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

Abstract

The present invention provides a vehicle trajectory prediction method, device, storage medium and electronic device, relating to the technical field of vehicle trajectory prediction. When predicting the lane-changing trajectory of a target vehicle in an embodiment of the present invention, the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane-changing completion duration are obtained based on a car-following model. Subsequently, several control points for constructing a lane-changing prediction path are selected according to the obtained displacements, and the control points are interpolated using a Bessel function to obtain a first lane-changing prediction path. Then, a supplementary path is obtained according to the lane centerline between the lane-changing prediction end point and the end control point, and finally a second lane-changing prediction path is obtained. Finally, the speeds of each path point are obtained using the car-following model, so that the finally obtained lane-changing prediction trajectory is more in line with the actual vehicle driving trajectory points and has a smaller deviation, which is more conducive to accurate decision-making by autonomous driving vehicles, thereby ensuring driving safety.

Description

Technical Field

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

[0002] In vehicle prediction, it is divided into two processes: intention prediction and trajectory prediction. Among them, intention prediction is to give the probability of each behavior intention through a deep learning neural network according to different scenarios. And trajectory prediction is to consider the interaction with the host vehicle.

[0003] For trajectory prediction, current methods mostly use the method of sampling + cost function selection, where the cost function parameters are trained using a neural network, and then a reasonable trajectory prediction is given.

[0004] However, the existing methods do not consider the influence of the interaction between adjacent vehicles in a driving fleet on a one-way street with restricted overtaking on trajectory prediction, resulting in the predicted trajectory not being able to well match the actual driving scenario. Summary of the Invention

[0005] (1) Technical Problems to be Solved

[0006] Aiming at the deficiencies of the existing technology, the present invention provides a vehicle trajectory prediction method, device, storage medium and electronic device, which solves the problem of how to obtain a vehicle trajectory that better matches the actual driving scenario.

[0007] (2) Technical Solutions

[0008] To achieve the above object, the present invention is realized through the following technical solutions:

[0009] In the first aspect, a vehicle trajectory prediction method is provided, and the method includes:

[0010] Obtain the behavior intention of the target vehicle;

[0011] When the behavior intention of the target vehicle is to change lanes, obtain the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane change completion duration based on the car-following model; and the prediction duration is greater than the lane change completion duration;

[0012] Obtain a number of control points based on the displacement of the target vehicle within the lane change completion duration, where the control points include a starting control point, an ending control point and a number of intermediate control points, and the control points are all located on the center line of the lane; and then perform interpolation on the control points based on the Bessel function to obtain the first lane change prediction path;

[0013] Obtain the lane change prediction end point based on the virtual point and the displacement of the target vehicle within the prediction duration; the virtual point is the point on the center line of the adjacent lane of the target vehicle that is closest to the target vehicle;

[0014] A supplementary path is obtained based on the lane centerline between the lane-changing prediction end point and the end control point;

[0015] The first lane-changing prediction path and the supplementary path are spliced to obtain a second lane-changing prediction path;

[0016] Based on the car-following model, the speeds of each path point in the second lane-changing prediction path are obtained to get a lane-changing prediction trajectory.

[0017] Furthermore, in the car-following model at time t, the calculation formulas for the desired headway of vehicle i and the acceleration of vehicle i are as follows:

[0018]

[0019]

[0020] where

[0021] represents the desired headway of vehicle i at time t;

[0022] d0 represents the static safety headway;

[0023] T represents the safe time headway;

[0024] v i，t represents the speed of vehicle i in the moving vehicle platoon at time t;

[0025] Δv i，t represents the speed difference between vehicle i and the vehicle in front at time t;

[0026] a max represents the maximum acceleration;

[0027] a c represents the comfortable deceleration;

[0028] a i，t represents the acceleration of vehicle i corresponding to the target vehicle in the moving vehicle platoon in the car-following model at time t;

[0029] a max represents the maximum acceleration;

[0030] v0 represents the free flow speed;

[0031] δ represents the speed power coefficient;

[0032] represents the actual headway between vehicle i and the vehicle in front at time t;

[0033] and the actual headway of the target vehicle at time t + 1 has the following calculation formula:

[0034] v i，t+1 = v i，t + a i，t Δt

[0035]

[0036]

[0037] wherein,

[0038] v i，t+1 represents the speed of vehicle i at time t + 1;

[0039] v i，t represents the speed of vehicle i at time t;

[0040] a i，t represents the acceleration of vehicle i at time t;

[0041] Δt represents the time interval between time t and time t + 1;

[0042] s i，t+1 represents the displacement of vehicle i from time t to time t + 1;

[0043] represents the actual inter-vehicle distance between vehicle i and the preceding vehicle i - 1 at time t + 1;

[0044] represents the actual inter-vehicle distance between vehicle i and the preceding vehicle i - 1 at time t;

[0045] v i-1 represents the speed of the preceding vehicle i - 1.

[0046] Furthermore, the calculation methods for the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane change completion duration are as follows:

[0047]

[0048]

[0049] wherein, t0 represents the prediction duration;

[0050] s0 represents the displacement of the target vehicle within the prediction duration t0;

[0051] t1 represents the lane change completion duration;

[0052] s1 represents the displacement of the target vehicle within the lane change completion duration t1.

[0053] Further, interpolation is performed on the control points based on the fifth-order Bessel function to obtain the first lane-changing prediction path; and the obtained several control points include:

[0054] Take the position P1(x1, y1) of the target vehicle in the vector map as the first control point;

[0055] Obtain the second control point P2(x2, y2) based on the first control point P1(x1, y1) and the first distance β1; and the first distance β1 represents the displacement of the target vehicle from the first control point P1(x1, y1) after moving for a duration of t2 based on the car-following model; t2 represents the first time period;

[0056] Obtain the third control point P3(x3, y3) based on the first control point P1(x1, y1), the second control point P2(x2, y2) and the second distance β2; and the second distance β2 represents the displacement of the target vehicle from the second control point P2(x2, y2) after moving for a duration of t3 based on the car-following model; t3 represents the second time period, which is a preset value set according to human experience;

[0057] Obtain the sixth control point P6(x6, y6) based on the virtual point P0(x0, y0) and the displacement s1 of the target vehicle within the lane-changing completion duration t1;

[0058] Obtain the fifth control point P5(x5, y5) based on the sixth control point P6(x6, y6) and the third distance β3; and the third distance β3 represents the displacement of the target vehicle from the sixth control point P6(x6, y6) when moving backward for a duration of t2 based on the car-following model;

[0059] Obtain the fourth control point P4(x4, y4) based on the sixth control point P6(x6, y6), the fifth control point P5(x5, y5) and the fourth distance β4, and the fourth distance β4 represents the displacement of the target vehicle from the fifth control point P5(x5, y5) when moving backward for a duration of t3 based on the car-following model;

[0060] Among them, the first control point P1(x1, y1) is the starting control point; the sixth control point P6(x6, y6) is the ending control point; the fifth control point P5(x5, y5), the fourth control point P4(x4, y4), the third control point P3(x3, y3), and the second control point P2(x2, y2) are all intermediate control points, and x and y represent the coordinate values corresponding to the control point.

[0061] Further, the speed calculation method for each path point in the second lane-changing prediction path is:

[0062]

[0063] Among them, v i，t+1Represents the speed of the corresponding path point at time t+1.

[0064] Further, the method further includes:

[0065] When the behavior intention of the target vehicle is to go straight,

[0066] Obtain the displacement s0 of the target vehicle within the prediction duration t0 and the first control point P1(x1, y1); the first control point is the position of the target vehicle in the vector map;

[0067] Take the point on the center line at a distance of s0 from the first control point P1(x1, y1) in the driving direction as the straight-line prediction end point P8(x8, y8), and then obtain the straight-line prediction path from P1(x1, y1) to P8(x8, y8);

[0068] Then, based on the car-following model, iterate to obtain the speeds of each path point in the straight-line prediction path to obtain the straight-line prediction trajectory.

[0069] Further, both the supplementary path and the straight-line prediction path use first-order Bezier for uniform interpolation between two points.

[0070] In a second aspect, a vehicle trajectory prediction device is provided, and the device includes:

[0071] A behavior intention acquisition module, configured to acquire the behavior intention of the target vehicle;

[0072] A displacement calculation module, configured to, when the behavior intention of the target vehicle is to change lanes, based on the car-following model, obtain the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane change completion duration; and the prediction duration is greater than the lane change completion duration;

[0073] A first lane change prediction path generation module, configured to obtain a plurality of control points based on the displacement of the target vehicle within the lane change completion duration, the control points include a starting control point, an ending control point, and a plurality of intermediate control points, and the control points are all located on the lane center line; then, based on the Bezier function, interpolate the control points to obtain the first lane change prediction path;

[0074] A lane change prediction end point acquisition module, configured to obtain a lane change prediction end point based on a virtual point and the displacement of the target vehicle within the prediction duration; the virtual point is the point on the center line of the adjacent lane of the target vehicle that is closest to the target vehicle;

[0075] A supplementary path generation module, configured to obtain a supplementary path based on the center line between the lane change prediction end point and the ending control point;

[0076] A second lane change prediction path generation module, configured to splice the first lane change prediction path and the supplementary path to obtain the second lane change prediction path;

[0077] The lane-changing prediction trajectory generation module is used to obtain the speeds of the respective path points in the second lane-changing prediction path based on the car-following model, and obtain the lane-changing prediction trajectory.

[0078] In a third aspect, a storage medium is provided, which stores a computer program for vehicle trajectory prediction, wherein the computer program causes a computer to execute the above-mentioned vehicle trajectory prediction method.

[0079] In a fourth aspect, an electronic device is provided, including:

[0080] One or more processors;

[0081] A memory; and

[0082] One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs include those for executing the above-mentioned vehicle trajectory prediction method.

[0083] (III) Advantageous Effects

[0084] The present invention provides a vehicle trajectory prediction method, device, storage medium and electronic device. Compared with the prior art, the following advantageous effects are achieved:

[0085] When predicting the lane-changing trajectory of a target vehicle in an embodiment of the present invention, the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane-changing completion duration are obtained based on the car-following model. Subsequently, several control points for constructing the lane-changing prediction path are selected according to the obtained displacements, and the control points are interpolated using the Bezier function to obtain the first lane-changing prediction path. Then, a supplementary path is obtained based on the lane centerline between the lane-changing prediction end point and the end control point, and finally, the second lane-changing prediction path is obtained. Finally, the speeds of the respective path points are obtained using the car-following model, so that the finally obtained lane-changing prediction trajectory is more in line with the actual vehicle driving trajectory points and has a smaller deviation, which is more conducive to accurate decision-making by autonomous driving vehicles and thus ensures driving safety. Description of the Drawings

[0086] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0087] Figure 1 It is a flowchart of an embodiment of the present invention;

[0088] Figure 2Schematic diagram of the lane-changing intention in the embodiment of the present invention;

[0089] Figure 3 Schematic diagram of the straight-ahead intention in the embodiment of the present invention. Detailed implementation manners

[0090] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are described clearly and completely. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0091] The embodiments of the present application provide a vehicle trajectory prediction method, device, storage medium, and electronic device, which solve the problem of how to obtain a prediction trajectory that is more in line with the actual situation.

[0092] The technical solutions in the embodiments of the present application to solve the above technical problems are generally as follows:

[0093] In this paper, an intelligent driving model (IDM) is used to develop an algorithm for the ACC function; considering that the longitudinal speed planning of this algorithm is more in line with the driving habits of drivers, an iterative calculation formula of "time - speed - position" is added; the prediction distance and the speed information corresponding to the path points are solved.

[0094] The fifth-order Bezier curve has more control points than the third-order Bezier curve, which can control the generated path shape more flexibly; among them, the calculation of the control points is related to the speed, so that the path curvature becomes smaller as the speed increases, and the area is smoother, better conforming to the driving path of the driver.

[0095] To better understand the above technical solutions, the above technical solutions will be described in detail below in conjunction with the accompanying drawings of the specification and specific implementation manners.

[0096] Embodiment 1:

[0097] As Figure 1 shown, the present invention provides a vehicle trajectory prediction method, which includes:

[0098] Obtain the behavior intention of the target vehicle;

[0099] When the behavior intention of the target vehicle is to change lanes, based on the car-following model, obtain the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane-changing completion duration; and the prediction duration is greater than the lane-changing completion duration;

[0100] Obtain a number of control points based on the displacement of the target vehicle within the lane change completion duration. The control points include a starting control point, an ending control point, and a number of intermediate control points, and all the control points are located on the center line of the lane. Then, perform interpolation on the control points based on the Bessel function to obtain the first lane change prediction path.

[0101] Obtain the lane change prediction end point based on the virtual point and the displacement of the target vehicle within the prediction duration. The virtual point is the point on the center line of the adjacent lane of the target vehicle that is closest to the target vehicle.

[0102] Obtain a supplementary path based on the center line of the lane between the lane change prediction end point and the ending control point.

[0103] Stitch the first lane change prediction path and the supplementary path to obtain the second lane change prediction path.

[0104] Obtain the speed of each path point in the second lane change prediction path based on the car-following model to obtain the lane change prediction trajectory.

[0105] The beneficial effects of this embodiment are as follows:

[0106] When predicting the lane change trajectory of the target vehicle in the embodiment of the present invention, the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane change completion duration are obtained based on the car-following model. Subsequently, a number of control points for constructing the lane change prediction path are selected according to the obtained displacements, and interpolation is performed on the control points using the Bessel function to obtain the first lane change prediction path. Then, a supplementary path is obtained based on the center line of the lane between the lane change prediction end point and the ending control point, and finally, the second lane change prediction path is obtained. Finally, the speed of each path point is obtained using the car-following model, so that the finally obtained lane change prediction trajectory is more in line with the actual vehicle driving trajectory points and has a smaller deviation, which is more conducive to accurate decision-making of autonomous driving vehicles and thus ensures driving safety.

[0107] The implementation process of the embodiment of the present invention will be described in detail below:

[0108] S1. Obtain the behavior intention of the target vehicle.

[0109] In the specific implementation of this embodiment, the specific method for obtaining the behavior intention is not limited. For example, according to different scenarios, the probability of each behavior intention can be given by a deep learning neural network.

[0110] S2. When the behavior intention of the target vehicle is to change lanes, obtain the displacement s0 of the target vehicle within the prediction duration t0 and the displacement s1 of the target vehicle within the lane change completion duration t1 based on the car-following model; and t0 > t1.

[0111] In the specific implementation: t0 can be set to 8 - 10 seconds, t1 can be set to 6 - 8 seconds, and the displacement s1 is calculated in the following manner:

[0112] S2.1. Obtain the point on the center line of the adjacent lane of the target vehicle that is closest to the target vehicle as the virtual point representing the target vehicle;

[0113] Assume that the position of the vehicle is the vehicle's center of mass or the position of the rear axle of the vehicle. If the position of the target vehicle in the vector map is P1(x1, y1), then the virtual point is denoted as P0(x0, y0).

[0114] S2.2. Based on the virtual point P0(x0, y0) and the car-following model, obtain the acceleration a of the target vehicle at time t i，t ; and the acceleration a i，t The calculation formula is:

[0115]

[0116]

[0117] where,

[0118] represents the desired headway of vehicle i at time t; (unknown, calculation result);

[0119] d0 represents the static safety distance; (its value is preset in advance, for example, it can be set to 5 - 10 meters);

[0120] T represents the safe time headway; (its value is preset in advance, for example, it can be set to 1.5 seconds);

[0121] v i，t represents the speed of vehicle i in the moving vehicle platoon at time t; (known quantity, obtained by reading vehicle information);

[0122] Δv i，t represents the speed difference between vehicle i and the speed of the vehicle in front at time t; (known, collected by sensors for the vehicle in front / adjacent vehicle);

[0123] a max represents the maximum acceleration; (its value is preset in advance, for example, it is set to 1.4 m·s -2 ) in existing references;

[0124] a c represents the comfortable deceleration; (its value is preset in advance, for example, it is set to 2 m·s -2 ) in existing references;

[0125] a i，t represents the acceleration of vehicle i corresponding to the target vehicle in the moving vehicle platoon in the car-following model at time t; (unknown, calculation result);

[0126] a maxDenote the maximum acceleration; (its value is preset, for example, it is set to 1.4 m·s in existing references -2 );

[0127] v0 denotes the free flow speed; (a known quantity, the traffic rules set the cruising speed, for example, it is set to 120 km·h in existing references -1 );

[0128] δ denotes the speed power coefficient; (its value can be obtained by calibrating the model parameters, for example, δ = 4 in existing references);

[0129] Denote the actual distance between vehicle i and the vehicle in front at time t; (a known quantity, collected by the vehicle's sensors).

[0130] S2.3. Based on the acceleration a of the target vehicle at time t i，t Iteratively obtain the vehicle information of the target vehicle at time t + 1 until the displacement s0 of the target vehicle within the prediction duration t0 and the displacement s1 of the target vehicle within the lane change completion duration t1 are obtained.

[0131] In specific implementation, the following formula can be used to obtain the real-time speed, displacement and actual distance between vehicles.

[0132] Among them, iteratively obtaining the vehicle information of the target vehicle at time t + 1 is as follows:

[0133] v i，t+1 = v i，t + a i，t Δt

[0134]

[0135]

[0136] Among them,

[0137] v i，t+1 Denotes the speed of vehicle i at time t + 1;

[0138] v i，t Denotes the speed of vehicle i at time t;

[0139] a i，t Denotes the acceleration of vehicle i at time t;

[0140] Δt denotes the interval duration between time t and time t + 1;

[0141] s i，t+1 Denotes the displacement of vehicle i from time t to time t + 1;

[0142] Denote the actual inter-vehicle distance between vehicle \(i\) and the preceding vehicle \(i - 1\) at time \(t+1\).

[0143] Denote the actual inter-vehicle distance between vehicle \(i\) and the preceding vehicle \(i - 1\) at time \(t\).

[0144] v i-1 Denote the speed of the preceding vehicle \(i - 1\), assuming that the speed of the preceding vehicle remains constant and moves at a uniform speed.

[0145] Thus, through the iterative calculation using the above formula, the displacement \(s1\) of the target vehicle within the lane-changing completion duration \(t1\) can be obtained.

[0146]

[0147] And the calculation process of the displacement \(s0\) of the target vehicle within the prediction duration \(t0\) is similar. Continue the iterative calculation until time \(t0\). The formula is as follows:

[0148]

[0149] S3. Based on the displacement \(s1\) of the target vehicle within the lane-changing completion duration \(t1\), obtain a number of control points. The control points include the starting control point \(P1(x1, y1)\), the ending control point \(P6(x6, y6)\) and several intermediate control points, and all the control points are located on the center line of the lane; then, based on the Bessel function, interpolate the control points to obtain the first lane-changing prediction path.

[0150] In specific implementation, the control points are used to generate a path using the Bessel function later. The number of them depends on the type of Bessel function used. An \(n\)-th order Bessel curve has \(n + 1\) vertices. However, in order to obtain a path that more conforms to the actual vehicle driving trajectory, a fifth-order Bessel curve is adopted in this embodiment, that is, as Figure 2 shown, 6 control points need to be set: \(P1(x1, y1)\), \(P2(x2, y2)\), \(P3(x3, y3)\), \(P4(x4, y4)\), \(P5(x5, y5)\), \(P6(x6, y6)\); where \(x\) and \(y\) represent the coordinate values corresponding to the control point.

[0151] Then, obtaining a number of control points based on the displacement \(s1\) of the target vehicle within the lane-changing completion duration \(t1\) includes the following steps:

[0152] S3.1. Obtaining a number of control points based on the displacement \(s1\) of the target vehicle within the lane-changing completion duration \(t1\) includes:

[0153] ① For the first control point \(P1(x1, y1)\), that is, the starting control point:

[0154] Take the position \(P1(x1, y1)\) of the target vehicle in the vector map as the starting control point / the first control point.

[0155] ②For the second control point P2(x2, y2):

[0156] Obtain the second control point P2(x2, y2) based on the first control point P1(x1, y1) and the first distance β1.

[0157] And taking Figure 2 as an example, the specific calculation formula is:[[]]END]]

[0158] x2 = x0 + β1

[0159] y2 = y0

[0160] Wherein, the first distance β1 represents the displacement of the target vehicle based on the car-following model after moving for a duration of t2 from the first control point P1(x1, y1); t2 represents the first time period, which is a preset value set according to human experience.

[0161] ③For the third control point P3(x3, y3):

[0162] Obtain the third control point P3(x3, y3) based on the first control point P1(x1, y1), the second control point P2(x2, y2) and the second distance β2.

[0163] And taking Figure 2 as an example, the specific calculation formula is:[[]]END]]

[0164] x3 = x2 + β2 * sin(θ1)

[0165] y3 = y2 + β2 * cos(θ1)

[0166]

[0167] Wherein, the second distance β2 represents the displacement of the target vehicle based on the car-following model after moving for a duration of t3 from the second control point P2(x2, y2); t3 represents the second time period, which is a preset value set according to human experience.

[0168] ④For the sixth control point P6(x6, y6), i.e., the end control point:

[0169] Obtain the sixth control point P6(x6, y6) based on the virtual point P0(x0, y0) and the displacement s1 of the target vehicle within the lane change completion duration t1.

[0170] And taking Figure 2 as an example, the calculation formula for the sixth control point P6(x6, y6) is:[[]]END]]

[0171] x6 = x0 + s1

[0172] y6 = y0

[0173] ⑤For the fifth control point P5(x5, y5):

[0174] Obtain the fifth control point P5(x5, y5) based on the sixth control point P6(x6, y6) and the third distance β3.

[0175] And taking Figure 2 as an example, the specific calculation formula is:[[]]

[0176] x5 = x6 - β3

[0177] y6 = y6

[0178] Among them, the third distance β3 represents the displacement of the target vehicle based on the car-following model when reversing from the sixth control point P6(x6, y6) for a duration of t2.

[0179] ⑥For the fourth control point P4(x4, y4):

[0180] Obtain the fourth control point P4(x4, y4) based on the sixth control point P6(x6, y6), the fifth control point P5(x5, y5) and the fourth distance β4.

[0181] x4 = x5 - β4 * sin(θ2)

[0182] y4 = y5 - β4 * cos(θ2)

[0183]

[0184] Among them, the fourth distance β4 represents the displacement of the target vehicle based on the car-following model when reversing from the fifth control point P5(x5, y5) for a duration of t3.

[0185] Obviously, in this embodiment, the first control point is the starting control point, the second to fifth control points are four intermediate control points, and the sixth control point is the ending control point.

[0186] 3.2 Interpolate the control points based on the Bessel function to obtain the first lane-changing prediction path.

[0187] As Figure 2 shown, in the specific implementation, use the fifth-order Bessel function to combine the above 6 control points to construct the first lane-changing prediction path passing through the first control point P1(x1, y1) to the sixth control point P6(x6, y6), and the first lane-changing prediction path contains several path points.

[0188] S4. Obtain the lane-changing prediction end point P7(x7, y7) based on the virtual point P0(x0, y0) and the displacement s0 of the target vehicle within the prediction duration t0; the virtual point P0(x0, y0) is the point on the center line of the adjacent lane of the target vehicle that is closest to the target vehicle.

[0189] In specific implementation, taking Figure 2 as an example, the calculation method of the lane-changing prediction end point P7(x7, y7) is as follows:

[0190] x7 = x0 + s0

[0191] y7 = y0

[0192] S5. Obtain a supplementary path based on the center line of the lane between the lane-changing prediction end point P7(x7, y7) and the end control point P6(x6, y6).

[0193] In specific implementation, the generation method of the path points of the supplementary path is not limited. For example, uniform interpolation between two points can be performed through first-order Bezier.

[0194] S6. Concatenate the first lane-changing prediction path and the supplementary path to obtain a second lane-changing prediction path, where the second lane-changing prediction path is the path from the first control point P1(x1, y1) to the lane-changing prediction end point P7(x7, y7).

[0195] S7. Based on the car-following model, obtain the speeds of the respective path points in the second lane-changing prediction path to obtain a lane-changing prediction trajectory.

[0196] In specific implementation, it includes the following steps:

[0197] S7.1. Obtain the speed of the path point corresponding to the first control point P1(x1, y1) in the second lane-changing prediction path, that is, v i，0 , which is a known quantity;

[0198] S7.2. Based on the car-following model, iteratively obtain the next path point in the second lane-changing prediction path in sequence until the speeds of all path points are obtained to obtain a lane-changing prediction trajectory.

[0199]

[0200] At this time, s i，t+1 is the distance between adjacent two path points, a known quantity;

[0201] v i，t is calculated from the previous step, a known quantity;

[0202] a i，t is calculated by the car-following model, a known quantity;

[0203] v i，t+1 represents the speed of the path point corresponding to the (t + 1) moment.

[0204] Embodiment 2:

[0205] On the basis of Embodiment 1, the following steps may further be included:

[0206] As Figure 3 shown, when the behavior intention of the target vehicle is to go straight,

[0207] obtain the displacement s0 of the target vehicle within the prediction duration t0 and the first control point P1(x1, y1) according to the steps of Embodiment 1;

[0208] Take the point on the center line at a distance of s0 from the first control point P1(x1, y1) in the driving direction as the straight-line prediction end point P8(x8, y8), and then obtain the straight-line prediction path from P1(x1, y1) to P8(x8, y8);

[0209] Then, based on the car-following model, iterate to obtain the speeds of each path point in the straight-line prediction path to obtain the straight-line prediction trajectory.

[0210] Embodiment 3:

[0211] A vehicle trajectory prediction device, the device includes:

[0212] A behavior intention acquisition module, used to acquire the behavior intention of the target vehicle;

[0213] A displacement calculation module, used to, when the behavior intention of the target vehicle is to change lanes, obtain the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane change completion duration based on the car-following model; and the prediction duration is greater than the lane change completion duration;

[0214] A first lane change prediction path generation module, used to obtain a number of control points based on the displacement of the target vehicle within the lane change completion duration, the control points include a starting control point, an ending control point, and a number of intermediate control points, and the control points are all located on the lane center line; then interpolate the control points based on the Bessel function to obtain the first lane change prediction path;

[0215] A lane change prediction end point acquisition module, used to obtain the lane change prediction end point based on the virtual point and the displacement of the target vehicle within the prediction duration; the virtual point is the point on the center line of the adjacent lane of the target vehicle that is closest to the target vehicle;

[0216] A supplementary path generation module, used to obtain a supplementary path based on the center line between the lane change prediction end point and the ending control point;

[0217] A second lane change prediction path generation module, used to splice the first lane change prediction path and the supplementary path to obtain the second lane change prediction path;

[0218] A lane change prediction trajectory generation module, used to obtain the speeds of each path point in the second lane change prediction path based on the car-following model to obtain the lane change prediction trajectory.

[0219] In addition, the device may further include:

[0220] A straight - line trajectory prediction module, which is used to take the point on the center line at a distance of s0 from the first control point P1(x1, y1) in the driving direction as the straight - line prediction end point P8(x8, y8), and obtain the straight - line prediction path from P1(x1, y1) to P8(x8, y8); then iteratively obtain the speeds of each path point in the straight - line prediction path based on the car - following model to obtain the straight - line prediction trajectory.

[0221] Example 4:

[0222] A storage medium stores a computer program for vehicle trajectory prediction, wherein the computer program causes a computer to perform the following steps:

[0223] Obtain the behavior intention of the target vehicle;

[0224] When the behavior intention of the target vehicle is to change lanes, obtain the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane - change completion duration based on the car - following model; and the prediction duration is greater than the lane - change completion duration;

[0225] Obtain a number of control points based on the displacement of the target vehicle within the lane - change completion duration, where the control points include a starting control point, an ending control point, and a number of intermediate control points, and the control points are all located on the lane center line; then perform interpolation on the control points based on the Bessel function to obtain the first lane - change prediction path;

[0226] Obtain the lane - change prediction end point based on the virtual point and the displacement of the target vehicle within the prediction duration; the virtual point is the point on the center line of the adjacent lane of the target vehicle that is closest to the target vehicle;

[0227] Obtain a supplementary path based on the lane center line between the lane - change prediction end point and the ending control point;

[0228] Splice the first lane - change prediction path and the supplementary path to obtain the second lane - change prediction path;

[0229] Obtain the speeds of each path point in the second lane - change prediction path based on the car - following model to obtain the lane - change prediction trajectory.

[0230] Example 5:

[0231] An electronic device includes:

[0232] One or more processors;

[0233] A memory; and

[0234] One or more programs, where the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the programs include steps for performing the following:

[0235] Obtain the behavior intention of the target vehicle;

[0236] When the behavior intention of the target vehicle is to change lanes, based on the car-following model, obtain the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane change completion duration; and the prediction duration is greater than the lane change completion duration;

[0237] Based on the displacement of the target vehicle within the lane change completion duration, obtain a number of control points. The control points include a starting control point, an ending control point, and several intermediate control points, and all the control points are located on the center line of the lane; then, based on the Bessel function, interpolate the control points to obtain the first lane change prediction path;

[0238] Based on the virtual point and the displacement of the target vehicle within the prediction duration, obtain the lane change prediction end point; the virtual point is the point on the center line of the adjacent lane of the target vehicle that is closest to the target vehicle;

[0239] Based on the center line of the lane between the lane change prediction end point and the ending control point, obtain the supplementary path;

[0240] Splice the first lane change prediction path and the supplementary path to obtain the second lane change prediction path;

[0241] Based on the car-following model, obtain the speeds of each path point in the second lane change prediction path to obtain the lane change prediction trajectory.

[0242] It can be understood that the vehicle trajectory prediction device, storage medium, and electronic device provided in the embodiments of the present invention correspond to the above vehicle trajectory prediction method. For the explanations, examples, beneficial effects, etc. of the relevant content, reference can be made to the corresponding content in the vehicle trajectory prediction method, which will not be elaborated here.

[0243] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0244] When predicting the lane change trajectory of the target vehicle, the present invention obtains the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane change completion duration based on the car-following model, and then selects the control points for constructing the lane change prediction path, and uses the Bessel function to interpolate the control points to obtain the first lane change prediction path. Then, according to the center line of the lane between the lane change prediction end point and the ending control point, obtain the supplementary path, and finally obtain the second lane change prediction path. Finally, use the car-following model to obtain the speeds of each path point, so that the finally obtained lane change prediction trajectory is more in line with the actual vehicle driving trajectory points and has a smaller deviation, which is more conducive to the accurate decision-making of autonomous driving vehicles, thereby ensuring driving safety.

[0245] It should be noted that through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments. In this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, article or device including the said element.

[0246] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A vehicle trajectory prediction method, characterized in that, The method includes: Obtaining the behavior intention of the target vehicle; When the behavior intention of the target vehicle is to change lanes, obtaining the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane change completion duration based on the car-following model; and the prediction duration is greater than the lane change completion duration; Obtaining a number of control points based on the displacement of the target vehicle within the lane change completion duration, where the control points include a starting control point, an ending control point, and a number of intermediate control points, and the control points are all located on the center line of the lane; and then interpolating the control points based on the Bessel function to obtain the first lane change prediction path; Obtaining the lane change prediction end point based on the virtual point and the displacement of the target vehicle within the prediction duration; the virtual point is the point on the center line of the adjacent lane of the target vehicle that is closest to the target vehicle; Obtaining a supplementary path based on the center line of the lane between the lane change prediction end point and the ending control point; Splicing the first lane change prediction path and the supplementary path to obtain the second lane change prediction path; Obtaining the speeds of the respective path points in the second lane change prediction path based on the car-following model to obtain the lane change prediction trajectory.

2. The vehicle trajectory prediction method according to claim 1, characterized in that, In the car-following model at time t, the calculation formula for the desired headway of vehicle i and the acceleration of vehicle i is: Where, Denote the desired inter-vehicle distance of vehicle i at time t; d0 represents the static safety headway; T represents the safety time headway; v i，t represents the speed of vehicle i in the moving vehicle fleet at time t; Δv i，t represents the speed difference obtained by subtracting the speed of the vehicle in front from the speed of vehicle i at time t; a max represents the maximum acceleration; a c Indicates a comfortable deceleration; a i，t represents the acceleration of vehicle i corresponding to the target vehicle in the platoon at time t in the car-following model; a max represents the maximum acceleration; v0 represents the free flow speed; δ represents the speed power coefficient; Denote the actual inter-vehicle distance between vehicle i and the preceding vehicle at time t; And the actual inter-vehicle distance of the target vehicle at time t+1 is calculated as follows: v i，t+1 = v i，t + a i，t Δt Where, v i，t+1 represents the speed of vehicle i at time t + 1; v i，t represents the speed of vehicle i at time t; a i，t Represents the acceleration of vehicle i at time t; Δt represents the time interval between time t and time t + 1; s i，t+1 represents the displacement of vehicle i from time t to time t + 1; Denote the actual inter-vehicle distance between vehicle i and the preceding vehicle i-1 at time t+1; Denote the actual inter-vehicle distance between vehicle i and the preceding vehicle i - 1 at time t; v i-1 represents the speed of the leading vehicle i - 1.

3. A vehicle trajectory prediction method according to claim 2, characterized in that, The calculation method for the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane change completion duration is: Where, t0 represents the prediction duration; s0 represents the displacement of the target vehicle within the prediction duration t0; t1 represents the lane change completion duration; s1 represents the displacement of the target vehicle within the lane change completion duration t1.

4. The vehicle trajectory prediction method according to claim 3, wherein Interpolating the control points based on the fifth-order Bessel function to obtain the first lane change prediction path; and the obtained number of control points includes: Taking the position P1(x1, y1) of the target vehicle in the vector map as the first control point; Obtaining the second control point P2(x2, y2) based on the first control point P1(x1, y1) and the first distance β1; and the first distance β1 represents the displacement of the target vehicle from the first control point P1(x1, y1) after moving for a duration of t2 based on the car-following model; t2 represents the first time period; Obtaining the third control point P3(x3, y3) based on the first control point P1(x1, y1), the second control point P2(x2, y2), and the second distance β2; and the second distance β2 represents the displacement of the target vehicle from the second control point P2(x2, y2) after moving for a duration of t3 based on the car-following model; t3 represents the second time period, which is a preset value set according to human experience; Obtaining the sixth control point P6(x6, y6) based on the virtual point P0(x0, y0) and the displacement s1 of the target vehicle within the lane change completion duration t1; Obtaining the fifth control point P5(x5, y5) based on the sixth control point P6(x6, y6) and the third distance β3; and the third distance β3 represents the displacement of the target vehicle before moving back for a duration of t2 based on the car-following model from the sixth control point P6(x6, y6); Obtain the fourth control point P4(x4, y4) based on the sixth control point P6(x6, y6), the fifth control point P5(x5, y5), and the fourth distance β4, and the fourth distance β4 represents the displacement of the target vehicle based on the car-following model when it moves backward for a duration of t3 from the fifth control point P5(x5, y5). Among them, the first control point P1(x1, y1) is the starting control point; the sixth control point P6(x6, y6) is the ending control point; the fifth control point P5(x5, y5), the fourth control point P4(x4, y4), the third control point P3(x3, y3), and the second control point P2(x2, y2) are all intermediate control points, and x and y represent the coordinate values corresponding to the control point.

5. The vehicle trajectory prediction method according to claim 2, wherein The method for calculating the speed of each path point in the second lane-changing prediction path is as follows: Among them, v i，t+1 represents the speed of the corresponding path point at time t + 1.

6. The vehicle trajectory prediction method according to claim 3, characterized in that This method further includes: When the behavior intention of the target vehicle is to go straight, Obtain the displacement s0 of the target vehicle within the prediction duration t0 and the first control point P1(x1, y1); the first control point is the position of the target vehicle in the vector map; Take the point on the center line at a distance of s0 from the first control point P1(x1, y1) in the driving direction as the straight-line prediction end point P8(x8, y8), and then obtain the straight-line prediction path from P1(x1, y1) to P8(x8, y8); Then, based on the car-following model, iterate to obtain the speeds of each path point in the straight-line prediction path to obtain the straight-line prediction trajectory.

7. A vehicle trajectory prediction method according to claim 1 or 6, characterized in that, Both the supplementary path and the straight-line prediction path use first-order Bezier for uniform interpolation between two points.

8. A vehicle trajectory prediction device, characterized in that, This device includes: A behavior intention acquisition module, used to acquire the behavior intention of the target vehicle; A displacement calculation module, used to, when the behavior intention of the target vehicle is to change lanes, obtain the displacement of the target vehicle within the prediction duration and the displacement of the target vehicle within the lane-changing completion duration based on the car-following model; and the prediction duration is greater than the lane-changing completion duration; A first lane-changing prediction path generation module, used to obtain a number of control points based on the displacement of the target vehicle within the lane-changing completion duration, the control points including a starting control point, an ending control point, and a number of intermediate control points, and the control points are all located on the center line of the lane; then, based on the Bezier function, interpolate the control points to obtain the first lane-changing prediction path; A lane-changing prediction end point acquisition module, used to obtain the lane-changing prediction end point based on the virtual point and the displacement of the target vehicle within the prediction duration; the virtual point is the point on the center line of the adjacent lane of the target vehicle that is closest to the target vehicle; A supplementary path generation module, used to obtain the supplementary path based on the center line between the lane-changing prediction end point and the ending control point; A second lane-changing prediction path generation module, used to splice the first lane-changing prediction path and the supplementary path to obtain the second lane-changing prediction path; A lane-changing prediction trajectory generation module, used to obtain the speeds of each path point in the second lane-changing prediction path based on the car-following model to obtain the lane-changing prediction trajectory.

9. A storage medium, characterized in that, It stores a computer program for vehicle trajectory prediction, where the computer program causes the computer to execute the vehicle trajectory prediction method according to any one of claims 1-7.

10. An electronic device, characterized in that, Including: One or more processors; A memory; And One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs include those for executing the vehicle trajectory prediction method according to any one of claims 1-7.

Images