Motion intention estimation method, device, computer equipment, and storage medium

Abstract

The present application relates to a motion intention estimation method, device, computer equipment, storage medium and computer program product. The method includes: obtaining the road surface information and obstacle space information of the road on which the autonomous driving vehicle is traveling in the current frame, dividing the obstacles according to the road surface information and obstacle space information to obtain a first-level group; for each first-level group, dividing according to the degree of spatial influence between obstacles to obtain a second-level group; according to the preset motion intention category, determining the motion intention probability distribution corresponding to the second-level group; according to the motion intention probability distribution corresponding to the second-level group and the motion intention probability distribution corresponding to the second-level group in the historical frame, determining the target intention probability distribution of the second-level group; according to the target intention probability distribution, determining the target motion intention of the second-level group. This method can achieve accurate estimation of the motion intention of the second-level group, thereby guiding the autonomous driving vehicle to accurately make a motion decision on whether to follow or change lanes to bypass obstacles.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a motion intention estimation method, apparatus, computer equipment, storage medium and computer program product. Background Art

[0002] Self-driving cars, also known as driverless cars, rely on the collaborative efforts of artificial intelligence, visual computing, radar, monitoring devices and global positioning systems to allow computers to automatically and safely operate motor vehicles without any active human operation.

[0003] Making a reasonable assessment of the movement intention of road obstacles is an important prerequisite for autonomous vehicles to make safe and efficient decisions. In traditional technologies, the movement prediction of obstacles in the field of autonomous driving mostly predicts the movement state in the next few seconds based on the movement data of the obstacles in the past few seconds and the relevant information of the lanes. However, the same movement state may correspond to different movement intentions in different scenarios. Therefore, the movement intention of the obstacle in the corresponding movement state is not clear. Summary of the invention

[0004] Based on this, it is necessary to provide a motion intention estimation method, device, computer equipment, computer readable storage medium and computer program product that can accurately determine the motion intention of road obstacles in order to solve the above technical problems.

[0005] In a first aspect, the present application provides a method for estimating motion intention. The method comprises:

[0006] Obtaining road surface information and obstacle space information of the road where the autonomous driving vehicle is traveling in the current frame;

[0007] Dividing obstacles according to the road surface information and the obstacle space information to obtain a primary grouping;

[0008] For each of the first-level groups, divide them according to the spatial influence degree between the obstacles to obtain second-level groups;

[0009] Determining the probability distribution of movement intention corresponding to the secondary grouping according to the preset movement intention category;

[0010] Determine the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame;

[0011] The target motion intention of the secondary grouping is determined according to the target intention probability distribution.

[0012] In one embodiment, for each of the first-level groups, the second-level groups are divided according to the spatial influence degree between the obstacles, including:

[0013] For each of the first-level groups, one obstacle is selected from the obstacles in turn as a first obstacle, and a spatial influence degree between the first obstacle and a second obstacle is calculated; the second obstacle is an obstacle other than the first obstacle among all obstacles in the first-level group;

[0014] Selecting a second obstacle that has the greatest spatial influence on the first obstacle as a target obstacle;

[0015] The target obstacle and the first obstacle are classified into the same secondary group.

[0016] In one embodiment, selecting one obstacle from the obstacles in sequence as a first obstacle and calculating the degree of spatial influence between the first obstacle and the second obstacle includes:

[0017] A Fresnel coordinate system is established with the autonomous driving vehicle as the origin and the road driving direction as the positive direction s;

[0018] The first obstacles are selected in order from small to large according to the s coordinates of the obstacles, and the spatial influence degree between the first obstacle and the second obstacle is calculated.

[0019] In one embodiment, the calculating the spatial influence degree between the first obstacle and the second obstacle includes:

[0020] Calculating an overlapping area between the first obstacle and the second obstacle;

[0021] The selecting the second obstacle corresponding to the maximum spatial influence between the first obstacle and the second obstacle as the target obstacle includes:

[0022] A second obstacle having the largest overlapping area with the first obstacle is selected as the target obstacle.

[0023] In one embodiment, determining the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame includes:

[0024] The motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame are fused to obtain the target intention probability distribution of the secondary group.

[0025] In one embodiment, the fusion of the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame to obtain the target intention probability distribution of the secondary group includes:

[0026] The target intention probability distribution of the secondary group is obtained by fusing the motion intention probability distribution corresponding to the secondary group, the motion intention probability distribution corresponding to the secondary group in the historical frame, the traffic attributes corresponding to the secondary group, the road information and the surrounding traffic attributes.

[0027] In a second aspect, the present application also provides a motion intention estimation device. The device comprises:

[0028] An information acquisition module, used to obtain road surface information and obstacle space information of the road on which the autonomous driving vehicle is traveling in the current frame;

[0029] A first division module, used for dividing obstacles according to the road surface information and the obstacle space information to obtain a primary group;

[0030] A second division module is used to divide each of the first-level groups according to the spatial influence degree between the obstacles to obtain a second-level group;

[0031] A first probability determination module, used to determine the probability distribution of motion intention corresponding to the secondary grouping according to a preset motion intention category;

[0032] A second probability determination module is used to determine the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame;

[0033] The motion intention determination module is used to determine the target motion intention of the secondary grouping according to the target intention probability distribution.

[0034] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0035] Obtaining road surface information and obstacle space information of the road where the autonomous driving vehicle is traveling in the current frame;

[0036] Dividing obstacles according to the road surface information and the obstacle space information to obtain a primary grouping;

[0037] For each of the first-level groups, divide them according to the spatial influence degree between the obstacles to obtain second-level groups;

[0038] Determining the probability distribution of movement intention corresponding to the secondary grouping according to the preset movement intention category;

[0039] Determine the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame;

[0040] The target motion intention of the secondary grouping is determined according to the target intention probability distribution.

[0041] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0042] Obtaining road surface information and obstacle space information of the road where the autonomous driving vehicle is traveling in the current frame;

[0043] Dividing obstacles according to the road surface information and the obstacle space information to obtain a primary grouping;

[0044] For each of the first-level groups, divide them according to the spatial influence degree between the obstacles to obtain second-level groups;

[0045] Determining the probability distribution of movement intention corresponding to the secondary grouping according to the preset movement intention category;

[0046] Determine the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame;

[0047] The target motion intention of the secondary grouping is determined according to the target intention probability distribution.

[0048] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0049] Obtaining road surface information and obstacle space information of the road where the autonomous driving vehicle is traveling in the current frame;

[0050] Dividing obstacles according to the road surface information and the obstacle space information to obtain a primary grouping;

[0051] For each of the first-level groups, divide them according to the spatial influence degree between the obstacles to obtain second-level groups;

[0052] Determining the probability distribution of movement intention corresponding to the secondary grouping according to the preset movement intention category;

[0053] Determine the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame;

[0054] The target motion intention of the secondary grouping is determined according to the target intention probability distribution.

[0055] The above-mentioned motion intention estimation method, device, computer equipment, storage medium and computer program product obtain the road surface information and obstacle space information of the road on which the autonomous driving vehicle is traveling in the current frame, divide the obstacles according to the road surface information and the obstacle space information to obtain a first-level group; for each of the first-level groups, divide them according to the spatial influence degree between the obstacles to obtain a second-level group; determine the motion intention probability distribution corresponding to the second-level group according to the preset motion intention category; determine the target intention probability distribution of the second-level group according to the motion intention probability distribution corresponding to the second-level group and the motion intention probability distribution corresponding to the second-level group in the historical frame; determine the target motion intention of the second-level group according to the target intention probability distribution. The present application groups obstacles twice according to the road surface information and obstacle space information, and can divide obstacles with the same motion intention into the same group, and determine the target motion intention probability distribution of the second-level group according to the motion intention probability distribution corresponding to the same second-level group in the current frame and the historical frame, and further determine the final motion intention of the second-level group, so as to achieve accurate estimation of the motion intention of the traffic flow of the second-level group, thereby guiding the autonomous driving vehicle to accurately make a motion decision whether to follow or change lanes to bypass obstacles. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A diagram showing an application environment of a method for estimating motion intention in an embodiment;

[0057] Figure 2 is a schematic flow chart of a method for estimating motion intention in one embodiment;

[0058] Figure 3 A visual schematic diagram of a primary grouping in an embodiment;

[0059] Figure 4 is a flow chart of step 206 in one embodiment;

[0060] Figure 5 A visual schematic diagram of secondary grouping in one embodiment;

[0061] Figure 6 is a flow chart of step 402 in one embodiment;

[0062] Figure 7A schematic diagram of a fusion process for obtaining a target intention probability distribution in one embodiment;

[0063] Figure 8 is a flow chart of a motion intention estimation method in another embodiment;

[0064] Fig. 9 is a structural block diagram of a motion intention estimation device in one embodiment;

[0065] Fig.10 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0067] In the field of autonomous driving, obstacle motion prediction is mostly based on the obstacle's motion data from the past few seconds and lane-related information to predict the motion state in the next few seconds. However, the same motion state may have different motion intentions in different scenarios, and no corresponding motion intention prediction is given. For example, a vehicle that slowly stops on the right ahead may be parked by the side of the road, may be stuck in traffic on the ramp following the flow of traffic, or may be stopped because of a right turn at a traffic light.

[0068] Due to the existence of road traffic signs, obstacles that move normally on the road will form a traffic flow on the lane, that is, the rear vehicle follows the front vehicle. Due to the limitation of physical space, the movement of traffic flow is conductive from the front to the rear. For example, at a traffic light, the front vehicle slows down and stops, and the red light will cause the rear vehicle to slow down and stop. Therefore, it is more reasonable and efficient to evaluate the movement intention of the entire traffic flow than a single obstacle.

[0069] In view of the technical problem that traditional technologies cannot accurately predict the movement intention of road obstacles, based on considerations of the movement phenomenon of traffic on lanes, this application provides a motion intention estimation method that can accurately estimate the movement intention of obstacles.

[0070] The motion intention estimation method provided in the embodiment of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The server 104 obtains the road surface information and obstacle space information of the road on which the autonomous driving vehicle is traveling in the current frame sent by the terminal 102; the obstacles are divided according to the road surface information and obstacle space information to obtain a first-level group; for each first-level group, the obstacles are divided according to the degree of spatial influence between the obstacles to obtain a second-level group; according to the preset motion intention category, the motion intention probability distribution corresponding to the second-level group is determined; according to the motion intention probability distribution corresponding to the second-level group and the motion intention probability distribution corresponding to the second-level group in the historical frame, the target intention probability distribution of the second-level group is determined; according to the target intention probability distribution, the target motion intention of the second-level group is determined.

[0071] The terminal 102 may be an autonomous driving terminal. In one embodiment, the terminal 102 may be an unmanned vehicle or a manned vehicle with autonomous driving capability. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers.

[0072] It should be noted that the embodiments of the present application can also be applied to a terminal alone or to a server alone, which is not specifically limited here.

[0073] In one embodiment, Figure 2 As shown, a motion intention estimation method is provided, which is applied to Figure 1 The server in is taken as an example to illustrate, including the following steps 202 to 212.

[0074] Step 202, obtaining road surface information and obstacle space information of the road on which the autonomous driving vehicle is traveling in the current frame.

[0075] The server obtains the road surface information and obstacle space information corresponding to the road on which the autonomous driving vehicle is traveling in the current frame image. Among them, obstacles can be objects that can move within a certain range around the autonomous driving vehicle. In this embodiment, such as obstacle vehicles, obstacle pedestrians, etc. The obstacle space information can be the current position information of each obstacle, which can be coordinate information in a fixed coordinate system, longitude and latitude information, etc., or position information with the same measurement standard as the autonomous driving vehicle. Road surface information refers to information on the road surface that serves as a motion marker. In this embodiment, such as lane line information, traffic light position information, sidewalk information, etc. In general, it is possible to obtain the road surface information of the road on which the autonomous driving vehicle is traveling in the current frame image and the obstacle space information within a certain range in front of the autonomous driving vehicle in the driving direction.

[0076] Optionally, the road surface information of the road on which the autonomous driving vehicle is traveling in the current frame image can be obtained through the on-board high-precision map, wherein the on-board high-precision map can be a high-precision map loaded in the on-board system of the autonomous driving vehicle, and the on-board high-precision map can include high-precision absolute coordinates, more detailed and rich road traffic information, etc. In addition, the spatial information of surrounding obstacles can be identified through sensors installed on the autonomous driving vehicle, such as cameras and radars, and the position information of obstacles around the autonomous driving vehicle and the movement information of obstacles can be accurately located through the sensors.

[0077] Optionally, a high-definition camera installed on the road can be used to obtain the road surface information corresponding to the road on which the autonomous driving vehicle is traveling and the obstacle space information around it in real time. For example, the high-definition camera can extract the autonomous driving vehicle from the monitoring video or the captured monitoring image according to the characteristics of the autonomous driving vehicle, and then identify the obstacle space information around the autonomous driving vehicle.

[0078] Step 204: divide the obstacles into primary groups according to the road surface information and the obstacle space information.

[0079] The server divides the obstacles according to the road surface information and the obstacle space information to obtain the first-level group. Optionally, according to the obstacle space information and the corresponding road surface information, obstacles with the same obstacle space information and road surface information are divided into the same first-level group. That is, obstacles with potential common driving intentions are divided into the same first-level group.

[0080] For example, according to whether the road surface information is a straight lane, an off-ramp lane, in front of and behind a traffic light in a red light state, in front of and behind a sidewalk, etc., the obstacles at positions corresponding to the road surface information are divided into corresponding first-level groups, that is, obstacles on the straight lane are one first-level group, obstacles on the off-ramp lane on one side are one first-level group, obstacles in front of the sidewalk are one first-level group, obstacles behind the sidewalk are one first-level group, obstacles in front of the traffic light are one first-level group, etc., as shown in the figure. Figure 3 The schematic diagram of the first-level grouping is shown in the figure. The square boxes in the figure represent obstacles, and the obstacles in the elliptical circles belong to the same first-level grouping.

[0081] Step 206: For each first-level group, divide it according to the spatial influence degree between obstacles to obtain a second-level group.

[0082] For each first-level group, the server can further divide the obstacles according to the degree of spatial influence between two obstacles to obtain second-level groups, so as to obtain traffic groups with more consistent movement intentions. Among them, the degree of spatial influence between obstacles can be understood as the degree of mutual occupation or obstruction of obstacles in space. For example, the degree of obstruction of the front car to the rear car, etc. Each first-level group includes at least one second-level group.

[0083] It can be understood that in the same traffic flow in this embodiment, the preceding obstacle has the greatest impact on the movement range of the following vehicle compared to other obstacles, the distance between two consecutive obstacles is not greater than the preset distance range, and in the same lane, there are no two overlapping obstacles, that is, only one obstacle is allowed to stay at the same position on the same lane.

[0084] Step 208: Determine the movement intention probability distribution corresponding to the secondary grouping according to the preset movement intention category.

[0085] The server can determine the probability distribution of the motion intention of the secondary grouping in the current frame according to the preset motion intention category. According to the attributes of the traffic corresponding to each secondary grouping and the characteristics of the surrounding traffic, the method of data statistics combined with machine learning can be used to determine the probability distribution of the motion intention corresponding to the secondary grouping. The preset motion intention category characterizes all possible motion intentions of the secondary grouping, which can be a list of motion intentions collected based on experience. For example, the motion intention list includes straight ahead, down ramp, traffic merging into a jam, waiting for a red light before a traffic light, waiting for pedestrians before a sidewalk, pulling over, slowing down in a traffic jam, etc. It should be noted that the content in the preset motion intention list can be different according to the actual application scenario, and can also be independently set and defined as needed, and added, modified or deleted on this basis, all of which belong to the content protected by the embodiments of the present application.

[0086] Optionally, the probability distribution of the secondary group corresponding to the preset motion intention category in the current frame can be obtained by manually recording the traffic attributes of the secondary group itself and the traffic characteristics around the secondary group.

[0087] Each secondary group corresponds to a motion intention probability distribution, and the motion intention probability distributions corresponding to different secondary groups are generally different.

[0088] Step 210 , determining the target intention probability distribution of the secondary grouping according to the motion intention probability distribution corresponding to the secondary grouping and the motion intention probability distribution corresponding to the secondary grouping in the historical frame.

[0089] In this embodiment, the server can determine the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group in the current frame and the motion intention probability distribution corresponding to the secondary group in the historical frame. That is, the final motion intention estimation result of the secondary group is determined comprehensively by the motion intention estimation result of the secondary group in the current frame and the motion intention estimation result of the secondary group in the historical frame, so that the target intention probability distribution can integrate the estimation information in the current frame and the historical frame, and the estimation result will be more accurate.

[0090] Optionally, a Bayesian network may be used to fuse the motion intention probability distribution corresponding to the secondary grouping with the motion intention probability distribution corresponding to the secondary grouping in the historical frame, for example, weighted fusion, to obtain the target intention probability distribution of the secondary grouping.

[0091] Step 212, determining the target motion intention of the secondary grouping according to the target intention probability distribution.

[0092] In this embodiment, the target motion intention of the secondary grouping is determined according to the target intention probability distribution based on preset conditions.

[0093] Optionally, the motion intention corresponding to the maximum probability in the target intention probability distribution may be used as the target motion intention of the secondary grouping.

[0094] The above-mentioned motion intention estimation method obtains the road surface information and obstacle space information of the road on which the autonomous driving vehicle is traveling in the current frame, divides the obstacles according to the road surface information and obstacle space information to obtain a first-level group; for each first-level group, divides according to the degree of spatial influence between obstacles to obtain a second-level group; determines the motion intention probability distribution corresponding to the second-level group according to the preset motion intention category; determines the target intention probability distribution of the second-level group according to the motion intention probability distribution corresponding to the second-level group and the motion intention probability distribution corresponding to the second-level group in the historical frame; determines the target motion intention of the second-level group according to the target intention probability distribution. This embodiment groups obstacles twice according to the road surface information and obstacle space information, and can divide obstacles with the same motion intention into the same group, and determines the target intention probability distribution of the second-level group according to the motion intention probability distribution corresponding to the same second-level group in the current frame and the historical frame, so as to achieve accurate estimation of the motion intention of the traffic flow of the second-level group, thereby guiding the autonomous driving vehicle to accurately make a motion decision whether to follow or change lanes to bypass obstacles.

[0095] In one embodiment, Figure 4 As shown, for each first-level group, the obstacles are divided according to the spatial influence between them to obtain the step 206 of second-level grouping, including the following steps 402 to 406.

[0096] Step 402: for each first-level group, select one obstacle from the obstacles in turn as the first obstacle, and calculate the spatial influence degree between the first obstacle and the second obstacle; wherein the second obstacle is the obstacle other than the first obstacle among all obstacles in the first-level group.

[0097] In each first-level group, one of the obstacles in the first-level group is selected as the first obstacle, and the spatial influence between the first obstacle and the second obstacle is calculated. The second obstacle is the obstacle other than the first obstacle in the first-level group. For example, there are four obstacles A, B, C, and D in the first-level group. When obstacle A is selected as the first obstacle, the spatial influence between obstacle A and obstacle B, the spatial influence between obstacle A and obstacle C, and the spatial influence between obstacle A and obstacle D are calculated; and so on, when obstacle B is selected as the first obstacle, the spatial influence between obstacle B and obstacle A, obstacle C, and obstacle D is calculated.

[0098] In this embodiment, for each first-level group, one obstacle is selected from the obstacles in turn as the first obstacle, and the spatial influence between the first obstacle and the second obstacle is calculated. Here, one obstacle may be selected from the obstacles in the first-level group in turn as the first obstacle in a preset order, or one obstacle may be randomly selected as the first obstacle without following the order, until all obstacles in the first-level group are selected.

[0099] Step 404: Select the second obstacle that has the greatest spatial influence on the first obstacle as the target obstacle.

[0100] For each first obstacle, a second obstacle corresponding to the spatial influence degree with the largest spatial influence degree between the first obstacle and the second obstacle is selected as the target obstacle. In other words, for each first obstacle, there will be a corresponding second obstacle in the first-level group as the target obstacle.

[0101] Step 406: group the target obstacle and the first obstacle into the same secondary group.

[0102] In this embodiment, the second obstacle corresponding to the largest spatial influence between the first obstacle and the target obstacle is selected as the target obstacle, and the target obstacle and the first obstacle are classified into the same secondary group, and finally multiple different secondary groups can be obtained. Optionally, when all obstacles in the primary group are classified into the corresponding secondary group, the secondary grouping work stops. In an example, the secondary grouping schematic diagram is as follows Figure 5 As shown, the obstacles in an elliptical circle are a secondary grouping.

[0103] The above-mentioned motion intention estimation method can divide obstacles with the same motion intention into the same secondary group to a large extent by selecting the second obstacle corresponding to the maximum spatial influence between the second obstacle and the first obstacle in the same primary group as the target obstacle, and classifying the target obstacle and the first obstacle into the same secondary group, thereby improving the accuracy of the motion intention probability distribution estimation of the corresponding secondary group.

[0104] In some embodiments, Figure 6 As shown, step 402 of selecting one obstacle from the obstacles as the first obstacle and calculating the spatial influence degree between the first obstacle and the second obstacle includes the following steps 602 to 604.

[0105] Step 602, establish a Fresnel coordinate system with the autonomous driving vehicle as the origin and the road driving direction as the positive direction s.

[0106] Compared with the Cartesian coordinate system, the Freynesian coordinate system can simplify the problem of motion trajectory prediction. In the Freynesian coordinate system, the center line of the road is used as the reference line. In this embodiment, the autonomous driving vehicle is used as the origin. The s direction of the coordinate axis is along the center line of the road, which is called the longitudinal direction. The l direction of the coordinate axis is the normal direction on the center line of the road passing through the origin, which is called the transverse direction. The s direction and the l direction are perpendicular to each other. That is, the Freynesian coordinate system uses the tangent vector and the normal vector of the center line of the road to establish the coordinate system. See Figure 5 shown.

[0107] Step 604: select the first obstacle in order according to the s coordinates of the obstacles from small to large, and calculate the spatial influence degree between the first obstacle and the second obstacle.

[0108] The first obstacle can be selected in order from small to large according to the s coordinate of the obstacle, that is, starting from the obstacle with the smaller s coordinate in the first-level group, they are selected as the first obstacle in turn, and then the spatial influence degree between the first obstacle and the second obstacle in the first-level group is calculated.

[0109] In this embodiment, by establishing a Freynes coordinate system, the first obstacle is selected in order from small to large according to the s coordinates of the obstacles, and the spatial influence between the first obstacle and the second obstacle is calculated, so that the first obstacle can be located more conveniently and quickly, the amount of calculation can be reduced, and the operation speed can be improved.

[0110] In one embodiment, calculating the spatial influence degree between the first obstacle and the second obstacle in step 402 includes: calculating the overlapping area between the first obstacle and the second obstacle.

[0111] Step 404 of selecting the second obstacle corresponding to the maximum spatial influence between the second obstacle and the first obstacle as the target obstacle includes: selecting the second obstacle having the largest overlapping area with the first obstacle as the target obstacle.

[0112] In this embodiment, the spatial influence degree between the first obstacle and the second obstacle is characterized by the overlapping area between the first obstacle and the second obstacle. Optionally, the obstacle can be characterized by a box diagram of corresponding proportion or a point of corresponding size, and the overlapping area between the first obstacle and the second obstacle can be calculated, or by setting the first protection area and the second protection area respectively before and after the obstacle, the sum of the first protection area before the first obstacle, the first obstacle and the second protection area after the first obstacle is taken as the first area, and the sum of the first protection area before the second obstacle, the second obstacle and the second protection area after the second obstacle is taken as the second area, then the overlapping area between the first area and the second area can also be calculated, so as to select the second obstacle with the largest overlapping area with the first obstacle as the target obstacle.

[0113] For example, in the s direction in the Freynes coordinate system, for the front and rear obstacles, the overlapping area of ​​the rear of the front obstacle and the front of the rear obstacle can be calculated. For the first obstacle, the overlapping area of ​​the front of the first obstacle and the rear of each second obstacle in front can be calculated, and the second obstacle with the largest overlapping area with the first obstacle is selected as the target obstacle.

[0114] In one embodiment, determining the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame includes:

[0115] The motion intention probability distribution corresponding to the secondary grouping and the motion intention probability distribution corresponding to the secondary grouping in the historical frame are fused to obtain the target intention probability distribution of the secondary grouping.

[0116] In this embodiment, the motion intention probability distribution corresponding to the secondary grouping and the motion intention probability distribution corresponding to the secondary grouping in the historical frame can be fused to obtain the target intention probability distribution of the secondary grouping. The number of historical frames can be selected according to demand, for example, it can be 3 frames, 4 frames or 5 frames. The historical frames can be continuous frames before the current frame, or frames selected at a certain interval.

[0117] Optionally, the motion intention probability distribution corresponding to the secondary group in the current frame and the motion intention probability distribution corresponding to the secondary group in the historical frame can be weighted and fused to obtain the target intention probability distribution of the secondary group. For example, a larger weight is assigned to the motion intention probability distribution corresponding to the current frame, and a smaller weight is assigned to the motion intention probability distribution in the historical frame. After obtaining the target intention probability distribution of the secondary group, the motion intention corresponding to the maximum probability in the target intention probability distribution can be used as the final motion intention estimation result of the secondary group, so that the autonomous driving vehicle can make a timely decision on whether to follow the secondary group in front or change lanes.

[0118] In one embodiment, the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame are fused to obtain the target intention probability distribution of the secondary group, including:

[0119] The target intention probability distribution of the secondary group is obtained by integrating the motion intention probability distribution corresponding to the secondary group, the motion intention probability distribution corresponding to the secondary group in the historical frame, the traffic attributes corresponding to the secondary group, the road information and the surrounding traffic attributes.

[0120] In this embodiment, the target intention probability distribution of the secondary group can be obtained by fusion based on the motion intention probability distribution corresponding to the secondary group, the motion intention probability distribution corresponding to the secondary group in the historical frame, the traffic attributes corresponding to the secondary group, the road information and the surrounding traffic attributes. Among them, the traffic attributes, road information and surrounding traffic attributes corresponding to the secondary group include the traffic attributes, road information and surrounding traffic attributes corresponding to the secondary group in the current frame and the traffic attributes, road information and surrounding traffic attributes corresponding to the secondary group in the historical frame. The traffic attributes corresponding to the secondary group can be, for example, the location of the secondary group, the speed of movement, etc.; the road information can be, for example, whether there is a traffic light ahead, whether there is a sidewalk, whether it is an off-ramp, etc.; the surrounding traffic attributes can be, for example, the overall movement speed of the surrounding traffic. If the surrounding traffic is slow or stationary, and there is no room for detour, then there may be a larger probability distribution in the usual traffic jam scene.

[0121] In one example, a dynamic Bayesian network is used to weightedly fuse the motion intention probability distribution, traffic attributes, road information and surrounding traffic attributes of the secondary group in the current frame and the motion intention probability distribution, traffic attributes, road information and surrounding traffic attributes of the secondary group in the historical frame to obtain the target intention probability distribution of the secondary group.

[0122] In one embodiment, the motion intention probability distribution, traffic flow attributes, road information and surrounding traffic flow attributes corresponding to the secondary grouping can be fused to obtain the first motion intention probability distribution corresponding to the current frame, and the first motion intention probability distribution corresponding to the current frame and the second motion intention probability distribution corresponding to the historical frame can be fused again to obtain the target intention probability distribution of the secondary grouping. The fusion process diagram can be seen in Figure 7 As shown, Y represents the probability distribution of motion intention, F represents the traffic attributes, M represents the road information and surrounding traffic attributes, BE represents the Bayesian formula, O represents the probability distribution of motion intention of the corresponding frame, t represents the corresponding frame number, and the current frame is the nth frame.

[0123] In the above motion intention estimation method, the motion intention probability distribution, traffic attributes, road information and surrounding traffic attributes corresponding to the current frame and the historical frame are fused, and the target intention probability distribution is obtained by fusing more comprehensive fusion information, making the target intention probability distribution more accurate.

[0124] In some embodiments, the target motion intention of the secondary grouping is determined according to the target intention probability distribution, including: taking the motion intention corresponding to the maximum probability in the target intention probability distribution as the target motion intention of the secondary grouping.

[0125] Exemplarily, in the target intention probability distribution corresponding to the secondary grouping, the probability corresponding to the down ramp in the preset motion intention category is the largest, then the down ramp can be used as the target motion intention corresponding to the secondary grouping.

[0126] The target intention probability distribution includes the probability of each motion intention in the preset motion intention category. Generally speaking, the greater the probability, the more likely the corresponding motion intention is to occur. Therefore, the motion intention with the highest probability is taken as the target motion intention of the secondary grouping, and the motion intention estimation result is more accurate.

[0127] In one embodiment, the motion intention estimation method includes the following steps 802 to 818 .

[0128] Step 802, obtaining the road surface information and obstacle space information of the road on which the autonomous driving vehicle is traveling in the current frame.

[0129] Step 804: divide the obstacles according to the road surface information and the obstacle space information to obtain primary groups.

[0130] Step 806, establish a Fresnel coordinate system with the autonomous driving vehicle as the origin and the road driving direction as the positive direction s.

[0131] Step 808: For each first-level group, select the first obstacle in order according to the s coordinates of the obstacles from small to large, and calculate the overlapping area between the first obstacle and the second obstacle, wherein the second obstacle is the obstacle other than the first obstacle among all obstacles in the first-level group.

[0132] Step 810: Select the second obstacle with the largest overlapping area with the first obstacle as the target obstacle.

[0133] Step 812: classify the target obstacle and the first obstacle into the same secondary group.

[0134] Step 814: Determine the motion intention probability distribution corresponding to the secondary grouping according to the preset motion intention category.

[0135] Step 816, the motion intention probability distribution corresponding to the secondary grouping and the motion intention probability distribution corresponding to the secondary grouping in the historical frame are fused through a dynamic Bayesian network to obtain the target intention probability distribution of the secondary grouping.

[0136] Step 818, determining the target motion intention of the secondary grouping according to the target intention probability distribution.

[0137] The above-mentioned motion intention estimation method performs a first grouping of obstacles through road obstacle information and obstacle space information, and performs a second grouping according to the overlap between obstacles, so as to obtain secondary groupings with common motion intentions to the greatest extent possible, and then performs motion intention estimation based on the secondary grouping information in the current frame and the secondary grouping information in the historical frames, so as to obtain an accurate target intention probability distribution, and further determine the target motion intention of the secondary grouping, thereby guiding the autonomous driving vehicle to accurately make motion decisions on whether to follow or change lanes to bypass obstacles.

[0138] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0139] Based on the same inventive concept, the embodiment of the present application also provides a motion intention estimation device for implementing the motion intention estimation method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the motion intention estimation device provided below can refer to the limitations of the motion intention estimation method above, and will not be repeated here.

[0140] In one embodiment, Fig. 9 As shown, a motion intention estimation device is provided, including: an information acquisition module 902, a first division module 904, a second division module 906, a first probability determination module 908, a second probability determination module 910 and a motion intention determination module 912, wherein:

[0141] The information acquisition module 902 is used to obtain the road surface information and obstacle space information of the road where the autonomous driving vehicle is traveling in the current frame;

[0142] A first division module 904 is used to divide obstacles according to the road surface information and the obstacle space information to obtain a primary group;

[0143] A second division module 906 is used to divide each of the first-level groups according to the spatial influence degree between the obstacles to obtain a second-level group;

[0144] A first probability determination module 908, configured to determine the probability distribution of motion intention corresponding to the secondary grouping according to a preset motion intention category;

[0145] A second probability determination module 910 is used to determine the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame;

[0146] The motion intention determination module 912 is used to determine the target motion intention of the secondary grouping according to the target intention probability distribution.

[0147] In one embodiment, the second partitioning module 906 is further configured to:

[0148] For each of the first-level groups, one obstacle is selected from the obstacles in turn as the first obstacle, and the degree of spatial influence between the first obstacle and the second obstacle is calculated; the second obstacle is the obstacle other than the first obstacle among all the obstacles in the first-level group; the second obstacle corresponding to the maximum degree of spatial influence between the second obstacle and the first obstacle is selected as the target obstacle; and the target obstacle and the first obstacle are classified into the same second-level group.

[0149] In one embodiment, the second partitioning module 906 is further configured to:

[0150] A Freyna coordinate system is established with the autonomous driving vehicle as the origin and the road driving direction as the positive s direction; a first obstacle is selected in order from small to large according to the s coordinates of the obstacles, and the spatial influence degree between the first obstacle and the second obstacle is calculated.

[0151] In one embodiment, the second division module 906 is further used to: calculate the overlapping area between the first obstacle and the second obstacle; and select the second obstacle having the largest overlapping area with the first obstacle as the target obstacle.

[0152] In one embodiment, the second probability determination module 910 is further configured to:

[0153] The motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame are fused to obtain the target intention probability distribution of the secondary group.

[0154] In one embodiment, the second probability determination module 910 is further configured to:

[0155] The target intention probability distribution of the secondary group is obtained by fusing the motion intention probability distribution corresponding to the secondary group, the motion intention probability distribution corresponding to the secondary group in the historical frame, the traffic attributes corresponding to the secondary group, the road information and the surrounding traffic attributes.

[0156] In one embodiment, the motion intention determination module 912 is further configured to:

[0157] The motion intention corresponding to the maximum probability in the target intention probability distribution is used as the target motion intention of the secondary grouping.

[0158] Each module in the above-mentioned motion intention estimation device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0159] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig.10As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a motion intention estimation method is implemented.

[0160] Those skilled in the art will understand that Fig.10 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0161] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps of the motion intention estimation method in the above embodiment when executing the computer program.

[0162] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the motion intention estimation method in the above embodiment are implemented.

[0163] In one embodiment, a computer program product is provided, including a computer program, which implements the steps of the motion intention estimation method in the above embodiment when executed by a processor.

[0164] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0165] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0166] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0167] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for estimating motion intention, characterized in that: The method comprises: Obtaining road surface information and obstacle space information of the road where the autonomous driving vehicle is traveling in the current frame; Dividing obstacles according to the road surface information and the obstacle space information to obtain a primary grouping; For each of the first-level groups, divide them according to the spatial influence degree between the obstacles to obtain second-level groups; Determining the probability distribution of movement intention corresponding to the secondary grouping according to the preset movement intention category; Determine the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame; The target motion intention of the secondary grouping is determined according to the target intention probability distribution.

2. The method according to claim 1, characterized in that For each of the first-level groups, the second-level groups are obtained by dividing the first-level groups according to the spatial influence degree between the obstacles, including: For each of the first-level groups, one obstacle is selected from the obstacles in turn as a first obstacle, and a spatial influence degree between the first obstacle and a second obstacle is calculated; the second obstacle is an obstacle other than the first obstacle among all obstacles in the first-level group; Selecting a second obstacle that has the greatest spatial influence on the first obstacle as a target obstacle; The target obstacle and the first obstacle are classified into the same secondary group.

3. The method according to claim 2, characterized in that The selecting one obstacle from the obstacles in turn as the first obstacle and calculating the spatial influence degree between the first obstacle and the second obstacle includes: A Fresnel coordinate system is established with the autonomous driving vehicle as the origin and the road driving direction as the positive direction s; The first obstacles are selected in order from small to large according to the s coordinates of the obstacles, and the spatial influence degree between the first obstacle and the second obstacle is calculated.

4. The method according to claim 2 or 3, characterized in that: The calculating the spatial influence degree between the first obstacle and the second obstacle includes: Calculating an overlapping area between the first obstacle and the second obstacle; The selecting the second obstacle corresponding to the maximum spatial influence between the first obstacle and the second obstacle as the target obstacle includes: A second obstacle having the largest overlapping area with the first obstacle is selected as the target obstacle.

5. The method according to claim 1, characterized in that The determining the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame includes: The motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame are fused to obtain the target intention probability distribution of the secondary group.

6. The method according to claim 5, characterized in that The step of fusing the motion intention probability distribution corresponding to the secondary group with the motion intention probability distribution corresponding to the secondary group in the historical frame to obtain the target intention probability distribution of the secondary group includes: The target intention probability distribution of the secondary group is obtained by fusing the motion intention probability distribution corresponding to the secondary group, the motion intention probability distribution corresponding to the secondary group in the historical frame, the traffic attributes corresponding to the secondary group, the road information and the surrounding traffic attributes.

7. The method according to claim 1, characterized in that The determining the target movement intention of the secondary grouping according to the target intention probability distribution includes: The motion intention corresponding to the maximum probability in the target intention probability distribution is used as the target motion intention of the secondary grouping.

8. A motion intention estimation device, characterized in that: The device comprises: An information acquisition module, used to obtain road surface information and obstacle space information of the road on which the autonomous driving vehicle is traveling in the current frame; A first division module, used for dividing obstacles according to the road surface information and the obstacle space information to obtain a primary group; A second division module is used to divide each of the first-level groups according to the spatial influence degree between the obstacles to obtain a second-level group; A first probability determination module, used to determine the probability distribution of motion intention corresponding to the secondary grouping according to a preset motion intention category; A second probability determination module is used to determine the target intention probability distribution of the secondary group according to the motion intention probability distribution corresponding to the secondary group and the motion intention probability distribution corresponding to the secondary group in the historical frame; The motion intention determination module is used to determine the target motion intention of the secondary grouping according to the target intention probability distribution.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images