Intersection turning path generation method and device, electronic equipment, vehicle, storage medium and computer program product

Abstract

The invention relates to an intersection turning path generation method and device, electronic equipment, a storage medium and a computer program product. The intersection turning path generation method comprises the steps that after it is determined that the front of an automatic driving vehicle is an intersection turning scene according to navigation information, state information of the automatic driving vehicle is acquired; determining a target turning model according to the state information; according to the target turning model, extracting input information from the state information; determining parameters of a target turning model according to the input information; inputting the parameters of the target turning model into the target turning model, and obtaining at least one control point through the target turning model; and generating a turning path according to at least one control point. According to the technical scheme, the accurate left-right turning path can be generated without using a high-precision map.

Description

Technical Field

[0001] The present invention relates to the field of vehicle automatic driving technology, and in particular to a method, device, electronic device, storage medium and computer program product for generating a turning path at an intersection. Background Art

[0002] The gradual replacement of human drivers by autonomous driving is an inevitable trend. However, mainstream autonomous driving technologies based on high-precision map solutions are increasingly constrained by various practical factors. For example, lane markings in some areas adjust and change with actual traffic volume, HD map updates lag significantly, and even many areas lack HD map information. This severely limits the application and development of autonomous driving technology.

[0003] Therefore, autonomous driving technology based solely on vision and navigation maps has been explored. This technology is still in its early stages of development, particularly in handling intersections, which poses a major challenge. Intersections lack lane markings, unlike non-intersections. For autonomous driving solutions with high-precision maps, the map provider manually draws virtual lane centerlines for intersection turns based on the two lanes to be connected. These algorithms rely heavily on the virtual lane centerlines provided by the HD map. Turning paths are generated by smoothing the virtual lane centerlines provided by the map provider to create a reference line, which is then used to optimize the driving trajectory. This process lacks overall flexibility, rendering the vehicle a rigid rail car with poor ride comfort.

[0004] Pure visual perception solutions only provide the lane lines and curbs of the current lane, as well as some navigation map information. The visual perception output is limited and unstable, requiring downstream planning to create a safe and stable turning path based on the limited information available and generate a smooth reference line. Summary of the Invention

[0005] The purpose of this application is to provide a method, device, electronic device, vehicle, storage medium and computer program product for generating a turning path at an intersection, which can generate accurate left and right turning paths even when there is no high-precision map.

[0006] According to a first aspect of an embodiment of the present application, a method for generating a turning path at an intersection is provided, comprising:

[0007] After determining, based on the navigation information, that there is a turning scene at an intersection ahead of the autonomous driving vehicle, obtaining state information of the autonomous driving vehicle;

[0008] determining a target turning model according to the state information;

[0009] extracting input information from the state information according to the target turn model;

[0010] determining parameters of the target turning model according to the input information;

[0011] Inputting parameters of the target turning model into the target turning model, and obtaining at least one control point through the target turning model;

[0012] A turning path is generated according to the at least one control point.

[0013] In one embodiment, the status information includes: vehicle information, navigation map information, and visual perception module information;

[0014] The vehicle information includes: vehicle model parameters and vehicle positioning information;

[0015] The navigation map information includes next intersection turn information, lane number information and traffic signal information;

[0016] The visual perception module information includes lane line information, road curb information and road obstacle information.

[0017] In one embodiment, determining a target turning model according to the state information includes:

[0018] Identify a target turning scene according to the state information; the target turning scene is a left turn scene in a waiting area, a left turn scene without a waiting area, a right turn scene with a waiting area, a right turn scene with a small curvature, or a right turn scene with a large curvature;

[0019] The target turning model is determined according to the target turning scenario and the corresponding relationship between the turning scenario and the model.

[0020] In one embodiment, generating a turning path according to the at least one control point includes:

[0021] The at least one control point is smoothly connected using a specified curve to generate the turning path.

[0022] In one embodiment, the specified curve is a quintic spline curve.

[0023] In one embodiment, the next intersection turn information includes: a first end point of the turn path and a starting point of the turn path; the lane line information includes a lane line of the exit road;

[0024] The obtaining of the state information of the autonomous driving vehicle includes:

[0025] determining a turn-in vector according to the navigation information, wherein a starting point of the turn-in vector is used to indicate a first end point of the turning path, and a direction of the turn-in vector is used to indicate a driving direction after the turn;

[0026] The lane line of the exit road is obtained from the visual perception module of the autonomous driving vehicle, and an exit vector is determined based on the lane line, the end point of the exit vector is used to indicate the starting point of the turning path, and the direction of the exit vector is used to indicate the driving direction before the turn; the input information includes the starting point of the turning path and the first end point of the turning path; the parameters of the target turning model include the starting point of the turning path and the first end point of the turning path.

[0027] In one embodiment, after extracting input information from the state information according to the target turning model and before determining parameters of the target turning model according to the input information, the method further includes:

[0028] A stable update process is performed on the input information.

[0029] In one embodiment, the roadside information includes a valid sensed roadside; the input information also includes the valid sensed roadside;

[0030] The performing stable update processing on the input information includes:

[0031] Extracting the effective sensed road edge from the input information;

[0032] Performing a time-series linear smoothing update on the endpoint of the effective sensing edge;

[0033] The second end point of the turning path is determined based on the end point of the effective perceived road edge after smooth update; wherein, the input information after stable update processing includes the starting point of the turning path and the second end point of the turning path; the parameters of the target turning model include the starting point of the turning path and the second end point of the turning path.

[0034] According to a second aspect of an embodiment of the present application, a device for generating a turning path at an intersection is provided, comprising:

[0035] A first acquisition module is configured to acquire state information of the autonomous driving vehicle after determining, based on navigation information, that a road ahead of the autonomous driving vehicle is a turning scene at an intersection;

[0036] a first determining module, configured to determine a target turning model according to the state information;

[0037] an extraction module, configured to extract input information from the state information according to the target turning model;

[0038] a second determining module, configured to determine parameters of the target turning model according to the input information;

[0039] a second acquisition module, configured to input parameters of the target turning model into the target turning model, and acquire at least one control point through the target turning model;

[0040] A generating module is used to generate a turning path according to the at least one control point.

[0041] According to a third aspect of an embodiment of the present application, an electronic device is provided, comprising a memory and a processor, wherein the memory is used to store a computer program executable by the processor; and the processor is used to execute the computer program in the memory to implement the above method.

[0042] According to a fourth aspect of an embodiment of the present application, a vehicle is provided, comprising the above-mentioned electronic device.

[0043] According to a fifth aspect of an embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored, characterized in that when the executable computer program in the storage medium is executed by a processor, the above method can be implemented.

[0044] According to a sixth aspect of an embodiment of the present application, a computer program product is provided, comprising a computer program, which can implement the above method when executed by a processor.

[0045] Compared with the prior art, the present invention has the following advantages:

[0046] After determining that the autonomous vehicle is at an intersection and turning ahead based on navigation information, the autonomous vehicle's state information is obtained, and then a target turning model is determined based on the state information. Then, input information is extracted from the state information based on the target turning model. Then, parameters of the target turning model are determined based on the input information, the parameters of the target turning model are input into the target turning model, and at least one control point is obtained from the target turning model. Then, a turning path is generated based on the at least one control point. The technical solution provided by this application can generate accurate left and right turning paths without using high-precision maps. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 The figure is a flowchart of a method for generating a turning path at an intersection according to an exemplary embodiment.

[0048] Figure 2 The figure is a flowchart of a method for generating a turning path at an intersection according to another exemplary embodiment.

[0049] Figure 3 It is a schematic diagram of the turning path generated at the intersection when turning right.

[0050] Figure 4It is a schematic diagram of the turning path generated at the intersection when turning left.

[0051] Figure 5 The figure is a flowchart of a method for generating a turning path at an intersection according to another exemplary embodiment.

[0052] Figure 6 The figure is a flowchart of a method for generating a turning path at an intersection according to another exemplary embodiment.

[0053] Figure 7 The figure is a flowchart of a method for generating a turning path at an intersection according to another exemplary embodiment.

[0054] Figure 8 The figure is a block diagram of a device for generating a turning path at an intersection according to another exemplary embodiment.

[0055] Figure 9 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0056] Unless otherwise defined, the technical or scientific terms used in this specification and claims shall have the ordinary meanings understood by persons having ordinary skills in the technical field to which the invention belongs. Specific embodiments of the present invention will be described below in conjunction with the accompanying drawings. It should be noted that in the specific description of these embodiments, for the sake of clarity and clarity, this specification cannot provide a detailed description of all features of the actual embodiments. Without departing from the spirit and scope of the present invention, those skilled in the art may modify and replace the embodiments of the present invention, and the resulting embodiments are also within the scope of protection of the present invention.

[0057] Among related technologies, autonomous driving based on pure vision and navigation maps presents a major challenge in handling intersection scenarios. Intersections lack lane markings, unlike non-intersections. For autonomous driving solutions with high-precision maps, the map provider manually draws virtual lane centerlines for intersection turns based on the two lanes to be connected. These algorithms rely heavily on the virtual lane centerlines provided by the HD map. Turning paths are generated by smoothing the virtual lane centerlines provided by the map provider to create a reference line, which is then used to optimize the driving trajectory. This process lacks overall flexibility, rendering the vehicle a rigid rail car with extremely poor ride comfort.

[0058] Pure visual perception solutions only provide the lane lines and curbs of the current lane, as well as some navigation map information. The visual perception output is limited and unstable, requiring downstream planning to create a safe and stable turning path based on the limited information available and generate a smooth reference line.

[0059] In order to solve the above technical problems, the present application proposes a method, device, electronic device, storage medium and computer program product for generating intersection turning paths, which can generate accurate left and right turning paths without using high-precision maps.

[0060] Figure 1 This is a flow chart showing a method for generating a turning path at an intersection according to an exemplary embodiment. This method for generating a turning path at an intersection can be applied to an electronic device with computing capabilities, which can be installed on an autonomous driving vehicle. Figure 1 The method for generating a turning path at an intersection may include the following steps:

[0061] Step 101: After determining that there is a turning intersection ahead of the autonomous driving vehicle based on navigation information, obtain the status information of the autonomous driving vehicle.

[0062] In one embodiment, the electronic device can obtain navigation information of the vehicle. After determining that the autonomous vehicle is at a turning intersection, the electronic device obtains status information of the autonomous vehicle from upstream.

[0063] In one embodiment, the aforementioned status information may include vehicle information, navigation map information, and visual perception module information. The vehicle information includes vehicle model parameters and vehicle location information. The navigation map information includes next intersection turn information, lane number information, and traffic signal information. The visual perception module information includes lane marking information, road curb information, and road obstacle information. The electronic device may obtain vehicle information from a device storing vehicle information, navigation map information from a navigation device, and visual perception module information from a visual perception module.

[0064] In one embodiment, the next intersection turning information includes: a first end point of the turning path and a starting point of the turning path; and the lane line information includes a lane line for turning out of the road.

[0065] In one embodiment, Figure 2 As shown, step 101 may include the following steps:

[0066] Step 201 : determining a turn-in vector based on navigation information. The starting point of the turn-in vector is used to indicate the first end point of the turning path, and the direction of the turn-in vector is used to indicate the driving direction after the turn.

[0067] In one embodiment, the electronic device may obtain information about whether the next intersection will turn left or right from the navigation map, and process the information into a vector: Vector-A.

[0068] In an exemplary embodiment, Figure 3As shown, when the next intersection is about to turn right, the electronic device can determine the turning vector Vector-A based on the navigation information. The starting point of the turning vector Vector-A is used to represent the first end point ctp-A' of the turning path, and the direction of the turning vector Vector-A is used to represent the driving direction after the turn.

[0069] In an exemplary embodiment, Figure 4 As shown, when the next intersection is about to turn left, the electronic device can determine the turning vector Vector-A based on the navigation information. The starting point of the turning vector Vector-A is used to represent the first end point ctp-A' of the turning path, and the direction of the turning vector Vector-A is used to represent the driving direction after the turn.

[0070] The starting point of the turn-in vector Vector-A can generally accurately determine the target road (Road) to be turned into, but not the specific lane (Lane). The specific turn-in location needs to be determined by the roadside obtained in a subsequent step. The direction of the turn-in vector Vector-A can generally accurately determine the direction of the turn-in lane (InLane) with an error of less than 30 degrees. The electronic device can also obtain the lane number of the turn-in lane from the navigation map.

[0071] In step 202, the lane line of the exit road is obtained from the visual perception module of the autonomous driving vehicle, and an exit vector is determined based on the lane line. The end point of the exit vector is used to indicate the starting point of the turning path, and the direction of the exit vector is used to indicate the driving direction before the turn.

[0072] like Figure 3 As shown, when the next intersection is about to turn right, the electronic device obtains the first lane line 31 of the first exit road from the visual perception module. This first lane line 31 can also be called the perception lane line. Then, the first lane line 31 of the first exit lane is offset to the left by half a lane width (1.875m) to generate the first center line 32 of the first exit lane. Then, the exit vector Vector-D is determined based on the first center line 32. The end point of the exit vector Vector-D is used to represent the starting point cpt-D of the turning path, and the direction of the exit vector Vector-D is used to represent the driving direction before the turn.

[0073] like Figure 4 As shown, when the next left turn is about to occur at the intersection, the second lane line 41 of the second exit lane is offset to the right by half a lane width (1.875m) to generate the second center line 42 of the second exit lane. Then, an exit vector Vector-D is determined based on the second center line 42. The end point of the exit vector Vector-D is used to represent the starting point cpt-D of the turning path, and the direction of the exit vector Vector-D is used to represent the driving direction before the turn.

[0074] Step 102: Determine a target turning model based on the state information.

[0075] In one embodiment, the electronic device may determine the target turning model through model recognition based on the state information, and is not limited to obtaining the state information of the vehicle and then making logical judgments based on data and rules.

[0076] In another embodiment, Figure 5 As shown, step 102 may include the following steps:

[0077] Step 501: Identify a target turning scenario based on the state information, wherein the target turning scenario is a left turn scenario in a waiting area, a left turn scenario without a waiting area, a right turn scenario with a waiting area, a right turn scenario with a small curvature, or a right turn scenario with a large curvature.

[0078] Step 502: Determine a target turning model according to the target turning scenario and the corresponding relationship between the turning scenario and the model.

[0079] In one embodiment, the electronic device may identify the target turning scene through model recognition based on the state information, and then determine the target turning model based on the target turning scene and the corresponding relationship between the turning scene and the model.

[0080] Step 103: extract input information from the state information according to the target turning model.

[0081] In one embodiment, the information required to be input for each turning model may be different. The electronic device can extract the required input information from the state information according to the target turning model.

[0082] In one embodiment, the input information may include a starting point of the turning path and a first end point of the turning path.

[0083] Step 104: Determine parameters of the target turning model based on the input information.

[0084] In one embodiment, the electronic device may extract parameters of the target turning model from the input information.

[0085] In one embodiment, the parameters of the target turning model may include a starting point of the turning path and a first end point ctp-A′ of the turning path.

[0086] Step 105: Input the parameters of the target turning model into the target turning model, and obtain at least one control point through the target turning model.

[0087] In one embodiment, the obtained parameters of the target turning model can be input into the target turning model. Specifically, the positions of the starting point cpt-D of the turning path and the first end point ctp-A' of the turning path can be input into the target turning model. In other words, the target turning model can be constructed based on the parameters of the target turning model.

[0088] In one embodiment, Figure 3 or Figure 4 As shown, the perpendicular line L1 passing through the starting point ctp-D of the turning path and perpendicular to the turning vector Vector-D can be calculated, and then the foot of the perpendicular ctp-O of the first end point ctp-A' on the perpendicular line can be calculated as the initial center of the arc segment r. The distances from the first end point ctp-A' and the starting point cpt-D of the turning path to ctp-O are calculated to obtain two distance values, and then half of the minimum of the two distance values ​​is taken as the arc radius. Then, the distance from the first end point ctp-A' to the foot of the perpendicular ctp-O minus the arc radius is used as the length of the cut-out straight line segment L2, and the distance from the starting point ctp-D of the turning path to the foot of the perpendicular ctp-O minus the arc radius is used as the length of the cut-in value segment L3. The length ratio of the cut-out straight line segment L2 Figure 3 or Figure 4 The blue line segment shown is short, which is the part of the blue line segment close to the arc segment r.

[0089] For a right turn with a small curvature, the foot of the perpendicular ctp-O can be used as the center of the arc, and the right turn path is composed of only the arc segment r.

[0090] In one embodiment, the electronic device may obtain at least one control point through a target turning model.

[0091] In one embodiment, Figure 3 or Figure 4 As shown, seven points in total can be taken, including the three quarter points of the arc segment r, the two endpoints of the straight line segment L2, and the two endpoints of the entry line segment L3 (for a right turn with small curvature, five points in total are used, including the three quarter points and the two endpoints of the arc segment r). Some of these points are selected as control points to generate a turning path.

[0092] If the visual perception module identifies the lane markings of the incoming lane during a turn, it selects a relatively normal lane marking to generate a vector and updates the final direction of the turning path. Curb avoidance is performed based on the lane marking attribute. Finally, the turning path is extended along the incoming vector (vector-A) to smoothly enter the target lane.

[0093] Among them, the method for selecting more normal lane lines is: the number of lane lines for turning into the lane needs to be greater than 1, for each lane line of the turning-in lane, the vector of the lane line's end point minus the beginning point is used as the vector direction of the lane line, then, the average vector direction of the vector directions of all lane lines is calculated, and then the lane line closest to the average vector direction is selected as the above-mentioned more normal lane line.

[0094] Step 106: Generate a turning path according to at least one control point.

[0095] In one embodiment, a designated curve is used to smoothly connect the at least one obtained control point to generate a turning path. The designated curve may be a quintic spline curve, but is not limited thereto.

[0096] In one embodiment, after determining that there is a turning scene at an intersection ahead of the autonomous driving vehicle based on navigation information, the status information of the autonomous driving vehicle can be obtained in real time to achieve real-time updating and optimization of the turning path.

[0097] In this embodiment, after determining that the autonomous vehicle is at an intersection and turning ahead based on navigation information, the state information of the autonomous vehicle is obtained. Then, a target turning model is determined based on the state information. Then, input information is extracted from the state information based on the target turning model. Then, parameters of the target turning model are determined based on the input information. The parameters of the target turning model are input into the target turning model, and at least one control point is obtained from the target turning model. Then, a turning path is generated based on the at least one control point. The technical solution provided by this application can generate accurate left and right turning paths without using high-precision maps.

[0098] Figure 6 This is a flow chart of a method for generating a turning path at an intersection according to another exemplary embodiment. In this embodiment, the is processed to generate accurate and stable left and right turning paths. Figure 6 The method for generating a turning path at an intersection may include the following steps:

[0099] Step 601: After determining that there is a turning intersection ahead of the autonomous driving vehicle based on navigation information, obtain the status information of the autonomous driving vehicle.

[0100] This step is similar to the above-mentioned step 101 and will not be described again here.

[0101] Step 602: Determine a target turning model based on the state information.

[0102] This step is similar to the above-mentioned step 102 and will not be described again here.

[0103] Step 603: extract input information from the state information according to the target turning model.

[0104] This step is similar to the above-mentioned step 103 and will not be described again here.

[0105] Step 604: Perform stable update processing on the input information.

[0106] In one embodiment, the roadside information of the road may include valid perceived roadside. The input information may also include valid perceived roadside.

[0107] In one embodiment, Figure 7 As shown, step 604 may include the following steps:

[0108] Step 701: extract valid sensed edges from input information.

[0109] like Figure 3 As shown, when the next intersection is about to turn right, the first sensing edge 33 located to the right of the exit vector Vector-D and to the right of the entry vector Vector-A is selected as the right reference edge. The first sensing edge 33 is a valid sensing edge extracted from the input information.

[0110] like Figure 4 As shown, when the next intersection is about to turn left, the median strip curb 43 is first searched. The specific method is: select the curb located to the left of the exit vector Vector-D and to the left of the entry vector Vector-A, and the approximately U-shaped curb closest to the entry vector Vector-A as the median strip curb 43 for the left turn. If the median strip curb 43 is not found, the right side curb 44 of the target lane to be turned into, located to the left of the exit vector Vector-D and to the right of the entry vector Vector-A, is searched. The median strip curb 43 or the right side curb 44 is the valid perception curb extracted from the input information. When turning left, the second perception curb 45 and the double yellow line 46 can be ignored.

[0111] In summary, electronic devices can extract effective perception curbs by intelligently filtering input information.

[0112] In addition, electronic devices can extract valid lane lines by intelligently filtering input information.

[0113] Step 702: Perform temporal linear smoothing update on the endpoints of the valid sensing edges.

[0114] In this step, a time-series linear smoothing update is performed on the endpoints of the valid sensing edge to prevent path mutations caused by abnormal sensing recognition, thereby generating an accurate and stable left and right turning path.

[0115] Step 703: Determine the second end point of the turning path based on the end point of the valid sensed curb after the smooth update. The input information after the stable update includes the start point and the second end point of the turning path; and the parameters of the target turning model include the start point and the second end point of the turning path.

[0116] like Figure 3 As shown, when a right turn is about to be made at the next intersection, the entry vector Vector-A is rotated 90 degrees counterclockwise to obtain a temporary vector direction. Then, the end point (right end point) of the first sensing curb 33 is translated along the temporary vector direction by a distance less than one lane width (3 meters) to obtain the right turn entry point ctp-A. The entry point ctp-A is the second end point of the turning path.

[0117] like Figure 4 As shown, when the next intersection is about to turn left, the end point of the left-turn median strip curb 43 is translated 90 degrees clockwise along the turn-in vector Vector-A by a distance less than one lane width (3 meters) to obtain the left-turn entry point ctp-A. Alternatively, the end point of the right curb 44 of the left-turn target lane is translated 90 degrees counterclockwise along the turn-in vector Vector-A by a distance equal to the number of target road lanes minus 1 multiplied by the lane width (e.g., 3.7 meters) to obtain the left-turn entry point ctp-A. The entry point ctp-A is the second end point of the turning path.

[0118] It should be noted that Figure 3 and Figure 4 The point ctp-A' in the figure may not actually coincide with the point ctp-A.

[0119] At this point, the starting and ending points of the turning path have been determined. The next step is to use the intersection path starting and ending point parameters to build an intersection turning model, and use the model to calculate the optimal smooth point set consisting of straight segments and arc segments.

[0120] Step 605 : Determine the parameters of the target turning model based on the input information after the stable update process.

[0121] This step is similar to the above-mentioned step 104 and will not be described again here.

[0122] Step 606: Input the parameters of the target turning model into the target turning model, and obtain at least one control point through the target turning model.

[0123] This step is similar to the above-mentioned step 105 and will not be described again here.

[0124] Step 607: Generate a turning path according to at least one control point.

[0125] This step is similar to the above-mentioned step 106 and will not be described again here.

[0126] The technical solution provided by this application can also generate accurate and stable left and right turning paths when there is no high-precision map and when the lane line of the target lane cannot be recognized, is recognized late, or is recognized unstably. At the same time, it can make the autonomous driving vehicle more intelligent and greatly reduce the optimization task and difficulty of subsequent trajectories.

[0127] This application is based on pure visual perception, identifies and extracts intersection scene information, processes valid data, and generates intelligent turning paths that adapt to different special scenarios.

[0128] The technical solution provided by this application is a new intersection turning mode based on pure vision and navigation map information. Different optimal turning paths are provided based on different scenarios. For example, in a left-turn scenario with a waiting area: when turning left in the second left lane, the vehicle should go through the waiting area, that is, the turning path should be cut along the waiting area; when turning left in the first left lane, if the left turn light is red, the turning path is cut normally, but if the light is green, the turning path should not be cut out in the middle of the waiting area, choosing a shorter and more convenient path that is closer to human driving habits.

[0129] The technical solutions provided in this application can effectively expand the scope and application scenarios of autonomous driving technology, effectively improving the intelligent turning of autonomous vehicles at intersections and passenger comfort. In left-turn scenarios in the waiting area, the generated turning trajectory is more closely aligned with human driving habits. In right-turn scenarios, different right-turn scenarios are judged based on information fed back from the curb, generating a right-turn path that is more in line with human driving habits.

[0130] In addition, pure vision also has the problem of insufficient and unstable visual perception recognition input data. The embodiment of the present application provides a method for generating a more intelligent and comfortable turning path under limited input, which is used to achieve autonomous driving.

[0131] Figure 8 This is a device for generating a turning path at an intersection according to an exemplary embodiment. Figure 8 As shown, in this embodiment, the device for generating the intersection turning path includes:

[0132] A first acquisition module 81 is configured to acquire state information of the autonomous driving vehicle after determining, based on navigation information, that the scene ahead of the autonomous driving vehicle is a turning intersection;

[0133] A first determining module 82 is configured to determine a target turning model according to the state information;

[0134] an extraction module 83 for extracting input information from the state information according to the target turning model;

[0135] a second determining module 84, configured to determine parameters of the target turning model according to the input information;

[0136] A second acquisition module 85 is configured to input the parameters of the target turning model into the target turning model, and acquire at least one control point through the target turning model;

[0137] The generating module 86 is configured to generate a turning path according to the at least one control point.

[0138] An embodiment of the present application also proposes an electronic device, including a processor and a memory; the memory is used to store a computer program executable by the processor; the processor is used to execute the computer program in the memory to implement the method for generating an intersection turning path in any of the above embodiments.

[0139] The embodiments of the present application further provide a computer-readable storage medium. When the executable computer program in the storage medium is executed by a processor, the method for generating a turning path at an intersection of any of the above embodiments can be implemented.

[0140] An embodiment of the present application also provides a vehicle comprising the above-mentioned electronic device.

[0141] An embodiment of the present application further provides a computer program product, including a computer program, which implements the above-mentioned method for generating a turning path at an intersection when executed by a processor.

[0142] Regarding the apparatus in the above embodiment, the specific manner in which the processor performs operations has been described in detail in the embodiment of the method, and will not be elaborated on here.

[0143] Figure 9 9 is a block diagram of an electronic device according to an exemplary embodiment. For example, the electronic device 900 can be provided as a server or a computing device. Figure 9 Device 900 includes a processing component 922, which further includes one or more processors, and memory resources represented by memory 932 for storing instructions executable by processing component 922, such as applications. The applications stored in memory 932 may include one or more modules, each corresponding to a set of instructions. Furthermore, processing component 922 is configured to execute the instructions to perform the above-described method for generating a turn path at an intersection.

[0144] The device 900 may also include a power supply component 926 configured to perform power management of the device 900, a wired or wireless network interface 950 configured to connect the device 900 to a network, and an input / output (I / O) interface 958. The device 900 may operate based on an operating system stored in the memory 932, such as Windows Server™, MacOS X™, Unix™, Linux™, FreeBSD™, or the like.

[0145] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 932 including instructions, which can be executed by the processing component 922 of the device 900 to perform the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0146] In the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance. The term "plurality" refers to two or more, unless otherwise clearly defined.

[0147] The above description of the embodiments is intended to facilitate understanding and application of the present application by those skilled in the art. It will be apparent that those skilled in the art can readily make various modifications to these embodiments and apply the general principles described herein to other embodiments without expending any creative effort. Therefore, the present application is not limited to the embodiments described herein, and improvements and modifications made by those skilled in the art based on the disclosure of this application without departing from the scope and spirit of this application are within the scope of this application.

Claims

1. A method for generating a turning path at an intersection, characterized in that: include: After determining, based on the navigation information, that there is a turning scene at an intersection ahead of the autonomous driving vehicle, obtaining state information of the autonomous driving vehicle; determining a target turning model according to the state information; extracting input information from the state information according to the target turn model; determining parameters of the target turning model according to the input information; Inputting parameters of the target turning model into the target turning model, and obtaining at least one control point through the target turning model; A turning path is generated according to the at least one control point.

2. The method for generating a turning path at an intersection according to claim 1, wherein: The state information includes: vehicle information, navigation map information and visual perception module information; The vehicle information includes: vehicle model parameters and vehicle positioning information; The navigation map information includes next intersection turn information, lane number information and traffic signal information; The visual perception module information includes lane line information, road curb information and road obstacle information.

3. The method for generating a turning path at an intersection according to claim 1, wherein: The determining of the target turning model according to the state information includes: Identify a target turning scene according to the state information; the target turning scene is a left turn scene in a waiting area, a left turn scene without a waiting area, a right turn scene with a waiting area, a right turn scene with a small curvature, or a right turn scene with a large curvature; The target turning model is determined according to the target turning scenario and the corresponding relationship between the turning scenario and the model.

4. The method for generating a turning path at an intersection according to claim 1, wherein: Generating a turning path according to the at least one control point includes: The at least one control point is smoothly connected using a specified curve to generate the turning path.

5. The method for generating a turning path at an intersection according to claim 4, wherein: The specified curve is a quintic spline curve.

6. The method for generating a turning path at an intersection according to claim 2, wherein: The next intersection turning information includes: the first end point of the turning path and the starting point of the turning path; the lane line information includes the lane line of the exit road; The obtaining of the state information of the autonomous driving vehicle includes: determining a turn-in vector according to the navigation information, wherein a starting point of the turn-in vector is used to indicate a first end point of the turning path, and a direction of the turn-in vector is used to indicate a driving direction after the turn; The lane line of the exit road is obtained from the visual perception module of the autonomous driving vehicle, and an exit vector is determined based on the lane line, the end point of the exit vector is used to indicate the starting point of the turning path, and the direction of the exit vector is used to indicate the driving direction before the turn; the input information includes the starting point of the turning path and the first end point of the turning path; the parameters of the target turning model include the starting point of the turning path and the first end point of the turning path.

7. The method for generating a turning path at an intersection according to claim 6, wherein: After extracting input information from the state information according to the target turning model and before determining parameters of the target turning model according to the input information, the method further includes: A stable update process is performed on the input information.

8. The method for generating a turning path at an intersection according to claim 7, wherein: The roadside information includes a valid sensed roadside; the input information also includes the valid sensed roadside; The performing stable update processing on the input information includes: Extracting the effective sensed road edge from the input information; Performing a time-series linear smoothing update on the endpoint of the effective sensing edge; The second end point of the turning path is determined based on the end point of the effective perceived road edge after smooth update; wherein, the input information after stable update processing includes the starting point of the turning path and the second end point of the turning path; the parameters of the target turning model include the starting point of the turning path and the second end point of the turning path.

9. A device for generating a turning path at an intersection, characterized in that: include: A first acquisition module is configured to acquire state information of the autonomous driving vehicle after determining, based on navigation information, that a road ahead of the autonomous driving vehicle is a turning scene at an intersection; a first determining module, configured to determine a target turning model according to the state information; an extraction module, configured to extract input information from the state information according to the target turning model; a second determining module, configured to determine parameters of the target turning model according to the input information; a second acquisition module, configured to input parameters of the target turning model into the target turning model, and acquire at least one control point through the target turning model; A generating module is used to generate a turning path according to the at least one control point.

10. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory is used to store a computer program executable by the processor; and the processor is used to execute the computer program in the memory to implement the method according to any one of claims 1 to 8.

11. A vehicle, characterized in that: Comprising the electronic device as claimed in claim 10.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the executable computer program in the storage medium is executed by a processor, the method according to any one of claims 1 to 8 can be implemented.

13. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images