Vehicle trajectory prediction method, device, vehicle, equipment and medium

Abstract

The present disclosure provides a vehicle trajectory prediction method, apparatus, vehicle, electronic device, and computer-readable storage medium, relating to the field of autonomous driving technology. The method comprises: obtaining a first vehicle trajectory based on a constant turn rate and acceleration model (CTRA); obtaining the vehicle's driving intent based on a hidden Markov model (HMM); obtaining the vehicle's lane change endpoint coordinates based on the vehicle's driving intent using a Gaussian mixture model (GMM); obtaining a second vehicle trajectory based on the vehicle's current coordinates and the vehicle's lane change endpoint coordinates using a quintic polynomial; and obtaining a composite trajectory of the vehicle based on the first and second trajectories. The method provided by the embodiments of the present disclosure can achieve more accurate prediction of vehicle trajectories.

Description

Technical Field

[0001] The present disclosure relates to the field of autonomous driving technology, and in particular to a method, device, vehicle, electronic device, and computer-readable storage medium for predicting vehicle trajectory. Background Art

[0002] The rapid development of perception technology is laying the foundation for the implementation of autonomous driving. As a crucial component of autonomous vehicles, the perception system undertakes the crucial task of perceiving the traffic / road environment. By sensing the movement of surrounding vehicles, the vehicle "understands" the traffic / road environment, enabling it to make informed decisions and complete specific driving tasks. However, one of the main technical difficulties hindering the implementation of autonomous driving technology is the vehicle's inability to deeply "understand" the traffic environment, particularly its inability to accurately predict the movement of surrounding vehicles. This results in inconsistencies between autonomous driving behavior and that of real drivers, reducing the trust of surrounding drivers and passengers in autonomous driving technology. Summary of the Invention

[0003] The embodiments of the present disclosure provide a vehicle trajectory prediction method, device, vehicle, electronic device, and computer-readable storage medium, which relate to the field of vehicle technology and can realize the prediction of vehicle trajectory.

[0004] An embodiment of the present disclosure provides a method for predicting a vehicle trajectory, comprising: obtaining a first trajectory of the vehicle based on a constant turn rate and acceleration model (CTRA); obtaining a driving intention of the vehicle based on a hidden Markov model (HMM); obtaining the coordinates of a lane change end point of the vehicle based on the driving intention of the vehicle through a Gaussian mixture model (GMM); obtaining a second trajectory of the vehicle based on the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle through a quintic polynomial; and obtaining a composite trajectory of the vehicle based on the first and second trajectories.

[0005] In one embodiment, obtaining a first trajectory of a vehicle according to a constant turn and acceleration model (CTRA) includes: obtaining a longitudinal coordinate, a lateral coordinate, a yaw angle, a longitudinal velocity, an acceleration, and a yaw angular velocity of the vehicle; and obtaining the first trajectory of the vehicle based on the CTRA according to the longitudinal coordinate, the lateral coordinate, the yaw angle, the longitudinal velocity, the acceleration, and the yaw angular velocity of the vehicle.

[0006] In one embodiment, obtaining the driving intention of the vehicle according to a hidden Markov model HMM includes: obtaining the lateral coordinate and lateral speed of the vehicle; and obtaining the driving intention of the vehicle based on the HMM according to the lateral coordinate and lateral speed of the vehicle.

[0007] In one embodiment, obtaining the lane change end point coordinates of the vehicle through a Gaussian mixture model (GMM) based on the driving intention of the vehicle includes: using the GMM to perform statistics based on a collected real driver data set to obtain a driving behavior model of the driver; obtaining a vehicle longitudinal coordinate and a transverse coordinate curve with the maximum probability of varying with speed based on the driver's driving behavior model; and obtaining the lane change end point coordinates of the vehicle based on the vehicle speed and the vehicle longitudinal coordinate and transverse coordinate curve with the maximum probability of varying with speed.

[0008] In one embodiment, obtaining the second trajectory of the vehicle through a quintic polynomial based on the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle includes: obtaining a coefficient matrix of the quintic polynomial based on the coordinates of the lane change end point of the vehicle; and determining the second trajectory based on the coefficient matrix of the quintic polynomial and the quintic polynomial.

[0009] In one embodiment, obtaining a composite trajectory of the vehicle based on the first trajectory and the second trajectory includes: when the driving intention of the vehicle is to turn left or right and the predicted time is less than the lag time of intention recognition, synthesizing the composite trajectory of the vehicle based on the first trajectory; when the driving intention of the vehicle is to turn left or right and the predicted time is greater than or equal to the lag time of intention recognition, synthesizing the composite trajectory of the vehicle based on the second trajectory.

[0010] An embodiment of the present disclosure provides a vehicle trajectory prediction device, comprising: an acquisition unit for acquiring a first trajectory of a vehicle based on a constant turn rate and acceleration model (CTRA); the acquisition unit is further configured to acquire a driving intention of the vehicle based on a hidden Markov model (HMM); the acquisition unit is further configured to acquire the coordinates of a lane change end point of the vehicle through a Gaussian mixture model (GMM) based on the driving intention of the vehicle; the acquisition unit is further configured to acquire a second trajectory of the vehicle through a quintic polynomial based on the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle; and a synthesis unit is configured to acquire a synthesized trajectory of the vehicle based on the first and second trajectories.

[0011] An embodiment of the present disclosure provides a vehicle, comprising the vehicle trajectory prediction device as described in the above embodiment.

[0012] An embodiment of the present disclosure provides an electronic device, comprising: one or more processors; and a storage device configured to store one or more programs, which, when executed by the one or more processors, enables the one or more processors to implement the method described in any one of the above embodiments.

[0013] An embodiment of the present disclosure provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the method described in any one of the above embodiments is implemented.

[0014] The vehicle trajectory prediction method of the present application obtains the first trajectory of the vehicle based on the constant turn rate and acceleration model CTRA; obtains the driving intention of the vehicle based on the hidden Markov model HMM; obtains the lane change end point coordinates of the vehicle through the Gaussian mixture model GMM based on the driving intention of the vehicle; obtains the second trajectory of the vehicle based on the current coordinates of the vehicle and the lane change end point coordinates of the vehicle through a fifth-order polynomial; and obtains the composite trajectory of the vehicle based on the first trajectory and the second trajectory, which can more accurately predict the vehicle trajectory. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0016] Figure 1 is a flow chart of a vehicle trajectory prediction method provided by an embodiment of the present disclosure;

[0017] Figure 2 is a flowchart of a method for obtaining a first trajectory of a vehicle according to a constant turn rate and acceleration model CTRA provided by an embodiment of the present disclosure;

[0018] Figure 3 is a flowchart of a method for obtaining the driving intention of the vehicle according to a hidden Markov model HMM provided by an embodiment of the present disclosure;

[0019] Figure 4 The driving intention recognition model of the embodiment of the present disclosure is based on HMM;

[0020] Figure 5 is a flowchart of a method for obtaining the coordinates of the lane change end point of the vehicle through a Gaussian mixture model (GMM) according to the driving intention of the vehicle, provided by an embodiment of the present disclosure;

[0021] Figure 6 shows driving data in an LCL scenario according to an embodiment of the present application;

[0022] Figure 7 shows driving data in an LK scenario according to an embodiment of the present application;

[0023] Figure 8shows driving data in an LCR scenario according to an embodiment of the present application;

[0024] Figure 9 D of an embodiment of the present application is shown x , D y Schematic diagram of the meaning;

[0025] Figure 10 D of an embodiment of the present application is shown x , D y Schematic diagram of statistical laws;

[0026] Figure 11 is a flowchart of a method for obtaining a second trajectory of the vehicle using a quintic polynomial according to the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle, provided by an embodiment of the present disclosure;

[0027] Figure 12 is a flowchart of a method for obtaining a composite trajectory of the vehicle according to the first trajectory and the second trajectory provided by an embodiment of the present disclosure;

[0028] Figure 13 A schematic diagram showing the change of weighting coefficients with prediction time in one embodiment of the present application is shown;

[0029] Figure 14 A schematic diagram showing a comparison of prediction results in an LCL scenario according to an embodiment of the present application is shown;

[0030] Figure 15 A schematic diagram showing a comparison of prediction results in an LCR scenario according to an embodiment of the present application is shown;

[0031] Figure 16 is a structural diagram of a vehicle trajectory prediction device provided by an embodiment of the present disclosure;

[0032] Figure 17 It is a structural diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0033] The following will be combined with the accompanying drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the embodiments described are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present disclosure.

[0034] Figure 1 The method provided by the embodiment of the present disclosure can be executed by any computer terminal or server with computing capabilities, or by the terminal and the server interacting to execute the method.

[0035] like Figure 1 As shown, the vehicle trajectory prediction method provided by the embodiment of the present disclosure may include the following steps.

[0036] In step S110 , a first trajectory of the vehicle is obtained according to a constant turn rate and acceleration model CTRA.

[0037] In this step, the terminal or the server may obtain the first trajectory of the vehicle according to a constant turn rate and acceleration model CTRA (Constant Turn Rate and Acceleration).

[0038] In step S120 , the driving intention of the vehicle is obtained according to a Hidden Markov Model (HMM).

[0039] In this step, the terminal or the server may obtain the driving intention of the vehicle according to a hidden Markov model HMM.

[0040] In step S130, the coordinates of the lane change end point of the vehicle are obtained through a Gaussian mixture model (GMM) according to the driving intention of the vehicle.

[0041] In this step, the terminal or server may obtain the lane change end point coordinates of the vehicle through a Gaussian mixture model (GMM) according to the driving intention of the vehicle.

[0042] In step S140 , a second trajectory of the vehicle is obtained by using a quintic polynomial according to the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle.

[0043] In this step, the terminal or the server may obtain the second trajectory of the vehicle through a quintic polynomial according to the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle.

[0044] In step S150 , a composite trajectory of the vehicle is obtained according to the first trajectory and the second trajectory.

[0045] In this step, the terminal or the server may obtain a composite trajectory of the vehicle according to the first trajectory and the second trajectory.

[0046] This application Figure 1A method for predicting a vehicle trajectory is provided, wherein a first trajectory of the vehicle is obtained according to a constant turn rate and acceleration model CTRA; a driving intention of the vehicle is obtained according to a hidden Markov model HMM; the coordinates of a lane change end point of the vehicle are obtained according to the driving intention of the vehicle through a Gaussian mixture model GMM; a second trajectory of the vehicle is obtained according to the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle through a quintic polynomial; and a composite trajectory of the vehicle is obtained according to the first trajectory and the second trajectory, thereby obtaining a more accurate predicted vehicle trajectory.

[0047] Figure 2 This is a flowchart of a method for obtaining a first trajectory of a vehicle according to a constant turn rate and acceleration model CTRA provided by an embodiment of the present disclosure.

[0048] like Figure 2 As shown, the method for obtaining the first trajectory provided by the embodiment of the present disclosure may include the following steps.

[0049] In step S210 , the longitudinal coordinate, lateral coordinate, yaw angle, longitudinal velocity, acceleration and yaw angular velocity of the vehicle are obtained.

[0050] In this step, the terminal or the server may obtain the longitudinal coordinate, the transverse coordinate, the yaw angle, the longitudinal velocity, the acceleration and the yaw angular velocity of the vehicle.

[0051] In step S220 , a first trajectory of the vehicle is acquired based on the CTRA according to the longitudinal coordinate, the lateral coordinate, the yaw angle, the longitudinal velocity, the acceleration, and the yaw angular velocity of the vehicle.

[0052] In this step, the terminal or the server may obtain the first trajectory of the vehicle based on the CTRA according to the longitudinal coordinate, the lateral coordinate, the yaw angle, the longitudinal velocity, the acceleration, and the yaw angular velocity of the vehicle.

[0053] Among them, when predicting vehicle trajectory based on the CTRA model, the CTRA model includes a three-degree-of-freedom model of longitudinal, lateral and yaw motion. In this model, the yaw angular velocity and acceleration of the vehicle are considered constant in a short period of time, and the state variables of the vehicle are recorded as (T is the transpose), where x is the longitudinal coordinate, y is the transverse coordinate, θ is the yaw angle, v is the longitudinal velocity, a is the acceleration, and ω is the yaw angular velocity. Therefore, the vehicle motion state calculation formula (1) at time t+T is:

[0054]

[0055] Δx(T)=[(v(t)ω+aωT)sin(θ(t)+ωT)+

[0056] acos(θ(t)+ωT)-v(t)ωsinθ(t)-acosθ(t)] / ω 2 (2)

[0057] Δy(T)=[(-v(t)ω-aωT)cos(θ(t)+ωT)+

[0058] asin(θ(t)+ωT)+v(t)ωcosθ(t)-asinθ(t)] / ω 2 (3)

[0059] The predicted trajectory of the vehicle obtained according to the above formulas (1), (2), and (3) is recorded as the first trajectory Trajectory1.

[0060] Figure 3 It is a flowchart of a method for obtaining the driving intention of the vehicle based on a hidden Markov model HMM provided in an embodiment of the present disclosure.

[0061] like Figure 3 As shown, the method for obtaining driving intention provided by the embodiment of the present disclosure may include the following steps.

[0062] In step S310 , the lateral coordinate and lateral speed of the vehicle are obtained.

[0063] In this step, the terminal or the server obtains the lateral coordinates and lateral speed of the vehicle.

[0064] In step S320 , the driving intention of the vehicle is obtained based on the HMM according to the lateral coordinate and lateral speed of the vehicle.

[0065] In this step, the terminal or the server obtains the driving intention of the vehicle based on the HMM according to the lateral coordinate and lateral speed of the vehicle.

[0066] Figure 4 The embodiment of the present disclosure is based on the HMM driving intention recognition model.

[0067] Among them, the driving intention recognition model based on HMM is as follows Figure 4 As shown: The observation set V represents the lateral coordinate d and lateral velocity of the vehicle (The dot represents the derivative), that is Where d is denoted as v1, Denoted as v2, the hidden state set Q = {S1, S2, S3} (Note: S1 represents LCL, i.e., left turn, S2 represents LK, i.e., straight ahead, and S3 represents LCR, i.e., right turn three states, N represents the number of states of the HMM model, which is 3 in this application); A refers to the state transition probability distribution. In this application, Among them, a ij =P[qt+1 = S j |q t = S i , 1 ≤ i, j ≤ N, the state can be transferred from the current state to any state, and aij ≥ 0 is satisfied. At the same time, it satisfies B refers to the observation probability distribution, and M represents the dimension of the observed variable. In this application, M = 2. And it satisfies According to N, M, A, B, and π, as time goes by, a series of observation sequences can be generated and denoted as: O = O1O2…O T , T is the length of the observation sequence, and O t is a certain value in V.

[0068] The process of generating the observation sequence of HMM is as follows:

[0069] (1) According to the initial state probability distribution matrix π, select q1 = S i .

[0070] (2) Set the current time t = 1.

[0071] (3) According to the observation value b i of S i select the observation value O t = v k .

[0072] (4) According to a ij calculate the probability of transferring from q t = S i to the new state q t+1 = S j .

[0073] (5) Set the current time to t = t + 1; repeat step (3) if t < T (T is the number of iterations), otherwise end the program. Through the above steps, a complete HMM model can be obtained and denoted as λ = (A, B, π). After obtaining the complete HMM model, the Baum-Welch algorithm is used for model training to obtain a set of HMM model parameters λ that maximize P(O / λ). After training the HMM model, the state d of the vehicle can be observed and the numerical value of

[0074] Figure 5 is the flowchart of the method provided by the embodiment of the present disclosure for obtaining the lane change end point coordinates of the vehicle through the Gaussian mixture model GMM according to the driving intention of the vehicle.

[0075] As Figure 5As shown, the method for obtaining the coordinates of the lane change end point provided by the embodiment of the present disclosure may include the following steps.

[0076] In step S510, the GMM is used to perform statistics based on the collected real driver data set to obtain a driving behavior model of the driver.

[0077] In this step, the terminal or server uses the GMM to perform statistics based on the collected real driver data set to obtain the driver's driving behavior model.

[0078] In step S520 , the vehicle longitudinal coordinate and transverse coordinate curves with the maximum probability of changing with speed are obtained according to the driver's driving behavior model.

[0079] In this step, the terminal or the server obtains the vehicle longitudinal coordinate and the transverse coordinate curve with the maximum probability of changing with speed according to the driver's driving behavior model.

[0080] In step S530, the coordinates of the lane change end point of the vehicle are obtained according to the vehicle speed and the vehicle longitudinal coordinate with the maximum probability of speed change and the transverse coordinate curve.

[0081] In this step, the terminal or the server obtains the coordinates of the lane-changing end point of the vehicle according to the vehicle speed and the vehicle longitudinal coordinate with the maximum probability of speed change and the transverse coordinate curve.

[0082] Among them, when the driving behavior data statistics based on GMM are mainly based on the collected real driver data set, the GMM model is used to perform statistics to obtain the driver's driving behavior model, that is, the maximum probability longitudinal driving distance D of the driver in the LCL and LCR scenarios. x The relationship curve with vehicle speed v and the maximum probability of lateral travel distance D y The relationship between the vehicle speed v is denoted as D xLCL (v) D yLCL (v) D xLCR (v) D yLCR (v) (Note: In this application, it is considered that the driver is driving in the LCL and LCR scenarios. x , D y There is a certain regularity with the vehicle speed v.

[0083] Figure 6 Driving data in an LCL scenario according to an embodiment of the present application is shown.

[0084] Figure 7 Driving data in an LK scenario according to an embodiment of the present application is shown.

[0085] Figure 8shows driving data in an LCR scenario according to an embodiment of the present application.

[0086] Figure 9 D of an embodiment of the present application is shown x , D y Schematic diagram of the meaning.

[0087] Figure 10 D of an embodiment of the present application is shown x , D y Schematic diagram of statistical laws.

[0088] The core idea of ​​GMM is to use multiple Gaussian distribution functions to approximate any probability distribution. Each Gaussian is called a "component", and the linear sum of these "components" is the probability density function of GMM. The maximum likelihood estimation method is used to estimate the parameters of the Gaussian distribution. The GMM mathematical model can be expressed as:

[0089]

[0090] In formula (4), a i , b i and c i is the distribution parameter of the model, which can be obtained by maximum likelihood estimation, and e is a natural constant. Assuming there are K data points, the probability distribution of each K data point is p(K), then the Gaussian mixture model composed of n Gaussian models is formula (5):

[0091]

[0092] In formula (5), π k is the prior probability of each mixture component; N k (x|μ k ,Σ k ) is the posterior probability, Σ k represents the covariance, μ k Represents the average value, and the vertical line represents the conditional probability expression. For the GMM model parameters, the extracted D x , D y The data of and v are introduced into the model, and the EM (expectation maximization) method is used to iteratively obtain four groups of GMM models, which are denoted as: f LCL (v,D x ),f LCL (v,D y ),f LCR (v,D x ),f LCR (v,D y ). Among them, f LCL (v,D x ) indicates LCL, Dx and the probability density distribution of v; f LCL (v,D y ) indicates LCL, D y and the probability density distribution of v.

[0093] According to the four groups of GMM models, f LCL (v,D x ),f LCL (v,D y ),f LCR (v,D x ),f LCR (v,D y ) can be further solved to obtain the maximum probability D that changes with the speed v x and D y Curve: D xLCL (v) D yLCL (v) D xLCR (v) D yLCR (v).

[0094] Figure 11 This is a flowchart of a method for obtaining a second trajectory of the vehicle through a quintic polynomial based on the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle, provided by an embodiment of the present disclosure.

[0095] like Figure 11 As shown, the method for obtaining the second trajectory provided by the embodiment of the present disclosure may include the following steps.

[0096] In step S1110 , the coefficient matrix of the quintic polynomial is obtained according to the lane change end point coordinates of the vehicle.

[0097] In this step, the terminal or the server obtains the coefficient matrix of the quintic polynomial according to the coordinates of the lane change end point of the vehicle.

[0098] In step S1120 , the second trajectory is determined according to the coefficient matrix of the quintic polynomial and the quintic polynomial.

[0099] In this step, the terminal or the server determines the second trajectory according to the coefficient matrix of the quintic polynomial and the quintic polynomial.

[0100] Among them, when the fifth-order polynomial trajectory prediction is obtained, the maximum probability D x and D y . The calculated D x and D yPerform quintic polynomial trajectory prediction. Studies have shown that the lane-changing trajectory of the vehicle conforms to the quintic polynomial curve. In this step, the trajectory of the traffic vehicle will be predicted based on the quintic polynomial. At time t0, the current position of the vehicle is recorded as Speed ​​is Then the trajectory coordinates predicted at time t0+nΔT are n is a positive integer, and ΔT is the time step.

[0101] If the driving intention identified by the HMM model is LK, the second trajectory is calculated according to formula (6):

[0102]

[0103] If the driving intention identified by the HMM model is LCR or LCL, set the lane change end point coordinates to Lane change process time At t0~t f Within the time range, the vehicle trajectory satisfies the fifth-order polynomial equation (7):

[0104]

[0105] Where x(t) is the longitudinal trajectory equation and y(t) is the transverse trajectory equation. xishu =[a5 a4 a3 a2 a1 a0],B xishu =[b5 b4 b3 b2 b1 b0] is the coefficient vector of the quintic polynomial. Derivative equation (7) yields the velocity equation (8):

[0106]

[0107] Derivative equation (8) yields the acceleration equation (9):

[0108]

[0109] Next, we will focus on solving the coefficients of the quintic polynomial, assuming that the initial state of the lane change and the state after the lane change are (10):

[0110]

[0111] Lane change end point coordinates The maximum possible value can be obtained by combining the probability density function based on GMM.

[0112] So

[0113]

[0114]

[0115] make:

[0116]

[0117] After finishing, we can get:

[0118]

[0119] Solving the above formula (14) we can get the determinant A xishu ,B xishu The value of time t0+nΔT and A xishu ,B xishu Substituting the coefficient matrix into formula (7) can obtain the vehicle's Similarly, the predicted coordinates at any other time can be obtained, and the predicted trajectory is recorded as the second trajectory Trajectory2.

[0120] Figure 12 This is a flowchart of a method for obtaining a composite trajectory of the vehicle based on the first trajectory and the second trajectory provided by an embodiment of the present disclosure.

[0121] like Figure 12 As shown, the method for obtaining a synthetic trajectory provided by an embodiment of the present disclosure may include the following steps.

[0122] In step S1210 , when the driving intention of the vehicle is to turn left or right and the predicted time is less than the lag time of intention recognition, a synthetic trajectory of the vehicle is synthesized based on the first trajectory.

[0123] In this step, when the driving intention of the vehicle is to turn left or right and the prediction time is less than the lag time of intention recognition, the terminal or server synthesizes a synthetic trajectory of the vehicle based on the first trajectory.

[0124] In step S1220, when the driving intention of the vehicle is to turn left or right and the prediction time is greater than or equal to the lag time of intention recognition, a synthetic trajectory of the vehicle is synthesized based on the second trajectory.

[0125] In this step, when the driving intention of the vehicle is to turn left or right and the prediction time is greater than or equal to the lag time of intention recognition, the terminal or server synthesizes a synthetic trajectory of the vehicle based on the second trajectory.

[0126] Among them, when the trajectory is fused. On the one hand: due to the short term (0-T lag seconds)(T lagis the lag time of intention recognition), it is difficult to accurately judge the driving intention, and the trajectory prediction error of the trajectory prediction model based on driving intention is large. This application believes that the vehicle is more in line with the assumptions of the CTRA model. Therefore, the predicted trajectory of the vehicle is based on the calculation results of the CTRA model (denoted as: Trajectory1). On the other hand: as time goes by, the prediction error of the CTRA model gradually increases. At this time, the driving intention can be accurately judged based on the historical trajectory information, and the trajectory prediction result based on driving intention is more in line with the driver's actual trajectory. Therefore, when the time exceeds T lag When the vehicle's predicted trajectory is based on the trajectory prediction result of the driving intention (recorded as: Trajectory2). lag The CTRA model is more suitable for the vehicle trajectory within seconds; when it exceeds the predicted future T lag If the trajectory is 10 seconds later, the trajectory prediction model based on driving intention is more appropriate. Therefore, the two models need to be fused. According to the previous steps, we know that the trajectory predicted by the CTRA model is Trajectory1, and the trajectory predicted by the trajectory prediction model based on driving intention is Trajectory2. The trajectories will be fused to obtain the final composite trajectory Trajectory3. The specific fusion method is as follows:

[0127] (1) In the LK mode, the trajectory prediction result Trajectory1 based on CTRA and the predicted trajectory Trajectory2 based on driving intention recognition are linearly weighted to obtain the final synthetic trajectory Trajectory3.

[0128] Trajectory3=(Trajectory2+ Trajectory1) / 2 (15)

[0129] (2) In the LCL and LCR modes, it is believed that the vehicle is more consistent with the assumptions of the kinematic model in the short term, and the trajectory prediction result Trajectory1 of the CTRA model is relatively accurate. However, as time goes by, the prediction error of the CTRA model is larger. The trajectory prediction model based on driving intention recognition has the problem of intention recognition lag in a short period of time, resulting in a large error in the predicted trajectory Trajectory2. The lag time constant is T lag , that is to say, in the future 0~T lag The prediction error is large, T lag The error is smaller in the future. Therefore, this paper fuses Trajectory2 and Trajectory1 to obtain the final predicted trajectory Trajectory2. The specific fusion principle is as follows:

[0130]

[0131] Where G represents the weight coefficient of Trajectory1, which is determined by the following formula (17):

[0132]

[0133] When the prediction time t is less than T lag When the trajectory prediction result of the CTRA model is used as the final prediction trajectory, it is greater than T lag , using the first-order inertial transition (T res represents the transition time constant) to prevent the problem of discontinuous predicted trajectory caused by sudden changes in the weighting coefficients.

[0134] Figure 13 A schematic diagram showing the change of weighting coefficient with prediction time in one embodiment of the present application is shown.

[0135] Figure 14 A schematic diagram showing comparison of prediction results in an LCL scenario according to an embodiment of the present application is shown.

[0136] Figure 15 A schematic diagram showing comparison of prediction results in an LCR scenario according to an embodiment of the present application is shown.

[0137] Figure 16 It is a structural diagram of a vehicle trajectory prediction device provided by an embodiment of the present disclosure.

[0138] like Figure 16 As shown, the vehicle trajectory prediction device 1600 provided in the embodiment of the present disclosure may include:

[0139] an acquiring unit 1610 , configured to acquire a first trajectory of the vehicle according to a constant turn rate and acceleration model CTRA;

[0140] The acquisition unit 1610 is further configured to acquire the driving intention of the vehicle according to a Hidden Markov Model (HMM);

[0141] The acquisition unit 1610 is further configured to acquire the coordinates of the lane change end point of the vehicle through a Gaussian mixture model (GMM) according to the driving intention of the vehicle;

[0142] The acquiring unit 1610 is further configured to acquire a second trajectory of the vehicle using a quintic polynomial according to the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle;

[0143] The synthesis unit 1620 is configured to obtain a synthesized trajectory of the vehicle according to the first trajectory and the second trajectory.

[0144] Figure 16A vehicle trajectory prediction device is provided, which is used to obtain a first trajectory of the vehicle according to a constant turn rate and acceleration model CTRA through an acquisition unit; the acquisition unit is also used to obtain the driving intention of the vehicle according to a hidden Markov model HMM; the acquisition unit is also used to obtain the lane change end point coordinates of the vehicle through a Gaussian mixture model GMM according to the driving intention of the vehicle; the acquisition unit is also used to obtain the second trajectory of the vehicle according to the current coordinates of the vehicle and the lane change end point coordinates of the vehicle through a fifth-order polynomial; and a synthesis unit is used to obtain a synthesized trajectory of the vehicle according to the first trajectory and the second trajectory, which can achieve more accurate prediction of the vehicle trajectory.

[0145] In one embodiment, the acquisition unit 1610 is further configured to acquire the longitudinal coordinate, lateral coordinate, yaw angle, longitudinal velocity, acceleration, and yaw angular velocity of the vehicle; and acquire the first trajectory of the vehicle based on the CTRA according to the longitudinal coordinate, lateral coordinate, yaw angle, longitudinal velocity, acceleration, and yaw angular velocity of the vehicle.

[0146] In one embodiment, the acquisition unit 1610 is further configured to acquire the lateral coordinate and lateral speed of the vehicle; and acquire the driving intention of the vehicle based on the HMM according to the lateral coordinate and lateral speed of the vehicle.

[0147] In one embodiment, the acquisition unit 1610 is further used to obtain the driver's driving behavior model by using the GMM based on the collected real driver data set; obtain the vehicle's lateral coordinates with the maximum probability of changing with speed and the lateral coordinate curve based on the driver's driving behavior model; and obtain the coordinates of the vehicle's lane change end point based on the vehicle speed and the vehicle's lateral coordinates with the maximum probability of changing with speed and the lateral coordinate curve.

[0148] In one embodiment, the acquiring unit 1610 is further configured to acquire a coefficient matrix of the quintic polynomial according to the lane change end point coordinates of the vehicle; and determine the second trajectory according to the coefficient matrix of the quintic polynomial and the quintic polynomial.

[0149] In one embodiment, the synthesis unit 1620 is further configured to synthesize the synthetic trajectory of the vehicle based on the first trajectory when the driving intention of the vehicle is to turn left or right and the predicted time is less than the lag time of intention recognition; and to synthesize the synthetic trajectory of the vehicle based on the second trajectory when the driving intention of the vehicle is to turn left or right and the predicted time is greater than or equal to the lag time of intention recognition.

[0150] The present application also provides a vehicle, including Figure 16 The vehicle trajectory prediction device is shown.

[0151] The vehicle of the present application can realize the prediction of the vehicle trajectory, thereby realizing the accurate prediction of the vehicle trajectory.

[0152] See also Figure 17 , Figure 17 FIG. 1 is a schematic diagram of the structure of an electronic device 1700 provided by an embodiment of the present disclosure. Figure 17 As shown, the electronic device in the embodiment of the present disclosure may include: one or more processors 1701, a memory 1702, and an input / output interface 1703. The processor 1701, the memory 1702, and the input / output interface 1703 are connected via a bus 1704. The memory 1702 is used to store computer programs, which include program instructions. The input / output interface 1703 is used to receive and output data, such as for data exchange between a host computer and the electronic device, or for data exchange between virtual machines in the host computer. The processor 1701 is used to execute the program instructions stored in the memory 1702.

[0153] The processor 1701 may perform the following operations:

[0154] A first trajectory of the vehicle is obtained according to a constant turn rate and acceleration model CTRA; a driving intention of the vehicle is obtained according to a hidden Markov model HMM; the coordinates of the lane change end point of the vehicle are obtained through a Gaussian mixture model GMM according to the driving intention of the vehicle; a second trajectory of the vehicle is obtained according to the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle through a fifth-order polynomial; and a composite trajectory of the vehicle is obtained according to the first trajectory and the second trajectory.

[0155] In some feasible implementations, the processor 1701 may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor, or the processor may be any conventional processor, etc.

[0156] The memory 1702 may include a read-only memory and a random access memory, and provides instructions and data to the processor 1701 and the input / output interface 1703. A portion of the memory 1702 may also include a non-volatile random access memory. For example, the memory 1702 may also store device type information.

[0157] In a specific implementation, the electronic device can execute the implementation methods provided in the various steps in the above embodiments through its built-in functional modules. For details, please refer to the implementation methods provided in the various steps in the above embodiments, which will not be repeated here.

[0158] The embodiment of the present disclosure provides an electronic device including a processor, an input / output interface, and a memory. The processor obtains a computer program in the memory, executes each step of the method shown in the above embodiment, and performs a transmission operation.

[0159] The embodiments of the present disclosure also provide a computer-readable storage medium, which stores a computer program, and the computer program is suitable for being loaded by the processor and executing the methods provided in the various steps of the above embodiments. For details, please refer to the implementation methods provided in the various steps of the above embodiments, which will not be repeated here. In addition, the description of the beneficial effects of adopting the same method will not be repeated. For technical details not disclosed in the computer-readable storage medium embodiments involved in the present disclosure, please refer to the description of the method embodiments of the present disclosure. As an example, the computer program can be deployed to be executed on an electronic device, or on multiple electronic devices located in one place, or on multiple electronic devices distributed in multiple places and interconnected by a communication network.

[0160] The computer-readable storage medium may be the device provided in any of the aforementioned embodiments or the internal storage unit of the electronic device, such as the hard disk or memory of the electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart memory card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device. Furthermore, the computer-readable storage medium may also include both the internal storage unit of the electronic device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the electronic device. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0161] The present disclosure also provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the methods provided in various optional embodiments described above.

[0162] The terms "first", "second", etc. in the description, claims, and drawings of the embodiments of the present disclosure are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the listed steps or modules, but may optionally include steps or modules that are not listed, or may optionally include other steps and units inherent to these processes, methods, apparatuses, products, or devices.

[0163] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in this description according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this disclosure.

[0164] The methods and related devices provided by the embodiments of the present disclosure are described with reference to the method flow charts and / or structural diagrams provided by the embodiments of the present disclosure. Specifically, each process and / or block in the method flow charts and / or structural diagrams, as well as the combination of processes and / or blocks in the flow charts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable transmission device to generate a machine, so that the instructions executed by the processor of the computer or other programmable transmission device generate instructions for implementing the process. Figure 1 Schematic diagram of one or more processes and / or structures Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable transmission device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the function specified in the process. Figure 1 Schematic diagram of one or more processes and / or structures Figure 1These computer program instructions can also be loaded onto a computer or other programmable transmission device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide the functions for implementing the process. Figure 1 The flow or flows and / or structures illustrate the steps of the functions specified in one block or multiple blocks.

[0165] The above disclosure is merely a preferred embodiment of the present disclosure and certainly cannot be used to limit the scope of the present disclosure. Therefore, equivalent changes made according to the claims of the present disclosure are still within the scope of the present disclosure.

Claims

1. A method for predicting vehicle trajectory, characterized in that: include: Obtaining a first trajectory of the vehicle according to a constant turn rate and acceleration model CTRA; Obtaining the driving intention of the vehicle according to a Hidden Markov Model (HMM); Obtaining the lane change end point coordinates of the vehicle through a Gaussian mixture model (GMM) according to the driving intention of the vehicle; Obtaining a second trajectory of the vehicle using a quintic polynomial according to the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle; Obtaining a composite trajectory of the vehicle according to the first trajectory and the second trajectory; Obtaining the first trajectory of the vehicle according to the constant turn rate and acceleration model CTRA includes: Obtaining the longitudinal coordinate, lateral coordinate, yaw angle, longitudinal velocity, acceleration, and yaw angular velocity of the vehicle; Obtaining a first trajectory of the vehicle based on the CTRA according to the longitudinal coordinate, the lateral coordinate, the yaw angle, the longitudinal velocity, the acceleration, and the yaw rate of the vehicle; Obtaining the driving intention of the vehicle according to the Hidden Markov Model HMM includes: Obtaining the lateral coordinate and lateral speed of the vehicle; Obtaining a driving intention of the vehicle based on the HMM according to the lateral coordinate and lateral speed of the vehicle; Obtaining the lane change end point coordinates of the vehicle through a Gaussian mixture model (GMM) according to the driving intention of the vehicle includes: Based on the collected real driver data set, the GMM is used to perform statistics to obtain the driver's driving behavior model; Obtaining the vehicle longitudinal coordinate and lateral coordinate curves with the maximum probability of speed change according to the driver's driving behavior model; Obtaining the lane change end point coordinates of the vehicle according to the vehicle speed and the vehicle longitudinal coordinate and transverse coordinate curve with the maximum probability of change with the speed; Obtaining a second trajectory of the vehicle using a quintic polynomial according to the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle includes: Obtaining a coefficient matrix of the quintic polynomial according to the lane change end point coordinates of the vehicle; determining the second trajectory according to the coefficient matrix of the quintic polynomial and the quintic polynomial; Obtaining a composite trajectory of the vehicle according to the first trajectory and the second trajectory includes: When the driving intention of the vehicle is to turn left or right and the prediction time is less than the lag time of intention recognition, synthesizing a synthetic trajectory of the vehicle based on the first trajectory; When the driving intention of the vehicle is to turn left or right and the prediction time is greater than or equal to the lag time of intention recognition, a synthetic trajectory of the vehicle is synthesized based on the second trajectory.

2. A vehicle trajectory prediction device, characterized in that: include: an acquiring unit, configured to acquire a first trajectory of the vehicle according to a constant turn rate and acceleration model CTRA; The acquisition unit is further configured to acquire the driving intention of the vehicle according to a Hidden Markov Model (HMM); The acquisition unit is further configured to acquire the lane change end point coordinates of the vehicle through a Gaussian mixture model (GMM) according to the driving intention of the vehicle; The acquisition unit is further configured to acquire a second trajectory of the vehicle using a quintic polynomial according to the current coordinates of the vehicle and the coordinates of the lane change end point of the vehicle; a synthesis unit, configured to obtain a synthetic trajectory of the vehicle based on the first trajectory and the second trajectory; The vehicle trajectory prediction device is used to implement the vehicle trajectory prediction method as claimed in claim 1.

3. A vehicle, characterized in that: The vehicle trajectory prediction device according to claim 2 is included.

4. An electronic device, characterized in that: include: one or more processors; A storage device configured to store one or more programs, which, when executed by the one or more processors, enable the one or more processors to implement the vehicle trajectory prediction method according to claim 1.

5. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the vehicle trajectory prediction method according to claim 1 is implemented.

Images