (e.g., a vehicle-mounted camera and LiDAR, etc.) and a positioning module, and the obstacle information may include obstacle type information, obstacle size information, and obstacle location information, etc. The obstacle type information may be an obstacle type identifier, and obstacle types may be distinguished by pre-defining different obstacle type identifiers; the obstacle type information may also be an obstacle data format, and after the vehicle detection module detects an obstacle, it processes the obstacle data and stores it in a different obstacle data format, and when the decision planning module acquires the obstacle information, it may distinguish the obstacle type through the obstacle data format, for example, the obstacle data format for a grid obstacle is “.ogm”, the obstacle data format for a convex hull obstacle is “.mot”, etc. As described above, based on the obstacle type information, a first grid obstacle and a first convex hull obstacle can be determined to obtain first grid obstacle information and first convex hull obstacle information.

[0060] In step (S220), based on road information and first grid obstacle information, the first grid obstacle is preprocessed to obtain the second grid obstacle.

[0061] In step (S230), the second grid obstacle is converted into a second convex hull obstacle.

[0062] FIG. 5 is a block diagram of a functional module of an obstacle decision-maker. As shown in FIG. 5, the obstacle decision-maker (113) may include a grid obstacle processor (1131), a passage method decision-maker (1132), and a convex hull obstacle filter (1133).

[0063] Here, the grid obstacle processor (1131) performs the steps of: preprocessing the first grid obstacle based on road information and the first grid obstacle information to obtain a second grid obstacle (S220); and converting the second grid obstacle into a second convex hull obstacle (S230).

[0064] In step (S220), the preprocessing of the first grid obstacle can be used to reduce the amount of data computation and to simplify the obstacle decision flow, and may include at least one of the steps of: generating a grid obstacle contour of the first grid obstacle; generating an obstacle bounding box of the first grid obstacle; filtering the first grid obstacle located outside the road; and performing aggregation processing on the first grid obstacle located inside the road.

[0065] In some embodiments, preprocessing is performed on the first grid obstacles so that the number of second grid obstacles obtained by preprocessing is less than the number of first grid obstacles, thereby facilitating the calculation of obstacles by the downstream module.

[0066] An embodiment of the present invention can reduce the number of first grid obstacles by filtering first grid obstacles located outside the road. In some embodiments, the step of obtaining second grid obstacles by preprocessing first grid obstacles based on road information and first grid obstacle information may include the following steps.

[0067] In step (S221), first grid obstacles located outside the road are filtered based on road information and first grid obstacle information.

[0068] In some embodiments, the step of filtering first grid obstacles located outside the road based on road information and first grid obstacle information may include the following steps.

[0069] In step (S2211), based on road information, a road bounding box following the road's direction of travel is generated.

[0070] In some embodiments, based on road information, road boundaries are discretized into boundary points; and based on the boundary points, a road bounding box is generated. In an embodiment of the present invention, in order to easily determine later whether a first grid obstacle is located outside the road, the shape of the road bounding box is an axisymmetric bounding box based on the right coordinate system of the unmanned vehicle body, provided that it can cover the road by passing through the boundary points.

[0071] Specifically, referring to FIG. 6, road boundaries and road boundary curvatures are determined based on road information; based on road boundary curvatures, road boundaries are discretized to obtain groups of boundary points spaced apart along the road traffic direction, each group of boundary points includes a left boundary point a and a right boundary point a' corresponding in the transverse direction, the transverse direction being perpendicular to the road traffic direction; based on any two adjacent boundary point groups, a rectangular box b is generated passing through each boundary point of any two adjacent boundary point groups, and the rectangular box b is used as a road bounding box.

[0072] In some embodiments, one of the two adjacent sides of the rectangular box b is parallel to the driving direction of the vehicle, i.e., the driving direction x of the autonomous vehicle (100), and the other side is perpendicular to the driving direction of the vehicle, i.e., the normal direction y of the driving of the autonomous vehicle (100). Additionally, the distance between two adjacent boundary points on the same road boundary has a negative relationship with the road boundary curvature; in other words, the greater the road boundary curvature, the greater the degree of curvature and the smaller the distance between two adjacent boundary points on the road boundary. In this way, it is ensured that the road bounding box completely covers the road, and by avoiding the first grid obstacle, which is partially located inside the road, being filtered out as being located outside the road, it is possible to avoid affecting the obstacle decision.

[0073] In some embodiments, the step of discretizing the road boundary based on the road boundary curvature to obtain a group of boundary points spaced apart along the road travel direction comprises: using the current position of the ego vehicle as a starting waypoint; obtaining one group of boundary points corresponding to the starting waypoint in the lateral direction; selecting the next waypoint along the road travel direction based on the road boundary curvature—wherein the distance between two adjacent waypoints has a negative relationship with the road boundary curvature—; returning to the step of obtaining one group of boundary points corresponding to the starting waypoint in the lateral direction using the next waypoint as a starting waypoint until the distance from the next waypoint to the current position of the ego vehicle in the road travel direction is greater than a preset distance threshold, and determining all currently obtained groups of boundary points as boundary point groups. Herein, the preset distance threshold may be determined according to the maximum range of obstacles detected by the vehicle.

[0074] Based on the above technical solution, in a specific embodiment of the present invention, one road bounding box for every four boundary points (two adjacent boundary points on the left side of the road and two corresponding adjacent boundary points on the right side of the road) It can generate, and are the minimum and maximum coordinate points of the road bounding box, respectively, and are the left and right coordinate points of the road, respectively, and the entire road is represented by n road bounding boxes, a road bounding box sequence It can generate. In an embodiment of the present invention, the road may be one of the route segments of a vehicle driving path, and the road boundary can be discretized by determining the route segment where the ego vehicle is located according to vehicle positioning information, that is, the boundary of the route segment where the ego vehicle is located can be discretized. For example, a list of road boundary points is defined as S and initialized as an empty list, and the road bounding box sequence Initialize to empty, start from the path segment where the ego vehicle is located, discretize the road boundaries to obtain the first waypoint of the path segment (which may be the ego vehicle's current location), obtain the left and right boundary points corresponding to the first waypoint in the lateral direction, and add the current left and right boundary points to List S; based on the road boundary curvature, select the next waypoint along the road travel direction, and check whether the distance from the first waypoint to the next waypoint along the road travel direction is less than or equal to a preset distance threshold; if the distance is less than or equal to the preset distance threshold, obtain the left and right boundary points corresponding to the next waypoint in the lateral direction and add them to List S; continue selecting the next waypoint along the road travel direction based on the road boundary curvature until the distance from the first waypoint to the next waypoint in the road travel direction is greater than the preset distance threshold, stop obtaining the left and right boundary points, and based on the last updated List S, the network bounding box sequence Creates.

[0075] In step (S2212), a grid obstacle bounding box of the first grid obstacle is generated based on the first grid obstacle information.

[0076] In some embodiments, based on the first grid obstacle information, a grid obstacle contour of the first grid obstacle is generated; and based on the grid obstacle contour, a grid obstacle bounding box is generated.

[0077] Specifically, based on the first grid obstacle information, the Suzuki contour tracking algorithm is used to generate the first grid obstacle as a closed contour graphic, i.e., a grid obstacle contour. In this way, processing of all raw point cloud grid obstacle data is avoided, thereby significantly reducing hardware requirements for processors and sensors. For example, grid obstacle contour at, is a single coordinate point of the grid obstacle contour, and the grid obstacle contour consists of n coordinate points. Grid obstacle bounding box In this case, the 4 vertices of the grid obstacle bounding box are 2 coordinate points and It is composed of the coordinate values ​​of, and here, is,

[0078] x0,x1,… , x n is the x-coordinate of n coordinate points on the grid obstacle contour, and y0, y1, … , y n is the y-coordinate of n coordinate points in the grid obstacle contour.

[0079] In step (S2213), a first grid obstacle located outside the road is determined based on the grid obstacle bounding box and the road bounding box.

[0080] An embodiment of the present invention can quickly and accurately determine a first grid obstacle located outside the road by performing two-stage collision detection on a first grid obstacle. For example, by first performing coarse collision detection on the grid obstacle, the first grid obstacle located outside the road can be quickly filtered out to reduce the amount of computation for collision detection; and by performing fine collision detection on the first grid obstacle where a collision occurred determined through coarse collision detection, the first grid obstacle located outside the road can be further determined to ensure that all remaining filtered first grid obstacles are located inside the road.

[0081] In the case of the above rough collision detection, in some embodiments, for each grid obstacle bounding box, a target road bounding box is determined based on the grid obstacle bounding box and the road bounding box, such that the Euclidean distance from the road bounding box to the grid obstacle bounding box is the smallest; collision detection is performed between the grid obstacle bounding box and the corresponding target road bounding box; and if the grid obstacle bounding box and the corresponding target road bounding box do not collide, it is determined that the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road. If the Euclidean distance from the road bounding box to the grid obstacle bounding box is relatively small, it indicates that there is a greater probability of a collision occurring between the road bounding box and the grid obstacle bounding box, and if the road bounding box and the grid obstacle bounding box corresponding to the case where the Euclidean distance is small do not collide, then the road bounding box and the grid obstacle corresponding to the case where the Euclidean distance is small do not collide even further. Accordingly, by determining the target road bounding box with the smallest Euclidean distance from the road bounding box to the grid obstacle bounding box and performing collision detection with the grid obstacle bounding box, the computational load of collision detection can be reduced, thereby improving the speed of obstacle decision-making. In some embodiments, it is only necessary to detect whether the vertices of the grid obstacle bounding box are located on or inside the target road bounding box. For example, when all vertices of the grid obstacle bounding box are located outside the target road bounding box, the first grid obstacle corresponding to the grid obstacle bounding box is determined to be located outside the road; when the vertices of the grid obstacle bounding box are located on or inside the target road bounding box, the first grid obstacle corresponding to the grid obstacle bounding box is determined to be located inside the road.

[0082] In the case of the above micro-collision detection, in some embodiments, when a collision occurs between a grid obstacle bounding box and a corresponding target road bounding box, it is determined whether a first grid obstacle corresponding to the grid obstacle bounding box is located outside the road. In some embodiments, collision detection is performed through the cross product of vectors based on the boundary point of the target road bounding box and the grid obstacle bounding box to determine whether a first grid obstacle corresponding to the grid obstacle bounding box is located outside the road. Specifically, a boundary point vector is determined; a vertex vector of the grid obstacle bounding box is determined; and if the cross product of the vertex vector of the grid obstacle bounding box and the boundary point vector is both greater than 0, it is determined that a first grid obstacle corresponding to the grid obstacle bounding box is located outside the road. If the cross product of the vertex vector of the grid obstacle bounding box and the boundary point vector is less than or equal to 0, it is determined that a first grid obstacle corresponding to the grid obstacle bounding box is located inside the road. Here, the boundary point vector includes a left boundary vector composed of two left boundary points of the target road bounding box and a right boundary vector composed of two right boundary points of the target road bounding box, and the vertex vector of the grid obstacle bounding box is a vector composed of a vertex of the grid obstacle bounding box and a single boundary point of the target road bounding box, and the said boundary point is a single boundary point corresponding to the boundary point vector participating in the cross product operation; for example, when the vertex vector and the right boundary vector are cross products, the boundary point of the vertex vector is a single boundary point corresponding to the right boundary vector. As an example, the grid obstacle bounding box is , the target road bounding box is , the left boundary vector is , the right boundary vector is ..., the four vertices of the grid obstacle bounding box B are traversed to form a four-vertex vector, the four-vertex vector is cross-producted with either the left boundary vector or the right boundary vector, respectively, and based on the cross-product result, it is determined whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road. For example, one vertex of B is If so, the cross product of the right boundary vector and the vertex vector is and, if c1 > 0, the vertex Located to the right of the right boundary of this road; otherwise, the vertex It is located at the top of the right boundary of the road or to the left of the right boundary of the road. Likewise, it can be determined whether another vertex of the grid obstacle bounding box is located at the top of the right boundary of the road or to the left or right of the right boundary of the road. In this way, it can be determined whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.

[0083] In step (S2214), the first grid obstacle located outside the road is filtered.

[0084] In step (S222), the remaining first grid obstacle is used as the second grid obstacle.

[0085] Additionally, embodiments of the present invention can reduce the number of first grid obstacles by aggregating and processing first grid obstacles located within the road. In some embodiments, the step of obtaining second grid obstacles by preprocessing first grid obstacles based on road information and first grid obstacle information may include the following steps.

[0086] In step (S223), based on road information and first grid obstacle information, a first grid obstacle located inside the road is determined.

[0087] In this embodiment, the first grid obstacle located inside the road may be determined according to the method of determining whether the first grid obstacle is located outside the road in the above embodiment, which is not described repeatedly herein.

[0088] In step (S224), the first grid obstacle located inside the road is collected and processed.

[0089] In some embodiments, based on the first grid obstacle information of the first grid obstacle located inside the road, a first obstacle bounding box of the first grid obstacle is generated; if the Euclidean distance between two adjacent first obstacle bounding boxes is smaller than the width of the vehicle, the two adjacent first obstacle bounding boxes are merged to generate a second obstacle bounding box; until the Euclidean distance between the second obstacle bounding box and the adjacent first obstacle bounding box is greater than or equal to the width of the vehicle, or until there is no first obstacle bounding box adjacent to the second obstacle bounding box, the second obstacle bounding box is used as the first obstacle bounding box, and when the Euclidean distance between two adjacent first obstacle bounding boxes is smaller than the width of the vehicle, the two adjacent first obstacle bounding boxes are merged to generate a second obstacle bounding box.

[0090] For example, create a CLOSED table and initialize it as an empty table, and the first obstacle bounding box set Take one first obstacle bounding box from and add it to the CLOSED table, and set Delete the above first obstacle bounding box, and next, the set Iterating through the set If the Euclidean distance between the first obstacle bounding box of and the first obstacle bounding box of the CLOSED table is smaller than the ego vehicle width, then the set The above-mentioned first obstacle bounding box is added to the CLOSED table, a new first obstacle bounding box is created by combining it with the first obstacle bounding box with the smaller Euclidean distance in the CLOSED table, and the first obstacle bounding box added to the CLOSED table is again set Delete from the set If this cycle is repeated until it is emptied, the aggregation process for the first grid obstacle located inside the road is completed.

[0091] In step (S225), the first grid obstacle processed through aggregation is used as the second grid obstacle.

[0092] In addition, embodiments of the present invention may filter first grid obstacles located outside the road based on road information and first grid obstacle information, and may also determine first grid obstacles located inside the road based on road information and first grid obstacle information, and aggregate and process first grid obstacles located inside the road. In this way, the number of first grid obstacles can be further reduced.

[0093] Based on the above embodiment, after obtaining a second grid obstacle, the second grid obstacle can be converted into a second convex hull obstacle using a fast convex hull algorithm. In this way, integrated decision-making regarding the grid obstacle and the convex hull obstacle can be realized.

[0094] In step (S240), based on the target convex hull obstacle information of the target convex hull obstacle, an avoidance decision is made for the target convex hull obstacle.

[0095] This step may be performed by the traffic method determiner (1132) of FIG. 5. In some embodiments, the step of making an avoidance decision for a target convex hull obstacle based on the target convex hull obstacle information is to apply a tag that does not require avoidance or a tag that does not require lateral avoidance to the target convex hull obstacle satisfying a preset filtering condition based on the target convex hull obstacle information. The embodiment of the present invention, by applying a tag that does not require avoidance or a tag that does not require lateral avoidance to the target convex hull obstacle satisfying a preset filtering condition, causes the trajectory generation unit (12) of FIG. 2 to ignore the target convex hull obstacle, thereby reducing the burden of processing obstacles on the trajectory generation unit (12), as well as improving the speed of trajectory generation and the rationality of trajectory generation.

[0096] In some embodiments, the preset filtering conditions include at least one of: the target convex hull obstacle being located outside the road; the motion state of the target convex hull obstacle satisfying the condition that lateral avoidance is not required; and the target convex hull obstacle being located on the guideline of the ego vehicle. Accordingly, the step of applying a tag that does not require avoidance or a tag that does not require lateral avoidance to a target convex hull obstacle satisfying the preset filtering conditions based on the target convex hull obstacle information includes: the step of applying a tag that does not require avoidance to the target convex hull obstacle when the target convex hull obstacle is located outside the road; and the step of applying a tag that does not require lateral avoidance to the target convex hull obstacle when the motion state of the target convex hull obstacle satisfies the condition that lateral avoidance is not required or when the target convex hull obstacle is located on the guideline of the ego vehicle. For example, referring to FIG. 7, when the target convex hull obstacle is located outside the road, such as obstacle 1, the target convex hull obstacle does not affect the normal driving of the autonomous vehicle (100), and in this case, the target convex hull obstacle is ignored and a tag is applied that does not require avoidance of the target convex hull obstacle.When the motion state of the target convex hull obstacle satisfies the condition that lateral avoidance is not required, for example, when the target convex hull obstacle crosses the road, such as when a pedestrian crosses the road, the autonomous vehicle (100) only needs to wait for the pedestrian to pass and does not need to generate a trajectory that bypasses the pedestrian, and can apply a tag to the pedestrian that lateral avoidance is not required; additionally, when the target convex hull obstacle changes lanes into the lane of the vehicle or when the longitudinal speed of the target convex hull obstacle is greater than the speed of the vehicle (e.g., an obstacle moving at high speed in an adjacent lane), if it does not affect the safety of the vehicle's lane, the vehicle does not need to avoid laterally, and applies a tag to the target convex hull obstacle that lateral avoidance is not required; Additionally, when the target convex hull obstacle is located on the guideline of the vehicle, such as when obstacle 2 is located on the guideline c of the vehicle, the autonomous vehicle (100) can choose to follow obstacle 2 without needing to avoid obstacle 2 in the lateral direction, and can understand that obstacle 2 is a dynamic obstacle moving in the same direction as the autonomous vehicle (100).

[0097] In the above embodiment, target convex hull obstacles satisfying preset filtering conditions can be filtered through the convex hull obstacle filter (1133) of FIG. 5. In some embodiments, the convex hull obstacle filter (1133) may include at least one of an obstacle frenet bounding box-based road network filter, a behavioral semantic information filter, and a guideline filter. The obstacle frenet bounding box-based road network filter can quickly filter target convex hull obstacles located outside the road, and the obstacle frenet bounding box-based road network filter filters target convex hull obstacles located outside the road through the two-stage collision detection method of the above embodiment; the behavioral semantic information filter filters target convex hull obstacles that do not require avoidance or do not require lateral avoidance according to the semantic information contained in the target convex hull obstacle; and the guideline filter can filter target convex hull obstacles that collide with the guideline.

[0098] In addition to the case where a decision is made that no avoidance is required by applying a tag that no avoidance is required to the target convex hull obstacle, and the case where a decision is made that no lateral avoidance is required by applying a tag that no lateral avoidance is required to the target convex hull obstacle, the embodiment of the present invention may make an avoidance decision of following, passing to the left, or passing to the right to the target convex hull obstacle. In some embodiments, the step of applying an avoidance tag to the target convex hull obstacle based on the target convex hull obstacle information and the ego vehicle guideline is to apply a following tag to the target convex hull obstacle if the target convex hull obstacle is located on the ego vehicle guideline; if the target convex hull obstacle is not located on the ego vehicle guideline, a passing tag to the right is applied to the target convex hull obstacle when the center of mass of the target convex hull obstacle is located to the left of the ego vehicle guideline, and a passing tag to the left is applied to the target convex hull obstacle when the center of mass of the target convex hull obstacle is located to the right of the ego vehicle guideline.

[0099] Referring further to FIG. 7, obstacle 2 is located on the guideline c of the vehicle, in which case the autonomous vehicle (100) follows obstacle 2 and applies a follow tag to obstacle 2. Obstacle 3 is located inside the road rather than on the guideline c of the vehicle and affects the lane safety of the autonomous vehicle (100), in which case the vehicle must avoid obstacle 3 by passing to the left or to the right of obstacle 3. Based on the technical method of the present invention, the center of mass of obstacle 3 and the relative position of the vehicle guideline c are detected, and if the center of mass of obstacle 3 is located to the right of the vehicle guideline c (see FIG. 7), the vehicle passes to the left of obstacle 3 and applies a pass tag to the left of obstacle 3; if the center of mass of obstacle 3 is located to the left of the vehicle guideline c, the vehicle passes to the right of obstacle 3 and applies a pass tag to the right of obstacle 3.

[0100] The vehicle decision planning method provided by an embodiment of the present invention preprocesses a first grid obstacle to obtain a second grid obstacle, and then converts the second grid obstacle into a second convex hull obstacle—that is, converts a grid-type obstacle into a convex hull-type obstacle—thereby realizing integrated decision-making for two types of obstacles (i.e., mixed-type obstacles)—a grid-type obstacle and a convex hull-type obstacle. This simplifies the obstacle decision-making flow for mixed-type obstacles, accelerates the obstacle decision-making process, and enables the decision planning module to conveniently and quickly perform obstacle decision-making.

[0101] With the advancement of vehicle intelligence technology, automatic control technology for autonomous vehicles is gradually emerging as a hotspot in the field of vehicle research. To prevent vehicles from colliding with obstacles, autonomous driving systems must plan a smooth, safe, and traversable path. Regarding optimization-based planning algorithms, Julius Ziegler proposed an optimization method in Dikard space that transforms planning problems into optimization problems. However, the above method significantly increases the computational burden of the planning module, fails to solve the problem of rapid obstacle avoidance, reduces the speed of trajectory generation, and furthermore, cannot be applied to handling obstacles in dynamic environments.

[0102] In response to the technical problems described above, FIG. 8 is a partial flowchart of another vehicle decision planning method provided by an embodiment of the present invention. The method may be applied to situations where an autonomous vehicle generates a drivable area for static and / or dynamic obstacles, and the method may be executed by a driving space generator. As illustrated in FIG. 8, the step of generating a drivable area (or, a method for generating a drivable area of ​​a vehicle) according to obstacle decision includes the following steps.

[0103] In step (S310), environmental sensing information is obtained.

[0104] The above environment sensing information includes at least two of lane information, obstacle information, and ego vehicle information, and the obstacle information includes static obstacle information and / or dynamic obstacle information. In some embodiments, lane information may include lane line information and road boundary information and may be acquired using a vehicle-mounted camera; obstacle information may include obstacle location information, obstacle size information, and obstacle motion information, wherein the obstacle location information may be collected using a high-precision map and a vehicle-mounted camera / LiDAR, obstacle size information may be acquired using a vehicle-mounted camera, and obstacle motion information may be acquired using a vehicle-mounted camera and / or LiDAR; and ego vehicle information may include ego vehicle location information and ego vehicle motion information, wherein the ego vehicle location information may be acquired using a high-precision map and an ego vehicle positioning module (e.g., GPS), and the ego vehicle motion information may be acquired using an ego vehicle motion sensor (e.g., speed sensor and acceleration sensor, etc.).

[0105] In step (S320), lane decision semantic information for each lane is determined based on environmental sensing information.

[0106] Here, lane decision semantic information includes transit time costs and safety costs. Transit time costs represent the traffic conditions of a lane; for example, if a vehicle can pass through a lane quickly, the transit time for that lane is fast; and safety costs represent the safety of the lane.

[0107] In some embodiments, the passing time cost of each lane may be determined according to the magnitude relationship between the longitudinal speed of the ego vehicle and the longitudinal speed of the obstacle. Accordingly, when the passing time cost is included in the lane decision semantic information, the step of determining the lane decision semantic information of each lane based on the environment sensing information includes, for each lane, the step of determining the collision time between the ego vehicle and the first obstacle in front of the ego vehicle based on the environment sensing information; and the step of determining the collision time as the passing time cost.

[0108] Specifically, the environmental sensing information includes the position information and longitudinal speed information of the ego vehicle, the position information and longitudinal speed information of the front obstacle closest to the ego vehicle in each lane, and based on the position information and obstacle position information of the ego vehicle, the longitudinal distance of the front obstacle closest to the ego vehicle in each lane is calculated; and based on the longitudinal speed information of the ego vehicle and the longitudinal speed information of the obstacle, it is determined whether the longitudinal speed of the obstacle is less than the longitudinal speed of the ego vehicle. When the longitudinal speed of the obstacle is less than the longitudinal speed of the ego vehicle, the collision time when a collision occurs between the ego vehicle and the front obstacle is predicted based on the longitudinal distance, the longitudinal speed information of the ego vehicle, and the longitudinal speed information of the obstacle, and the said collision time is determined as the passing time cost. Additionally, if there is no obstacle in front of the ego vehicle, or if the longitudinal speed of the first obstacle in front of the ego vehicle is greater than or equal to the longitudinal speed of the ego vehicle, a preset time is determined as the passing time cost. Based on the above technical method, the passing time cost can be calculated using the following formula.

[0109] ;

[0110] Here, The transit time cost, is the longitudinal speed of the vehicle, is the longitudinal velocity of the obstacle, and is a preset time, which is a fixed value and is greater than the collision time, for example, 1000 (this value is a numeric value, and the unit may be seconds or milliseconds, etc., the same as the collision time unit). As can be seen from the above formula, the lower the longitudinal speed of the first obstacle ahead of the ego vehicle, the lower the passing time cost and the worse the passability of the corresponding lane; when there is no obstacle ahead of the ego vehicle or when the longitudinal speed of the first obstacle ahead of the ego vehicle is greater than or equal to the longitudinal speed of the ego vehicle, the ego vehicle does not collide with the obstacle in the corresponding lane, so the passability of the corresponding lane is optimal.

[0111] For example, as illustrated in FIG. 9, in lanes of the same direction, the first obstacle in front of the ego vehicle in the lane is obstacle 1, and the first obstacle in front of the ego vehicle in the adjacent lane is obstacle 2. The longitudinal speed of the ego vehicle is 5 m / s, the longitudinal speed of obstacle 1 is 1 m / s, and the longitudinal speed of obstacle 2 is 10 m / s. In the case of the ego vehicle's lane, the longitudinal speed of obstacle 1 is smaller than the longitudinal speed of the ego vehicle, so the ego vehicle and obstacle 1 collide. In this case, if the distance D between the ego vehicle and obstacle 1 is determined to be 16 m, the collision time between the ego vehicle and obstacle 1 can be determined to be 4 s according to the above formula, so the passing time cost of the ego vehicle's lane is 4. In the case of an adjacent lane, the longitudinal speed of obstacle 2 is greater than the longitudinal speed of the ego vehicle, so no collision occurs between the ego vehicle and obstacle 2. In this case, the time-to-pass cost of the adjacent lane is a preset time, such as 10,000. Therefore, it can be determined that the time-to-pass cost of the adjacent lane is greater than the time-to-pass cost of the ego vehicle's lane; in other words, the passability of the adjacent lane is better.

[0112] In some embodiments, to ensure the safety of the vehicle, the safety cost of each lane must be determined simultaneously. Accordingly, when the safety cost is included in the lane decision semantic information, the step of determining the lane decision semantic information of each lane based on environmental sensing information comprises: determining the lane of the self-vehicle and other lanes based on lane information and self-vehicle information; determining a first preset safety cost as the safety cost for the lane of the self-vehicle; and for other lanes, determining a second preset safety cost as the safety cost if, based on environmental sensing information, it is determined that an obstacle has entered the danger zone of the self-vehicle within a preset time, and determining the first preset safety cost as the safety cost if, based on environmental sensing information, it is determined that an obstacle has not entered the danger zone of the self-vehicle within a preset time, wherein the second preset safety cost is different from the first preset safety cost.

[0113] Specifically, environmental sensing information includes lane information, ego vehicle information, and obstacle information. Based on the lane information and ego vehicle information, the ego vehicle's lane and other lanes are determined. In the case of the ego vehicle's lane, the ego vehicle fundamentally holds the absolute right of way, meaning that the safety of the ego vehicle's lane is the highest. For other lanes, a safety assessment can be performed regarding obstacles within the ego vehicle's observation area. If it is predicted that an obstacle within the ego vehicle's observation area will enter the ego vehicle's danger zone within a specific future time (i.e., a preset time), it means that the safety of the lane where the obstacle is currently located is low; conversely, if it is predicted that an obstacle within the ego vehicle's observation area will not enter the ego vehicle's danger zone within a specific future time, it means that the safety of the lane where the obstacle is currently located is high.

[0114] It can be understood that the lane safety corresponding to the second preset safety cost is lower than the lane safety corresponding to the first preset safety cost. In some embodiments, the second preset safety cost is smaller than the first preset safety cost. In some embodiments, a penalty mechanism may be used to assign values ​​to the first preset safety cost and the second preset safety cost, for example, the first preset safety cost is 0 and the second preset safety cost is -100000.

[0115] Based on the above technical method, an ST diagram (longitudinal displacement-time diagram) can be used to determine whether an obstacle in another lane will enter the danger zone of the self-vehicle within a specific time in the future. In some embodiments, based on environmental sensing information, the ST diagram curve of the self-vehicle and the ST diagram curve of the obstacle can be determined; based on the ST diagram curve of the self-vehicle, the danger zone of the self-vehicle can be determined; it is determined whether the ST diagram curve of the obstacle overlaps with the danger zone of the self-vehicle within a preset time; if the ST diagram curve of the obstacle overlaps with the danger zone of the self-vehicle within a preset time, it is determined that the obstacle has entered the danger zone of the self-vehicle within the preset time; otherwise, it is determined that the obstacle has not entered the danger zone of the self-vehicle within the preset time. For example, as illustrated in FIG. 10, the ST diagram curve of the self-vehicle is a curve labeled as the self-vehicle in the drawing, and the obstacle ST diagram curve includes curves labeled as obstacle 1, obstacle 2, obstacle 3, and obstacle 4, respectively, in the drawing. The danger area of ​​the vehicle includes the rear danger area of ​​the vehicle (the area corresponding to section L2) and the front danger area of ​​the vehicle (the area corresponding to section L3), and the observation area of ​​the vehicle includes the rear observation area of ​​the vehicle (the area corresponding to section L1) and the front observation area of ​​the vehicle (the area corresponding to section L4), and the preset time is T_e. Alternatively, L1 is 100 meters, L2 is 20 meters, L3 is 10 meters, L4 is 100 meters, and T_e is 6 seconds.Referring to FIG. 10, as can be seen from the ST diagram curve of the self-vehicle and the ST diagram curve of each obstacle, the obstacle ST diagram curve corresponding to obstacle 2 in the rear observation area of ​​the self-vehicle overlaps with the rear danger area of ​​the self-vehicle within a preset time T_e, the obstacle ST diagram curve corresponding to obstacle 1 in the rear observation area of ​​the self-vehicle does not overlap with the rear danger area of ​​the self-vehicle within a preset time T_e, the obstacle ST diagram curve corresponding to obstacle 3 in the front observation area of ​​the self-vehicle overlaps with the front danger area of ​​the self-vehicle within a preset time T_e, and the obstacle ST diagram curve corresponding to obstacle 4 in the front observation area of ​​the self-vehicle does not overlap with the front danger area of ​​the self-vehicle within a preset time T_e. As can be seen from this, obstacles 2 and 3 have entered the danger area of ​​the self-vehicle within a preset time, and the safety of the lane where obstacles 2 and 3 are currently located is low, that is, the safety cost of the corresponding lane is the second preset safety cost; Obstacle 1 and Obstacle 4 did not enter the danger zone of the self-vehicle within a preset time, and the safety of the lane where Obstacle 1 and Obstacle 4 are currently located is high, that is, the safety cost of the corresponding lane is the first preset safety cost. The fact that the obstacle ST diagram curve and the danger zone of the self-vehicle overlap includes cases where the obstacle ST diagram curve is located entirely within the danger zone of the self-vehicle or where a part of the obstacle ST diagram curve is located within the danger zone of the self-vehicle.In the above embodiment, to simplify calculations, the obstacle is set to move at a constant speed, the obstacle ST diagram curve is a straight line, the danger zone of the ego vehicle is a parallelogram, and all are shapes of the convex hull type. Accordingly, a collision detection algorithm based on the Gilbert-Johnson-Keerthi algorithm can be used to quickly calculate whether the obstacle will enter the danger zone of the ego vehicle within a preset time T_e.

[0116] In step (S330), a drivable area is generated based on lane decision semantic information.

[0117] For each lane, an embodiment of the present invention can generate a drivable area by taking into account both the time-to-pass cost and the safety cost, and accordingly, a lane can be selected that takes into account both traversability and safety.

[0118] In some embodiments, a weighted summation may be performed for each cost of the lane decision semantic information, and a drivable area may be generated based on the weighted summation result. In this embodiment, a weighted summation result is obtained by performing a weighted summation on the time-to-pass cost and the safety cost. For example:

[0119] ;

[0120] Here, is the weighted sum result (or, weighted sum value), and is the transit time cost, and It is a safety cost, and It is the weight of the time-to-pass cost, and is the weight of the safety cost. and This can be obtained through simulation or actual vehicle test experiments. Based on the above technical method, an embodiment of the present invention can determine the lane with the largest weighted sum value as the drivable area.

[0121] In some embodiments, to enable a planner to easily receive the drivable area, the boundaries of the drivable area are discretized to form drivable area boundary points including left boundary points and right boundary points. For example, based on a Frenet coordinate system, the drivable area can be discretized at a fixed resolution. As illustrated in FIG. 11, the left and right boundaries of the drivable area are generated at a fixed resolution based on a curved Frenet coordinate system constructed with respect to lane decision information and the lane center, where the left boundary represents the upper bound of the L value in the Frenet coordinate system, and the right boundary represents the lower bound of the L value in the Frenet coordinate system. Here, the longitudinal distance between two adjacent left boundary points or two adjacent right boundary points is the fixed resolution. FIG. 11 indicates that the vehicle is located in the right lane, and since the result of the lane decision is also the right lane, the left and right boundaries of the drivable area are as shown in FIG. 11. If the result of the lane decision is a lane change, then the drivable area includes two lanes.

[0122] The method for generating a drivable area of ​​a vehicle provided by an embodiment of the present invention determines lane decision semantic information for each lane according to environmental sensing information, converts the lane decision semantic information into constraint boundaries of a drivable area, and can rapidly generate a drivable area with high passability and safety by simultaneously considering passability and safety, and realizes rapid avoidance of obstacles by accelerating the generation of a driving trajectory; at the same time, since both the pass-time cost and the safety cost can represent the pass-cost of dynamic obstacles and the pass-time cost can represent the pass-cost of static obstacles, the technical solution of the present invention can simultaneously realize a pass-plan for dynamic obstacles and static obstacles by generating a drivable area based on the pass-time cost and the safety cost, and is applicable to the handling of obstacles in a dynamic environment.

[0123] Based on the above technical method, if at least two lanes are included in a drivable area determined according to lane decision semantic information, and static obstacles exist in both of the at least two lanes, the most optimal lane can be further selected according to the traffic width cost. In some embodiments, the lane decision semantic information further includes a traffic width cost, and the step of determining lane decision semantic information for each lane based on environmental sensing information includes: a step of determining the minimum traffic width of the lane based on lane information and static obstacle information; and a step of determining the minimum traffic width as a traffic width cost. The traffic width cost indicates that the lane is blocked by a static obstacle in front of the vehicle. In some embodiments, the step of determining the minimum traffic width of the lane based on lane information and static obstacle information includes: a step of determining the maximum traffic width of each static obstacle on the lane based on lane information and static obstacle information; and a step of determining the minimum value among the maximum traffic widths of each static obstacle as the minimum traffic width of the lane.

[0124] Specifically, a Frenet coordinate system is constructed, and each static obstacle is projected onto the Frenet coordinate system to generate an SL bounding box for each obstacle. For each lane, the left and right traffic widths of each static obstacle are calculated, the maximum traffic width of each static obstacle is selected from the left and right traffic widths, and the smallest maximum traffic width among all static obstacles is selected and used as the minimum traffic width of the lane, and the said minimum traffic width is determined as the traffic width cost. The larger the traffic width cost, the smaller the degree of lane congestion caused by static obstacles. For example, as illustrated in FIG. 12, obstacles 1 and 2 are included on the lane of the ego vehicle, and obstacle 3 is included on the adjacent lane (obstacles 1, 2, and 3 are all static obstacles). The maximum traffic width of obstacle 1 is d1, the maximum traffic width of obstacle 2 is d2, and the maximum traffic width of obstacle 3 is d3. Since d1 is smaller than d2 and d2 is smaller than d3, the minimum traffic width of the lane of the ego vehicle is d1, i.e., the traffic width cost of the lane of the ego vehicle is d1, and the minimum traffic width of the adjacent lane is d3, i.e., the traffic width cost of the adjacent lane is d3. In this case, the degree of obstacle blockage in the adjacent lane is smaller than the degree of obstacle blockage in the lane of the ego vehicle. Therefore, the adjacent lane can be further selected to create a drivable area.

[0125] Based on the above technical method, if the optimal drivable area cannot still be determined based on each cost among the lane decision semantic information, a stability cost is added to the lane decision semantic information to prioritize the selection of the ego vehicle's lane and generate a drivable area in order to secure the stability of the vehicle's driving trajectory. Accordingly, in some embodiments, the lane decision semantic information further includes a stability cost, and the step of determining the lane decision semantic information of each lane based on environment sensing information includes: a step of determining the ego vehicle's lane and other lanes based on lane information and ego vehicle information; a step of determining a first preset stability cost as a stability cost for the ego vehicle's lane; and a step of determining a second preset stability cost as a stability cost for other lanes; wherein the second preset stability cost is different from the first preset stability cost. In the above technical method, the first preset stability cost may be greater than the second preset stability cost, where the first preset stability cost is 100 and the second preset stability cost is 0.

[0126] Based on each of the above embodiments, a weighted sum is obtained by performing a weighted sum for each cost among the lane decision semantic information, and can be calculated through the following formula.

[0127]

[0128] Here, is the traffic width cost, and is a stability cost, and is the weight of the traffic width cost, and is the weight of the stability cost.

[0129] Based on the above technical method, in some embodiments, after the step of generating a drivable area based on lane decision semantic information, the method,

[0130] A step of updating the drivable area based on preset traffic rules;

[0131] A step of updating the drivable range based on the kinematic and dynamic constraints of the vehicle;

[0132] A step of updating the drivable area based on obstacle semantic information and a preset safe area - the preset safe area is connected to the drivable area - ; and

[0133] A step of updating a drivable area based on static obstacle information, obstacle decision semantic information, and a ray tracing algorithm, wherein the obstacle decision semantic information includes passing to the left of the obstacle or passing to the right of the obstacle; and at least one of these.

[0134] Specifically, in some embodiments, it can be determined whether a drivable area violates a traffic rule based on a preset traffic rule, and the drivable area that violates the traffic rule can be trimmed to update the drivable area. In embodiments of the present invention, the preset traffic rule may include general traffic rules such as yellow dashed solid lines, white dashed solid lines, and lane marking lines. For example, as illustrated in FIG. 13, after lane decision, the vehicle selects a lane change, so the two lanes of the lane change process are first used as the drivable area, i.e., the initial drivable area. However, since the end of the lane is a solid line, the drivable area is trimmed based on the preset traffic rule to obtain a drivable area limited to the black dot in FIG. 13 as a boundary point, i.e., an updated drivable area.

[0135] In some embodiments, the technical solution of the present invention may update the drivable area based on the kinematic and dynamic constraints of the vehicle. For example, based on the kinematic and dynamic constraints of the vehicle, when the vehicle temporarily borrows a road, the drivable area is updated by separately adding a drivable area. As illustrated in FIG. 14, the in-between angle between the heading angle of the vehicle and the heading angle of the road network is And, the curvature at the lane position of the own vehicle is And, since the L-coordinate value of the ego vehicle in the Frenet coordinate system is d, according to the Frenet kinematic equations, the lateral velocity of the ego vehicle with respect to the Frenet coordinate system is And, the lateral acceleration is

[0136] is;

[0137] Here, The wheel angle of the unmanned vehicle bicycle model It is the curvature of the ego vehicle trajectory calculated using, and the wheelbase of the ego vehicle is B, and in this case, And, is the rate of change of the curvature of the road network, and for approximate calculation, Set to, and a separate drivable area To calculate, assuming the final Frenet lateral velocity of the ego vehicle is 0, the kinematics in the Frenet coordinate system is In this way, the drivable area is updated by extending the drivable area outward through a separate calculated drivable area.

[0138] In some embodiments, considering the situation where a dynamic obstacle affecting the drivable area may exist in an adjacent lane, absolute safety of the drivable area cannot be secured; therefore, the drivable area that may be affected by the dynamic obstacle is trimmed to secure the safety of the remaining drivable area.

[0139] Specifically, based on obstacle semantic information, an obstacle to be avoided laterally is determined; if the movement trajectory of the obstacle to be avoided laterally occupies a preset safe area, a portion of the drivable area occupying the corresponding position of the preset safe area is trimmed. In this embodiment, the obstacle semantic information may include information indicating the obstacle motion state, such as obstacle lane merging, obstacle crossing, driving parallel to the obstacle, and driving in reverse. The ethic vehicle automatically determines whether to avoid the obstacle laterally based on the obstacle semantic information. For example, if obstacle lane merging is determined based on the obstacle semantic information, the ethic vehicle does not need to avoid the obstacle laterally, and if it is determined based on the obstacle semantic information that the obstacle is too close to the ethic vehicle's lane, the ethic vehicle must avoid the obstacle laterally. For example, as illustrated in FIG. 15, a preset safety area (if the drivable area is adjacent to a road boundary, a preset safety area may be added only on the inner side of the drivable area) is added to both sides of the drivable area to determine whether the movement trajectory of an obstacle output by an obstacle prediction module (as a prior module, not related to the present invention) occupies the preset safety area; if the movement trajectory occupies the preset safety area, for example, a portion of the drivable area that occupies the corresponding area of ​​the preset safety area, such as the trimming area in FIG. 15, is trimmed, and the drivable area is updated.

[0140] In some embodiments, since the drivable area obtained by each of the above embodiments still includes an area where static obstacles are located and thus does not satisfy the constraints for obstacle avoidance, the drivable area must be updated by further trimming the area where static obstacles are located from the drivable area. To avoid the problem of reduced vehicle traversability caused by generating an approximate drivable area using a Frenet bounding box in existing methods, the technical solution of the present invention accurately determines the area where static obstacles are located by combining obstacle decision semantic information and a ray tracing algorithm.

[0141] Specifically, the step of updating the drivable area based on static obstacle information, obstacle decision semantic information, and a ray tracing algorithm includes the step of determining the collision point between the ray and the obstacle based on static obstacle information, obstacle decision semantic information, and a ray tracing algorithm—where the collision point is located in the drivable area—; and the step of updating the drivable area based on the collision point. In some embodiments, the step of determining the collision point between the ray and the obstacle based on static obstacle information, obstacle decision semantic information, and a ray tracing algorithm includes the step of determining the light source point and the ray projection direction based on obstacle decision semantic information; the step of determining the ray projection range based on static obstacle information; the step of performing a ray scan on the obstacle based on the light source point, the ray projection direction, and the ray projection range; and the step of determining the collision point between the ray and the obstacle. In some embodiments, the ray tracing algorithm may use a Gilbert-Johnson-Keerthi algorithm-based ray projection algorithm to improve precision.

[0142] For example, obstacle decision semantic information may include passing to the left of the obstacle or passing to the right of the obstacle. If the obstacle decision semantic information is passing to the left of the obstacle, it is determined that the light source point is located to the left of the static obstacle, and the ray projection direction is perpendicular to the lane traffic direction and directed toward the static obstacle; if the obstacle decision semantic information is passing to the right of the obstacle, it is determined that the light source point is located to the right of the static obstacle, and the ray projection direction is perpendicular to the lane traffic direction and directed toward the static obstacle. Based on static obstacle information, such as location and size information of the static obstacle, the area where the static obstacle is located can be determined to determine the ray projection range. After determining the collision points between the ray and the obstacle, the drivable area defined by each collision point is trimmed. Embodiments of the present invention do not limit the specific location of the light source point, and in some embodiments, the light source point may be located at the boundary of the drivable area.

[0143] In a specific embodiment, as illustrated in FIG. 16, the Frenet bounding box box_sl={S_min,S_max,L_min,L_max} of the static obstacle is first determined, and the ID range of the static obstacle in the longitudinal direction of the drivable area can be determined based on the Frenet bounding box. When the resolution of the boundary point of the drivable area is △s, the ID range, i.e., the ray projection range, is (id_start, id_end), where id_start=floor(s_min / △s) and id_end=ceil(s_max / △s), floor represents the downward integer operation of a floating-point number, and ceil represents the upward integer operation of a floating-point number. An embodiment of the present invention determines the ray projection range that includes the entire static obstacle using only the Frenet bounding box to ensure a complete scan of the obstacle by a ray, and the collision point determined thereafter is located on the static obstacle rather than at the boundary of the Frenet bounding box. As illustrated in FIG. 17, the drivable area includes two static obstacles, namely Obstacle 1 and Obstacle 2. Based on the obstacle decision semantic information of Obstacle 1, it can be determined that the vehicle passes to the right of Obstacle 1, and based on the obstacle decision semantic information of Obstacle 2, it can be determined that the vehicle passes to the left of Obstacle 2. Taking Obstacle 1 as an example, based on the obstacle decision semantic information of Obstacle 1, it is determined that a point light source is located at the right boundary point of the drivable area, and based on the static obstacle information of Obstacle 1, the light source determines the ray projection range for Obstacle 1, i.e., the ID range, thereby allowing the point light source to sequentially scan Obstacle 1 according to the ID range. Specifically, when a ray collides with Obstacle 1, if the collision point is located within the drivable area, the drivable area on one side far from the point light source is trimmed until the scanning of the ray projection range is completed.By using the above technical solution, the drivable area satisfies the constraints on obstacle avoidance, while simultaneously improving the accuracy of resolving the area occupied by static obstacles within the drivable area, thereby improving the vehicle's passability.

[0144] An embodiment of the present invention further provides a vehicle decision planning device. Referring to FIG. 2, the vehicle decision planning device includes a basic coordinate system generator (111), a guideline generator (112), an obstacle decision maker (113), and a driving space generator (114).

[0145] The above basic coordinate system generator (111) is used to generate a basic coordinate system; the guideline generator (112) generates a guideline under the above basic coordinate system to determine the approximate future driving trajectory of the vehicle; the obstacle decision maker (113) performs obstacle decision under the constraints of the above guideline; and the driving space generator (114) generates a driving area according to the obstacle decision.

[0146] FIG. 18 is a block diagram of a functional module of an obstacle decision-maker in a vehicle decision planning device provided by an embodiment of the present invention. As shown in FIG. 18, the obstacle decision-maker (or, decision device for obstacle avoidance) includes an information acquisition module (401), a preprocessing module (402), a type conversion module (403), and an avoidance decision module (404).

[0147] The above information acquisition module (401) is used to acquire road information, first grid obstacle information of the first grid obstacle and first convex hull obstacle information of the first convex hull obstacle;

[0148] The preprocessing module (402) obtains a second grid obstacle by preprocessing the first grid obstacle based on road information and first grid obstacle information, and the number of the second grid obstacles is smaller than the number of the first grid obstacles;

[0149] The type conversion module (403) is used to convert the second grid obstacle into the second convex hull obstacle;

[0150] The avoidance decision module (404) is used to make an avoidance decision regarding a target convex hull obstacle based on target convex hull obstacle information of the target convex hull obstacle, and the target convex hull obstacle includes a first convex hull obstacle and / or a second convex hull obstacle.

[0151] In some embodiments, the preprocessing module (402) is,

[0152] An obstacle filtering unit that filters first obstacles located outside the road based on road information and first grid obstacle information - using the remaining first grid obstacles as second grid obstacles - ; and / or,

[0153] Based on road information and first grid obstacle information, an obstacle aggregation unit for determining a first grid obstacle located inside the road and performing aggregation processing on said first grid obstacle located inside the road—using the aggregated first grid obstacle as a second grid obstacle—is included.

[0154] In some embodiments, the obstacle filtering unit is,

[0155] A road bounding box generation sub-unit for generating a road bounding box along the road's traffic direction based on road information;

[0156] A grid obstacle bounding box generation sub-unit for generating a grid obstacle bounding box of a first grid obstacle based on first grid obstacle information;

[0157] A first grid obstacle sub-unit for determining a first grid obstacle located outside the road based on a grid obstacle bounding box and a road bounding box; and

[0158] It includes a first grid obstacle filtering sub-unit for filtering a first grid obstacle located outside the road.

[0159] In some embodiments, the road bounding box generating sub-unit specifically,

[0160] Based on road information, road boundaries are discretized into boundary points;

[0161] Creates a road bounding box based on the boundary points.

[0162] In some embodiments, the road bounding box generating sub-unit specifically,

[0163] Based on road information, determine road boundaries and road boundary curvature;

[0164] Based on the curvature of the road boundary, the road boundary is discretized to obtain a group of boundary points spaced apart along the road traffic direction, each group of boundary points including a corresponding left boundary point and a right boundary point in the transverse direction, and the transverse direction is perpendicular to the road traffic direction.

[0165] In some embodiments, the road bounding box generating sub-unit specifically,

[0166] Based on any group of two adjacent boundary points, generate a rectangular box passing through each boundary point of any group of two adjacent boundary points, and use the rectangular box as a road bounding box.

[0167] In some embodiments, the road bounding box generating sub-unit specifically,

[0168] Using the vehicle's current location as the starting waypoint;

[0169] Acquiring a group of one boundary point of a corresponding starting waypoint in the lateral direction;

[0170] Based on the road boundary curvature, the next waypoint is selected along the road traffic direction, and the distance between two adjacent waypoints has a negative relationship with the road boundary curvature;

[0171] Until the distance in the direction of travel from the next waypoint to the current position of the vehicle is greater than a preset distance threshold, the process returns to the step of obtaining one boundary point group of the corresponding starting waypoint in the lateral direction using the next waypoint as the starting waypoint, and determines all currently obtained boundary point groups as boundary point groups.

[0172] In some embodiments, the grid obstacle bounding box generation sub-unit specifically,

[0173] Based on the first grid obstacle information, the grid obstacle contour of the first grid obstacle is generated;

[0174] Generates grid obstacle bounding boxes based on grid obstacle contours.

[0175] In some embodiments, the first grid obstacle sub-unit specifically,

[0176] For each grid obstacle bounding box, based on the grid obstacle bounding box and the road bounding box, determine the target road bounding box with the smallest Euclidean distance from the road bounding box to the grid obstacle bounding box;

[0177] Performs collision detection between the grid obstacle bounding box and the corresponding target road bounding box;

[0178] If the grid obstacle bounding box and the corresponding target road bounding box do not collide, it is determined that the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.

[0179] In some embodiments, the device is,

[0180] It further includes a first grid obstacle location determination module for determining whether a first grid obstacle corresponding to the grid obstacle bounding box is located outside the road when a collision occurs between the grid obstacle bounding box and the corresponding target road bounding box.

[0181] In some embodiments, the first grid obstacle location determination module specifically,

[0182] Based on the boundary points of the target road bounding box and the grid obstacle bounding box, collision detection is performed through the cross product of vectors to determine whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.

[0183] In some embodiments, the first grid obstacle location determination module specifically,

[0184] Determining the boundary point vector;

[0185] Determining the vertex vectors of the grid obstacle bounding box;

[0186] When the cross product of the vertex vector and the boundary point vector of the grid obstacle bounding box is greater than 0, it is determined that the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.

[0187] In some embodiments, the first grid obstacle location determination module specifically,

[0188] Determining the boundary point vector;

[0189] Determining the vertex vectors of the grid obstacle bounding box;

[0190] If the cross product of the vertex vector and the boundary point vector of the grid obstacle bounding box is less than or equal to 0, it is determined that the first grid obstacle corresponding to the grid obstacle bounding box is located inside the road.

[0191] In some embodiments, the obstacle gathering unit is,

[0192] A first obstacle bounding box generating sub-unit for generating a first obstacle bounding box of a first grid obstacle based on first grid obstacle information of a first grid obstacle located inside a road;

[0193] A second obstacle bounding box generating sub-unit for merging two adjacent first obstacle bounding boxes to generate a second obstacle bounding box when the Euclidean distance between two adjacent first obstacle bounding boxes is smaller than the width of the vehicle; and

[0194] A return execution unit that returns to the step of merging two adjacent first obstacle bounding boxes to create a second obstacle bounding box, using the second obstacle bounding box as the first obstacle bounding box until the Euclidean distance between the second obstacle bounding box and the adjacent first obstacle bounding box is greater than or equal to the width of the ego vehicle, or until there is no first obstacle bounding box adjacent to the second obstacle bounding box.

[0195] In some embodiments, the avoidance decision module (404) is,

[0196] A first decision unit for applying a tag that does not require avoidance or a tag that does not require lateral avoidance to a target convex hull obstacle satisfying preset filtering conditions, based on target convex hull obstacle information; and / or,

[0197] It includes a second decision-making unit for applying avoidance tags to a target convex hull obstacle based on target convex hull obstacle information and self-vehicle guidelines; wherein the avoidance tags include a pass to the left tag, a pass to the right tag, or a follow tag.

[0198] In some embodiments, the preset filtering conditions are the following conditions, i.e.

[0199] Target convex hull obstacle located outside the road;

[0200] The motion state of the target convex hull obstacle satisfies the condition that lateral avoidance is not required; and

[0201] The target convex hull obstacle is located on the vehicle's guideline; includes at least one of these.

[0202] In some embodiments, the first decision-making unit specifically,

[0203] When the target convex hull obstacle is located outside the road, apply a tag to the target convex hull obstacle that does not require avoidance;

[0204] When the motion state of the target convex hull obstacle satisfies the condition that lateral avoidance is not required, or when the target convex hull obstacle is located on the guide line of the vehicle, a tag that lateral avoidance is not required is applied to the target convex hull obstacle.

[0205] In some embodiments, the condition under which lateral avoidance is not required is the following condition, i.e.

[0206] Target convex hull obstacle crossing the road;

[0207] The target convex hull obstacle changes lanes into the vehicle's lane;

[0208] Conditions in which the longitudinal speed of the target convex hull obstacle is faster than the vehicle speed; include any one of the following.

[0209] In some embodiments, the second decision-making unit specifically,

[0210] If the target convex hull obstacle is located on the guideline of the self-vehicle, a follow tag is applied to the target convex hull obstacle;

[0211] If the target convex hull obstacle is not located on the ego vehicle's guideline, apply a pass tag to the right for the target convex hull obstacle when the center of mass of the target convex hull obstacle is located to the left of the ego vehicle's guideline, and apply a pass tag to the left for the target convex hull obstacle when the center of mass of the target convex hull obstacle is located to the right of the ego vehicle's guideline.

[0212] FIG. 19 is a block diagram of a functional module of a driving space generator of a vehicle decision planning device provided by an embodiment of the present invention. As shown in FIG. 19, the driving space generator (or, vehicle driving area generating device) includes a sensing information acquisition module (501), a lane decision semantic information determination module (502), and a driving area generating module (503).

[0213] The above detection information acquisition module (501) is used to acquire environmental detection information, and the environmental detection information includes at least two of lane information, obstacle information and self-vehicle information, and the obstacle information includes static obstacle information and / or dynamic obstacle information;

[0214] The lane decision semantic information determination module (502) is used to determine lane decision semantic information for each lane based on environmental sensing information, and the lane decision semantic information includes a time-to-go cost and a safety cost;

[0215] The drivable area generation module (503) is used to generate a drivable area based on lane decision semantic information.

[0216] In some embodiments, when the lane decision semantic information includes a passage time cost, the lane decision semantic information determination module (502) specifically,

[0217] For each lane, based on environmental sensing information, the collision time between the ego vehicle and the first obstacle in front of the ego vehicle is determined;

[0218] The collision time is determined as the pass time cost.

[0219] In some embodiments, the lane decision semantic information determination module (502) is,

[0220] Based on environmental sensing information, if it is determined that there are no obstacles in front of the vehicle, or that the longitudinal speed of the first obstacle in front of the vehicle is greater than or equal to the longitudinal speed of the vehicle, a preset time is determined as the passing time cost.

[0221] In some embodiments, when safety costs are included in lane decision semantic information, the lane decision semantic information determination module (502) specifically,

[0222] Based on lane information and ego vehicle information, determine the ego vehicle's lane and other lanes;

[0223] For the self-vehicle lane, a first preset safety cost is determined as the safety cost;

[0224] For other lanes, if it is determined based on environmental detection information that an obstacle enters the danger zone of the self-vehicle within a preset time, a second preset safety cost is determined as the safety cost, and if it is determined based on environmental detection information that an obstacle does not enter the danger zone of the self-vehicle within a preset time, a first preset safety cost is determined as the safety cost, and the second preset safety cost is different from the first preset safety cost.

[0225] In some embodiments, the lane decision semantic information determination module (502) specifically,

[0226] Based on environmental sensing information, the ST diagram curve of the ego vehicle and the ST diagram curve of the obstacle are determined;

[0227] Based on the ST diagram curve of the own vehicle, the risk area of ​​the own vehicle is determined;

[0228] Determining whether the obstacle's ST diagram curve overlaps with the self-vehicle's danger zone within a preset time;

[0229] If the obstacle's ST diagram curve overlaps with the self-vehicle's danger zone within a preset time, it is determined that the obstacle has entered the self-vehicle's danger zone within the preset time; otherwise, it is determined that the obstacle has not entered the self-vehicle's danger zone within the preset time.

[0230] In some embodiments, the lane decision semantic information further includes traffic width costs, and the lane decision semantic information determination module (502) specifically,

[0231] Based on lane information and static obstacle information, determine the minimum lane width;

[0232] The minimum traffic width is determined by the traffic width cost.

[0233] In some embodiments, the lane decision semantic information determination module (502) specifically,

[0234] Based on lane information and static obstacle information, the maximum passage width of each static obstacle on the lane is determined;

[0235] The minimum value among the maximum traffic widths of each static obstacle is determined as the minimum traffic width of the lane.

[0236] In some embodiments, the lane decision semantic information further includes stability costs, and the lane decision semantic information determination module (502) specifically,

[0237] Based on lane information and ego vehicle information, determine the ego vehicle's lane and other lanes;

[0238] For the lane of the self-vehicle, a first preset safety cost is determined as the safety cost;

[0239] For other lanes, a second preset stability cost is determined as the stability cost, and the second preset stability cost is different from the first preset stability cost.

[0240] In some embodiments, the drivable area generating module (503) specifically,

[0241] Performing a weighted summation for each cost of the lane decision semantic information;

[0242] Based on the weighted sum results, a drivable area is generated.

[0243] In some embodiments, the device is,

[0244] It further includes a discrete module for generating a drivable area based on lane decision semantic information and then discretizing the boundaries of the drivable area.

[0245] In some embodiments, the device further includes a drivable area update module, and the drivable area update module generates a drivable area based on lane decision semantic information, and then specifically

[0246] Based on preset traffic rules, it updates the drivable area;

[0247] Based on the vehicle's kinematic and dynamic constraints, it updates the drivable range;

[0248] Based on obstacle semantic information and a preset safety zone, the drivable zone is updated, wherein the preset safety zone is connected to the drivable zone; and

[0249] Based on static obstacle information, obstacle decision semantic information, and a ray tracing algorithm, it is used for at least one of the update operations to update the drivable area—the obstacle decision semantic information includes passing to the left of the obstacle or passing to the right of the obstacle.

[0250] In some embodiments, the drivable area update module specifically,

[0251] Based on obstacle semantic information, determine the obstacles to be avoided laterally;

[0252] If the movement trajectory of an obstacle that must be avoided laterally occupies a preset safe area, a portion of the drivable area that occupies the corresponding area of ​​the preset safe area is trimmed.

[0253] In some embodiments, the drivable area update module specifically,

[0254] Based on static obstacle information, obstacle decision semantic information, and ray tracing algorithms, the collision point between the ray and the obstacle is determined - the collision point is located within the drivable area - ;

[0255] Update the drivable area based on the collision point.

[0256] In some embodiments, the drivable area update module specifically,

[0257] Based on obstacle decision semantic information, determine the light source point and the ray projection direction;

[0258] Determining the ray projection range based on static obstacle information;

[0259] Based on the light source point, ray projection direction, and ray projection range, perform a ray scan against obstacles;

[0260] Determines the collision point between the beam and the obstacle.

[0261] The vehicle decision planning device disclosed in the above embodiment can perform the vehicle decision planning method disclosed in the above embodiment and can achieve equivalent beneficial effects, and a detailed description is omitted to prevent repetition of content.

[0262] Embodiments of the present invention also provide an electronic device, said electronic device comprising a memory and one or more processors, said memory being connected to one or more processors in communication, said memory storing instructions that can be executed by one or more processors, said instructions being executed by one or more processors, and thereby the method described in any embodiment of the present invention is performed by said electronic device.

[0263] FIG. 20 is a schematic diagram of the structure of an electronic device according to a preferred embodiment of the present invention. As shown in FIG. 20, the electronic device (600) includes a central processing unit (CPU) (601), which can implement various steps of the embodiment by executing a program stored in a read-only memory (ROM) (602) or a program loaded into a random access memory (RAM) (603) from a memory unit (608). Various programs and data required for the operation of the electronic device (600) are stored in the RAM (603). The CPU601, ROM602, and RAM603 are connected to each other via a bus (604). An input / output (I / O) interface (605) is also connected to the bus (604).

[0264] The following components are connected to the I / O interface (605): namely, an input section (606) of a keyboard and mouse; an output section (607) of a cathode ray tube (CRT), liquid crystal monitor (LCD), and speaker; a memory section (608) of a hard disk; and a communication section (609) of a network interface card of a LAN card and modem. The communication section (609) performs communication processing through a network such as the Internet. A drive (610) is also connected to the I / O interface (605) as needed. A removable medium (611), such as a disk, optical disk, magnetic disk, or semiconductor memory, is mounted on the drive (610) so that a computer program read from it is installed in the memory section (608) as needed.

[0265] In particular, according to an embodiment of the present invention, the above-described method may be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product comprising a computer program tangibly contained in a readable medium, said computer program including program code for performing the obstacle avoidance method. In such an embodiment, the computer program may be downloaded and installed from a network via a communication unit (609) and / or installed from a removable medium (611).

[0266] The flowcharts and block diagrams of the accompanying drawings illustrate the architecture, function, and operation of possible implementations of the apparatus, method, and computer program product of each embodiment of the present invention. Each block of a flowchart or block diagram may represent a module, program segment, or part of code, and the module, program segment, or part of code may contain one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions indicated in the blocks may occur in an order different from the order indicated in the drawings. For example, two consecutive blocks may actually be executed in parallel or in reverse order depending on the related functions. Additionally, each block of the block diagram and / or flowchart, and combinations of blocks of the block diagram and / or flowchart, may be implemented by a dedicated hardware-based system that performs the specified function or operation, or by a combination of dedicated hardware and computer instructions.

[0267] The units or modules described in the embodiments of the present invention may be implemented by software or hardware. The described units or modules may also be provided to a processor, and the names of such units or modules do not, in some cases, constitute a limitation of the units or modules themselves.

[0268] Additionally, embodiments of the present invention provide a computer-readable storage medium, said computer-readable storage medium may be a computer-readable storage medium included in the device of said embodiment, or a computer-readable storage medium existing separately and not assembled in the device. Computer-readable instructions are stored in the computer-readable storage medium, and the method described in any embodiment of the present invention may be implemented by executing the computer-readable instructions by a computing device.

[0269] Based on the above road information, the step of generating a road bounding box along the road traffic direction is:

[0270] Based on the above road information, a step of discretizing road boundaries into boundary points; and

[0271] The method includes the step of generating the road bounding box based on the above boundary points.

[0272] Based on the above road information, the step of discretizing road boundaries into boundary points is:

[0273] A step of determining a road boundary and road boundary curvature based on the above road information; and

[0274] Based on the curvature of the road boundary, the method includes the step of discretizing the road boundary to obtain a group of boundary points spaced apart along the road traffic direction, wherein each group of boundary points includes a left boundary point and a right boundary point corresponding in the transverse direction, and the transverse direction is perpendicular to the road traffic direction.

[0275] Based on the above boundary points, the step of generating the road bounding box is:

[0276] Based on any two adjacent boundary point groups, a rectangular box passing through each boundary point of said any two adjacent boundary point groups is generated, and said rectangular box is used as a road bounding box.

[0277] The step of discretizing the road boundary based on the curvature of the road boundary to obtain a group of boundary points arranged at intervals along the road traffic direction is,

[0278] Step of using the vehicle's current location as a starting waypoint;

[0279] A step of obtaining a group of one boundary point of the corresponding starting waypoint in the lateral direction;

[0280] A step of selecting the next waypoint along the road traffic direction based on the road boundary curvature - the distance between the two adjacent waypoints has a negative relationship with the road boundary curvature - ;

[0281] The method includes the step of returning to the step of obtaining a group of boundary points of a group of corresponding start waypoints in the lateral direction by using the next waypoint as the start waypoint until the distance from the next waypoint to the current position of the vehicle in the road traffic direction is greater than a preset distance threshold, and determining all currently obtained boundary point groups as boundary point groups.

[0282] Based on the first grid obstacle information, the step of generating a grid obstacle bounding box of the first grid obstacle is:

[0283] A step of generating a grid obstacle contour of the first grid obstacle based on the first grid obstacle information; and

[0284] The method includes the step of generating the grid obstacle bounding box based on the grid obstacle contour.

[0285] The step of determining the first grid obstacle located outside the road based on the grid obstacle bounding box and the road bounding box is:

[0286] For each of the above grid obstacle bounding boxes, a step of determining the target road bounding box having the smallest Euclidean distance from the road bounding box to the grid obstacle bounding box based on the grid obstacle bounding box and the road bounding box;

[0287] A step of performing collision detection between the grid obstacle bounding box and the corresponding target road bounding box; and

[0288] If the target road bounding box corresponding to the grid obstacle bounding box does not collide with the grid obstacle bounding box, the method includes the step of determining that the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.

[0289] The above method is,

[0290] The method further includes the step of determining whether a first grid obstacle corresponding to the grid obstacle bounding box is located outside the road when a collision occurs between the grid obstacle bounding box and the target road bounding box.

[0291] The step of determining whether a first grid obstacle corresponding to the grid obstacle bounding box is located outside the road is:

[0292] Based on the boundary points of the target road bounding box and the grid obstacle bounding box, the method includes the step of performing collision detection through the cross product of vectors to determine whether a first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.

[0293] The step of determining whether a first grid obstacle corresponding to the grid obstacle bounding box is located outside the road by performing collision detection through the cross product of vectors based on the boundary points of the target road bounding box and the grid obstacle bounding box,

[0294] Step of determining boundary point vectors;

[0295] Step of determining the vertex vectors of the grid obstacle bounding box; and

[0296] The method includes the step of determining that a first grid obstacle corresponding to the grid obstacle bounding box is located outside the road when the cross product of the vertex vector of the grid obstacle bounding box and the boundary point vector is both greater than 0.

[0297] The above method is,

[0298] Step of determining boundary point vectors;

[0299] Step of determining the vertex vectors of the grid obstacle bounding box; and

[0300] The method further includes the step of determining that a first grid obstacle corresponding to the grid obstacle bounding box is located inside the road when the cross product of the vertex vector of the grid obstacle bounding box and the boundary point vector is less than or equal to 0.

[0301] The step of performing aggregation processing on the first grid obstacle located inside the road is:

[0302] A step of generating a first obstacle bounding box of the first grid obstacle based on the first grid obstacle information of the first grid obstacle located inside the road;

[0303] If the Euclidean distance between two adjacent first obstacle bounding boxes is smaller than the width of the vehicle, the step of merging the two adjacent first obstacle bounding boxes to create a second obstacle bounding box; and

[0304] The method includes the step of returning to the step of merging the two adjacent first obstacle bounding boxes to create a second obstacle bounding box, until the Euclidean distance between the second obstacle bounding box and the adjacent first obstacle bounding box is greater than or equal to the width of the vehicle, or until there is no first obstacle bounding box adjacent to the second obstacle bounding box, by using the second obstacle bounding box as the first obstacle bounding box, and when the Euclidean distance between the two adjacent first obstacle bounding boxes is smaller than the width of the vehicle.

[0305] The step of making an avoidance decision regarding the target convex hull obstacle based on the target convex hull obstacle information of the target convex hull obstacle,

[0306] Based on the above target convex hull obstacle information, a step of applying a tag that does not require avoidance or a tag that does not require lateral avoidance to the target convex hull obstacle satisfying a preset filtering condition; and / or,

[0307] Based on the target convex hull obstacle information and the vehicle guideline, the method includes the step of applying an avoidance tag to the target convex hull obstacle, wherein the avoidance tag includes a pass to the left tag, a pass to the right tag, or a follow tag.

[0308] The above preset filtering conditions are the following conditions, namely

[0309] The above-mentioned convex hull obstacle is located outside the road;

[0310] The motion state of the above-mentioned target convex hull obstacle satisfies the condition that lateral avoidance is not required; and

[0311] The above-mentioned target convex hull obstacle is located on the guideline of the vehicle; at least one of which is included.

[0312] Based on the above target convex hull obstacle information, the step of applying a tag that does not require avoidance or a tag that does not require lateral avoidance to the target convex hull obstacle satisfying preset filtering conditions is as follows:

[0313] When the target convex hull obstacle is located outside the road, the step of applying a tag that does not require avoidance to the target convex hull obstacle; and

[0314] The method includes the step of applying a tag that does not require lateral avoidance to the target convex hull obstacle when the motion state of the target convex hull obstacle satisfies the condition that lateral avoidance is not required, or when the target convex hull obstacle is located on the guide line of the vehicle.

[0315] The condition under which lateral avoidance is not required, as mentioned above, is the following condition, namely

[0316] The above-mentioned convex hull obstacle crosses the road;

[0317] The above-mentioned target convex hull obstacle changes lanes into the lane of the vehicle; and

[0318] Conditions in which the longitudinal speed of the target convex hull obstacle is faster than the vehicle speed; any one of these conditions.

[0319] Based on the above target convex hull obstacle information and self-vehicle guidelines, the step of applying an avoidance tag to the above target convex hull obstacle is:

[0320] If the target convex hull obstacle is located on the guideline of the vehicle, the step of applying a tracking tag to the target convex hull obstacle; and

[0321] If the target convex hull obstacle is not located on the guideline of the self-vehicle, the method includes the step of applying a pass tag to the right for the target convex hull obstacle when the center of mass of the target convex hull obstacle is located to the left of the guideline of the self-vehicle, and applying a pass tag to the left for the target convex hull obstacle when the center of mass of the target convex hull obstacle is located to the right of the guideline of the self-vehicle.

[0322] The above method is,

[0323] Based on the above environment sensing information, if it is determined that there is no obstacle in front of the vehicle or that the longitudinal speed of the first obstacle in front of the vehicle is greater than or equal to the longitudinal speed of the vehicle, the method further includes the step of determining a preset time as the passing time cost.

[0324] The above method is,

[0325] A step of determining the ST diagram curve of the self-vehicle and the ST diagram curve of the obstacle based on the above-mentioned environment sensing information;

[0326] A step of determining the risk area of ​​the self-vehicle based on the ST diagram curve of the self-vehicle above;

[0327] A step of determining whether the obstacle ST diagram curve overlaps with the danger area of ​​the self-vehicle within the above preset time; and

[0328] The method further includes the step of determining that the obstacle enters the danger zone of the vehicle within the preset time if the ST diagram curve of the obstacle overlaps with the danger zone of the vehicle within the preset time, and otherwise determining that the obstacle does not enter the danger zone of the vehicle within the preset time.

[0329] The above lane decision semantic information further includes traffic width costs, and the step of determining lane decision semantic information for each lane based on the above environment sensing information is,

[0330] A step of determining the minimum traffic width of a lane based on the lane information and the static obstacle information; and

[0331] The step of determining the above minimum traffic width at the above traffic width cost; is included.

[0332] Based on the lane information and the static obstacle information, the step of determining the minimum traffic width of the lane is:

[0333] A step of determining the maximum passage width of each static obstacle on the lane based on the lane information and the static obstacle information; and

[0334] It includes the step of determining the minimum value among the maximum traffic widths of each static obstacle as the minimum traffic width of the lane.

[0335] The above lane decision semantic information further includes safety costs, and the step of determining lane decision semantic information for each lane based on the environment sensing information is,

[0336] A step of determining the lane of the self vehicle and other lanes based on the above lane information and the above self vehicle information;

[0337] For the lane of the above-mentioned vehicle, a step of determining a first preset safety cost as the safety cost; and

[0338] For the other lane mentioned above, the method includes the step of determining a second preset stability cost as the stability cost; wherein the second preset stability cost is different from the first preset stability cost.

[0339] Based on the above lane decision-making semantic information, the step of generating a drivable area is:

[0340] A step of performing a weighted sum for each cost of the above-mentioned lane decision semantic information; and

[0341] It includes the step of generating a drivable area based on the weighted sum result.

[0342] After the step of generating a drivable area based on the lane decision semantic information above, the method,

[0343] It further includes the step of discretizing the boundary of the drivable area.

[0344] After the step of generating a drivable area based on the lane decision semantic information above, the method,

[0345] A step of updating the drivable area based on preset traffic rules;

[0346] A step of updating the drivable area based on the kinematic and dynamic constraints of the vehicle;

[0347] A step of updating the drivable area based on obstacle semantic information and a preset safe area - the preset safe area is connected to the drivable area - ; and

[0348] The method further includes at least one of the step of updating the drivable area based on the static obstacle information, obstacle decision semantic information, and ray tracing algorithm, wherein the obstacle decision semantic information includes passing to the left of the obstacle or passing to the right of the obstacle.

[0349] The step of updating the drivable area based on the obstacle semantic information and the preset safety area is:

[0350] A step of determining obstacles to be avoided laterally based on obstacle semantic information; and

[0351] If the movement trajectory of an obstacle that must be avoided in the above lateral direction occupies the above preset safe area, the method includes the step of trimming a portion of the drivable area that occupies the corresponding area of ​​the above preset safe area.

[0352] Based on the static obstacle information, obstacle decision semantic information, and ray tracing algorithm, the step of updating the drivable area is:

[0353] A step of determining the collision point between a ray and an obstacle based on the above static obstacle information, obstacle decision semantic information, and ray tracing algorithm - the collision point is located in the above drivable area - ;

[0354] Based on the collision point, the method includes the step of updating the drivable area.

[0355] The step of determining the collision point between the ray and the obstacle based on the above static obstacle information, obstacle decision semantic information, and ray tracing algorithm is as follows:

[0356] A step of determining the light source point and the ray projection direction based on the above obstacle decision semantic information;

[0357] A step of determining a ray projection range based on the above static obstacle information;

[0358] A step of performing a ray scan on an obstacle based on the light source point, the ray projection direction, and the ray projection range; and

[0359] It includes the step of determining the collision point between the beam and the obstacle.

[0360] As a decision-making method for obstacle avoidance, the above method is,

[0361] A step of obtaining road information, first grid obstacle information of a first grid obstacle and first convex hull obstacle information of a first convex hull obstacle;

[0362] A step of obtaining a second grid obstacle by preprocessing the first grid obstacle based on the above road information and the above first grid obstacle information;

[0363] A step of converting the above-mentioned second grid obstacle into a second convex hull obstacle; and

[0364] The method includes the step of making an avoidance decision regarding the target convex hull obstacle based on the target convex hull obstacle information of the target convex hull obstacle, wherein the target convex hull obstacle includes the first convex hull obstacle and / or the second convex hull obstacle.

[0365] As a method for generating a drivable area of ​​a vehicle, the method comprises:

[0366] Step of acquiring environmental sensing information - the environmental sensing information includes at least two of lane information, obstacle information, and ego vehicle information, and the obstacle information includes static obstacle information and / or dynamic obstacle information - ;

[0367] A step of determining lane decision semantic information for each lane based on the above environment sensing information - the lane decision semantic information includes passage time costs and safety costs -; and

[0368] It includes the step of generating a drivable area based on the lane decision semantic information above.

[0369] The foregoing description is merely a specific embodiment of the invention for those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Accordingly, the invention is not limited by the foregoing embodiments and should be applied to the broadest scope consistent with the principles and novelty of the invention. Industrial applicability

[0370] The present invention converts grid-type obstacles into convex-hull type obstacles to realize integrated decision-making for both grid-type and convex-hull type obstacles, thereby simplifying the obstacle decision-making flow for mixed-type obstacles and accelerating the obstacle decision-making process, enabling the decision planning module to conveniently and quickly perform obstacle decisions, and has strong potential for industrial application.

Claims

Claim 1 A vehicle decision planning method comprises: a step of generating a basic coordinate system; a step of generating a guideline in the basic coordinate system to determine the future driving trajectory of a vehicle; a step of performing an obstacle decision under the constraints of the guideline; and a step of generating a drivable area according to the obstacle decision; wherein the step of generating a drivable area according to the obstacle decision comprises: a step of acquiring environmental sensing information - the environmental sensing information includes at least two of lane information, obstacle information, and ego vehicle information, and the obstacle information includes static obstacle information or dynamic obstacle information -; a step of determining lane decision semantic information for each lane based on the environmental sensing information - the lane decision semantic information includes passing time cost and safety cost -; and a step of generating a drivable area based on the lane decision semantic information. Claim 2 A vehicle decision planning method according to claim 1, wherein the step of performing obstacle decision-making comprises: a step of obtaining road information, first grid obstacle information of a first grid obstacle, and first convex hull obstacle information of a first convex hull obstacle; a step of preprocessing the first grid obstacle based on the road information and the first grid obstacle information to obtain a second grid obstacle; a step of converting the second grid obstacle into a second convex hull obstacle; and a step of making an avoidance decision regarding the target convex hull obstacle based on the target convex hull obstacle information of the target convex hull obstacle - wherein the target convex hull obstacle includes the first convex hull obstacle and the second convex hull obstacle - wherein the convex hull obstacle is an obstacle containing semantic information, and the grid obstacle is an obstacle not containing semantic information. Claim 3 A vehicle decision planning method according to claim 2, wherein the step of obtaining a second grid obstacle by preprocessing the first grid obstacle based on the road information and the first grid obstacle information comprises: filtering the first grid obstacle located outside the road based on the road information and the first grid obstacle information; using the remaining first grid obstacle as the second grid obstacle; or determining the first grid obstacle located inside the road based on the road information and the first grid obstacle information; performing aggregation processing on the first grid obstacle located inside the road; and using the aggregated first grid obstacle as the second grid obstacle. Claim 4 A vehicle decision planning method according to claim 3, wherein the step of filtering the first grid obstacle located outside the road based on the road information and the first grid obstacle information comprises: a step of generating a road bounding box along the direction of travel of the road based on the road information; a step of generating a grid obstacle bounding box of the first grid obstacle based on the first grid obstacle information; a step of determining the first grid obstacle located outside the road based on the grid obstacle bounding box and the road bounding box; and a step of filtering the first grid obstacle located outside the road. Claim 5 delete Claim 6 A vehicle decision planning method according to claim 1, wherein the step of determining lane decision semantic information for each lane based on the environment sensing information comprises: a step of determining, for each lane, the collision time between the self vehicle and the first obstacle in front of the self vehicle based on the environment sensing information; and a step of determining the collision time as the passing time cost. Claim 7 A vehicle decision planning method according to claim 1, wherein the step of determining lane decision semantic information for each lane based on the environment sensing information comprises: a step of determining the lane of the self-vehicle and other lanes based on the lane information and the self-vehicle information; a step of determining a first preset safety cost as the safety cost for the lane of the self-vehicle; and a step of determining a second preset safety cost as the safety cost for the other lane if, based on the environment sensing information, it is determined that an obstacle enters the danger zone of the self-vehicle within a preset time, and if, based on the environment sensing information, it is determined that the obstacle does not enter the danger zone of the self-vehicle within the preset time, the first preset safety cost as the safety cost - wherein the second preset safety cost is different from the first preset safety cost - Claim 8 delete Claim 9 delete Claim 10 A method for generating a drivable area of ​​a vehicle, comprising: a step of acquiring environmental sensing information, wherein the environmental sensing information includes at least two of lane information, obstacle information, and self-vehicle information, and wherein the obstacle information includes static obstacle information or dynamic obstacle information; a step of determining lane decision semantic information for each lane based on the environmental sensing information, wherein the lane decision semantic information includes a passing time cost and a safety cost; and a step of generating a drivable area based on the lane decision semantic information; characterized by comprising: a step of acquiring environmental sensing information, wherein the environmental sensing information includes at least two of lane information, obstacle information, and self-vehicle information, and wherein the obstacle information includes static obstacle information or dynamic obstacle information. Claim 11 A method for generating a drivable area of ​​a vehicle, characterized in that, in the case where the lane decision semantic information includes a passage time cost, the step of determining lane decision semantic information for each lane based on the environment sensing information comprises: for each lane, the step of determining the collision time between the self vehicle and the first obstacle in front of the self vehicle based on the environment sensing information; and the step of determining the collision time as the passage time cost. Claim 12 A method for generating a drivable area of ​​a vehicle, characterized in that, in the case where the lane decision semantic information includes safety costs, the step of determining lane decision semantic information for each lane based on the environment sensing information comprises: a step of determining the lane of the self-vehicle and other lanes based on the lane information and the self-vehicle information; a step of determining a first preset safety cost as the safety cost for the lane of the self-vehicle; and for the other lane, a step of determining a second preset safety cost as the safety cost if it is determined based on the environment sensing information that an obstacle enters the danger area of ​​the self-vehicle within a preset time, and determining the first preset safety cost as the safety cost if it is determined based on the environment sensing information that an obstacle does not enter the danger area of ​​the self-vehicle within the preset time - the second preset safety cost is different from the first preset safety cost - Claim 13 A vehicle decision planning device comprising: a basic coordinate system generator for generating a basic coordinate system; a guideline generator for generating a guideline in the basic coordinate system to determine the approximate future driving trajectory of a vehicle; an obstacle decision maker for performing obstacle decision-making under the constraints of the guideline; and a driving space generator for generating a driving area according to the obstacle decision-making, wherein the driving space generator comprises: a sensing information acquisition module for acquiring environmental sensing information - the environmental sensing information includes at least two of lane information, obstacle information, and ego vehicle information, and the obstacle information includes static obstacle information or dynamic obstacle information -; a lane decision semantic information determination module for determining lane decision semantic information for each lane based on the environmental sensing information - the lane decision semantic information includes a passing time cost and a safety cost -; and a driving area generation module for generating a driving area based on the lane decision semantic information. Claim 14 In claim 13, the obstacle decision-maker comprises: an information acquisition module for acquiring road information, first grid obstacle information of a first grid obstacle, and first convex hull obstacle information of a first convex hull obstacle; a preprocessing module for obtaining a second grid obstacle by preprocessing the first grid obstacle based on the road information and the first grid obstacle information, wherein the number of the second grid obstacles is smaller than the number of the first grid obstacles; a type conversion module for converting the second grid obstacle into a second convex hull obstacle; and an avoidance decision module for making an avoidance decision regarding the target convex hull obstacle based on the target convex hull obstacle information of the target convex hull obstacle, wherein the target convex hull obstacle includes the first convex hull obstacle and the second convex hull obstacle; and the vehicle decision planning device is characterized in that the convex hull obstacle is an obstacle containing semantic information, and the grid obstacle is an obstacle not containing semantic information. Claim 15 delete Claim 16 delete Claim 17 A vehicle driving range generation device comprising: a sensing information acquisition module for acquiring environmental sensing information, wherein the environmental sensing information includes at least two of lane information, obstacle information, and ego vehicle information, and wherein the obstacle information includes static obstacle information or dynamic obstacle information; a lane decision semantic information determination module for determining lane decision semantic information for each lane based on the environmental sensing information, wherein the lane decision semantic information includes passing time cost and safety cost; and a driving range generation module for generating a driving range based on the lane decision semantic information; characterized by comprising: a driving range generation device for a vehicle. Claim 18 An electronic device comprising a memory and one or more processors; wherein the memory is communicationally connected to the one or more processors, and an instruction that can be executed by the one or more processors is stored in the memory, and wherein the method of any one of claims 1 to 4, 6, 7, and 10 to 12 is performed by the electronic device by executing the instruction by the one or more processors. Claim 19 A computer-readable storage medium in which a computer-executable instruction is stored, wherein the method of any one of claims 1 to 4, 6, 7, and 10 to 12 is performed by executing the computer-executable instruction by a computing device. Claim 20 A computer program stored on a computer-readable storage medium, wherein the computer program is configured to implement any one of the methods of claims 1 to 4, 6, 7, and 10 to 12 when executed by a computing device.

Images