Trajectory prediction method, device, equipment, vehicle, medium and product

Abstract

The application discloses a trajectory prediction method and device, equipment, vehicle, medium and product. The method comprises the following steps: acquiring motion information of a target intelligent agent at a current collection time, wherein the target intelligent agent refers to a moving object in an environment of a target vehicle; determining a motion label of the target intelligent agent according to the motion information of the target intelligent agent and motion information of the target vehicle, wherein the motion label is used for identifying a motion scene of the target intelligent agent; and performing trajectory prediction on the target intelligent agent according to the motion label of the target intelligent agent and the motion information of the target intelligent agent, so as to obtain a target prediction trajectory of the target intelligent agent at the current collection time, wherein the target prediction trajectory is used for providing a reference for driving path planning of the target vehicle. The accuracy of trajectory prediction of the target intelligent agent can be improved.

Description

Technical Field

[0001] This application belongs to the field of autonomous driving technology, and in particular relates to a trajectory prediction method, device, equipment, vehicle, medium and product. Background Technology

[0002] With the evolution of autonomous driving technology, end-to-end systems have become a hot research topic in the industry due to their advantage of simplifying the perception-decision-planning chain. Target vehicle trajectory prediction, as the core module of the system, directly affects driving safety and decision-making rationality.

[0003] Existing end-to-end systems use deep learning models to capture the motion patterns and intentions of intelligent agents (including target vehicles and pedestrians) and output predicted trajectories accordingly. However, the predicted trajectories have defects. For example, the fitting characteristics of neural networks can easily lead to trajectories that violate physical rules, such as accelerations or positional jumps that exceed the dynamic limits of the target vehicle. Training data noise and insufficient generalization to extreme scenarios can cause trajectories to contain high-frequency jitter or exhibit anomalies such as "passing through walls".

[0004] In other words, when using deep learning models to predict the trajectory of an agent, there may be situations where the predicted trajectory does not match the actual scenario, resulting in low trajectory prediction accuracy. Summary of the Invention

[0005] This application provides a trajectory prediction method, apparatus, device, vehicle, medium, and product that can improve the accuracy of trajectory prediction for intelligent agents.

[0006] In a first aspect, embodiments of this application provide a trajectory prediction method, the method comprising:

[0007] Acquire the motion information of the target intelligent agent at the current acquisition time, wherein the target intelligent agent refers to a moving object in the environment in which the target vehicle is located;

[0008] Based on the motion information of the target intelligent agent and the motion information of the target vehicle, a motion tag for the target intelligent agent is determined, and the motion tag is used to identify the motion scene of the target intelligent agent;

[0009] Based on the motion tag and motion information of the target intelligent agent, trajectory prediction is performed on the target intelligent agent to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time. The target predicted trajectory is used to provide a reference for the target vehicle to plan its driving path.

[0010] Secondly, embodiments of this application provide a trajectory prediction device, the device comprising:

[0011] The acquisition module is used to acquire the motion information of the target intelligent agent at the current acquisition time. The target intelligent agent refers to the moving object in the environment where the target vehicle is located.

[0012] The determination module is used to determine the motion tag of the target intelligent agent based on the motion information of the target intelligent agent and the motion information of the target vehicle, wherein the motion tag is used to identify the motion scene of the target intelligent agent;

[0013] The prediction module is used to predict the trajectory of the target intelligent agent based on the motion tag and motion information of the target intelligent agent, and obtain the target predicted trajectory of the target intelligent agent at the current acquisition time. The target predicted trajectory is used to provide a reference for the target vehicle to plan the driving path.

[0014] Thirdly, embodiments of this application provide an electronic device, including: a processor and a memory storing computer program instructions; the processor executes the computer program instructions to implement the trajectory prediction method as described in the first aspect.

[0015] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the trajectory prediction method as described in the first aspect.

[0016] Fifthly, embodiments of this application provide a computer program product in which instructions, when executed by a processor of an electronic device, cause the electronic device to perform the trajectory prediction method as described in the first aspect.

[0017] In this embodiment, the motion tags of the target intelligent agent can identify different motion scenarios of the target intelligent agent, thereby performing trajectory prediction for the motion scenarios of the target intelligent agent, avoiding the predicted trajectory from not conforming to the actual scenario, and improving the accuracy of the predicted trajectory. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic flowchart of the trajectory prediction method provided in the embodiments of this application;

[0020] Figure 2 This is a schematic diagram of the trajectory prediction device provided in the embodiments of this application;

[0021] Figure 3This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0022] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0023] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0024] In all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. Additionally, when embodiments of this application require access to sensitive personal information, separate permission or consent from the user is obtained through pop-ups or redirects to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments obtained.

[0025] To address the problems of the prior art, embodiments of this application provide a trajectory prediction method, apparatus, device, vehicle, medium, and product. The trajectory prediction method provided in this application embodiment is described below first.

[0026] Figure 1 A schematic flowchart of a trajectory prediction method provided in one embodiment of this application is shown. Figure 1As shown, the trajectory prediction method provided in this application embodiment is applied to an electronic device, which can be installed on the target vehicle or outside the target vehicle, and the electronic device is communicatively connected to the target vehicle; the trajectory prediction method includes the following steps 101-103:

[0027] Step 101: Obtain the motion information of the target intelligent agent at the current acquisition time. The target intelligent agent refers to the moving object in the environment where the target vehicle is located.

[0028] The moving objects in the target vehicle's environment can refer to moving vehicles or pedestrians. The target agent's motion information can be obtained through sensors installed on the target vehicle, or it can be calculated based on information collected by the sensors on the target vehicle. The sensors collect data at regular intervals; the collection time refers to the moment the data is collected.

[0029] Step 102: Based on the motion information of the target intelligent agent and the motion information of the target vehicle, determine the motion tag of the target intelligent agent. The motion tag is used to identify the motion scene of the target intelligent agent.

[0030] The motion tags include crossing, going straight, or changing lanes. The motion information of the target agent may include the target agent's current position, speed, acceleration, heading angle, etc.; the motion information of the target vehicle may include the target vehicle's current position, speed, acceleration, heading angle, etc.

[0031] Based on the motion information of the target agent and the operation information of the target vehicle, it can be determined whether the target agent's motion tag is crossing, going straight, or changing lanes. For example, the motion angle of the target agent relative to the target vehicle can be used to determine whether the target agent is crossing; the speed of the target agent relative to the lane where the target vehicle is located can be used to determine whether the target agent is going straight or changing lanes. There are no restrictions on this.

[0032] Step 103: Based on the motion tag and motion information of the target intelligent agent, perform trajectory prediction on the target intelligent agent to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time. The target predicted trajectory is used to provide a reference for the target vehicle to plan its driving path.

[0033] Based on the motion tags of the target agent, a corresponding method is used to predict the trajectory of the target agent according to its motion information. For example, if the motion tag of the target agent is "crossing," in order to avoid the predicted trajectory of the target agent output by the end-to-end system in related technologies from crossing the curb, the predicted trajectory of the target agent can be adjusted to avoid the predicted trajectory crossing the curb and not matching the real scene.

[0034] For example, if the target agent's motion label is straight, the positions of multiple trajectory points can be calculated based on the target agent's longitudinal velocity and longitudinal acceleration, thereby determining the predicted trajectory of the target agent.

[0035] For example, if the target agent's motion tag is lane change, the longitudinal coordinates of multiple trajectory points can be calculated based on the target agent's longitudinal velocity and longitudinal acceleration; the lateral coordinates of multiple trajectory points can be calculated based on the target agent's lateral velocity and lateral acceleration; then, the longitudinal and lateral coordinates of the same trajectory point can be calculated to obtain the coordinates of that trajectory point. By calculating the positions of multiple trajectory points in the above manner, the target predicted trajectory of the target agent at the current acquisition time can be obtained.

[0036] After obtaining the target agent's predicted trajectory, the target vehicle can be provided with a reference for driving path planning based on the target agent's predicted trajectory, thereby avoiding collisions between the target vehicle and the target agent and improving the driving safety of the target vehicle.

[0037] The trajectory prediction method in this embodiment can be applied to the field of autonomous driving, where the target vehicle can be an autonomous vehicle. Alternatively, the target vehicle in this embodiment can also be a non-autonomous vehicle. In this case, the purpose of planning the driving path for the target vehicle based on the target agent's predicted trajectory can be to assist the driver in driving the vehicle and avoid collisions.

[0038] In this embodiment, the motion information of the target intelligent agent at the current acquisition time is obtained. The target intelligent agent refers to a moving object in the environment where the target vehicle is located. Based on the motion information of the target intelligent agent and the motion information of the target vehicle, a motion tag is determined for the target intelligent agent. The motion tag is used to identify the motion scene of the target intelligent agent. Based on the motion tag and the motion information of the target intelligent agent, trajectory prediction is performed on the target intelligent agent to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time. The target predicted trajectory is used to provide a reference for the target vehicle's driving path planning. Through the above method, different motion scenes of the target intelligent agent can be identified based on the motion tag of the target intelligent agent, thereby performing trajectory prediction based on the motion scene of the target intelligent agent, avoiding the predicted trajectory from not conforming to the actual scene, and improving the accuracy of the target predicted trajectory.

[0039] In the above-described step 101, before acquiring the motion information of the target agent at the current acquisition time, the method further includes:

[0040] Acquire multiple moving objects in the environment where the target vehicle is located at the current acquisition time;

[0041] For any one of multiple moving objects, if the moving object satisfies the third preset condition, then the moving object is taken as the target intelligent agent;

[0042] The third preset condition includes:

[0043] The moving object is located within the target area, and the target area is determined based on the position of the target vehicle;

[0044] The moving object is located within the curb of the environment in which the target vehicle is situated.

[0045] For example, using the center of the target vehicle as a reference, the forward scanning distance is set to 100m, the backward scanning distance to 30m, the left scanning distance to 12m, and the right scanning distance to 12m. This rectangle, 130m long and 24m wide, is the target area. The target area can also be set to other shapes or sizes according to actual conditions, and is not limited here. Specifically, the coordinates of the target vehicle and the moving object in the world coordinate system can be obtained. The position of the target vehicle can be the location of the center of its rear axle, and the position of the moving object can be the coordinates of its center. Through coordinate transformation, the position of the moving object relative to the target vehicle can be obtained. The target area can also be represented using coordinates relative to the target vehicle. If the coordinates of the moving object fall within the target area, then the moving object is determined to be within the target area.

[0046] The trajectory of a moving object falling within the target area has a significant impact on the driving path planning of the target vehicle, while the trajectory of a moving object falling outside the target area has a smaller impact. To improve processing efficiency, in this embodiment, multiple moving objects are screened, and those falling within the target area are selected. Furthermore, even if a moving object located outside the curb is within the target area, its trajectory will not intersect with the target vehicle due to the curb's obstruction; that is, its trajectory has a smaller impact on the target vehicle's driving path planning. Based on the above objectives, a third preset condition is used to screen multiple moving objects, and those meeting the third preset condition are selected as the target intelligent agent.

[0047] The inside of the curb can be understood as the side of the lane where the target vehicle is located. For example, if there are lanes on both sides of the curb, namely the forward lane and the reverse lane, and the target vehicle is traveling in the forward lane, then the forward lane is inside the curb. Or, for example, if there is a lane on one side of the curb and a green belt on the other side, and the target vehicle is traveling in the lane, then the side where the lane is located is inside the curb, and the side where the green belt is located is outside the curb.

[0048] In this embodiment, multiple moving objects in the environment where the target vehicle is located are filtered using a third preset condition. Subsequently, it is only necessary to determine the target prediction trajectory of the filtered target intelligent agent, which can improve data processing efficiency and reduce computing power consumption.

[0049] In another embodiment of this application, determining the motion tag of the target intelligent agent based on the motion information of the target intelligent agent and the motion information of the target vehicle includes:

[0050] If the absolute value of the difference between the heading angle of the target agent and the heading angle of the target vehicle is within a preset angle range, then the motion tag of the target agent is determined to be traversing.

[0051] For example, the preset angle range can be from 45° to 135°, and the preset angle range can also be adjusted according to the actual situation, which is not limited here. If the absolute value of the difference between the heading angles of the target agent and the target vehicle is within the preset angle range, it is considered that the target agent is traversing the target vehicle, and the motion tag of the target agent is set to traversing.

[0052] In this embodiment, the target intelligent agent is determined to be crossing by the difference between the heading angle of the target intelligent agent and the heading angle of the target vehicle falling within a preset angle range, thereby determining the motion tag of the target intelligent agent. The determination method is simple and efficient, does not require a complex calculation process, can save computing power, and improve the efficiency of motion tag determination.

[0053] In another embodiment of this application, determining the motion tag of the target intelligent agent based on the motion information of the target intelligent agent and the motion information of the target vehicle includes:

[0054] If the lateral speed of the target intelligent agent meets the first preset condition, then the motion tag of the target intelligent agent is determined to be lane change, and the lateral speed refers to the moving speed of the target intelligent agent relative to the lane where the target vehicle is located.

[0055] The first preset condition includes one of the following:

[0056] The lateral speed is less than a first speed threshold, and the target agent presses onto the target lane line, where the target lane line is the boundary line of the lane where the target agent is located;

[0057] The lateral speed is less than the first speed threshold, the target agent does not cross the target lane line, and the closest distance between the target agent and the target lane line is less than the first distance threshold, wherein the first distance threshold is a preset value, or the first distance threshold is determined based on the longitudinal distance between the target agent and the target vehicle;

[0058] The lateral velocity is greater than or equal to the first velocity threshold.

[0059] In the above, lateral speed can specifically refer to the movement speed of the target intelligent agent relative to the centerline of the lane where the target vehicle is located.

[0060] The target lane line refers to the boundary line of the lane where the target agent is located in the direction of the target agent's lateral movement. For example, if the lane where the target agent is located includes boundary line 1 (left side) and boundary line 2 (right side), and the direction of the target agent's lateral movement is towards the left, then boundary line 1 is the target lane line.

[0061] The first speed threshold can be set according to the actual situation, such as 0.2m / s, and is not limited here.

[0062] The target agent has a geometric shape. For example, if the target agent is a vehicle, the vehicle's geometry is similar to a rectangle. Using the four corner points of the vehicle as references, the distances between these four corner points and the target lane line are calculated. If the distance between the left front corner point of the vehicle and the target lane line is closest, and this distance is less than a first distance threshold, then the vehicle is considered to be changing lanes. The first distance threshold can be set according to the actual situation. The first distance threshold can be a preset value, i.e., used in all situations; or the first distance threshold can be determined flexibly based on the longitudinal distance between the target agent and the target vehicle, without using a preset value. The first distance threshold is negatively correlated with the longitudinal distance; that is, the smaller the longitudinal distance, the larger the first distance threshold; and vice versa.

[0063] In this embodiment, if the lateral speed of the target intelligent agent meets the first preset condition, the motion tag of the target intelligent agent is determined to be a lane change. In this way, it is possible to determine whether the target intelligent agent is changing lanes, providing a basis for subsequent trajectory prediction based on the motion tag of the target intelligent agent.

[0064] In one embodiment of this application, trajectory prediction is performed on the target intelligent agent based on its motion tag and motion information to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time, including:

[0065] If the motion tag of the target intelligent agent is lane change, then based on the motion information of the target intelligent agent, determine the longitudinal information corresponding to multiple trajectory points of the target intelligent agent in a future preset time period;

[0066] The lateral distance from the target agent to the first reference line is used as the initial lateral distance. The first reference line is the reference line of the lane to which the target agent is about to change lanes.

[0067] Based on the initial lateral distance, determine the lateral information corresponding to multiple trajectory points of the target agent within a future preset time period;

[0068] Based on the longitudinal and lateral information of multiple trajectory points, the target predicted trajectory of the target agent at the current acquisition time is obtained.

[0069] In this embodiment, the motion information of the target intelligent agent can refer to the current longitudinal velocity and the current longitudinal acceleration.

[0070] By using the above method, when the target agent's motion tag is lane change, the trajectory of the target agent within a preset time period can be predicted based on the target agent's motion information and the lateral distance from the target agent to the first reference line, thereby improving the accuracy of trajectory prediction.

[0071] Specifically, the future preset time period can refer to T seconds in the future. T can be set according to the actual situation. For example, if T is 8 seconds and the number of trajectory points is 80, then the time interval between each trajectory point is 0.1 seconds.

[0072] For example, when the motion label of the target agent is lane change, the longitudinal information corresponding to multiple trajectory points can be determined according to the following process:

[0073] Based on the current longitudinal velocity and current longitudinal acceleration of the target intelligent agent, the longitudinal velocity corresponding to each of the multiple trajectory points of the target intelligent agent within a future preset time period is calculated.

[0074] The longitudinal coordinate position of each trajectory point is obtained by calculating the longitudinal velocity and motion duration of the target agent at each trajectory point. The longitudinal information of the trajectory point includes the longitudinal coordinate position of the trajectory point.

[0075] Specifically, the motion in the longitudinal direction of the target agent can be regarded as uniform acceleration, that is, the acceleration remains constant. The current longitudinal velocity is v0, and the current longitudinal acceleration is a. The longitudinal velocity V in the t-th second is calculated using the following formula (1). t :

[0076] V t = V0 + at(1)

[0077] The value of t ranges from 0 to T.

[0078] The longitudinal distance of the target agent at each trajectory point can be obtained by integrating the velocity over the corresponding t (i.e., motion time) at the trajectory point. This longitudinal distance is based on the first trajectory point. Then, using the longitudinal coordinate of the first trajectory point as a reference, the longitudinal coordinate position of each trajectory point can be determined. This coordinate position can be a coordinate in the world coordinate system or another coordinate system, as long as it is in the same coordinate system as the target vehicle.

[0079] In addition to the longitudinal coordinates of the trajectory points, the longitudinal information corresponding to the trajectory points can also include the longitudinal velocity, longitudinal acceleration, and so on, providing more information for the driving path planning of the target vehicle.

[0080] In this embodiment, the longitudinal coordinate position of each trajectory point is determined in the manner described above in order to predict the trajectory of the target intelligent agent.

[0081] If the target vehicle's environment includes four lanes, designated lanes 1 to 4, and the target agent is about to change lanes to lane 2, then the first reference line is the reference line for lane 2. The lane reference line can be understood as the lane's centerline, which divides the lane into two equal parts, each including a lane boundary line. The lane the target agent is about to change to can be determined based on its lateral velocity. For example, if the target agent's lateral velocity is towards its right, and there are lanes 1 and 2 to its right, the lane whose reference line is closest to the target agent is selected; this lane is the lane the target agent is about to change to.

[0082] Based on the initial lateral distance, the lateral information corresponding to multiple trajectory points is determined. For example, the initial lateral distance is used as the lateral distance corresponding to the first trajectory point among the multiple trajectory points. For other trajectory points besides the first trajectory point, a distance value can be added based on the lateral distance corresponding to the previous trajectory point. This added distance value can be determined based on the lateral velocity of the target agent.

[0083] For example, when the motion label of the target agent is lane change, the lateral information corresponding to multiple trajectory points can be determined according to the following process:

[0084] The initial lateral distance is taken as the lateral distance corresponding to the first trajectory point among multiple trajectory points;

[0085] For multiple trajectory points other than the first trajectory point, the lateral distance corresponding to the previous trajectory point of the trajectory point is multiplied by a preset coefficient to obtain the lateral distance corresponding to the trajectory point. The preset coefficient is used to characterize the speed of lane change.

[0086] Based on the current position of the target agent and the lateral distance corresponding to each trajectory point, the lateral coordinate position corresponding to each trajectory point is determined, wherein the lateral information corresponding to the trajectory point includes the lateral coordinate position corresponding to the trajectory point.

[0087] Specifically, the preset coefficient is negatively correlated with the lateral velocity of the target intelligent agent, where the lateral velocity refers to the moving speed of the target intelligent agent relative to the lane where the target vehicle is located.

[0088] For example, when the lateral speed is less than 0.2 m / s, the target agent is considered to be changing lanes slowly, and the corresponding preset coefficient can be 0.98; when the lateral speed is between 0.2 m / s and 0.5 m / s, the target agent is considered to be changing lanes at a medium speed, and the corresponding preset coefficient can be 0.96; when the lateral speed is greater than 0.5 m / s, the target agent is considered to be changing lanes quickly, and the corresponding preset coefficient can be 0.94.

[0089] The greater the lateral distance corresponding to the trajectory point, the farther the distance between the trajectory point and the reference line of the target lane (i.e., the lane that the target agent is about to change lanes to).

[0090] Using the coordinates of the target agent's current position (in the same coordinate system as the target vehicle) as a reference, the lateral coordinate position of each trajectory point can be determined based on the lateral distance corresponding to each trajectory point.

[0091] In addition to the lateral coordinates of the trajectory points, the lateral information corresponding to each trajectory point can also include lateral velocity, lateral acceleration, and other parameters, providing more information for the target vehicle's driving path planning. After determining the lateral distance, the lateral velocity and lateral acceleration of each trajectory point can be obtained based on the movement duration of the target agent at each trajectory point.

[0092] In this embodiment, the lateral coordinate position of each trajectory point is determined in the manner described above in order to predict the trajectory of the target intelligent agent.

[0093] After obtaining the longitudinal and lateral information of each trajectory point, the longitudinal and lateral information are fused. For example, for the same trajectory point, the lateral and longitudinal coordinates are integrated to determine the coordinates of the trajectory point, thus determining its specific location. Similarly, the velocity of the trajectory point is calculated based on its lateral and longitudinal velocities; similarly, the acceleration of the trajectory point can be calculated from its lateral and longitudinal accelerations. It should be noted that after determining the position of each trajectory point, the heading angle corresponding to the current trajectory point can be calculated using the current trajectory point's position and the position of the adjacent previous trajectory point. The target prediction trajectory can include the coordinates of multiple trajectory points, as well as information such as the velocity, acceleration, and heading angle corresponding to each trajectory point.

[0094] In another embodiment of this application, determining the motion tag of the target intelligent agent based on the motion information of the target intelligent agent and the motion information of the target vehicle includes:

[0095] If the lateral speed of the target intelligent agent meets the second preset condition, then the motion tag of the target intelligent agent is determined to be straight, and the lateral speed refers to the moving speed of the target intelligent agent relative to the lane where the target vehicle is located;

[0096] The second preset condition includes:

[0097] The lateral speed is less than a first speed threshold, the target agent does not cross the target lane line, and the closest distance between the target agent and the target lane line is greater than or equal to a first distance threshold, wherein the first distance threshold is a preset value, or the first distance threshold is determined based on the longitudinal distance between the target agent and the target vehicle, and the target lane line is the boundary line of the lane where the target agent is located.

[0098] In the above, lateral speed can specifically refer to the movement speed of the target intelligent agent relative to the centerline of the lane where the target vehicle is located.

[0099] The target lane line refers to the boundary line of the lane where the target agent is located in the direction of the target agent's lateral movement. For example, if the lane where the target agent is located includes boundary line 1 (left side) and boundary line 2 (right side), and the direction of the target agent's lateral movement is towards the left, then boundary line 1 is the target lane line.

[0100] The first speed threshold can be set according to the actual situation, such as 0.2m / s, and is not limited here.

[0101] The target agent has a geometric shape. For example, if the target agent is a vehicle, the vehicle's geometry is similar to a rectangle. Using the four corner points of the vehicle as references, the distances between these four corner points and the target lane line are calculated. If the distance between the left front corner point of the vehicle and the target lane line is closest, and this distance is less than a first distance threshold, then the vehicle is considered to be changing lanes. The first distance threshold can be set according to the actual situation. The first distance threshold can be a preset value, i.e., used in all situations; or the first distance threshold can be determined flexibly based on the longitudinal distance between the target agent and the target vehicle, without using a preset value. The first distance threshold is negatively correlated with the longitudinal distance; that is, the smaller the longitudinal distance, the larger the first distance threshold; and vice versa.

[0102] In this embodiment, if the lateral velocity of the target intelligent agent meets the second preset condition, the motion tag of the target intelligent agent is determined to be straight. In this way, it is possible to determine whether the target intelligent agent is moving straight, providing a basis for subsequent trajectory prediction based on the motion tag of the target intelligent agent.

[0103] In another embodiment of this application, based on the motion tag and motion information of the target intelligent agent, trajectory prediction is performed on the target intelligent agent to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time, including:

[0104] If the motion label of the target intelligent agent is straight, and the lateral distance from the target intelligent agent to the second reference line is less than a preset distance threshold, then based on the motion information of the target intelligent agent, the longitudinal information corresponding to multiple trajectory points of the target intelligent agent in a future preset time period is determined, and the second reference line is the reference line of the lane closest to the target intelligent agent;

[0105] The lateral distance from the target agent to the second reference line is taken as the initial lateral distance;

[0106] Based on the initial lateral distance, determine the lateral information corresponding to multiple trajectory points of the target agent within a future preset time period;

[0107] Based on the longitudinal and lateral information of multiple trajectory points, the target predicted trajectory of the target agent at the current acquisition time is obtained.

[0108] Specifically, when the target agent's motion tag is "going straight," the target agent's motion information can refer to its current position. The lane in which the target agent is located is determined based on its current position. The lane in which the target agent is located is the lane closest to the target agent. The lane reference line can be understood as the lane's centerline (also called the center line), which divides the lane into two equal parts, each including a lane boundary line.

[0109] The preset distance threshold can be set according to the actual situation, for example, it can be set to 1.75 meters, and there is no limit here.

[0110] In this embodiment, the motion information of the target intelligent agent can refer to the current longitudinal velocity and the current longitudinal acceleration.

[0111] By using the above method, when the target agent's motion label is straight, the trajectory of the target agent within a preset time period can be predicted based on the target agent's motion information and the lateral distance from the target agent to the second reference line, thereby improving the accuracy of trajectory prediction.

[0112] Specifically, the future preset time period can refer to T seconds in the future. T can be set according to the actual situation. For example, if T is 8 seconds and the number of trajectory points is 80, then the time interval between each trajectory point is 0.1 seconds.

[0113] For example, when the motion label of the target agent is straight, the longitudinal information corresponding to multiple trajectory points can be determined according to the following process:

[0114] Based on the current longitudinal velocity and current longitudinal acceleration of the target intelligent agent, the longitudinal velocity corresponding to each of the multiple trajectory points of the target intelligent agent within a future preset time period is calculated.

[0115] The longitudinal coordinate position of each trajectory point is obtained by calculating the longitudinal velocity and motion duration of the target agent at each trajectory point. The longitudinal information of the trajectory point includes the longitudinal coordinate position of the trajectory point.

[0116] Specifically, the motion in the longitudinal direction of the target agent can be regarded as uniform acceleration, that is, the acceleration remains constant. The current longitudinal velocity is v0, and the current longitudinal acceleration is a. The longitudinal velocity V in the t-th second is calculated using equation (1). t :

[0117] The longitudinal distance of the target agent at each trajectory point can be obtained by integrating the velocity over the corresponding t (i.e., motion time) at the trajectory point. This longitudinal distance is based on the first trajectory point. Then, using the longitudinal coordinate of the first trajectory point as a reference, the longitudinal coordinate position of each trajectory point can be determined. This coordinate position can be a coordinate in the world coordinate system or another coordinate system, as long as it is in the same coordinate system as the target vehicle.

[0118] In addition to the longitudinal coordinates of the trajectory points, the longitudinal information corresponding to the trajectory points can also include the longitudinal velocity, longitudinal acceleration, and so on, providing more information for the driving path planning of the target vehicle.

[0119] In this embodiment, the longitudinal coordinate position of each trajectory point is determined in the manner described above in order to predict the trajectory of the target intelligent agent.

[0120] Based on the initial lateral distance, the lateral information corresponding to multiple trajectory points is determined. For example, the initial lateral distance is used as the lateral distance corresponding to the first trajectory point among the multiple trajectory points. For other trajectory points besides the first trajectory point, a distance value can be added based on the lateral distance corresponding to the previous trajectory point. This added distance value can be determined based on the lateral velocity of the target agent.

[0121] For example, when the motion label of the target agent is straight, the lateral information corresponding to multiple trajectory points can be determined according to the following process:

[0122] The initial lateral distance is taken as the lateral distance corresponding to the first trajectory point among multiple trajectory points;

[0123] For multiple trajectory points other than the first trajectory point, the lateral distance corresponding to the previous trajectory point of the trajectory point is multiplied by a preset coefficient to obtain the lateral distance corresponding to the trajectory point. The preset coefficient is used to characterize the speed of lane change.

[0124] Based on the current position of the target agent and the lateral distance corresponding to each trajectory point, the lateral coordinate position corresponding to each trajectory point is determined, wherein the lateral information corresponding to the trajectory point includes the lateral coordinate position corresponding to the trajectory point.

[0125] For example, if the target agent's motion label is "going straight", the corresponding preset coefficient can be 1, or a value close to 1, such as 0.99, which is not limited here. In this way, the lateral distance corresponding to each trajectory point does not change much, that is, the distance between the trajectory point and the reference line of the lane where the target agent is located does not change much, which matches the target agent being in a straight state.

[0126] Using the lateral coordinate position of the second reference line (in the same coordinate system as the target vehicle) as a reference, the lateral coordinate position of each trajectory point can be determined based on the lateral distance corresponding to each trajectory point.

[0127] In addition to the lateral coordinates of the trajectory points, the lateral information corresponding to each trajectory point can also include lateral velocity, lateral acceleration, and other parameters, providing more information for the target vehicle's driving path planning. After determining the lateral distance, the lateral velocity and lateral acceleration of each trajectory point can be obtained based on the movement duration of the target agent at each trajectory point.

[0128] In this embodiment, the lateral coordinate position of each trajectory point is determined in the manner described above in order to predict the trajectory of the target intelligent agent.

[0129] After obtaining the longitudinal and lateral information of each trajectory point, the longitudinal and lateral information are fused. For example, for the same trajectory point, the lateral and longitudinal coordinates are integrated to determine the coordinates of the trajectory point, thus determining its specific location. Similarly, the velocity of the trajectory point is calculated based on its lateral and longitudinal velocities; similarly, the acceleration of the trajectory point can be calculated from its lateral and longitudinal accelerations. It should be noted that after determining the position of each trajectory point, the heading angle corresponding to the current trajectory point can be calculated using the current trajectory point's position and the position of the adjacent previous trajectory point. The target prediction trajectory can include the coordinates of multiple trajectory points, as well as information such as the velocity, acceleration, and heading angle corresponding to each trajectory point.

[0130] In one embodiment of this application, trajectory prediction is performed on the target intelligent agent based on its motion tag and motion information to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time, including:

[0131] If the motion tag of the target agent is traversing, and the existing predicted trajectory of the target agent intersects with the curb in the environment where the target vehicle is located, then the motion information of the first trajectory point that crosses the curb in the existing predicted trajectory is adjusted according to a preset strategy to obtain the target predicted trajectory of the target agent at the current collection time. The preset strategy is used to adjust the first trajectory point so that it does not cross the curb while keeping the number of trajectory points in the existing predicted trajectory unchanged.

[0132] In this embodiment, the existing predicted trajectory of the target agent can be the trajectory output by the deep learning model of an end-to-end system in related technologies. This trajectory may have defects; for example, the existing predicted trajectory of the target agent may cross the curb, which is inconsistent with the actual scenario. This embodiment provides a processing method to solve this defect. A preset strategy is used to adjust the motion information of the first trajectory point that crosses the curb. While keeping the number of trajectory points in the existing predicted trajectory unchanged, the first trajectory point is adjusted to not cross the curb. For example, there are 80 existing predicted trajectory points, of which 50 trajectory points are located on the left side of the curb and 30 trajectory points are located on the right side of the curb. The target agent is currently located on the left side of the curb. Thus, it can be determined that the 30 trajectory points located on the right side of the curb have crossed the curb. For ease of description, these 30 trajectory points are called the first trajectory points. The motion information of these 30 trajectory points is adjusted by a preset strategy, for example, adjusting the position, velocity, acceleration, etc. of each trajectory point, so that these 30 trajectory points are located on the left side of the curb, that is, they do not cross the curb.

[0133] The above methods can be used to correct the existing predicted trajectory of the target agent, making it consistent with the actual scene and improving the accuracy of trajectory prediction for the target agent in traversing situations.

[0134] The preset strategy mentioned above includes at least one of the following:

[0135] Set the position corresponding to the first trajectory point as the position corresponding to the second trajectory point, where the second trajectory point is the last trajectory point among the trajectory points that did not cross the curb;

[0136] Set the heading angle corresponding to the first trajectory point as the heading angle corresponding to the second trajectory point;

[0137] Set the velocity corresponding to the first trajectory point to a first preset value;

[0138] Set the acceleration corresponding to the first trajectory point to a second preset value.

[0139] For example, given 80 predicted trajectory points, 50 of which are located on the left side of the curb and 30 on the right side, and the target agent is currently on the left side of the curb, the 30 trajectory points on the right side are determined to cross the curb; each of these points is called a first trajectory point. The last of the 50 trajectory points on the left side is called a second trajectory point. The positions corresponding to the 30 first trajectory points are all set to the positions corresponding to the second trajectory points; that is, the target agent's position remains fixed at the second trajectory point.

[0140] Both the first and second preset values ​​can be set to 0. In this case, the target agent remains stationary at the second trajectory point. The first and second preset values ​​can also be set to other values, which are not limited here.

[0141] In another embodiment of this application, after predicting the trajectory of the target intelligent agent based on its motion tag and motion information to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time, the method further includes:

[0142] The target predicted trajectory of the target agent at the current acquisition time is taken as the initial predicted trajectory of the target agent at the current acquisition time.

[0143] Based on the initial predicted trajectory of the target agent at the current acquisition time and the target predicted trajectory of the target agent at the previous acquisition time, the first predicted trajectory of the target agent is obtained.

[0144] The target intelligent agent's predicted trajectory at the current acquisition time is updated to the first predicted trajectory.

[0145] In this embodiment, after obtaining the target predicted trajectory of the target agent at the current acquisition time, a smoothing process is performed using the target predicted trajectory obtained at the previous acquisition time. For example, the average of the two target predicted trajectories is calculated to obtain the first predicted trajectory, and the target predicted trajectory of the target agent at the current acquisition time is updated to the first predicted trajectory. In other words, the first predicted trajectory becomes the target predicted trajectory of the target agent at the current acquisition time. This smoothing process reduces the jitter of the target predicted trajectory and improves its stability.

[0146] In another embodiment of this application, the first predicted trajectory of the target agent is obtained based on the initial predicted trajectory of the target agent at the current acquisition time and the target predicted trajectory of the target agent at the previous acquisition time at the current acquisition time, including:

[0147] If the initial predicted trajectory and the target predicted trajectory at the previous acquisition time have the same number of trajectory points, then obtain the first weight corresponding to the current acquisition time and the second weight corresponding to the previous acquisition time.

[0148] For the i-th trajectory point in the initial predicted trajectory and the i-th trajectory point in the target predicted trajectory at the previous acquisition time, the horizontal and vertical coordinate positions of the two i-th trajectory points are weighted according to the first weight and the second weight respectively to obtain the target horizontal and vertical coordinate positions of the i-th trajectory point, where i is an integer less than or equal to N, and N is the number of trajectory points in the initial predicted trajectory.

[0149] Based on the target lateral coordinate position and the target longitudinal coordinate position of the i-th trajectory point, determine the predicted coordinate position of the i-th trajectory point;

[0150] The first predicted trajectory of the target agent is obtained based on the predicted coordinate positions of the N trajectory points.

[0151] It should be noted that if the initial predicted trajectory and the target predicted trajectory at the previous acquisition time have different numbers of trajectory points, then they are not within the scope of discussion of the embodiments of this application.

[0152] The first and second weights can be set according to the actual situation, and there is no limitation here. Just make sure that the sum of the first and second weights is 1.

[0153] For example, the lateral coordinate position of the first trajectory point of the initial predicted trajectory and the lateral coordinate position of the first trajectory point of the target predicted trajectory at the previous acquisition time are obtained. The two lateral coordinate positions are weighted and summed according to the first weight and the second weight to obtain the target lateral coordinate position of the first trajectory point.

[0154] Similarly, the longitudinal coordinate position of the first trajectory point of the initial predicted trajectory and the longitudinal coordinate position of the first trajectory point of the target predicted trajectory at the previous acquisition time are obtained. The two longitudinal coordinate positions are weighted and summed according to the first weight and the second weight to obtain the target longitudinal coordinate position of the first trajectory point.

[0155] Based on the target's horizontal and vertical coordinates of the first trajectory point, the predicted coordinates of the i-th trajectory point are obtained. For example, if the target's horizontal coordinate is x1 and its vertical coordinate is y1, then the predicted coordinates are (x1, y1).

[0156] By processing each trajectory point in the above manner, the predicted coordinate position of each trajectory point can be obtained.

[0157] In addition to processing the coordinates of the trajectory points, the velocities and accelerations corresponding to the trajectory points can be smoothed using a first weight and a second weight. Considering the periodicity of angles, the two heading angles corresponding to the i-th trajectory point are averaged.

[0158] After determining the predicted coordinates of each trajectory point, the heading angle can be determined based on these coordinates. The heading angle for the current trajectory point is determined by comparing its predicted coordinates with those of the adjacent previous trajectory point. When the displacement between the predicted coordinates of these two trajectory points is too small (<0.5 meters), or the calculated heading angle changes too much (>0.1 radians), the heading angle for the current trajectory point is taken from the previous trajectory point to reduce jitter in the predicted trajectory and improve its stability.

[0159] In this embodiment, the initial predicted trajectory at the current acquisition time is smoothed using the target predicted trajectory at the previous acquisition time to obtain the first predicted trajectory of the target agent at the current acquisition time. This can reduce the jitter of the target predicted trajectory and improve its stability.

[0160] It should be noted that the motion tag also includes "other". When the motion tag of the target agent is not one of crossing, going straight, or changing lanes, it can be set to "other". When the motion tag of the target agent is "other", the existing predicted trajectory from the previous acquisition time can be directly smoothed against the existing predicted trajectory from the current acquisition time to obtain the target predicted trajectory of the target agent at the current acquisition time. The existing predicted trajectory of the target agent can be the trajectory output by a deep learning model of an end-to-end system in related technologies.

[0161] Figure 2 A structural diagram of the trajectory prediction device provided in an embodiment of this application is shown. Figure 2 As shown, the trajectory prediction device 200 includes:

[0162] The first acquisition module 201 is used to acquire the motion information of the target intelligent agent at the current acquisition time, wherein the target intelligent agent refers to a moving object in the environment where the target vehicle is located.

[0163] The determining module 202 is used to determine the motion tag of the target intelligent agent based on the motion information of the target intelligent agent and the motion information of the target vehicle, wherein the motion tag is used to identify the motion scene of the target intelligent agent;

[0164] The prediction module 203 is used to predict the trajectory of the target intelligent agent based on the motion tag and motion information of the target intelligent agent, and obtain the target predicted trajectory of the target intelligent agent at the current collection time. The target predicted trajectory is used to provide a reference for the target vehicle to plan the driving path.

[0165] In one embodiment of this application, the determining module 202 includes:

[0166] The first determining submodule is used to determine the motion tag of the target intelligent agent as traversing if the absolute value of the difference between the heading angle of the target intelligent agent and the heading angle of the target vehicle is within a preset angle range.

[0167] In one embodiment of this application, the determining module 202 includes:

[0168] The second determining submodule is used to determine the motion tag of the target intelligent agent as lane change if the lateral speed of the target intelligent agent meets the first preset condition. The lateral speed refers to the movement speed of the target intelligent agent relative to the lane where the target vehicle is located.

[0169] The first preset condition includes one of the following:

[0170] The lateral speed is less than a first speed threshold, and the target agent presses onto the target lane line, where the target lane line is the boundary line of the lane where the target agent is located;

[0171] The lateral speed is less than the first speed threshold, the target agent does not cross the target lane line, and the closest distance between the target agent and the target lane line is less than the first distance threshold, wherein the first distance threshold is a preset value, or the first distance threshold is determined based on the longitudinal distance between the target agent and the target vehicle;

[0172] The lateral velocity is greater than or equal to the first velocity threshold.

[0173] In one embodiment of this application, the determining module 202 includes:

[0174] The third determining submodule is used to determine the motion tag of the target intelligent agent as straight if the lateral speed of the target intelligent agent meets the second preset condition. The lateral speed refers to the movement speed of the target intelligent agent relative to the lane where the target vehicle is located.

[0175] The second preset condition includes:

[0176] The lateral speed is less than a first speed threshold, the target agent does not cross the target lane line, and the closest distance between the target agent and the target lane line is greater than or equal to a first distance threshold, wherein the first distance threshold is a preset value, or the first distance threshold is determined based on the longitudinal distance between the target agent and the target vehicle, and the target lane line is the boundary line of the lane where the target agent is located.

[0177] In one embodiment of this application, the prediction module 203 includes:

[0178] The first acquisition submodule is used to determine the longitudinal information corresponding to multiple trajectory points of the target intelligent agent within a future preset time period based on the target intelligent agent's motion information if the motion tag of the target intelligent agent is lane change.

[0179] The second acquisition submodule is used to take the lateral distance from the target intelligent agent to the first reference line as the initial lateral distance, where the first reference line is the reference line of the lane to which the target intelligent agent is about to change lanes;

[0180] The third acquisition submodule is used to determine the lateral information corresponding to multiple trajectory points of the target intelligent agent within a future preset time period based on the initial lateral distance;

[0181] The first prediction submodule is used to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time based on the longitudinal and lateral information of multiple trajectory points.

[0182] In one embodiment of this application, the prediction module 203 includes:

[0183] The fourth acquisition submodule is used to determine the longitudinal information corresponding to multiple trajectory points of the target intelligent body in a future preset time period based on the motion information of the target intelligent body if the motion tag of the target intelligent body is straight and the lateral distance of the target intelligent body to the second reference line is less than a preset distance threshold. The second reference line is the reference line of the lane closest to the target intelligent body.

[0184] The fifth acquisition submodule is used to take the lateral distance from the target agent to the second reference line as the initial lateral distance;

[0185] The sixth acquisition submodule is used to determine the lateral information corresponding to multiple trajectory points of the target intelligent agent within a future preset time period based on the initial lateral distance;

[0186] The second prediction submodule is used to obtain the target predicted trajectory of the target agent at the current acquisition time based on the longitudinal and lateral information of multiple trajectory points.

[0187] In one embodiment of this application, the first acquisition submodule and / or the fourth acquisition submodule are specifically used for:

[0188] Based on the current longitudinal velocity and current longitudinal acceleration of the target intelligent agent, the longitudinal velocity corresponding to each of the multiple trajectory points of the target intelligent agent within a future preset time period is calculated.

[0189] The longitudinal coordinate position of each trajectory point is obtained by calculating the longitudinal velocity and motion duration of the target agent at each trajectory point. The longitudinal information of the trajectory point includes the longitudinal coordinate position of the trajectory point.

[0190] In one embodiment of this application, the third acquisition submodule and / or the sixth acquisition submodule are specifically used for:

[0191] The initial lateral distance is taken as the lateral distance corresponding to the first trajectory point among multiple trajectory points;

[0192] For multiple trajectory points other than the first trajectory point, the lateral distance corresponding to the previous trajectory point of the trajectory point is multiplied by a preset coefficient to obtain the lateral distance corresponding to the trajectory point. The preset coefficient is used to characterize the speed of lane change.

[0193] Based on the current position of the target agent and the lateral distance corresponding to each trajectory point, the lateral coordinate position corresponding to each trajectory point is determined, wherein the lateral information corresponding to the trajectory point includes the lateral coordinate position corresponding to the trajectory point.

[0194] In one embodiment of this application, the preset coefficient is negatively correlated with the lateral velocity of the target intelligent agent, where the lateral velocity refers to the moving speed of the target intelligent agent relative to the lane where the target vehicle is located.

[0195] In one embodiment of this application, the prediction module 203 includes:

[0196] The third prediction submodule is used to adjust the motion information of the first trajectory point that crosses the road edge in the existing prediction trajectory according to a preset strategy if the motion tag of the target intelligent agent is traversing and the existing predicted trajectory of the target intelligent agent intersects with the road edge in the environment where the target vehicle is located, so as to obtain the target predicted trajectory of the target intelligent agent at the current collection time. The preset strategy is used to adjust the first trajectory point to not cross the road edge while keeping the number of trajectory points in the existing predicted trajectory unchanged.

[0197] In one embodiment of this application, the preset strategy includes at least one of the following:

[0198] Set the position corresponding to the first trajectory point as the position corresponding to the second trajectory point, where the second trajectory point is the last trajectory point among the trajectory points that did not cross the curb;

[0199] Set the heading angle corresponding to the first trajectory point as the heading angle corresponding to the second trajectory point;

[0200] Set the velocity corresponding to the first trajectory point to a first preset value;

[0201] Set the acceleration corresponding to the first trajectory point to a second preset value.

[0202] In one embodiment of this application, the trajectory prediction device 200 further includes:

[0203] The second acquisition module is used to take the target prediction trajectory of the target agent at the current acquisition time as the initial prediction trajectory of the target agent at the current acquisition time.

[0204] The third acquisition module is used to obtain the first predicted trajectory of the target intelligent agent based on the initial predicted trajectory of the target intelligent agent at the current acquisition time and the target predicted trajectory of the target intelligent agent at the previous acquisition time at the current acquisition time.

[0205] The update module is used to update the target predicted trajectory of the target agent at the current acquisition time to the first predicted trajectory.

[0206] In one embodiment of this application, the third acquisition module includes:

[0207] The seventh acquisition submodule is used to acquire the first weight corresponding to the current acquisition time and the second weight corresponding to the previous acquisition time if the initial predicted trajectory and the target predicted trajectory at the previous acquisition time have the same number of trajectory points.

[0208] The weighted processing submodule is used to perform weighted processing on the horizontal and vertical coordinate positions of the i-th trajectory point in the initial predicted trajectory and the i-th trajectory point in the target predicted trajectory at the previous acquisition time, according to the first weight and the second weight, to obtain the target horizontal and vertical coordinate positions of the i-th trajectory point, where i is an integer less than or equal to N, and N is the number of trajectory points in the initial predicted trajectory;

[0209] The fourth determination submodule is used to determine the predicted coordinate position of the i-th trajectory point based on the target lateral coordinate position and the target longitudinal coordinate position of the i-th trajectory point;

[0210] The eighth acquisition submodule is used to obtain the first predicted trajectory of the target intelligent agent based on the predicted coordinate positions of the N trajectory points.

[0211] In one embodiment of this application, the trajectory prediction device 200 further includes:

[0212] The fourth acquisition module is used to acquire multiple moving objects in the environment where the target vehicle is located at the current acquisition time;

[0213] The fifth acquisition module is used to, for any one of a plurality of moving objects, if the moving object meets the third preset condition, then take the moving object as the target intelligent agent;

[0214] The third preset condition includes:

[0215] The moving object is located within the target area, and the target area is determined based on the position of the target vehicle;

[0216] The moving object is located within the curb of the environment in which the target vehicle is situated.

[0217] The trajectory prediction device 200 provided in this application embodiment can implement the various processes implemented in the aforementioned trajectory prediction method embodiment and achieve the same technical effect. To avoid repetition, it will not be described again here.

[0218] Figure 3 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown.

[0219] The electronic device may include a processor 601 and a memory 602 storing computer program instructions.

[0220] Specifically, the processor 601 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0221] Memory 602 may include mass storage for data or instructions. For example, and not limitingly, memory 602 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 602 may include removable or non-removable (or fixed) media. Where appropriate, memory 602 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 602 is non-volatile solid-state memory.

[0222] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, generally, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the first aspect of this disclosure.

[0223] In one example, the electronic device may also include a communication interface 603 and a bus 610. For example, Figure 3 As shown, the processor 601, memory 602, and communication interface 603 are connected through bus 610 and complete communication with each other.

[0224] The communication interface 603 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0225] Bus 610 includes hardware, software, or both. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 610 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0226] Furthermore, in conjunction with the trajectory prediction methods in the above embodiments, this application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the trajectory prediction methods in the above embodiments.

[0227] This application also provides a computer program product, wherein the instructions in the computer program product, when executed by the processor of an electronic device, cause the electronic device to perform any of the trajectory prediction methods described in the above embodiments.

[0228] This application also provides a vehicle that includes the electronic devices described in the above embodiments.

[0229] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described as examples. However, the method process of this application is not limited to the specific steps described. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0230] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0231] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0232] The foregoing flowcharts and / or block diagrams of methods, apparatus (systems) according to embodiments of the present disclosure have described various aspects of the present disclosure. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to create a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowcharts and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block of the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0233] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A trajectory prediction method, characterized in that, The method includes: Acquire the motion information of the target intelligent agent at the current acquisition time, wherein the target intelligent agent refers to a moving object in the environment in which the target vehicle is located; Based on the motion information of the target intelligent agent and the motion information of the target vehicle, a motion tag for the target intelligent agent is determined, and the motion tag is used to identify the motion scene of the target intelligent agent; Based on the motion tag and motion information of the target intelligent agent, trajectory prediction is performed on the target intelligent agent to obtain the target predicted trajectory of the target intelligent agent at the current collection time. The target predicted trajectory is used to provide a reference for the target vehicle to plan the driving path. The target predicted trajectory of the target agent at the current acquisition time is taken as the initial predicted trajectory of the target agent at the current acquisition time. If the initial predicted trajectory and the target predicted trajectory at the previous acquisition time have the same number of trajectory points, then obtain the first weight corresponding to the current acquisition time and the second weight corresponding to the previous acquisition time. For the i-th trajectory point in the initial predicted trajectory and the i-th trajectory point in the target predicted trajectory at the previous acquisition time, the horizontal and vertical coordinate positions of the two i-th trajectory points are weighted according to the first weight and the second weight respectively to obtain the target horizontal and vertical coordinate positions of the i-th trajectory point, where i is an integer less than or equal to N, and N is the number of trajectory points in the initial predicted trajectory. Based on the target lateral coordinate position and the target longitudinal coordinate position of the i-th trajectory point, determine the predicted coordinate position of the i-th trajectory point; Based on the predicted coordinate positions of the N trajectory points, the first predicted trajectory of the target intelligent agent is obtained; The target intelligent agent's predicted trajectory at the current acquisition time is updated to the first predicted trajectory.

2. The trajectory prediction method according to claim 1, characterized in that, Based on the motion information of the target intelligent agent and the motion information of the target vehicle, the motion tag of the target intelligent agent is determined, including: If the absolute value of the difference between the heading angle of the target agent and the heading angle of the target vehicle is within a preset angle range, then the motion tag of the target agent is determined to be traversing.

3. The trajectory prediction method according to claim 1, characterized in that, Based on the motion information of the target intelligent agent and the motion information of the target vehicle, the motion tag of the target intelligent agent is determined, including: If the lateral speed of the target intelligent agent meets the first preset condition, then the motion tag of the target intelligent agent is determined to be lane change, and the lateral speed refers to the moving speed of the target intelligent agent relative to the lane where the target vehicle is located. The first preset condition includes one of the following: The lateral speed is less than a first speed threshold, and the target agent presses onto the target lane line, where the target lane line is the boundary line of the lane where the target agent is located; The lateral speed is less than the first speed threshold, the target agent does not cross the target lane line, and the closest distance between the target agent and the target lane line is less than the first distance threshold, wherein the first distance threshold is a preset value, or the first distance threshold is determined based on the longitudinal distance between the target agent and the target vehicle; The lateral velocity is greater than or equal to the first velocity threshold.

4. The trajectory prediction method according to claim 1, characterized in that, Based on the motion information of the target intelligent agent and the motion information of the target vehicle, the motion tag of the target intelligent agent is determined, including: If the lateral speed of the target intelligent agent meets the second preset condition, then the motion tag of the target intelligent agent is determined to be straight, and the lateral speed refers to the moving speed of the target intelligent agent relative to the lane where the target vehicle is located; The second preset condition includes: The lateral speed is less than a first speed threshold, the target agent does not cross the target lane line, and the closest distance between the target agent and the target lane line is greater than or equal to a first distance threshold, wherein the first distance threshold is a preset value, or the first distance threshold is determined based on the longitudinal distance between the target agent and the target vehicle, and the target lane line is the boundary line of the lane where the target agent is located.

5. The trajectory prediction method according to claim 1, characterized in that, Based on the motion tag and motion information of the target intelligent agent, trajectory prediction is performed on the target intelligent agent to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time, including: If the motion tag of the target intelligent agent is lane change, then based on the motion information of the target intelligent agent, determine the longitudinal information corresponding to multiple trajectory points of the target intelligent agent in a future preset time period; The lateral distance from the target agent to the first reference line is used as the initial lateral distance. The first reference line is the reference line of the lane to which the target agent is about to change lanes. Based on the initial lateral distance, determine the lateral information corresponding to multiple trajectory points of the target agent within a future preset time period; Based on the longitudinal and lateral information of multiple trajectory points, the target predicted trajectory of the target agent at the current acquisition time is obtained.

6. The trajectory prediction method according to claim 1, characterized in that, Based on the motion tag and motion information of the target intelligent agent, trajectory prediction is performed on the target intelligent agent to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time, including: If the motion label of the target intelligent agent is straight, and the lateral distance from the target intelligent agent to the second reference line is less than a preset distance threshold, then based on the motion information of the target intelligent agent, the longitudinal information corresponding to multiple trajectory points of the target intelligent agent in a future preset time period is determined, and the second reference line is the reference line of the lane closest to the target intelligent agent; The lateral distance from the target agent to the second reference line is taken as the initial lateral distance; Based on the initial lateral distance, determine the lateral information corresponding to multiple trajectory points of the target agent within a future preset time period; Based on the longitudinal and lateral information of multiple trajectory points, the target predicted trajectory of the target agent at the current acquisition time is obtained.

7. The trajectory prediction method according to claim 5 or 6, characterized in that, Based on the motion information of the target intelligent agent, determine the longitudinal information corresponding to multiple trajectory points of the target intelligent agent within a future preset time period, including: Based on the current longitudinal velocity and current longitudinal acceleration of the target intelligent agent, the longitudinal velocity corresponding to each of the multiple trajectory points of the target intelligent agent within a future preset time period is calculated. The longitudinal coordinate position of each trajectory point is obtained by calculating the longitudinal velocity and motion duration of the target agent at each trajectory point. The longitudinal information of the trajectory point includes the longitudinal coordinate position of the trajectory point.

8. The trajectory prediction method according to claim 5 or 6, characterized in that, Based on the initial lateral distance, determine the lateral information corresponding to multiple trajectory points of the target agent within a preset time period, including: The initial lateral distance is taken as the lateral distance corresponding to the first trajectory point among multiple trajectory points; For multiple trajectory points other than the first trajectory point, the lateral distance corresponding to the previous trajectory point of the trajectory point is multiplied by a preset coefficient to obtain the lateral distance corresponding to the trajectory point. The preset coefficient is used to characterize the speed of lane change. Based on the current position of the target agent and the lateral distance corresponding to each trajectory point, the lateral coordinate position corresponding to each trajectory point is determined, wherein the lateral information corresponding to the trajectory point includes the lateral coordinate position corresponding to the trajectory point.

9. The trajectory prediction method according to claim 8, characterized in that, The preset coefficient is negatively correlated with the lateral velocity of the target agent, where the lateral velocity refers to the movement speed of the target agent relative to the lane where the target vehicle is located.

10. The trajectory prediction method according to claim 1, characterized in that, Based on the motion tag and motion information of the target intelligent agent, trajectory prediction is performed on the target intelligent agent to obtain the target predicted trajectory of the target intelligent agent at the current acquisition time, including: If the motion tag of the target agent is traversing, and the existing predicted trajectory of the target agent intersects with the curb in the environment where the target vehicle is located, then the motion information of the first trajectory point that crosses the curb in the existing predicted trajectory is adjusted according to a preset strategy to obtain the target predicted trajectory of the target agent at the current collection time. The preset strategy is used to adjust the first trajectory point so that it does not cross the curb while keeping the number of trajectory points in the existing predicted trajectory unchanged.

11. The trajectory prediction method according to claim 10, characterized in that, The preset strategy includes at least one of the following: Set the position corresponding to the first trajectory point as the position corresponding to the second trajectory point, where the second trajectory point is the last trajectory point among the trajectory points that did not cross the curb; Set the heading angle corresponding to the first trajectory point as the heading angle corresponding to the second trajectory point; Set the velocity corresponding to the first trajectory point to a first preset value; Set the acceleration corresponding to the first trajectory point to a second preset value.

12. The trajectory prediction method according to claim 1, characterized in that, Before acquiring the motion information of the target agent at the current acquisition time, the method further includes: Acquire multiple moving objects in the environment where the target vehicle is located at the current acquisition time; For any one of multiple moving objects, if the moving object satisfies the third preset condition, then the moving object is taken as the target intelligent agent; The third preset condition includes: The moving object is located within the target area, and the target area is determined based on the position of the target vehicle; The moving object is located within the curb of the environment in which the target vehicle is situated.

13. A trajectory prediction device, characterized in that, The device includes: The first acquisition module is used to acquire the motion information of the target intelligent agent at the current acquisition time, wherein the target intelligent agent refers to a moving object in the environment where the target vehicle is located. The determination module is used to determine the motion tag of the target intelligent agent based on the motion information of the target intelligent agent and the motion information of the target vehicle, wherein the motion tag is used to identify the motion scene of the target intelligent agent; The prediction module is used to predict the trajectory of the target intelligent agent based on the motion tag and motion information of the target intelligent agent, and obtain the target predicted trajectory of the target intelligent agent at the current collection time. The target predicted trajectory of the target intelligent agent is used for driving path planning for the target vehicle. The second acquisition module is used to take the target prediction trajectory of the target agent at the current acquisition time as the initial prediction trajectory of the target agent at the current acquisition time. The third acquisition module is used to obtain the first predicted trajectory of the target intelligent agent based on the initial predicted trajectory of the target intelligent agent at the current acquisition time and the target predicted trajectory of the target intelligent agent at the previous acquisition time at the current acquisition time. The update module is used to update the target predicted trajectory of the target agent at the current acquisition time to the first predicted trajectory; The third acquisition module includes: The seventh acquisition submodule is used to acquire the first weight corresponding to the current acquisition time and the second weight corresponding to the previous acquisition time if the initial predicted trajectory and the target predicted trajectory at the previous acquisition time have the same number of trajectory points. The weighted processing submodule is used to perform weighted processing on the horizontal and vertical coordinate positions of the i-th trajectory point in the initial predicted trajectory and the i-th trajectory point in the target predicted trajectory at the previous acquisition time, according to the first weight and the second weight, to obtain the target horizontal and vertical coordinate positions of the i-th trajectory point, where i is an integer less than or equal to N, and N is the number of trajectory points in the initial predicted trajectory; The fourth determination submodule is used to determine the predicted coordinate position of the i-th trajectory point based on the target lateral coordinate position and the target longitudinal coordinate position of the i-th trajectory point; The eighth acquisition submodule is used to obtain the first predicted trajectory of the target intelligent agent based on the predicted coordinate positions of the N trajectory points.

14. An electronic device, characterized in that, include: Processor and memory storing computer program instructions; When the processor executes the computer program instructions, it implements the trajectory prediction method as described in any one of claims 1-12.

15. A vehicle, characterized in that, The vehicle includes the electronic equipment as described in claim 14.

16. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the trajectory prediction method as described in any one of claims 1-12.

17. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device causes the electronic device to perform the trajectory prediction method as described in any one of claims 1-12.

Images