Vehicle Trajectory Prediction Method, Device, Electronic Device and Storage Medium

Abstract

The present application relates to a vehicle trajectory prediction method, apparatus, electronic device, and storage medium. The vehicle trajectory prediction method includes: determining the lane centerline of a target vehicle and predicting the lane-changing intention of the target vehicle; generating a short-term prediction trajectory of the target vehicle based on a preset kinematic model, and compensating the lane centerline based on the lane-changing intention and the short-term prediction trajectory by combining a preset interpolation algorithm to obtain a target reference line cluster; generating a vehicle prediction trajectory of the target vehicle based on the target reference line cluster. Through the present application, the short-term prediction trajectory based on the kinematic information of the vehicle realizes the compensation of the reference line, so that a reference line conforming to the actual characteristics of the vehicle trajectory can be obtained, and further the accuracy of the vehicle trajectory prediction result can be improved.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving, and particularly to a vehicle trajectory prediction method, apparatus, electronic device, and storage medium. Background Art

[0002] In the application of autonomous vehicles, autonomous vehicles need to have the ability of active decision-making for behaviors such as lane changing, overtaking, and decelerating to achieve safe and efficient driving. Based on this, predicting the future trajectories of surrounding vehicles can help autonomous vehicles plan their own driving states in advance. Currently, the technology of predicting the driving trajectory of a target vehicle based on a vehicle trajectory prediction algorithm often uses the lane center line as a reference line, or selects multiple reference lines by sampling in space to achieve vehicle trajectory prediction. In the scenario where a vehicle maneuvers or changes lanes, the accuracy of the predicted vehicle trajectory by the above vehicle trajectory prediction method is relatively low.

[0003] Regarding the problem of relatively low accuracy of the predicted result of the vehicle trajectory in the related art, no effective solution has been proposed yet. Summary of the Invention

[0004] In the present embodiment, a vehicle trajectory prediction method, apparatus, electronic device, and storage medium are provided to solve the problem of relatively low accuracy of the predicted result of the vehicle trajectory in the related art.

[0005] In a first aspect, in the present embodiment, a vehicle trajectory prediction method is provided, including:

[0006] Determine the lane center line of the target vehicle and predict the lane-changing intention of the target vehicle;

[0007] Generate a short-term prediction trajectory of the target vehicle based on a preset kinematic model, and based on the lane-changing intention and the short-term prediction trajectory, combine a preset interpolation algorithm to compensate the lane center line to obtain a target reference line cluster;

[0008] Generate a vehicle prediction trajectory of the target vehicle based on the target reference line cluster.

[0009] In some embodiments, the determining the lane center line of the target vehicle includes:

[0010] Calculate the lane center line of the target vehicle according to the pre-acquired lane boundary line information of the target vehicle.

[0011] In some embodiments, the predicting the lane-changing intention of the target vehicle includes:

[0012] According to the motion state information of the target vehicle, respectively determine the first position relationship between the target vehicle and the lane boundary line at a historical moment, the second position relationship between the target vehicle and the lane boundary line at the current moment, and predict the third position relationship between the target vehicle and the lane boundary line after a preset time period;

[0013] By comparing the first position relationship, the second position relationship, and the third position relationship, determine the lane-changing intention of the target vehicle.

[0014] In some embodiments, the generating the short-term prediction trajectory of the target vehicle based on a preset kinematic model includes:

[0015] Input the current motion state information of the target vehicle into the preset kinematic model for predicting the motion state for a preset time period, and generate the short-term prediction trajectory of the target vehicle.

[0016] In some embodiments, the preset time period is a preset number of interval moments; the inputting the current motion state information of the target vehicle into the preset kinematic model for predicting the motion state for a preset time period and generating the short-term prediction trajectory of the target vehicle includes:

[0017] Input the current motion state information of the target vehicle into the preset kinematic model, and predict the motion state information at the preset number of interval moments;

[0018] Based on the motion state information at the preset number of interval moments, generate the short-term prediction trajectory.

[0019] In some embodiments, the compensating the lane center line based on the lane-changing intention and the short-term prediction trajectory and combining with a preset interpolation algorithm to obtain a target reference line cluster includes:

[0020] Modify the lane center line according to the lane-changing intention to obtain a target lane center line;

[0021] Based on the principle of the cubic spline interpolation algorithm, use the short-term prediction trajectory to compensate the target lane center line to obtain the target reference line cluster.

[0022] In some embodiments, the compensating the target lane center line based on the principle of the cubic spline interpolation algorithm and using the short-term prediction trajectory to obtain the target reference line cluster includes:

[0023] According to the coordinate information in the short-term prediction trajectory, continuously extract a preset number of connection points from the target lane center line;

[0024] Based on the principle of the cubic spline interpolation algorithm, compensate the target lane center line through the connection points using the short-term prediction trajectory to obtain the target reference line cluster.

[0025] In some embodiments, generating the vehicle prediction trajectory of the target vehicle based on the target reference line cluster includes:

[0026] Establish a Frenet coordinate system according to the target reference line cluster;

[0027] Based on the position information of the reference points selected from the target reference line cluster and the current motion state information of the target vehicle, obtain the starting point and ending point of the prediction trajectory of the target vehicle in the Frenet coordinate system;

[0028] Generate the vehicle prediction trajectory of the target vehicle according to the starting point and ending point.

[0029] In some embodiments, after generating the vehicle prediction trajectory of the target vehicle based on the target reference line cluster, the method further includes:

[0030] Calculate the cost values of all vehicle prediction trajectories of the target vehicle according to a preset cost function based on acceleration information and lateral overshoot information;

[0031] Screen the cost values of all vehicle prediction trajectories based on a preset screening condition to obtain the optimal vehicle prediction trajectory.

[0032] In a second aspect, a vehicle trajectory prediction device is provided in this embodiment, including: an acquisition module, a first generation module, and a second generation module; where:

[0033] The acquisition module is configured to determine the lane center line of the target vehicle and predict the lane-changing intention of the target vehicle;

[0034] The first generation module is configured to generate a short-term prediction trajectory of the target vehicle based on a preset kinematic model, and compensate the lane center line by combining a preset interpolation algorithm based on the lane-changing intention and the short-term prediction trajectory to obtain a target reference line cluster;

[0035] The second generation module is configured to generate a vehicle prediction trajectory of the target vehicle based on the target reference line cluster.

[0036] In a third aspect, an electronic device is provided in this embodiment, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, it implements the vehicle trajectory prediction method described in the first aspect above.

[0037] Fourthly, in this embodiment, a storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the vehicle trajectory prediction method described in the first aspect above is implemented.

[0038] Compared with the related art, in the vehicle trajectory prediction method, device, electronic device, and storage medium provided in this embodiment, the center line of the lane of the target vehicle is determined, and the lane-changing intention of the target vehicle is predicted; a short-term prediction trajectory of the target vehicle is generated based on a preset kinematic model, and based on the lane-changing intention and the short-term prediction trajectory, the center line of the lane is compensated by combining a preset interpolation algorithm to obtain a target reference line cluster; a vehicle prediction trajectory of the target vehicle is generated based on the target reference line cluster. It realizes the compensation of the reference line through the short-term prediction trajectory based on the kinematic information of the vehicle, so as to obtain a reference line that conforms to the actual characteristics of the vehicle trajectory, and further improve the accuracy of the prediction result of the vehicle trajectory.

[0039] The details of one or more embodiments of the present application are set forth in the following drawings and description, so that the other features, objects, and advantages of the present application will become more concise and understandable. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation of the present application. In the drawings:

[0041] Figure 1 is a hardware structure block diagram of a terminal of the vehicle trajectory prediction method in this embodiment;

[0042] Figure 2 is a flowchart of the vehicle trajectory prediction method in this embodiment;

[0043] Figure 3 is a schematic diagram of the conversion relationship between Frenet coordinates and Cartesian coordinates;

[0044] Figure 4 is a model diagram of the target reference line cluster in this embodiment;

[0045] Figure 5 is a flowchart of the vehicle trajectory prediction method in this preferred embodiment;

[0046] Figure 6 is a flowchart of the method for generating the target reference line cluster in this preferred embodiment;

[0047] Figure 7 is a flowchart of the method for generating the vehicle prediction trajectory in this preferred embodiment;

[0048] Figure 8It is a structural block diagram of the vehicle trajectory prediction device according to this embodiment. Detailed implementation manners

[0049] To understand the purpose, technical solution and advantages of this application more clearly, the following describes and explains this application in combination with the accompanying drawings and embodiments.

[0050] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meanings understood by those with ordinary skills in the technical field to which this application belongs. In this application, words such as "a", "one", "a kind of", "the", "these" and the like do not indicate a limitation in quantity, and they can be singular or plural. The terms "including", "comprising", "having" and any variants thereof involved in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product or device including a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent in these processes, methods, products or devices. The terms "connected", "coupled" and the like involved in this application do not limit to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" may represent: A exists alone, A and B exist simultaneously, and B exists alone. Usually, the character " / " indicates that the objects associated before and after are in an "or" relationship. The terms "first", "second", "third" and the like involved in this application only distinguish similar objects and do not represent a specific sorting for the objects.

[0051] The method embodiment provided in this embodiment can be executed on a terminal, a computer or a similar computing device. For example, running on a terminal, Figure 1 It is a hardware structural block diagram of the terminal of the vehicle trajectory prediction method according to this embodiment. As Figure 1 shown, the terminal may include one or more ( Figure 1 only one is shown in Figure 1 the figure) processors 102 and a memory 104 for storing data. Among them, the processor 102 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA. The above terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those of ordinary skill in the art can understand that Figure 1 the structure shown in Figure 1The different configurations shown.

[0052] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the vehicle trajectory prediction method in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implements the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memories. In some instances, the memory 104 may further include a memory remotely disposed relative to the processor 102, and these remote memories can be connected to the terminal through a network. Examples of the above network include but are not limited to the Internet, intranet, local area network, mobile communication network, and combinations thereof.

[0053] The transmission device 106 is used to receive or send data via a network. The above network includes the wireless network provided by the communication provider of the terminal. In one instance, the transmission device 106 includes a network adapter (Network Interface Controller, abbreviated as NIC), which can be connected to other network devices through a base station and thus communicate with the Internet. In one instance, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0054] In this embodiment, a vehicle trajectory prediction method is provided. Figure 2 It is the flowchart of the vehicle trajectory prediction method in this embodiment, as Figure 2 shown, and this process includes the following steps:

[0055] Step S210, determine the lane centerline of the target vehicle and predict the lane-changing intention of the target vehicle.

[0056] Among them, the above target vehicle is the vehicle for which trajectory prediction is to be performed. The lane centerline and lane-changing intention of the above target vehicle can be determined from the pre-acquired perception data. Among them, the perception data includes the historical information and environmental information of the target vehicle. The historical information includes information such as the historical trajectory, heading, speed, acceleration, and angular velocity of the vehicle. The environmental information may include lane boundary lines. Specifically, the lane centerline of the target vehicle can be calculated based on the coordinate information of the discrete points in the lane boundary line of the lane where the target vehicle is located that is pre-acquired; or the lane centerline of the target vehicle can be determined based on a high-precision map. In addition, the lane-changing intention of the target vehicle can be predicted from the historical information of the target vehicle. Among them, the lane-changing intention can specifically be the intention of the vehicle to change lanes to the left or to the right. Preferably, based on the historical trajectory of the target vehicle, the positional relationship between the target vehicle and the left and right lane boundary lines at the historical moment and the current moment can be determined, and the positional relationship between the target vehicle and the left and right lane boundary lines at the preset moment can be predicted. Furthermore, based on the positional relationships obtained at the above different moments, it can be predicted that the target vehicle will change lanes to the left, change lanes to the right, or go straight.

[0057] Step S220: Generate a short-term prediction trajectory of the target vehicle based on a preset kinematic model, and based on the lane-changing intention and the short-term prediction trajectory, combine a preset interpolation algorithm to compensate the lane centerline to obtain a target reference line cluster.

[0058] Among them, the vehicle trajectory of the target vehicle can be predicted for a short time based on a preset kinematic model to obtain a short-term prediction trajectory. Specifically, the current motion state information of the target vehicle, such as the current coordinate information, heading angle, speed, acceleration, angular velocity, etc. of the target vehicle, can be input into the preset kinematic model for processing, so as to predict the motion state information at several future moments, and then obtain the short-term prediction trajectory. The preset kinematic model can be selected according to the actual application scenario. For example, the CA (Constant Acceleration) model, the CV (Constant Velocity) model, the CT (Coordinated Turn) model, and the CYRA (Constant Yaw Rate and Acceleration) model, etc. Preferably, in order to accurately predict the short-term trajectory of the vehicle in the curve scenario, the kinematic model can specifically be the CYRA model. In addition, after determining the lane center line and the lane change intention, the lane center line can be corrected based on the lane change intention to obtain the target lane center line. Specifically, the coordinates of the discrete points in the lane center line can be corrected according to the predicted intention of the target vehicle to change lanes to the left or right. Exemplarily, if the target vehicle changes lanes to the left, the ordinate of the discrete points in the lane center line can be increased by a preset value. If the target vehicle changes lanes to the right, the ordinate of the discrete points in the lane center line can be decreased by a preset value, and then the correction of the lane center line is realized based on the lane change intention. After that, according to the preset interpolation algorithm, the short-term prediction trajectory is used to compensate the corrected target lane center line, so as to obtain the target reference line cluster.

[0059] Specifically, several consecutive points with abscissas greater than the abscissa in the short-term prediction trajectory can be extracted from the discrete points of the target lane center line as connection points. The short-term prediction trajectory generated above is used as compensation and spliced with the connection points based on the cubic spline interpolation algorithm, so as to realize the compensation of the target lane center line and obtain the target reference line cluster.

[0060] Compared with the related technology, directly using the lane center line as the reference line or obtaining the reference line by spatial sampling, the generated reference line will lead to a large trajectory prediction error for the vehicle that changes lanes. In this embodiment, a short-term prediction trajectory is generated based on the kinematic model, and the lane center line is compensated according to the short-term prediction trajectory to obtain a reference line cluster that is closer to the vehicle trajectory characteristics, which can solve the problem of large trajectory prediction deviation during lane change, thereby improving the accuracy of the vehicle trajectory prediction result.

[0061] Step S230, generating a vehicle prediction trajectory of the target vehicle based on the target reference line cluster.

[0062] First, a Frenet coordinate system can be established based on the target reference line cluster. Among them, taking the longitudinal extension direction of each reference line in the target reference line cluster as the s-axis and the normal direction of each point on the reference line as the d-axis, a Frenet coordinate system is established. Among them, the s-axis represents the longitudinal displacement of the target vehicle, and the d-axis represents the lateral displacement of the target vehicle. The Frenet coordinate system can decompose the position information of the target vehicle into two directions of the s-axis and the d-axis, facilitating subsequent trajectory prediction. Figure 3 It is a schematic diagram of the conversion relationship between Frenet coordinates and Cartesian coordinates. Among them, represents the coordinates of a discrete point on the vehicle trajectory in the Frenet coordinate system, represents the normal vector of this discrete point, represents the tangent vector of this discrete point, d(t) represents the lateral displacement of the target vehicle, represents the normal vector, represents the tangent vector, represents the vector from the reference point to the vehicle position, s(t) represents the longitudinal displacement of the target vehicle, and t can represent time or curve length. As Figure 3 shown, the relationship between the coordinates of the discrete points on the vehicle trajectory and the reference line is shown in the following formula. Based on the following formula, it can be known that the Frenet coordinates of the discrete points on the vehicle trajectory can be obtained according to the coordinates of the reference points selected from the known reference line and the Cartesian coordinates of the discrete points on the vehicle trajectory, or the Cartesian coordinates of the discrete points on the vehicle trajectory can be obtained according to the coordinates of the reference points and the Frenet coordinates of the discrete points on the vehicle trajectory.

[0063]

[0064] Among them, represents the Cartesian coordinates of the discrete points on the vehicle trajectory, represents the vector from the reference point to the vehicle position, represents the unit vector of.

[0065] After establishing the Frenet coordinate system based on the target reference line cluster, the point closest to the position of the target vehicle in the current frame can be selected from the reference line cluster as the reference point. Based on the coordinates of this reference point and the pre-obtained current motion state information of the target vehicle, the starting point and the ending point of the vehicle prediction trajectory in the Frenet coordinate system can be determined. Among them, the information of the starting point can include the s coordinate and the d coordinate of the initial position of the target vehicle in the Frenet coordinate system, as well as the corresponding speeds and accelerations in the longitudinal direction and the lateral direction. The information of the ending point can include the s coordinate and the d coordinate of the ending position of the target vehicle in the Frenet coordinate system, as well as the corresponding speeds and accelerations in the longitudinal direction and the lateral direction. After that, based on the determined starting point and ending point, the vehicle prediction trajectory of the target vehicle in the Frenet coordinate system can be solved. In addition, the generated vehicle prediction trajectory can be screened based on a preset cost function, so as to obtain the vehicle prediction trajectory with the minimum cost.

[0066] In addition, this embodiment can generate the vehicle prediction trajectory only based on numerical operation processing, and can accurately predict the trajectory of the target vehicle even without the support of a high-precision map. Therefore, it can be applied to various levels of autonomous driving systems.

[0067] In the above steps S210 to S230, the center line of the lane of the target vehicle is determined, and the lane-changing intention of the target vehicle is predicted; a short-term prediction trajectory of the target vehicle is generated based on a preset kinematic model, and based on the lane-changing intention and the short-term prediction trajectory, the center line of the lane is compensated by combining a preset interpolation algorithm to obtain a target reference line cluster; a vehicle prediction trajectory of the target vehicle is generated based on the target reference line cluster. It realizes the compensation of the reference line through the short-term prediction trajectory based on the kinematic information of the vehicle, so as to obtain a reference line that conforms to the actual characteristics of the vehicle trajectory, and further improve the accuracy of the prediction result of the vehicle trajectory.

[0068] Further, in one embodiment, based on the above step S210, determining the center line of the lane of the target vehicle may specifically include the following steps:

[0069] Step S211, calculate the center line of the lane of the target vehicle according to the pre-obtained lane boundary line information of the target vehicle.

[0070] The lane boundary line information of the target vehicle can be determined according to the environmental information in the perception data. Among them, the lane boundary line includes the left lane boundary line and the right lane boundary line. Assume that the speed of the vehicle on the x-axis in the Cartesian coordinate system is positive. Let the left and right lane boundary lines of the target vehicle be L 1 and L 2 . Among them, both L 1 and L 2 are composed of a number of discrete points, denoted as: Among them, is the coordinate of the i-th discrete point of the left lane boundary line in the Cartesian coordinate system, is the coordinate of the i-th discrete point of the right lane boundary line in the Cartesian coordinate system. n is the number of discrete points of the left lane boundary line, and m is the number of discrete points of the right lane boundary line. Denote the lane center line of the target vehicle as The lane center line calculated according to the left and right lane boundary lines can specifically be that the height difference between the left and right lane boundary lines is calculated as Among them, It can be understood that represents the y-axis coordinate of the l-th discrete point of the right lane boundary line, represents the x-axis coordinate of the l-th discrete point of the right lane boundary line. Since the index values of the corresponding discrete points are not necessarily the same when calculating the height difference between the left and right lane boundary lines, the i-th discrete point of the left lane boundary line and the l-th discrete point of the right lane boundary line are used to calculate respectively. Among them, the y-axis coordinate of the discrete point in the lane center line is: The x-axis coordinate is: Based on this, the coordinates of the discrete points of the lane center line can be determined according to the coordinates of the discrete points of the left and right lane boundary lines.

[0071] In addition, in one embodiment, based on the above step S210, predicting the lane-changing intention of the target vehicle may specifically include the following steps:

[0072] Step S212, according to the motion state information of the target vehicle, respectively determine the first position relationship between the target vehicle and the lane boundary line at the historical moment, the second position relationship between the target vehicle and the lane boundary line at the current moment, and the third position relationship between the predicted target vehicle and the lane boundary line after a preset time period.

[0073] Preferably, the position relationship mentioned in step S212 may be the distance difference. For example, the first distance difference l- between the target vehicle and the left lane boundary line in the previous 1 second can be calculated respectively T , as the first position relationship, calculate the second distance difference l between the target vehicle and the left lane boundary line at the current moment 0 , as the second position relationship, predict the third distance difference l between the target vehicle and the left lane boundary line after 1 second T , as the third position relationship.

[0074] Step S213, by comparing the first position relationship, the second position relationship, and the third position relationship, determine the lane-changing intention of the target vehicle.

[0075] Exemplarily, if the above three distance differences satisfy the following formula, it can be predicted that the lane-changing intention of the target vehicle is to change lanes to the left:

[0076]

[0077] Among them, the above α is a preset distance threshold, and w is the width of the lane boundary line. Similarly, based on the distance differences between the target vehicle and the right lane boundary line at historical moments, the current moment, and after a preset moment, it can also be predicted whether the vehicle has the intention to change lanes to the right.

[0078] In this embodiment, by determining the position relationship between the target vehicle and the lane boundary line at historical moments and the current moment, and predicting the position relationship between the target vehicle and the lane boundary line after a preset period, the lane-changing intention of the target vehicle is determined, which can improve the accuracy of predicting the lane-changing intention.

[0079] In addition, in one embodiment, based on the above step S220, a short-term prediction trajectory of the target vehicle is generated based on a preset kinematic model, which may specifically include the following steps:

[0080] Step S221, input the current motion state information of the target vehicle into the preset kinematic model to predict the motion state for a preset period, and generate a short-term prediction trajectory of the target vehicle.

[0081] Further, in one embodiment, based on the above step S221, the preset period is a preset number of interval moments; inputting the current motion state information of the target vehicle into the preset kinematic model to predict the motion state for a preset period and generating a short-term prediction trajectory of the target vehicle may specifically include: inputting the current motion state information of the target vehicle into the preset kinematic model to predict the motion state information at a preset number of interval moments; generating a short-term prediction trajectory based on the motion state information at a preset number of interval moments.

[0082] For example, the current motion state information of the target vehicle is X 0 , X 0 It can be shown as the following formula:

[0083] X 0 = [x 0 , y 0 , θ 0 , v 0 , a 0 , w 0 T (3)

[0084] Among them, x 0 , y 0 are respectively the x-axis coordinate and y-axis coordinate of the target vehicle at the current moment, θ 0 is the heading angle, v 0 is the speed, a 0 is the acceleration, w 0Let the angular velocity be ω. Then, after Δt seconds, using the CYRA model to predict the motion state of the target vehicle, the prediction result can be obtained as follows:

[0085] X Δt = X 0 + ΔX (4)

[0086] Among them, X Δt is the motion state of the target vehicle predicted after Δt seconds. ΔX is the motion state change information of the target vehicle during Δt seconds. Specifically, it is shown as the following formula:

[0087] X Δt = [x Δt , y Δt , θ Δt , v Δt , a Δt , w Δt T (5)

[0088]

[0089]

[0090] Performing a short-term prediction of the target vehicle for k·Δt seconds by formula (6), the short-term prediction trajectory X 0:k·Δt can be obtained as follows.

[0091] X 0:k·Δt = {X 0 , X Δt , …, X k·Δt} (7)

[0092] Additionally, in one embodiment, based on the above step S220, based on the lane change intention and the short-term prediction trajectory, combining with a preset interpolation algorithm to compensate the lane center line, a target reference line cluster can be obtained, which specifically may include the following steps:

[0093] Step S222, modifying the lane center line according to the lane change intention to obtain the target lane center line.

[0094] Among them, the y-axis coordinates of the discrete points in the lane center line can be set based on the lane change intention. Exemplarily, the target lane center line is where:

[0095]

[0096] Among them, is the x-axis coordinate of the discrete point in the target lane center line, is the y-axis coordinate of the discrete point in the target lane center line, ​$x$ is the x-axis coordinate of the discrete points in the lane centerline before correction. $h$ is the height difference between the left and right boundary lines of the lane. In the above formula, $h$ can also be replaced by other values according to the actual application scenario. When predicting that the target vehicle is changing lanes to the left, the y-axis coordinate of the discrete points in the lane centerline can be increased by a preset value, and when predicting that the target vehicle is changing lanes to the right, the y-axis coordinate of the discrete points in the lane centerline can be decreased by a preset value. In this embodiment, the lane centerline is corrected based on the lane-changing intention to obtain the target lane centerline, which can make the subsequent generated reference line clusters closer to the vehicle trajectory characteristics, thereby improving the accuracy of vehicle trajectory prediction.

[0097] Step S223: Based on the principle of the cubic spline interpolation algorithm, use the short-term prediction trajectory to compensate the target lane centerline to obtain the target reference line clusters. After correcting the lane centerline, compensating the target lane centerline based on the short-term prediction trajectory predicted in the above steps can generate a reference line cluster that fits the vehicle trajectory better, thereby improving the accuracy of vehicle trajectory prediction.

[0098] Further, in one embodiment, based on the above step S223, based on the principle of the cubic spline interpolation algorithm, using the short-term prediction trajectory to compensate the target lane centerline to obtain the target reference line clusters may specifically include: Continuously extract a preset number of connection points from the target lane centerline according to the coordinate information in the short-term prediction trajectory; Based on the principle of the cubic spline interpolation algorithm, use the short-term prediction trajectory to compensate the target lane centerline through the connection points to obtain the target reference line clusters.

[0099] Among them, it is assumed that the vehicle speed in the x-axis direction is positive in the Cartesian coordinate system. Extract $n$ consecutive points with x-axis coordinates greater than $x$ among the discrete points of the target lane centerline as connection points. That is: k·Δt The points are used as connection points. That is:

[0100]

[0101] After that, through the above $n$ connection points, use cubic spline interpolation to compensate the above short-term prediction trajectory to the target lane centerline to generate the target reference line clusters. Specifically, in the interval $[x$ 0 , formed by the x-axis coordinate $x$ of the target vehicle at the current moment, and the x-axis coordinate $x$ of the end point of the target lane centerline end $[x$ 0 , $x$ end , assume the cubic spline function is as follows:

[0102]

[0103] Among them, f(x) is a cubic polynomial with continuous second derivative in each interval. Each interval polynomial has 4 parameters, and there are end intervals in total. Therefore, there are 4end parameters to be solved by equations. According to the properties of the cubic spline interpolation function, it can be obtained that:

[0104]

[0105] Under the natural boundary conditions, it can be obtained that f i ″(x 0 ) = f″ i+1 (x end ) = 0. Therefore, combining the 4end equations in the above formula (11), interpolate the predicted short-term prediction trajectory, connection points, and the target lane centerline using formulas (10), (11), and the natural boundary conditions. Thus, for n connection points, a target reference line cluster with n reference lines can be obtained as Figure 4 shown, where Figure 4 is the model diagram of the target reference line cluster in this embodiment.

[0106] In addition, in one embodiment, based on the above step S230, generating a vehicle prediction trajectory of the target vehicle based on the target reference line cluster may specifically include the following steps:

[0107] Step S231, establish a Frenet coordinate system according to the target reference line cluster;

[0108] Step S232, based on the position information of the reference points selected from the target reference line cluster and the current motion state information of the target vehicle, obtain the starting point and ending point of the prediction trajectory of the target vehicle in the Frenet coordinate system.

[0109] Among them, the point closest to the current position of the target vehicle can be selected from the target reference line cluster as the reference point. Based on the coordinate information of this reference point and combined with the current motion state information of the target vehicle, calculate the initial state and termination state of the prediction trajectory of the target vehicle in the Frenet coordinate system, that is, the starting point and ending point. Specifically, it is shown as the following formula:

[0110]

[0111] Among them, k is the curvature of the reference line obtained by interpolation calculation, k x is the curvature of the historical trajectory, θ x is the heading angle of the historical trajectory. s 0 is the longitudinal coordinate of the starting point, is the longitudinal speed of the target vehicle at the starting point, is the longitudinal acceleration of the target vehicle at the starting point. d 0 is the lateral coordinate of the starting point, is the lateral velocity of the target vehicle at the initial point, is the lateral acceleration of the target vehicle at the initial point. θ 0 is the heading angle of the target vehicle at the current moment, v 0 is the speed of the target vehicle at the current moment, a 0 is the acceleration of the target vehicle at the current moment, and d is the d coordinate in the Frenet coordinate system. Based on the above formula, the initial point can be obtained as where the end point can be obtained by the following formula:

[0112]

[0113] where t is the predicted duration.

[0114] Step S233, generate the vehicle prediction trajectory of the target vehicle according to the initial point and the end point.

[0115] After determining the initial point and the end point, the coefficients of the fifth-degree polynomial of the lateral d and the coefficients of the fourth-degree polynomial of the longitudinal s in the following formula can be solved according to the initial point, the end point, and the predicted duration t, so as to determine the vehicle prediction trajectory in the Frenet coordinate system. The fifth-degree polynomial of the lateral d is:

[0116] d(t) = a 5 t 5 + a 4 t 2 + a 3 t 3 + a 2 t 2 + a 1 t 1 + a 0 (14)

[0117] The fourth-degree polynomial of the longitudinal s is:

[0118] s(t) = b 4 t 4 + b 3 t 3 + b 2 t 2 + b 1 t 1 + b 0 (15)

[0119] Solve all the above parameters a i and b i , and the vehicle prediction trajectory of the target vehicle with a duration of t can be obtained. Among them, the solution of the above parameter a i is as follows:

[0120]

[0121] The above parameter b i is solved as follows:

[0122]

[0123] The above steps S231 to S233 generate the vehicle prediction trajectory of the target vehicle in the Frenet coordinate system based on the above target reference line cluster that fits the vehicle trajectory characteristics and the conversion relationship between the Frenet coordinate system and the Cartesian coordinate system, which can improve the generation accuracy of the vehicle prediction trajectory.

[0124] In addition, in one embodiment, the above vehicle trajectory prediction method may further include the following steps:

[0125] Step S241: Calculate the cost values of all vehicle prediction trajectories of the target vehicle according to a preset cost function based on acceleration information and lateral overshoot information.

[0126] Among them, assuming that the generation frequency of the vehicle prediction trajectory is 10 Hz, the number of discrete points of the generated trajectory is num = t / 10. Substitute the time series into equations (16) and (17) to obtain the vehicle prediction trajectory in the Frenet coordinates as path f ={(s i , d i )} i＝1:num . Among them, the path f corresponding to different reference lines in the target reference line cluster is different. Therefore, a set of vehicle prediction trajectories and costs {path fi , cost i} i＝1:n can be obtained. Among them, cost i is the cost corresponding to the i-th vehicle prediction trajectory, and is specifically shown as the following formula:

[0127]

[0128] β 1 , β 2 , β 3 are coefficients that can be adaptively set based on the actual application scenario. Among them, the first term of the above formula measures the overshoot of the vehicle prediction trajectory, and the second and third terms measure the jerk in the lateral and longitudinal directions respectively.

[0129] Step S242: Screen the cost values of all vehicle prediction trajectories based on a preset screening condition to obtain the optimal vehicle prediction trajectory.

[0130] For example, the preset screening condition can be that the cost value is the smallest. Therefore, the vehicle prediction trajectory with the smallest cost value cost can be selected from the above vehicle prediction trajectory set as the optimal vehicle prediction trajectory, and then the optimal vehicle prediction trajectory obtained by screening is converted to the Cartesian coordinate system to obtain the final prediction trajectory of the target vehicle. i Among them, for the discrete points (s

[0131] , d i ) of the trajectory in the Frenet coordinate system, the process of converting it to the Cartesian coordinate system may include: selecting the projection point closest to the discrete point as the matching point, and calculating the curvature k i of the matching point, the heading angle θ r , and the matching point coordinates (x rj , y r,j ) according to the preset interpolation equation. Then, the following formula is used for coordinate transformation to obtain the trajectory in the Cartesian coordinate system: r,j ) of the trajectory in the Frenet coordinate system, the process of converting it to the Cartesian coordinate system may include: selecting the projection point closest to the discrete point as the matching point, and calculating the curvature k

[0132]

[0133] In the above steps S241 to S242, the vehicle prediction trajectories are screened through the cost function, and the most reasonable and safe prediction trajectory can be obtained while measuring the trajectory comfort and rationality of the target vehicle.

[0134] The following describes and illustrates this embodiment through preferred embodiments.

[0135] Figure 5 is the flowchart of the vehicle trajectory prediction method of this preferred embodiment. As Figure 5 shown, the vehicle trajectory prediction method includes the following steps:

[0136] Step S501, obtain the historical state information and lane information of the target vehicle from the perception module deployed on the vehicle;

[0137] Step S502, calculate the lane center line according to the lane boundary lines in the lane information;

[0138] Step S503, based on the historical state information, obtain the state information of the target vehicle at the current moment, 1 second before the current moment, and 1 second after the current moment, and predict the lane-changing intention of the target vehicle;

[0139] Step S504, input the motion state information of the target vehicle at the current moment into the preset vehicle kinematic model for short-term motion state prediction to obtain a short-term prediction trajectory;

[0140] Step S505: Based on the lane-changing intention, correct the lane centerline to obtain the target lane centerline. According to the preset interpolation algorithm, use the short-term prediction trajectory to compensate the target lane centerline to obtain the target reference line cluster;

[0141] Step S506: Generate the vehicle prediction trajectory according to the target reference line cluster, and select the vehicle prediction trajectory with the minimum cost value as the final prediction trajectory of the target vehicle based on the cost function.

[0142] Additionally, Figure 6 is the flowchart of the method for generating the target reference line cluster of this preferred embodiment. As Figure 6 shown, the method for generating the target reference line cluster includes the following steps:

[0143] Step S601: Obtain the motion state information of the target vehicle at the current moment;

[0144] Step S602: Input the motion state information of the target vehicle at the current moment into the CYRA model for motion state prediction to obtain the short-term prediction trajectory;

[0145] Step S603: Correct the lane centerline based on the lane-changing intention to obtain the target lane centerline;

[0146] Step S604: Determine whether the current discrete point in the target lane centerline meets Condition 1. If so, execute Step S605; otherwise, continue to traverse the next discrete point in the target lane centerline for judgment. Among them, Condition 1 is that the x-axis coordinate of the current discrete point in the target lane centerline is greater than the x-axis coordinate of the end point in the short-term prediction trajectory;

[0147] Step S605: Concatenate the short-term prediction trajectory and the target lane centerline, and generate the target reference line cluster based on cubic spline interpolation.

[0148] In addition, Figure 7 is the flowchart of the method for generating the vehicle prediction trajectory of this preferred embodiment. As Figure 7 shown, the method for generating the vehicle prediction trajectory includes the following steps:

[0149] Step S701: Obtain the target reference line cluster;

[0150] Step S702: Establish a Frenet coordinate system based on the reference lines in the target reference line cluster;

[0151] Step S703: Obtain the motion state of the target vehicle at the current moment;

[0152] Step S704: Determine the initial state and final state of the target vehicle in the Frenet coordinate system;

[0153] Step S705: Solve the fifth-degree polynomial of the lateral d(t) and the fourth-degree polynomial of the longitudinal s(t) based on the result of S704;

[0154] Step S706: Generate the predicted trajectory in the Frenet coordinate system and the corresponding cost value based on the solution result of S705;

[0155] Step S707: Select the predicted trajectory with the minimum cost value;

[0156] Step S708: Transform the predicted trajectory with the minimum cost from the Frenet coordinate system to the Cartesian coordinate system to obtain the vehicle predicted trajectory of the target vehicle.

[0157] In this embodiment, a vehicle trajectory prediction device is further provided. This device is used to implement the above embodiments and preferred implementation manners, and those that have been described will not be repeated here. The following terms such as "module", "unit", "sub-unit", etc. can be a combination of software and / or hardware that can achieve a predetermined function. Although the device described in the following embodiments is preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0158] Figure 8 is the structural block diagram of the vehicle trajectory prediction device 80 of this embodiment. As Figure 8 shown, the vehicle trajectory prediction device 80 includes: an acquisition module 82, a first generation module 84, and a second generation module 86; where:

[0159] The acquisition module 82 is used to determine the lane centerline of the target vehicle and predict the lane-changing intention of the target vehicle;

[0160] The first generation module 84 is used to generate the short-term predicted trajectory of the target vehicle based on a preset kinematic model, and compensate the lane centerline by combining a preset interpolation algorithm based on the lane-changing intention and the short-term predicted trajectory to obtain a target reference line cluster;

[0161] The second generation module 86 is used to generate the vehicle predicted trajectory of the target vehicle based on the target reference line cluster.

[0162] The above vehicle trajectory prediction device 80 determines the lane centerline of the target vehicle and predicts the lane-changing intention of the target vehicle; generates the short-term predicted trajectory of the target vehicle based on a preset kinematic model, and compensates the lane centerline by combining a preset interpolation algorithm based on the lane-changing intention and the short-term predicted trajectory to obtain a target reference line cluster; generates the vehicle predicted trajectory of the target vehicle based on the target reference line cluster. It realizes the compensation of the reference line through the short-term predicted trajectory based on the kinematic information of the vehicle, so as to obtain a reference line that conforms to the actual characteristics of the vehicle trajectory, and further improve the accuracy of the predicted result of the vehicle trajectory.

[0163] It should be noted that each of the above modules can be a functional module or a program module, and can be implemented either by software or by hardware. For the modules implemented by hardware, each of the above modules can be located in the same processor; or each of the above modules can also be located in different processors in any combination.

[0164] In this embodiment, an electronic device is further provided, including a memory and a processor. A computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0165] Optionally, the above electronic device may further include a transmission device and an input / output device. Among them, the transmission device is connected to the above processor, and the input / output device is connected to the above processor.

[0166] Optionally, in this embodiment, the above processor may be configured to execute the following steps through a computer program:

[0167] S1, determine the lane center line of the target vehicle and predict the lane-changing intention of the target vehicle;

[0168] S2, generate a short-term prediction trajectory of the target vehicle based on a preset kinematic model, and based on the lane-changing intention and the short-term prediction trajectory, combine a preset interpolation algorithm to compensate the lane center line to obtain a target reference line cluster;

[0169] S3, generate a vehicle prediction trajectory of the target vehicle based on the target reference line cluster.

[0170] It should be noted that specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation manners, and will not be repeated in this embodiment.

[0171] In addition, in combination with the vehicle trajectory prediction method provided in the above embodiment, a storage medium can also be provided to implement it in this embodiment. A computer program is stored on the storage medium; when the computer program is executed by a processor, it implements any one of the vehicle trajectory prediction methods in the above embodiment.

[0172] It should be understood that the specific embodiments described here are only used to explain this application, rather than to limit it. According to the embodiments provided in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this application.

[0173] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data that have been authorized by the user or fully authorized by all parties.

[0174] Obviously, the accompanying drawings are only some examples or embodiments of this application. For those of ordinary skill in the art, this application can also be applied to other similar situations based on these drawings without creative efforts. Additionally, it can be understood that although the work done during this development process may be complex and time-consuming, for those of ordinary skill in the art, certain design, manufacturing, or production changes based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient disclosure of this application.

[0175] The term "embodiment" in this application means that the specific features, structures, or characteristics described in connection with the embodiments may be included in at least one embodiment of this application. The phrase appears in various positions in the specification and does not necessarily mean the same embodiment, nor does it mean being independent or alternative to other embodiments and mutually exclusive. Those of ordinary skill in the art can clearly or implicitly understand that the embodiments described in this application can be combined with other embodiments without conflict.

[0176] The above-described embodiments merely represent several implementation manners of this application, and their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of patent protection. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several modifications and improvements can still be made, and these all fall within the protection scope of this application. Therefore, the protection scope of this application shall be subject to the appended claims.

Claims

1. A vehicle trajectory prediction method, characterized in that, it includes: Determine the lane center line of the target vehicle and predict the lane-changing intention of the target vehicle; Generate a short-term prediction trajectory of the target vehicle based on a preset kinematic model, and based on the lane-changing intention and the short-term prediction trajectory, combine a preset interpolation algorithm to compensate the lane center line to obtain a target reference line cluster; Generate a vehicle prediction trajectory of the target vehicle based on the target reference line cluster; Among them, the generating the short-term prediction trajectory of the target vehicle based on a preset kinematic model includes: Input the current motion state information of the target vehicle into a preset kinematic model, and predict the motion state information at a preset number of interval moments; Generate the short-term prediction trajectory based on the motion state information at the preset number of interval moments.

2. The vehicle trajectory prediction method according to claim 1, characterized in that, the determining the lane center line of the target vehicle includes: Calculate the lane center line of the target vehicle according to the pre-acquired lane boundary line information of the target vehicle.

3. The vehicle trajectory prediction method according to claim 1, characterized in that, the predicting the lane-changing intention of the target vehicle includes: According to the motion state information of the target vehicle, respectively determine the first position relationship between the target vehicle and the lane boundary line at a historical moment, the second position relationship between the target vehicle and the lane boundary line at the current moment, and predict the third position relationship between the target vehicle and the lane boundary line after a preset time period; Determine the lane-changing intention of the target vehicle by comparing the first position relationship, the second position relationship, and the third position relationship.

4. The vehicle trajectory prediction method according to claim 1, characterized in that, the combining the lane-changing intention and the short-term prediction trajectory, and combining a preset interpolation algorithm to compensate the lane center line to obtain a target reference line cluster includes: Correct the lane center line according to the lane-changing intention to obtain a target lane center line; Based on the principle of the cubic spline interpolation algorithm, use the short-term prediction trajectory to compensate the target lane center line to obtain the target reference line cluster.

5. The vehicle trajectory prediction method according to claim 4, characterized in that, the based on the principle of the cubic spline interpolation algorithm, using the short-term prediction trajectory to compensate the target lane center line to obtain the target reference line cluster includes: According to the coordinate information in the short-term prediction trajectory, continuously extract a preset number of connection points from the target lane center line; Based on the principle of the cubic spline interpolation algorithm, use the short-term prediction trajectory to pass through the connection points to compensate the target lane center line to obtain the target reference line cluster.

6. The vehicle trajectory prediction method according to claim 1, characterized in that, the generating the vehicle prediction trajectory of the target vehicle based on the target reference line cluster includes: Establish a Frenet coordinate system according to the target reference line cluster; Obtain the initial point and the end point of the predicted trajectory of the target vehicle in the Frenet coordinate system based on the position information of the reference points selected from the target reference line cluster and the current motion state information of the target vehicle; Generate the vehicle predicted trajectory of the target vehicle according to the initial point and the end point.

7. The vehicle trajectory prediction method according to any one of claims 1 to 6, characterized in that, after generating the vehicle predicted trajectory of the target vehicle based on the target reference line cluster, the method further includes: Calculate the cost values of all vehicle predicted trajectories of the target vehicle according to a preset cost function based on acceleration information and lateral overshoot information; Screen the cost values of all vehicle predicted trajectories based on a preset screening condition to obtain an optimal vehicle predicted trajectory.

8. A vehicle trajectory prediction device, characterized in that, comprising: An acquisition module, a first generation module, and a second generation module; wherein: The acquisition module is configured to determine the lane center line of the target vehicle and predict the lane-changing intention of the target vehicle; The first generation module is configured to generate a short-term predicted trajectory of the target vehicle based on a preset kinematic model, and based on the lane-changing intention and the short-term predicted trajectory, combine a preset interpolation algorithm to compensate the lane center line to obtain a target reference line cluster; The second generation module is configured to generate a vehicle predicted trajectory of the target vehicle based on the target reference line cluster; wherein, generating the short-term predicted trajectory of the target vehicle based on the preset kinematic model includes: Input the current motion state information of the target vehicle into a preset kinematic model to predict the motion state information at a preset number of interval moments; generate the short-term predicted trajectory based on the motion state information at the preset number of interval moments.

9. An electronic device, comprising a memory and a processor, characterized in that, A computer program is stored in the memory, and the processor is configured to run the computer program to execute the vehicle trajectory prediction method according to any one of claims 1 to 7.

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

Images