Obstacle trajectory prediction method and device

Abstract

The application provides an obstacle trajectory prediction method and device, the method comprising: feature extraction on vehicle perception data to obtain global feature information in a Cartesian coordinate system; determining a reference line according to the global feature information, and converting from the Cartesian coordinate system to a Frenet coordinate system, and then to an agent coordinate system to obtain second feature information in the agent coordinate system; predicting a future trajectory of an obstacle according to a pre-constructed trajectory prediction model and the second feature information to obtain future trajectory information; converting the future trajectory information from the agent coordinate system to the Cartesian coordinate system to obtain target future trajectory information in the Cartesian coordinate system. The method and device can convert curved road data into straight road data through Frenet coordinate system conversion, eliminate the differences in the map in the scene, and then predict the trajectory through the trajectory prediction model, which is highly accurate, and also improves the accuracy of curved road prediction and the experience of autonomous driving.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving, in particular to a method and device for predicting trajectory of an obstacle. BACKGROUND

[0002] As an important development direction of future transportation, the core of automatic driving technology lies in the accurate prediction of the motion trajectory of various objects in the surrounding environment. The automatic driving motion prediction technology realizes the accurate prediction of the future motion trajectory of a vehicle by comprehensively considering the historical trajectory of the vehicle, the current surrounding environment information and the future driving plan, which is of great significance to ensuring driving safety and improving traffic efficiency. Current trajectory prediction technologies generally rely on a physical model (e.g., a dynamics or kinematics model) to describe the motion pattern of a target, and predict the future position and dynamics of the target based on the model combined with the instantaneous state of the target. However, in practice, it is found that the existing methods perform well in straight road scenarios, but the prediction accuracy of the future trajectory of an obstacle in a curved road environment is greatly reduced. In addition, relying solely on a physical model for prediction often fails to achieve the desired accuracy level. SUMMARY

[0003] The purpose of the embodiments of the present application is to provide a method and device for predicting trajectory of an obstacle, which can convert curved road data into straight road data through Frenet coordinate system conversion, eliminate the differences in the map in the scene, and then perform trajectory prediction through a trajectory prediction model, thereby improving the accuracy of curved road prediction and the experience of automatic driving.

[0004] The first aspect of the present application provides a method for predicting trajectory of an obstacle, comprising:

[0005] collecting vehicle perception data of a target vehicle on a curved road through a sensor device;

[0006] performing feature extraction on the vehicle perception data to obtain global feature information in a Cartesian coordinate system;

[0007] determining a reference line according to the global feature information;

[0008] obtaining at least one to-be-predicted obstacle that needs to be predicted, and extracting obstacle features of the to-be-predicted obstacle from the global feature information;

[0009] converting the obstacle features from the Cartesian coordinate system to a Frenet coordinate system according to the reference line, to obtain first feature information of the to-be-predicted obstacle in the Frenet coordinate system;

[0010] converting the first feature information from the Frenet coordinate system to an agent coordinate system, to obtain second feature information of the to-be-predicted obstacle in the agent coordinate system;

[0011] performing future trajectory prediction of the obstacle according to the pre-constructed trajectory prediction model and the second feature information, to obtain future trajectory information of the obstacle to be inferred;

[0012] converting the future trajectory information from the agent coordinate system to the Cartesian coordinate system, to obtain target future trajectory information of the obstacle to be inferred in the Cartesian coordinate system.

[0013] Further, the vehicle perception data at least includes vehicle trajectory information and vehicle surrounding map information;

[0014] The global feature information includes historical trajectory information and surrounding map information of all obstacles within a preset range; wherein the historical trajectory information at least includes position information, speed information and motion direction; and the surrounding map information at least includes lane center point position information and lane type.

[0015] Further, the reference line is determined according to the global feature information, including:

[0016] determining a scene range in which the target vehicle is currently located and a target lane in which the target vehicle is located according to the global feature information;

[0017] searching for other lanes having a predecessor-successor reachable relationship with the target lane in the scene range by using a depth-first search algorithm;

[0018] if the other lanes include a plurality of predecessor-successor lanes, selecting one of the predecessor-successor lanes from the other lanes as a predecessor-successor reachable lane in the scene;

[0019] generating a reference line according to the target lane and the predecessor-successor reachable lane in the scene.

[0020] Further, the reference line is generated according to the target lane and the predecessor-successor reachable lane in the scene, including:

[0021] combining the target lane and the predecessor-successor reachable lane in the scene in the order from behind the target vehicle to in front of the target vehicle, to obtain a reference line.

[0022] Further, the obstacle feature is converted from the Cartesian coordinate system to the Frenet coordinate system according to the reference line, to obtain first feature information of the obstacle to be inferred in the Frenet coordinate system, including:

[0023] obtaining obstacle historical trajectory coordinates and surrounding lane coordinates of the obstacle to be inferred from the obstacle feature;

[0024] convert the obstacle historical trajectory coordinates from the Cartesian coordinate system to the Frenet coordinate system based on the reference line to obtain obstacle historical trajectory Frenet coordinates;

[0025] convert the surrounding lane coordinates from the Cartesian coordinate system to the Frenet coordinate system based on the reference line to obtain surrounding lane Frenet coordinates;

[0026] aggregate the obstacle historical trajectory Frenet coordinates and the surrounding lane Frenet coordinates to obtain first feature information of the to-be-reasoned obstacle in the Frenet coordinate system.

[0027] Further, the conversion of the first feature information from the Frenet coordinate system to the agent coordinate system to obtain second feature information of the to-be-reasoned obstacle in the agent coordinate system comprises:

[0028] extract target trajectory information of the to-be-reasoned obstacle from the obstacle historical trajectory Frenet coordinates; wherein the target trajectory information at least includes trajectory current time position, trajectory adjacent frame displacement, obstacle speed, and obstacle movement direction;

[0029] convert the obstacle historical trajectory Frenet coordinates from the Frenet coordinate system to the agent coordinate system with the to-be-reasoned obstacle as the center to obtain agent coordinate system trajectory features in the agent coordinate system;

[0030] convert the surrounding lane Frenet coordinates from the Frenet coordinate system to the agent coordinate system with the to-be-reasoned obstacle as the center to obtain lane features in the agent coordinate system;

[0031] aggregate the agent coordinate system trajectory features and the lane features to obtain second feature information of the to-be-reasoned obstacle in the agent coordinate system.

[0032] Further, the conversion of the obstacle historical trajectory Frenet coordinates from the Frenet coordinate system to the agent coordinate system with the to-be-reasoned obstacle as the center to obtain agent coordinate system trajectory features in the agent coordinate system comprises:

[0033] normalize the target trajectory information with the to-be-reasoned obstacle as the center to obtain normalized trajectory information;

[0034] obtain obstacle movement heading of the to-be-reasoned obstacle;

[0035] direction alignment processing is performed on the trajectory directions of all the obstacles in the normalized trajectory information based on the obstacle motion heading, to obtain trajectory features;

[0036] A mask is added to the trajectory features to obtain agent coordinate system trajectory features in the agent coordinate system.

[0037] Further, the surrounding lane Frenet coordinates are converted from the Frenet coordinate system to the agent coordinate system with the to-be-reasoned obstacle as the center, to obtain lane features in the agent coordinate system, which includes:

[0038] Target map information of the to-be-reasoned obstacle within a preset range at the current time is extracted from the surrounding lane Frenet coordinates; wherein the target map information at least includes lane center line node position, lane node displacement, lane connection relationship, road structure and traffic signal light;

[0039] The target map information is normalized with the to-be-reasoned obstacle as the center, to obtain normalized map information;

[0040] Direction alignment processing is performed on the lane directions in the normalized map information based on the obstacle motion heading, to obtain lane features in the agent coordinate system.

[0041] Further, the future trajectory information is converted from the agent coordinate system to the Cartesian coordinate system, to obtain target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system, which includes:

[0042] The future trajectory information is converted from the agent coordinate system to the Frenet coordinate system, to obtain future trajectory information of the to-be-reasoned obstacle in the Frenet coordinate system;

[0043] The future trajectory information is converted from the Frenet coordinate system to the Cartesian coordinate system, to obtain target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system.

[0044] The second aspect of the present application provides an obstacle trajectory prediction device, which includes:

[0045] The acquisition unit is configured to acquire vehicle perception data of a target vehicle on a curve through a sensor device;

[0046] The feature extraction unit is configured to perform feature extraction on the vehicle perception data, to obtain global feature information in a Cartesian coordinate system;

[0047] determining unit, configured to determine a reference line according to the global feature information;

[0048] an obtaining unit, configured to obtain at least one to-be-reasoned obstacle needing reasoning, and extract obstacle feature of the to-be-reasoned obstacle from the global feature information;

[0049] a first coordinate conversion unit, configured to convert the obstacle feature from the Cartesian coordinate system to a Frenet coordinate system according to the reference line, to obtain first feature information of the to-be-reasoned obstacle in the Frenet coordinate system;

[0050] a second coordinate conversion unit, configured to convert the first feature information from the Frenet coordinate system to an agent coordinate system, to obtain second feature information of the to-be-reasoned obstacle in the agent coordinate system;

[0051] a trajectory prediction unit, configured to perform obstacle future trajectory prediction according to a pre-constructed trajectory prediction model and the second feature information, to obtain future trajectory information of the to-be-reasoned obstacle;

[0052] a third coordinate conversion unit, configured to convert the future trajectory information from the agent coordinate system to the Cartesian coordinate system, to obtain target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system.

[0053] Further, the vehicle perception data at least includes vehicle trajectory information and vehicle surrounding map information;

[0054] The global feature information includes historical trajectory information and surrounding map information of all obstacles within a preset range; wherein, the historical trajectory information at least includes position information, speed information and motion direction; and the surrounding map information at least includes lane center point position information and lane type.

[0055] Further, the determining unit includes:

[0056] a determining subunit, configured to determine a scene range in which the target vehicle currently locates and a target lane in which the target vehicle locates according to the global feature information;

[0057] a searching subunit, configured to search other lanes having predecessor-successor reachable relationship with the target lane in the scene range by using a depth-first search algorithm;

[0058] a selecting subunit, configured to select one predecessor-successor lane from the other lanes as a predecessor-successor reachable lane in the scene when the other lanes include multiple predecessor-successor lanes;

[0059] The generating sub-unit is configured to generate a reference line according to the target lane and the front-preceding and rear-following reachable lane in the scene.

[0060] Further, the generating sub-unit is specifically configured to combine the target lane and the front-preceding and rear-following reachable lane in the scene in the order from the rear of the target vehicle to the front of the target vehicle to obtain the reference line.

[0061] Further, the first coordinate conversion sub-unit comprises:

[0062] The obtaining sub-unit is configured to obtain obstacle historical trajectory coordinates and surrounding lane coordinates of the to-be-reasoned obstacle from the obstacle features.

[0063] The first conversion sub-unit is configured to convert the obstacle historical trajectory coordinates from the Cartesian coordinate system to the Frenet coordinate system based on the reference line to obtain obstacle historical trajectory Frenet coordinates.

[0064] The first conversion sub-unit is further configured to convert the surrounding lane coordinates from the Cartesian coordinate system to the Frenet coordinate system based on the reference line to obtain surrounding lane Frenet coordinates.

[0065] The first summary sub-unit is configured to summarize the obstacle historical trajectory Frenet coordinates and the surrounding lane Frenet coordinates to obtain first feature information of the to-be-reasoned obstacle in the Frenet coordinate system.

[0066] Further, the second coordinate conversion sub-unit comprises:

[0067] The extraction sub-unit is configured to extract target trajectory information of the to-be-reasoned obstacle from the obstacle historical trajectory Frenet coordinates; wherein the target trajectory information at least includes a trajectory current time position, a trajectory adjacent frame displacement, an obstacle speed, and an obstacle motion direction.

[0068] The second conversion sub-unit is configured to convert the obstacle historical trajectory Frenet coordinates from the Frenet coordinate system to an agent coordinate system with the to-be-reasoned obstacle as the center to obtain agent coordinate system trajectory features in the agent coordinate system.

[0069] The second conversion sub-unit is further configured to convert the surrounding lane Frenet coordinates from the Frenet coordinate system to the agent coordinate system with the to-be-reasoned obstacle as the center to obtain lane features in the agent coordinate system.

[0070] The second aggregation subunit is configured to aggregate the agent coordinate system trajectory feature and the lane feature to obtain second feature information of the to-be-reasoned obstacle in the agent coordinate system.

[0071] Further, the second conversion subunit includes:

[0072] The processing module is configured to perform normalization processing on the target trajectory information with the to-be-reasoned obstacle as the center to obtain normalized trajectory information.

[0073] The acquisition module is configured to acquire an obstacle motion heading of the to-be-reasoned obstacle.

[0074] The processing module is further configured to perform direction alignment processing on a trajectory direction of all obstacles in the normalized trajectory information based on the obstacle motion heading to obtain a trajectory feature.

[0075] The adding module is configured to add a mask to the trajectory feature to obtain an agent coordinate system trajectory feature in the agent coordinate system.

[0076] Further, the second conversion subunit includes:

[0077] The extraction module is configured to extract target map information of the to-be-reasoned obstacle within a preset range at a current time from the surrounding lane Frenet coordinate, wherein the target map information at least includes a lane center line node position, a lane node displacement, a lane connection relationship, a road structure, and a traffic signal lamp.

[0078] The processing module is further configured to perform normalization processing on the target map information with the to-be-reasoned obstacle as the center to obtain normalized map information.

[0079] The processing module is further configured to perform direction alignment processing on a lane direction in the normalized map information based on the obstacle motion heading to obtain a lane feature in the agent coordinate system.

[0080] Further, the third coordinate conversion unit is specifically configured to convert the future trajectory information from the agent coordinate system to a Frenet coordinate system to obtain future trajectory information of the to-be-reasoned obstacle in the Frenet coordinate system.

[0081] The third coordinate conversion unit is specifically further configured to convert the future trajectory information from the Frenet coordinate system to the Cartesian coordinate system to obtain target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system.

[0082] The third aspect of the present application provides an electronic device, comprising a memory for storing a computer program and a processor for running the computer program to enable the electronic device to perform the obstacle trajectory prediction method according to any one of the first aspect of the present application.

[0083] The fourth aspect of the present application provides a computer readable storage medium storing computer program instructions, which, when read and run by a processor, perform the obstacle trajectory prediction method according to any one of the first aspect of the present application.

[0084] The beneficial effects of the present application are that the method and device can convert the curved road data into a straight road through Frenet coordinate system conversion, eliminate the differences in the map in the scene, and then perform trajectory prediction through a trajectory prediction model, which is highly accurate and improves the accuracy of curved road prediction and the experience of autonomous driving. BRIEF DESCRIPTION OF DRAWINGS

[0085] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0086] Figure 1 A flowchart of an obstacle trajectory prediction method provided by an embodiment of the present application is shown in the figure;

[0087] Figure 2 A flowchart of another obstacle trajectory prediction method provided by an embodiment of the present application is shown in the figure;

[0088] Figure 3 A result diagram of a trajectory prediction model predicting a future trajectory of a continuous curved road provided by an embodiment of the present application is shown in the figure;

[0089] Figure 4 A result diagram of converting model input features into Frenet coordinate system, then converting to agent coordinate system, and then performing trajectory prediction provided by an embodiment of the present application is shown in the figure;

[0090] Figure 5 A result diagram of converting a prediction result into a Cartesian coordinate system provided by an embodiment of the present application is shown in the figure;

[0091] Figure 6 A structural diagram of an obstacle trajectory prediction device provided by an embodiment of the present application is shown in the figure;

[0092] Figure 7Another structure diagram of an obstacle trajectory prediction device provided by the embodiment of the present application is shown. DETAILED DESCRIPTION

[0093] The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.

[0094] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. Meanwhile, in the description of the present application, the terms "first", "second", etc. are only used for differentiation, and cannot be understood as indicating or implying relative importance.

[0095] Embodiment 1

[0096] Please refer to Figure 1 , Figure 1 A flowchart of an obstacle trajectory prediction method provided by the embodiment is shown. The obstacle trajectory prediction method includes the following steps.

[0097] S101, collecting vehicle perception data of a target vehicle on a curve by a sensor device.

[0098] S102, performing feature extraction on the vehicle perception data to obtain global feature information in a Cartesian coordinate system.

[0099] S103, determining a reference line according to the global feature information.

[0100] S104, obtaining at least one to-be-reasoned obstacle that needs to be reasoned, and extracting obstacle features of the to-be-reasoned obstacle from the global feature information.

[0101] S105, converting the obstacle features from the Cartesian coordinate system to the Frenet coordinate system according to the reference line to obtain first feature information of the to-be-reasoned obstacle in the Frenet coordinate system.

[0102] S106, converting the first feature information from the Frenet coordinate system to the agent coordinate system to obtain second feature information of the to-be-reasoned obstacle in the agent coordinate system.

[0103] In the embodiment, each obstacle in the scene that needs to be reasoned is taken in turn as an agent, that is, the method converts the features in the Frenet coordinate system to the agent coordinate system in turn.

[0104] S107, performing future trajectory prediction of the obstacle according to a pre-constructed trajectory prediction model and the second feature information to obtain future trajectory information of the to-be-reasoned obstacle.

[0105] S108, convert the future trajectory information from the agent coordinate system to the Cartesian coordinate system to obtain target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system.

[0106] Generally, the amount of straight track data in the trajectory prediction public data set is much larger than the amount of curved track data, and the same is true for self-collected data. This makes the model's straight track prediction generalization ability better than curved track (straight track prediction is better than curved track prediction). In addition, the distribution of curved track map feature data is very different, which limits the generalization ability of the model.

[0107] Currently, the commonly used Cartesian coordinate system is difficult to express the relative position relationship between the vehicle and the road, which will lead to the ambiguity of the relative relationship between the two. As a kind of coordinate system based on curve, Frenet coordinate system (a kind of reference line based rectangular coordinate system) can better describe the position and motion state of the object on the curve.

[0108] Specifically, the Frenet coordinate system uses the road centerline as the reference line (the centerline is a series of discrete points output by the map module), and then uses the tangent vector and normal vector of the reference line to establish the coordinate system. The longitudinal coordinate represents the distance of the vehicle along the road, and the lateral coordinate represents the distance of the vehicle from the road centerline. In this way, the lateral coordinate and the longitudinal coordinate at each point on the road are perpendicular, and compared with the Cartesian coordinate system, the curved track can be regarded as a straight track in the Frenet coordinate system.

[0109] Therefore, the present application uses the Frenet coordinate system to convert the curved track into a straight track, reduces the distribution difference of the curved track map features, uses the strong generalization ability of the straight track prediction to improve the curved track prediction effect. Among them, the curved track data is converted into a straight track through the Frenet coordinate system conversion, which can eliminate the difference of the map in the scene, make the input more standardized, and then perform model reasoning, which can also reduce the requirements on the model. After the reasoning is completed, the prediction line is converted back to the Cartesian coordinate system, so as to improve the curved track prediction accuracy, effectively reduce the problems of brake caused by the curved straight prediction as cut-in or the cut-in error prediction as straight, and improve the automatic driving experience.

[0110] For example, the method can first use a sensor device to collect trajectory information of a vehicle on a curve and surrounding map information; then, global feature information required for trajectory prediction is extracted from the collected data; then, a reference line is selected, and the historical trajectory coordinates and surrounding lane coordinates in the global features are converted from the Cartesian coordinate system to the Frenet coordinate system based on the reference line, and other information not involving coordinates remains unchanged; then, the features in the Frenet coordinate system are converted to the agent coordinate system, and then input into a trajectory prediction model for trajectory prediction; finally, the prediction result is converted from the Frenet coordinate system to the Cartesian coordinate system.

[0111] In this embodiment, the execution subject of the method can be a computer, a server, or other computing devices, which are not limited in this embodiment.

[0112] In this embodiment, the execution subject of the method can also be a smart phone, a tablet computer, or other smart devices, which are not limited in this embodiment.

[0113] It can be seen that by implementing the obstacle trajectory prediction method described in this embodiment, the curve data can be converted to a straight road through Frenet coordinate system conversion, eliminating the differences in the map in the scene, and then trajectory prediction is performed through a trajectory prediction model, which is highly accurate and improves the accuracy of curve prediction and the experience of autonomous driving.

[0114] Embodiment 2

[0115] Please refer to Figure 2 , Figure 2 A flowchart of an obstacle trajectory prediction method provided in this embodiment is shown. The obstacle trajectory prediction method includes the following steps.

[0116] S201, collecting vehicle perception data of a target vehicle on a curve through a sensor device.

[0117] In this embodiment, the vehicle perception data at least includes vehicle trajectory information and vehicle surrounding map information.

[0118] In this embodiment, the method can use a sensor device (such as a laser radar, a camera, a millimeter wave radar, etc.) to perceive and locate the surrounding environment, collect surrounding traffic participant trajectory information and positioning map information.

[0119] S202, performing feature extraction on the vehicle perception data to obtain global feature information in the Cartesian coordinate system.

[0120] In this embodiment, the global feature information includes historical trajectory information of all obstacles in a preset range and surrounding map information; wherein, the historical trajectory information at least includes position information, speed information and movement direction; and the surrounding map information at least includes lane center point position information and lane type.

[0121] In this embodiment, the method can extract historical trajectory information of all obstacles in a specified range and surrounding map information from the perception data. Wherein, the historical trajectory information includes but is not limited to position information, speed information, and movement direction. The surrounding map information includes but is not limited to lane center point position information and lane type.

[0122] S203, determining the scene range in which the target vehicle is currently located and the target lane in which the target vehicle is located according to the global feature information.

[0123] S204, searching for other lanes having a predecessor-successor reachable relationship with the target lane in the scene range by using a depth-first search algorithm.

[0124] S205, when the other lanes include multiple predecessor-successor lanes, selecting one of the predecessor-successor lanes from the other lanes as a predecessor-successor reachable lane in the scene.

[0125] In this embodiment, if the lane is long enough, neither the predecessor lane nor the successor lane is needed, and only the target lane is used. It can be seen that the step exists to make the reference line cover the scene range as much as possible to facilitate Frenet projection.

[0126] S206, combining the target lane and the predecessor-successor reachable lane in the scene in the order from the rear of the target vehicle to the front of the target vehicle to obtain a reference line.

[0127] In this embodiment, since Frenet coordinate conversion needs to select a reference line, and the coverage range of the reference line needs to cover the range of the historical and trajectory of the to-be-reasoned obstacle, the future trajectory, and the lane feature as much as possible.

[0128] In this embodiment, the method uses the lane in which the host vehicle is located and the predecessor and successor lanes in the scene range as the reference line. The reason is that when the vehicle to be predicted is the vehicle near the host vehicle, using the lane in which the host vehicle is located and the predecessor and successor as the reference line can make the to-be-reasoned obstacle and the nearby map in the reference line range as much as possible. Wherein, if there are multiple predecessors and successors, only one is selected. In addition, the predecessors of the predecessors and the successors of the successors are determined by using the depth-first search until all the lanes in the scene range having a predecessor-successor reachable relationship with the lane in which the host vehicle is located are selected, and the multiple lanes are combined in a fixed order from the rear of the host vehicle to the front of the host vehicle to form the reference line.

[0129] S207, acquire at least one to-be-reasoned obstacle needing reasoning, and extract the obstacle feature of the to-be-reasoned obstacle from the global feature information.

[0130] S208, convert the obstacle feature from the Cartesian coordinate system to the Frenet coordinate system according to the reference line, and obtain the first feature information of the to-be-reasoned obstacle in the Frenet coordinate system.

[0131] As an optional implementation, converting the obstacle feature from the Cartesian coordinate system to the Frenet coordinate system according to the reference line to obtain the first feature information of the to-be-reasoned obstacle in the Frenet coordinate system includes:

[0132] Obtain the obstacle historical trajectory coordinates and surrounding lane coordinates of the to-be-reasoned obstacle from the obstacle feature;

[0133] Convert the obstacle historical trajectory coordinates from the Cartesian coordinate system to the Frenet coordinate system based on the reference line to obtain the obstacle historical trajectory Frenet coordinates;

[0134] Convert the surrounding lane coordinates from the Cartesian coordinate system to the Frenet coordinate system based on the reference line to obtain the surrounding lane Frenet coordinates;

[0135] Summarize the obstacle historical trajectory Frenet coordinates and the surrounding lane Frenet coordinates to obtain the first feature information of the to-be-reasoned obstacle in the Frenet coordinate system.

[0136] In this embodiment, the method can convert the historical trajectory coordinates and surrounding lane coordinate points (x, y) in the global feature from the Cartesian coordinate system to the Frenet coordinate system based on the reference line (reference line) in C, and each point is changed from (x, y) to (s, l). In a curved road scene, after Frenet coordinate system conversion, the curved road map coordinate information will become a straight road, and the curved road historical trajectory will become a straight road. Other global features not involving position, such as speed information, direction information, and lane type information, remain unchanged.

[0137] S209, convert the first feature information from the Frenet coordinate system to the agent coordinate system to obtain the second feature information of the to-be-reasoned obstacle in the agent coordinate system.

[0138] As an optional implementation, converting the first feature information from the Frenet coordinate system to the agent coordinate system to obtain the second feature information of the to-be-reasoned obstacle in the agent coordinate system includes:

[0139] extract target trajectory information of the obstacle to be reasoned from the Frenet coordinates of the historical trajectory of the obstacle; wherein the target trajectory information at least includes a current time position of the trajectory, a displacement of adjacent frames of the trajectory, a speed of the obstacle, and a moving direction of the obstacle;

[0140] convert the Frenet coordinates of the historical trajectory of the obstacle from the Frenet coordinate system to an agent coordinate system with the obstacle to be reasoned as the center, to obtain agent coordinate system trajectory features in the agent coordinate system;

[0141] convert the Frenet coordinates of the surrounding lane from the Frenet coordinate system to the agent coordinate system with the obstacle to be reasoned as the center, to obtain lane features in the agent coordinate system;

[0142] aggregate the agent coordinate system trajectory features and the lane features to obtain second feature information of the obstacle to be reasoned in the agent coordinate system.

[0143] As a further optional implementation, converting the Frenet coordinates of the historical trajectory of the obstacle from the Frenet coordinate system to the agent coordinate system with the obstacle to be reasoned as the center, to obtain agent coordinate system trajectory features in the agent coordinate system, includes:

[0144] perform normalization processing on the target trajectory information with the obstacle to be reasoned as the center, to obtain normalized trajectory information;

[0145] obtain an obstacle moving heading of the obstacle to be reasoned;

[0146] perform direction alignment processing on the trajectory directions of all obstacles in the normalized trajectory information based on the obstacle moving heading, to obtain trajectory features;

[0147] add a mask to the trajectory features, to obtain agent coordinate system trajectory features in the agent coordinate system.

[0148] As a further optional implementation, converting the Frenet coordinates of the surrounding lane from the Frenet coordinate system to the agent coordinate system with the obstacle to be reasoned as the center, to obtain lane features in the agent coordinate system, includes:

[0149] extract target map information of the obstacle to be reasoned within a preset range at a current time from the Frenet coordinates of the surrounding lane; wherein the target map information at least includes a lane center line node position, a lane node displacement, a lane connection relationship, a road structure, and a traffic signal light;

[0150] perform normalization processing on the target map information with the obstacle to be reasoned as the center, to obtain normalized map information;

[0151] The lane direction in the normalized map information is directionally aligned based on the obstacle motion heading, to obtain a lane feature in an agent coordinate system.

[0152] In this embodiment, the method can take the origin of the converted coordinates as the starting point of the reference line after Frenet coordinate system conversion, and select an agent. The perception information within a specified range around the agent and the map information constitute a scene. At this time, the scene is represented in a structured manner, and the trajectory information of all obstacles (including the current time position of the trajectory, the displacement of the adjacent frame of the trajectory, the obstacle speed, and the obstacle motion direction) is extracted from the perception data. The model reasoning needs to convert the features to the agent coordinate system. After selecting the agent, each scene coordinate feature is normalized to be centered on the agent (the obstacle to be reasoned) and the trajectory direction and lane direction of all actors (all obstacles in the scene) are aligned with the agent heading. The trajectory length is T, and the insufficient T is filled with 0. In addition, the method can also add a binary 1xT mask to the trajectory feature to represent whether there is a missing value. The map information within a certain range of the current time of the obstacle is extracted from the map data (including lane center line node position, lane node displacement, lane connection relationship, road structure, traffic signal, etc. to constitute map features), and the map position information in the map scene is converted to be centered on the agent and aligned with the agent's heading. These features can provide a basis for subsequent motion prediction.

[0153] S210, according to the pre-constructed trajectory prediction model and the second feature information, the future trajectory information of the to-be-reasoned obstacle is obtained.

[0154] In this embodiment, the method can integrate the above-mentioned features in the agent coordinate system into the trajectory prediction model to perform future trajectory prediction.

[0155] S211, converting the future trajectory information from the agent coordinate system to the Cartesian coordinate system to obtain the target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system.

[0156] As an optional implementation, converting the future trajectory information from the agent coordinate system to the Cartesian coordinate system to obtain the target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system, comprising:

[0157] Converting the future trajectory information from the agent coordinate system to the Frenet coordinate system to obtain the future trajectory information of the to-be-reasoned obstacle in the Frenet coordinate system;

[0158] Convert the future trajectory information from the Frenet coordinate system to the Cartesian coordinate system to obtain the target future trajectory information of the obstacle to be inferred in the Cartesian coordinate system.

[0159] In this embodiment, the method can convert the future trajectory information inferred by the model from the agent coordinate system back to the Frenet coordinate system, and then convert it from the Frenet coordinate system back to the Cartesian coordinate system, thereby obtaining the future trajectory information in the Cartesian global coordinate system.

[0160] Please refer to Figure 3 , Figure 4 and Figure 5 . Among them, Figure 3 shows a schematic diagram of the result of predicting the future trajectory of a continuous curve by a trajectory prediction model, including the lane center point, the historical trajectory of the obstacle in the scene, the historical obstacle trajectory of the agent to be predicted, and the future trajectory predicted by the model; wherein the intersection point of the historical obstacle trajectory of the agent to be predicted and the future trajectory predicted by the model is the current time coordinate position. It should be noted that, Figure 3 the future trajectory predicted in

[0161] Figure 4 To convert the model input features to the Frenet coordinate system and then to the agent coordinate system for trajectory prediction, the future trajectory prediction is correct.

[0162] Figure 5 is the schematic diagram of the result converted from Figure 4 to the Cartesian coordinate system, and the future trajectory prediction is correct.

[0163] As can be seen from Figures 3 to 5 , in more original curve unpredictable scenes, the method can correctly predict the future trajectory of the obstacle.

[0164] In this embodiment, the execution subject of the method can be a computer, a server, or other computing devices, which are not limited in this embodiment.

[0165] In this embodiment, the execution subject of the method can also be a smart phone, a tablet computer, or other smart devices, which are not limited in this embodiment.

[0166] As can be seen, by implementing the obstacle trajectory prediction method described in this embodiment, the curve data can be converted to a straight road through Frenet coordinate system conversion, eliminating the differences in the map in the scene, and then the trajectory prediction model is used for trajectory prediction, which is accurate. At the same time, it also improves the accuracy of curve prediction and improves the experience of autonomous driving.

[0167] Embodiment 3

[0168] Please refer to Figure 6 , Figure 6 A structural schematic diagram of an obstacle trajectory prediction device is provided in this embodiment. As shown in the figure, Figure 6 The obstacle trajectory prediction device comprises:

[0169] The collection unit 310 is configured to collect vehicle perception data of a target vehicle on a curve through a sensor device;

[0170] The feature extraction unit 320 is configured to perform feature extraction on the vehicle perception data to obtain global feature information in a Cartesian coordinate system;

[0171] The determination unit 330 is configured to determine a reference line according to the global feature information;

[0172] The acquisition unit 340 is configured to acquire at least one to-be-predicted obstacle that needs to be predicted, and extract obstacle features of the to-be-predicted obstacle from the global feature information;

[0173] The first coordinate conversion unit 350 is configured to convert the obstacle features from the Cartesian coordinate system to the Frenet coordinate system according to the reference line, to obtain first feature information of the to-be-predicted obstacle in the Frenet coordinate system;

[0174] The second coordinate conversion unit 360 is configured to convert the first feature information from the Frenet coordinate system to the agent coordinate system, to obtain second feature information of the to-be-predicted obstacle in the agent coordinate system;

[0175] The trajectory prediction unit 370 is configured to perform future trajectory prediction of the obstacle according to a pre-constructed trajectory prediction model and the second feature information, to obtain future trajectory information of the to-be-predicted obstacle;

[0176] The third coordinate conversion unit 380 is configured to convert the future trajectory information from the agent coordinate system to the Cartesian coordinate system, to obtain target future trajectory information of the to-be-predicted obstacle in the Cartesian coordinate system.

[0177] In this embodiment, the explanation and description of the obstacle trajectory prediction device can refer to the description in Embodiment 1 or Embodiment 2, and no more details are added in this embodiment.

[0178] It can be seen that the obstacle trajectory prediction device described in this embodiment can convert the curve data into a straight road through Frenet coordinate system conversion, eliminate the difference of the map in the scene, and then perform trajectory prediction through the trajectory prediction model, which is high in accuracy, improves the accuracy of curve prediction, and improves the experience of autonomous driving.

[0179] Embodiment 4

[0180] Please refer toFigure 7 , Figure 7 A structural schematic diagram of an obstacle trajectory prediction device is provided in this embodiment. As shown in the figure, the obstacle trajectory prediction device comprises: Figure 7

[0181] The collection unit 310 is configured to collect vehicle perception data of a target vehicle on a curve through a sensor device;

[0182] The feature extraction unit 320 is configured to perform feature extraction on the vehicle perception data to obtain global feature information in a Cartesian coordinate system;

[0183] The determination unit 330 is configured to determine a reference line according to the global feature information;

[0184] The acquisition unit 340 is configured to acquire at least one to-be-predicted obstacle that needs to be predicted and extract obstacle features of the to-be-predicted obstacle from the global feature information;

[0185] The first coordinate conversion unit 350 is configured to convert the obstacle features from the Cartesian coordinate system to a Frenet coordinate system according to the reference line to obtain first feature information of the to-be-predicted obstacle in the Frenet coordinate system;

[0186] The second coordinate conversion unit 360 is configured to convert the first feature information from the Frenet coordinate system to an agent coordinate system to obtain second feature information of the to-be-predicted obstacle in the agent coordinate system;

[0187] The trajectory prediction unit 370 is configured to perform future trajectory prediction of the obstacle according to a pre-constructed trajectory prediction model and the second feature information to obtain future trajectory information of the to-be-predicted obstacle;

[0188] The third coordinate conversion unit 380 is configured to convert the future trajectory information from the agent coordinate system to the Cartesian coordinate system to obtain target future trajectory information of the to-be-predicted obstacle in the Cartesian coordinate system.

[0189] In this embodiment, the vehicle perception data at least includes vehicle trajectory information and vehicle surrounding map information;

[0190] The global feature information includes historical trajectory information and surrounding map information of all obstacles within a preset range; wherein the historical trajectory information at least includes position information, speed information and motion direction; and the surrounding map information at least includes lane center point position information and lane type.

[0191] As an optional implementation, the determination unit 330 comprises:

[0192] ​The determining sub-unit 331 is configured to determine a scene range in which the target vehicle currently locates and a target lane in which the target vehicle locates according to the global feature information.

[0193] The searching sub-unit 332 is configured to search, in the scene range, other lanes having a predecessor-successor reachable relationship with the target lane by using a depth-first search algorithm.

[0194] The selecting sub-unit 333 is configured to select one predecessor-successor lane from the other lanes as a predecessor-successor reachable lane in the scene range when the other lanes include a plurality of predecessor-successor lanes.

[0195] The generating sub-unit 334 is configured to generate a reference line according to the target lane and the predecessor-successor reachable lane in the scene range.

[0196] As an optional implementation, the generating sub-unit 334 is specifically configured to combine the target lane and the predecessor-successor reachable lane in the scene range in a sequence from behind the target vehicle to in front of the target vehicle to obtain the reference line.

[0197] As an optional implementation, the first coordinate conversion unit 350 includes:

[0198] The obtaining sub-unit 351 is configured to obtain, from the obstacle feature, an obstacle historical trajectory coordinate and a surrounding lane coordinate of the to-be-reasoned obstacle.

[0199] The first conversion sub-unit 352 is configured to convert the obstacle historical trajectory coordinate from a Cartesian coordinate system to a Frenet coordinate system based on the reference line to obtain an obstacle historical trajectory Frenet coordinate.

[0200] The first conversion sub-unit 352 is further configured to convert the surrounding lane coordinate from the Cartesian coordinate system to the Frenet coordinate system based on the reference line to obtain a surrounding lane Frenet coordinate.

[0201] The first summarizing sub-unit 353 is configured to summarize the obstacle historical trajectory Frenet coordinate and the surrounding lane Frenet coordinate to obtain first feature information of the to-be-reasoned obstacle in the Frenet coordinate system.

[0202] As an optional implementation, the second coordinate conversion unit 360 includes:

[0203] The extracting sub-unit 361 is configured to extract target trajectory information of the to-be-reasoned obstacle from the obstacle historical trajectory Frenet coordinate, where the target trajectory information at least includes a trajectory current time position, a trajectory adjacent frame displacement, an obstacle speed, and an obstacle motion direction.

[0204] The second conversion sub-unit 362 is configured to convert the Frenet coordinate of the obstacle history track into an agent coordinate system under the agent coordinate system, to obtain an agent coordinate system track feature under the agent coordinate system, with the to-be-reasoned obstacle as the center.

[0205] The second conversion sub-unit 362 is further configured to convert the surrounding lane Frenet coordinate into the agent coordinate system under the agent coordinate system, to obtain a lane feature under the agent coordinate system, with the to-be-reasoned obstacle as the center.

[0206] The second aggregation sub-unit 363 is configured to aggregate the agent coordinate system track feature and the lane feature, to obtain second feature information of the to-be-reasoned obstacle under the agent coordinate system.

[0207] As an optional implementation, the second conversion sub-unit 362 includes:

[0208] The processing module is configured to perform normalization processing on the target track information with the to-be-reasoned obstacle as the center, to obtain normalized track information.

[0209] The acquisition module is configured to acquire an obstacle motion heading of the to-be-reasoned obstacle.

[0210] The processing module is further configured to perform directional alignment processing on the track direction of all obstacles in the normalized track information based on the obstacle motion heading, to obtain the track feature.

[0211] The adding module is configured to add a mask to the track feature, to obtain the agent coordinate system track feature under the agent coordinate system.

[0212] As an optional implementation, the second conversion sub-unit 362 includes:

[0213] The extraction module is configured to extract target map information of the to-be-reasoned obstacle within a preset range at a current time from the surrounding lane Frenet coordinate; wherein the target map information at least includes a lane center line node position, a lane node displacement, a lane connection relationship, a road structure, and a traffic signal lamp.

[0214] The processing module is further configured to perform normalization processing on the target map information with the to-be-reasoned obstacle as the center, to obtain normalized map information.

[0215] The processing module is further configured to perform directional alignment processing on the lane direction in the normalized map information based on the obstacle motion heading, to obtain the lane feature under the agent coordinate system.

[0216] As an optional implementation, the third coordinate conversion unit 380 is specifically configured to convert the future trajectory information from the agent coordinate system to the Frenet coordinate system to obtain the future trajectory information of the to-be-reasoned obstacle in the Frenet coordinate system.

[0217] The third coordinate conversion unit 380 is further specifically configured to convert the future trajectory information from the Frenet coordinate system to the Cartesian coordinate system to obtain the target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system.

[0218] In this embodiment, the description of the obstacle trajectory prediction device can refer to the description in Embodiment 1 or Embodiment 2, and the description will not be repeated here.

[0219] It can be seen that the obstacle trajectory prediction device described in this embodiment can convert the curved road data into a straight road through Frenet coordinate system conversion, eliminate the differences in the map in the scene, and then perform trajectory prediction through the trajectory prediction model, which is high in accuracy and improves the accuracy of curved road prediction and the experience of autonomous driving.

[0220] The embodiment of the present application provides an electronic device, including a memory and a processor, the memory is used to store a computer program, the processor runs the computer program to make the electronic device execute the obstacle trajectory prediction method in the embodiment 1 or the embodiment 2 of the present application.

[0221] The embodiment of the present application provides a computer readable storage medium, which stores computer program instructions, when the computer program instructions are read and run by a processor, the obstacle trajectory prediction method in the embodiment 1 or the embodiment 2 of the present application is executed.

[0222] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can also be implemented by other means. The apparatus embodiments described above are only illustrative, for example, the flowcharts and block diagrams in the drawings show the possible implementation architecture, function and operation of the apparatus, method and computer program product according to the embodiments of the present application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment or a part of code, which contains one or more executable instructions for implementing the specified logic function. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different order from that shown in the drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and sometimes they can be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0223] In addition, the functional modules in the embodiments of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0224] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0225] The above merely provides an example of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, and thus, once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.

[0226] The above merely provides an example of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, and thus, once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.

[0227] It should be noted that the relational terms herein such as first and second and the like are used solely to distinguish one entity or action from another, without necessarily requiring or implying any actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

Claims

1. A method of obstacle trajectory prediction, the method comprising: receiving a plurality of images of a scene; determining a plurality of obstacle trajectories based on the plurality of images; and outputting the plurality of obstacle trajectories. The method comprises the following steps: Collecting vehicle perception data of a target vehicle on a curve through a sensor device; Extracting features from the vehicle perception data to obtain global feature information in a Cartesian coordinate system; Determining a reference line according to the global feature information; Obtaining at least one to-be-reasoned obstacle that needs to be reasoned, and extracting obstacle features of the to-be-reasoned obstacle from the global feature information; Converting the obstacle features from the Cartesian coordinate system to a Frenet coordinate system according to the reference line to obtain first feature information of the to-be-reasoned obstacle in the Frenet coordinate system; Converting the first feature information from the Frenet coordinate system to an agent coordinate system to obtain second feature information of the to-be-reasoned obstacle in the agent coordinate system; Performing obstacle future trajectory prediction according to a pre-constructed trajectory prediction model and the second feature information to obtain future trajectory information of the to-be-reasoned obstacle; Converting the future trajectory information from the agent coordinate system to the Cartesian coordinate system to obtain target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system.

2. The obstacle trajectory prediction method of claim 1, wherein, The vehicle perception data at least includes vehicle trajectory information and vehicle surrounding map information; The global feature information includes historical trajectory information and surrounding map information of all obstacles within a preset range; wherein the historical trajectory information at least includes position information, speed information and motion direction; and the surrounding map information at least includes lane center point position information and lane type.

3. The obstacle trajectory prediction method of claim 1, wherein, The method for determining the reference line according to the global feature information comprises the following steps: Determining a scene range in which the target vehicle is currently located and a target lane in which the target vehicle is located according to the global feature information; Searching for other lanes having a predecessor-successor reachable relationship with the target lane in the scene range by using a depth-first search algorithm; If the other lanes include multiple predecessor-successor lanes, selecting one of the other lanes as a predecessor-successor reachable lane within the scene; Generating a reference line according to the target lane and the predecessor-successor reachable lane within the scene.

4. The obstacle trajectory prediction method of claim 3, wherein, The method for generating the reference line according to the target lane and the predecessor-successor reachable lane within the scene comprises the following steps: Combining the target lane and the predecessor-successor reachable lane within the scene in the order from behind the target vehicle to in front of the target vehicle to obtain the reference line.

5. The obstacle trajectory prediction method of claim 1, wherein, The method for converting the obstacle features from the Cartesian coordinate system to the Frenet coordinate system according to the reference line to obtain the first feature information of the to-be-reasoned obstacle in the Frenet coordinate system comprises the following steps: Obtaining obstacle historical trajectory coordinates and surrounding lane coordinates of the to-be-reasoned obstacle from the obstacle features; Converting the obstacle historical trajectory coordinates from the Cartesian coordinate system to the Frenet coordinate system based on the reference line to obtain obstacle historical trajectory Frenet coordinates; Converting the surrounding lane coordinates from the Cartesian coordinate system to the Frenet coordinate system based on the reference line to obtain surrounding lane Frenet coordinates; The Frenet coordinates of the obstacle historical trajectory and the Frenet coordinates of the surrounding lane are aggregated to obtain first feature information of the obstacle to be reasoned in the Frenet coordinate system.

6. The obstacle trajectory prediction method of claim 5, wherein, The first feature information is converted from the Frenet coordinate system to an agent coordinate system to obtain second feature information of the obstacle to be reasoned in the agent coordinate system. Target trajectory information of the obstacle to be reasoned is extracted from the Frenet coordinates of the obstacle historical trajectory, wherein the target trajectory information at least includes a trajectory current time position, a trajectory adjacent frame displacement, an obstacle speed, and an obstacle movement direction. The Frenet coordinates of the obstacle historical trajectory are converted from the Frenet coordinate system to the agent coordinate system with the obstacle to be reasoned as the center to obtain agent coordinate system trajectory features in the agent coordinate system. The Frenet coordinates of the surrounding lane are converted from the Frenet coordinate system to the agent coordinate system with the obstacle to be reasoned as the center to obtain lane features in the agent coordinate system. The agent coordinate system trajectory features and the lane features are aggregated to obtain the second feature information of the obstacle to be reasoned in the agent coordinate system.

7. The obstacle trajectory prediction method of claim 6, wherein, The Frenet coordinates of the obstacle historical trajectory are converted from the Frenet coordinate system to the agent coordinate system with the obstacle to be reasoned as the center to obtain agent coordinate system trajectory features in the agent coordinate system, including: The target trajectory information is normalized with the obstacle to be reasoned as the center to obtain normalized trajectory information. An obstacle movement heading of the obstacle to be reasoned is obtained. The trajectory directions of all obstacles in the normalized trajectory information are direction-aligned based on the obstacle movement heading to obtain trajectory features. A mask is added to the trajectory features to obtain the agent coordinate system trajectory features in the agent coordinate system.

8. The obstacle trajectory prediction method of claim 7, wherein, The Frenet coordinates of the surrounding lane are converted from the Frenet coordinate system to the agent coordinate system with the obstacle to be reasoned as the center to obtain lane features in the agent coordinate system, including: Target map information of the obstacle to be reasoned within a preset range at a current time is extracted from the Frenet coordinates of the surrounding lane, wherein the target map information at least includes a lane center line node position, a lane node displacement, a lane connection relationship, a road structure, and a traffic signal lamp. The target map information is normalized with the obstacle to be reasoned as the center to obtain normalized map information. The lane directions in the normalized map information are direction-aligned based on the obstacle movement heading to obtain the lane features in the agent coordinate system.

9. The obstacle trajectory prediction method of claim 1, wherein, The future trajectory information is converted from the agent coordinate system to the Cartesian coordinate system to obtain target future trajectory information of the obstacle to be reasoned in the Cartesian coordinate system, including: convert the future trajectory information from the agent coordinate system to the Frenet coordinate system to obtain the future trajectory information of the to-be-reasoned obstacle in the Frenet coordinate system; convert the future trajectory information from the Frenet coordinate system to the Cartesian coordinate system to obtain target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system.

10. An obstacle trajectory prediction device characterized by comprising: The obstacle trajectory prediction device comprises: a collection unit configured to collect vehicle perception data of a target vehicle on a curve by a sensor device; a feature extraction unit configured to perform feature extraction on the vehicle perception data to obtain global feature information in a Cartesian coordinate system; a determination unit configured to determine a reference line according to the global feature information; an acquisition unit configured to acquire at least one to-be-reasoned obstacle that needs to be reasoned and extract obstacle features of the to-be-reasoned obstacle from the global feature information; a first coordinate conversion unit configured to convert the obstacle features from the Cartesian coordinate system to a Frenet coordinate system according to the reference line to obtain first feature information of the to-be-reasoned obstacle in the Frenet coordinate system; a second coordinate conversion unit configured to convert the first feature information from the Frenet coordinate system to an agent coordinate system to obtain second feature information of the to-be-reasoned obstacle in the agent coordinate system; a trajectory prediction unit configured to perform obstacle future trajectory prediction according to a pre-constructed trajectory prediction model and the second feature information to obtain future trajectory information of the to-be-reasoned obstacle; a third coordinate conversion unit configured to convert the future trajectory information from the agent coordinate system to the Cartesian coordinate system to obtain target future trajectory information of the to-be-reasoned obstacle in the Cartesian coordinate system.

11. An electronic device, comprising: The electronic device comprises a memory and a processor, the memory is configured to store a computer program, and the processor is configured to run the computer program to enable the electronic device to perform the obstacle trajectory prediction method in any one of claims 1 to 9.

12. A readable storage medium, characterized by, The readable storage medium stores computer program instructions, and the computer program instructions are read and run by a processor to perform the obstacle trajectory prediction method in any one of claims 1 to 9.

Images