Vehicle control method and device, storage medium, electronic equipment and vehicle

Abstract

The present disclosure relates to a vehicle control method, device, storage medium, electronic equipment and vehicle to improve the control accuracy and stability of the vehicle. The method comprises: determining the pose information of the vehicle at a target time and the respective pose information of each prediction point as first pose information, the prediction point being determined from the prediction path corresponding to the vehicle; determining a plurality of reference points on the reference driving path corresponding to the vehicle, and determining the respective pose information of each reference point as second pose information, the plurality of reference points comprising a first reference point corresponding to the target time and a preset number of second reference points after the target time, and each reference point having a corresponding prediction point on the prediction path; constructing an LQR optimization model according to the first pose information and the second pose information; determining a target control parameter of a next time of the target time according to the LQR optimization model; and controlling the vehicle according to the target control parameter at the next time of the target time.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of vehicle control, in particular, to a vehicle control method and device, a storage medium, an electronic device and a vehicle. BACKGROUND

[0002] In the intelligent driving scene, in order to prevent the vehicle from deviating from the lane or colliding with the obstacle, the intelligent driving control system generally has a high requirement for the lateral control of the vehicle. The commonly used control methods are PID (Proportion Integral Differential) control, pure tracking control, LQR (Linear quadratic regulator) control and MPC (Model Prediction Control). Among them, the PID control and the pure tracking control have low accuracy, and the MPC consumes more computing power, so the LQR is more commonly used. However, in the related control scheme based on the LQR, there is a control instability. SUMMARY

[0003] The purpose of the present disclosure is to provide a vehicle control method, device, storage medium, electronic device and vehicle to improve the control accuracy and stability of the vehicle.

[0004] In order to achieve the above purpose, according to a first aspect of the present disclosure, a vehicle control method is provided, the method comprising:

[0005] determining the pose information of the vehicle at a target time and the pose information corresponding to each of a plurality of prediction points as first pose information, the prediction points being determined from a prediction path corresponding to the vehicle;

[0006] determining a plurality of reference points on the reference driving path corresponding to the vehicle, and determining the pose information corresponding to each of the reference points as second pose information, the plurality of reference points including a first reference point corresponding to the target time and a preset number of second reference points after the target time, and each of the reference points having a corresponding prediction point on the prediction path;

[0007] constructing an LQR optimization model according to the first pose information and the second pose information;

[0008] determining a target control parameter of a next time of the target time according to the LQR optimization model;

[0009] controlling the vehicle according to the target control parameter at the next time of the target time.

[0010] Optionally, the pose information of the vehicle at the target time point comprises a coordinate of the vehicle at the target time point.

[0011] The method further comprises:

[0012] determining a projection point of the coordinate of the vehicle at the target time point on the reference driving path as the first reference point;

[0013] determining the second reference points on the reference driving path one by one, starting from a position where the first reference point is located, and taking a preset distance as a distance between adjacent reference points.

[0014] Optionally, the reference driving path corresponds to a path equation, and the pose information comprises a coordinate and a heading angle.

[0015] The method further comprises:

[0016] determining the coordinate corresponding to each reference point according to the reference driving path;

[0017] for each reference point, determining the heading angle corresponding to the reference point according to the coordinate corresponding to the reference point and the path equation.

[0018] Optionally, the heading angle θi corresponding to the ith reference point Ri is determined by the following formula: Ri :

[0019] θ Ri = arctan(f'(x Ri ))

[0020] wherein 0≤i≤preset number, R0 is the first reference point, f(x Ri ) is the path equation, and f'(x Ri ) is a derivative of the path equation.

[0021] Optionally, the LQR optimization model comprises a target cost function.

[0022] The target cost function is determined by the following method:

[0023] obtaining a pose cost coefficient and a control cost coefficient;

[0024] generating a first cost function according to the first pose information, the second pose information, and the pose cost coefficient;

[0025] generating a second cost function according to a preset control parameter and the control cost coefficient;

[0026] A sum of the first cost function and the second cost function is determined as the target cost function.

[0027] Optionally, the target cost function F is generated according to the following formula:

[0028]

[0029] wherein, is the first cost function, is the second cost function; X i is the first cost function, Ri is the second cost function; X R0 is the first cost function, i is the control parameter of the i th prediction point.

[0030] Optionally, the pose information comprises coordinates and a heading angle, and the pose cost coefficient comprises a position cost coefficient and a heading cost coefficient.

[0031] Q is generated according to the following formula:

[0032] Q = diag [q xy , q xy , q θ ]

[0033] wherein, q xy is the position cost coefficient, and q θ is the heading cost coefficient.

[0034] Optionally, the LQR optimization model further comprises a target constraint condition, the target constraint condition comprising a first constraint condition for constraining a generation manner of the prediction path and a second constraint condition for constraining X0 to be the first pose information.

[0035] Optionally, the pose information comprises coordinates and a heading angle.

[0036] The first constraint condition is generated by the following manner:

[0037] A kinematics model is constructed with pose information as a state quantity and a preset control parameter as a control quantity.

[0038] According to a preset distance, the kinematics model is discretized to obtain a system equation for generating pose information of a second prediction point according to pose information of a first prediction point, the second prediction point being a next prediction point spaced apart from the first prediction point by a preset distance.

[0039] Linearize the system equation to obtain a processed target equation as the first constraint condition.

[0040] Optionally, the system equation is:

[0041]

[0042] wherein X i is the predicted pose information for the i-th prediction point, Δs is the preset distance, θ i is the heading angle corresponding to the i-th prediction point.

[0043] Optionally, the linearization processing is Taylor expansion.

[0044] The target equation is:

[0045] X i+1 = A i X i + B i κ i + C i

[0046] wherein A i , B i , C i are determined in the following manner:

[0047]

[0048]

[0049]

[0050] wherein θ Ri is the heading angle corresponding to the i-th reference point.

[0051] Optionally, the determining the target control parameter of the next moment of the target moment according to the LQR optimization model comprises:

[0052] determining a feedback gain matrix capable of minimizing the target cost function under the condition of satisfying the target constraint condition;

[0053] determining the target control parameter of the next moment of the target moment according to the first pose information and the feedback gain matrix.

[0054] Optionally, the determining the feedback gain matrix capable of minimizing the target cost function under the condition of satisfying the target constraint condition comprises:

[0055] solving the target cost function by using a dynamic programming solution method to obtain the feedback gain matrix.

[0056] Optionally, the determining the target control parameter of a next time point of the target time point according to the first pose information and the feedback gain matrix comprises:

[0057] The target control parameter κ is determined according to the following formula:

[0058] κ = -K i X i -L i

[0059] wherein K i , L i is the feedback gain matrix, and:

[0060]

[0061]

[0062]

[0063]

[0064] q i = -QX Ri

[0065] wherein P N = Q, F N = q N .

[0066] Optionally, the target control parameter is a steering curvature.

[0067] The controlling the vehicle according to the target control parameter comprises:

[0068] determining a front wheel steering angle corresponding to the steering curvature;

[0069] controlling the vehicle according to the front wheel steering angle.

[0070] According to a second aspect of the present disclosure, a vehicle control device is provided, the device comprising:

[0071] a first determining module configured to determine pose information of a vehicle at a target time point and pose information corresponding to a plurality of prediction points, as first pose information, the prediction points being determined from a prediction path corresponding to the vehicle;

[0072] a second determining module, configured to determine a plurality of reference points on a reference driving path corresponding to the vehicle, and determine respective pose information of each reference point as second pose information, the plurality of reference points including a first reference point corresponding to the target moment and a preset number of second reference points after the target moment, and each reference point having a corresponding prediction point on the prediction path;

[0073] a constructing module, configured to construct an LQR optimization model according to the first pose information and the second pose information;

[0074] a third determining module, configured to determine a target control parameter of a next moment of the target moment according to the LQR optimization model;

[0075] a control module, configured to control the vehicle according to the target control parameter at the next moment of the target moment.

[0076] According to a third aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the steps of the vehicle control method according to the first aspect of the present disclosure.

[0077] According to a fourth aspect of the present disclosure, there is provided an electronic device, comprising:

[0078] a memory having a computer program stored thereon;

[0079] a processor configured to execute the computer program in the memory to implement the steps of the vehicle control method according to the first aspect of the present disclosure.

[0080] According to a fifth aspect of the present disclosure, there is provided a vehicle comprising the electronic device according to the fourth aspect of the present disclosure.

[0081] By the technical solution, the pose information of the vehicle at the target moment and the pose information corresponding to each of the plurality of prediction points on the prediction path are determined as first pose information, a plurality of reference points and respective second pose information on the reference driving path corresponding to the vehicle are determined, the LQR optimization model is constructed according to the first pose information and the second pose information to determine the target control parameter for the next moment of the target moment, and then the vehicle is controlled according to the target control parameter at the next moment of the target moment. The reference points determined from the reference driving path include not only the first reference point corresponding to the target moment, but also the preset number of second reference points after the target moment, and these reference points have corresponding prediction points on the prediction path, and then the LQR optimization model is constructed by using these prediction points and reference points, which is equivalent to adding a preview mechanism in the LQR algorithm. When the target control parameter is determined, not only the current path tracking error is considered, but also the path tracking error in the future period is considered. Therefore, a more optimal control parameter can be obtained, and sudden turning driving can be avoided to ensure the driving stability of the vehicle.

[0082] Other features and advantages of the present disclosure will be described in detail in the following detailed description section. BRIEF DESCRIPTION OF DRAWINGS

[0083] The accompanying drawings are included to provide a further understanding of the present disclosure and constitute a part of the specification, and are used together with the following detailed description to explain the present disclosure, but do not constitute a limitation on the present disclosure. In the drawings:

[0084] Figure 1 is a flowchart of a vehicle control method according to an embodiment of the present disclosure;

[0085] Figure 2 is an exemplary schematic diagram of a reference driving path and a prediction path in a vehicle control method according to the present disclosure;

[0086] Figure 3 is a block diagram of a vehicle control device according to an embodiment of the present disclosure;

[0087] Figure 4 is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0088] The specific embodiments of the present disclosure are described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present disclosure, and are not used to limit the present disclosure.

[0089] Figure 1 is a flowchart of a vehicle control method according to an embodiment of the present disclosure. As Figure 1As shown, the method provided by the present disclosure can include steps 11 to 15.

[0090] In step 11, the pose information of the vehicle at the target time and the pose information corresponding to each of the plurality of prediction points are determined as first pose information.

[0091] For example, the target time can be the current time, and accordingly, the first pose information is the actual pose information of the vehicle at the current time. For another example, the target time can be a time after the current time.

[0092] The prediction points can be determined from the prediction path corresponding to the vehicle. In the prediction path corresponding to the vehicle, a plurality of prediction points are usually included, which can be generated by predicting the path of the vehicle for a series of times after the target time. Thus, the first pose information includes the pose information of the vehicle at the target time and a series of times after the target time.

[0093] Optionally, the pose information can include coordinates and a heading angle. For example, in a two-dimensional rectangular coordinate system constructed by an x-axis and a y-axis, the coordinates in the pose information include an x-axis coordinate and a y-axis coordinate.

[0094] In step 12, a plurality of reference points are determined on the reference driving path corresponding to the vehicle, and the pose information corresponding to each of the reference points is determined as second pose information.

[0095] The reference driving path can be directly obtained from other modules (for example, an upstream path planning module, etc.). For example, the reference driving path can be represented by a curve equation.

[0096] The plurality of reference points include a first reference point corresponding to the target time and a preset number of second reference points after the target time. In fact, the second reference points are a plurality of future reference points, which are equivalent to prediction points in the present disclosure. In addition, each reference point has a corresponding prediction point on the prediction path.

[0097] In one possible implementation, in step 12, determining the plurality of reference points on the reference driving path of the vehicle can include the following steps:

[0098] Determining a projection point of the coordinates of the vehicle at the target time on the reference driving path as the first reference point;

[0099] Taking the position of the first reference point as a starting position and taking a preset distance as the interval between adjacent reference points, a preset number of second reference points are determined on the reference driving path one by one.

[0100] The preset distances between the adjacent reference points can be equal, different, or partly equal and partly different.

[0101] As described above, the pose information can include coordinates, and thus the pose information of the vehicle at the target time includes coordinates of the vehicle at the target time.

[0102] By determining the projection point of the coordinates of the vehicle at the target time on the reference travel path, the reference point corresponding to the target time on the reference travel path, i.e., the first reference point, can be located. The manner of determining the projection point can be determined by a conventional manner of determining the projection point of a point to a curve, i.e., making the line connecting the point and the projection point perpendicular to the tangent of the projection point at the curve.

[0103] After the first reference point is determined, the other reference points, i.e., the second reference points, on the reference travel path can be determined one by one starting from the first reference point with a preset distance as the interval between the adjacent reference points. For example, the reference travel path can be as shown in FIG. 1M0, M1 is the first reference point, the next reference point M2 (the second reference point) of M1 on the reference travel path can be determined with a preset distance as the interval, the next reference point of M2 on the reference travel path can be determined with a preset distance as the interval, and so on until the number of the second reference points reaches a preset number. Figure 2

[0104] After the plurality of reference points are determined, the pose information of each reference point can be determined as the second pose information. As described above, the pose information can include coordinates and a heading angle. Meanwhile, the reference travel path can correspond to a path equation, for example, the curve equation. Based on this, determining the pose information corresponding to each reference point can include the following steps.

[0105] According to the reference travel path, the coordinates corresponding to each reference point are determined;

[0106] According to the coordinates corresponding to each reference point and the path equation, the heading angle corresponding to each reference point is determined.

[0107] After the reference points on the reference travel path are determined by the above manner, the coordinates of the reference points can be directly obtained. The heading angle of each reference point is determined respectively.

[0108] For example, the heading angle θi corresponding to the i-th reference point Ri can be determined by the following formula: Ri :

[0109] θ Ri = arctan(f'(x Ri ) ​

[0110] wherein 0≤i≤preset number, R0is a first reference point, f(x Ri ) is a path equation, and f'(x Ri ) is a derivative of the path equation.

[0111] Returning to Figure 1 , in step 13, an LQR optimization model is constructed according to the first pose information and the second pose information.

[0112] wherein the LQR optimization model can include a target cost function, and the purpose of the present disclosure is to determine a control parameter that can minimize the target cost function. Generally, the target cost function is accompanied by a constraint condition, so the purpose of the present disclosure is actually to determine a control parameter that can minimize the target cost function while meeting the constraint condition (target constraint condition of the present disclosure).

[0113] For example, the control parameter can be a steering curvature.

[0114] Optionally, the target cost function can be determined in the following manner:

[0115] obtaining a pose cost coefficient and a control cost coefficient;

[0116] generating a first cost function according to the first pose information, the second pose information, and the pose cost coefficient;

[0117] generating a second cost function according to the preset control parameter and the control cost coefficient;

[0118] determining a sum of the first cost function and the second cost function as the target cost function.

[0119] wherein the pose cost coefficient and the control cost coefficient can be set according to the actual scene requirements. If the pose information includes coordinates and a heading angle, the pose cost coefficient can include a position cost coefficient and a heading cost coefficient, and the sum of the position cost coefficient and the heading cost coefficient remains a stable value, that is, if the position cost coefficient is increased, the heading cost coefficient is correspondingly reduced.

[0120] Through the setting of the cost coefficient, the control of the vehicle can be balanced between accuracy and stability. For example, if a higher heading cost coefficient is set, the heading of the vehicle during the control process will be more accurate, and correspondingly, the position deviation will be slightly larger. For another example, if a higher control cost coefficient is set, the steering control of the vehicle during the control process will be relatively smooth, and correspondingly, the accuracy will be reduced.

[0121] Optionally, the target cost function F can be generated according to the following formula:

[0122]

[0123] wherein, is a first cost function, is a second cost function; X i is predicted pose information for the i th prediction point, X Ri is second pose information of the i th reference point, X R0 is second pose information of the first reference point, N is a preset number, Q is a diagonal matrix generated according to a pose cost coefficient; r is a control cost coefficient, K i is a control parameter of the i th prediction point.

[0124] As described above, the vehicle can correspond to a prediction path containing prediction points. The prediction path can be as shown in FIG. 0, each reference point has a corresponding prediction point on the prediction path, for example, the reference point M2 corresponds to the prediction point E2, and the reference point M4 corresponds to the prediction point E4. Figure 2

[0125] For example, Q can be generated according to the following formula:

[0126] Q = diag[q xy , q xy , q θ ]

[0127] wherein, q xy is a position cost coefficient, q θ is a heading cost coefficient.

[0128] Optionally, the LQR optimization model can further include a target constraint condition, which can include a first constraint condition for constraining the generation manner of the prediction path and a second constraint condition for constraining X0 to be the first pose information.

[0129] Optionally, the first constraint condition can be generated in the following manner:

[0130] constructing a kinematics model taking the pose information as a state quantity and a preset control parameter as a control quantity;

[0131] discretizing the kinematics model according to a preset distance to obtain a system equation for generating the pose information of a second prediction point according to the pose information of a first prediction point, the second prediction point being a next prediction point spaced apart from the first prediction point by a preset distance;

[0132] linearizing the system equation to obtain a processed target equation as the first constraint condition.

[0133] For example, the construction of the kinematics model can be as follows:

[0134] ​

[0135] wherein x, y are coordinates, θ is a heading angle, and κ is a preset control parameter.

[0136] After the kinematic model is constructed, the kinematic model is discretized according to a preset distance to obtain a system equation. Alternatively, the derivation of the discretization can use a trapezoidal integral formula.

[0137] For example, the system equation after the discretization can be:

[0138]

[0139] wherein X i is the predicted pose information for the ith prediction point, Δs is the preset distance, and θ i is the heading angle corresponding to the ith prediction point.

[0140] In the present disclosure, the above system equation of the LQR derived using the kinematic model can improve the lateral control accuracy, reduce the tracking error of the reference driving path, and compared to the parameters such as the lateral stiffness of the front and rear tires, the total vehicle mass, and the yaw moment of inertia around the center of mass, which are difficult to accurately obtain and are easily affected by the working conditions, the parameters required by the kinematic model do not change with the working conditions, and thus the uniformity is easier to maintain in subsequent control, and the control accuracy is improved. In addition, the kinematic model can also describe the large steering working condition, and has a wider application range.

[0141] Meanwhile, the system equation is discretized using a preset distance, so that the system equation obtained after the discretization is independent of the vehicle speed, which is beneficial to realize the decoupling of the vehicle lateral and longitudinal control, so that the path tracking effect in the low speed range (such as 0-30 km / h) is not affected by the vehicle speed, and the discretization using the trapezoidal integral formula is also beneficial to increase the control accuracy.

[0142] After the above system equation is obtained, linearization processing can be performed through Taylor expansion. For example, the target equation after Taylor expansion can be:

[0143] X i+1 = A i X i + B i κ i + C i

[0144] wherein A i , B i , and C i are determined in the following manner:

[0145]

[0146]

[0147]

[0148] wherein θ Ri is the heading angle corresponding to the ith reference point.

[0149] In step 14, the target control parameter at the next moment of the target moment is determined according to the LQR optimization model.

[0150] In a possible implementation, step 14 can include the following steps:

[0151] determining the feedback gain matrix capable of minimizing the target cost function under the condition of meeting the target constraint condition;

[0152] determining the target control parameter at the next moment of the target moment according to the first pose information and the feedback gain matrix.

[0153] Optionally, the target cost function can be solved by using a dynamic programming solving method to obtain the feedback gain matrix.

[0154] And the target control parameter κ can be determined according to the following formula:

[0155] κ = -K i X i -L i

[0156] wherein K i , L i is the feedback gain matrix, and:

[0157]

[0158]

[0159]

[0160]

[0161] q i = -QX Ri

[0162] wherein P N = Q, F N = q N .

[0163] In step 15, the vehicle is controlled according to the target control parameter at the next moment of the target moment.

[0164] In a possible implementation, if the target control parameter is the steering curvature, step 15 can include the following steps:

[0165] determining a front wheel steering angle corresponding to the steering curvature;

[0166] controlling the vehicle according to the front wheel steering angle.

[0167] For example, the front wheel steering angle δ can be determined according to the following formula:

[0168] δ = arctan (κL)

[0169] wherein κ is the steering curvature, and L is the wheelbase of the vehicle.

[0170] According to the above technical solution, the pose information of the vehicle at the target time and the pose information of each of the plurality of prediction points on the predicted path are determined as the first pose information, a plurality of reference points and the second pose information of each of the reference points are determined on the reference driving path corresponding to the vehicle, the LQR optimization model is constructed according to the first pose information and the second pose information, so as to determine the target control parameter for the next time of the target time, and then the vehicle is controlled according to the target control parameter at the next time of the target time. The reference points determined from the reference driving path include not only the first reference point corresponding to the target time, but also the second reference points of a preset number after the target time, and these reference points have corresponding prediction points on the predicted path. Then, the LQR optimization model is constructed by using these prediction points and reference points, which is equivalent to adding a preview mechanism in the LQR algorithm. When determining the target control parameter, not only the current path tracking error is considered, but also the path tracking error in the future period is considered. Therefore, a more optimal control parameter can be obtained, and the driving stability of the vehicle can be ensured.

[0171] Figure 3 is a block diagram of a vehicle control device according to an embodiment of the present disclosure. As shown in Figure 3 the device 30 includes:

[0172] A first determination module 31 is configured to determine the pose information of the vehicle at the target time and the pose information of each of the plurality of prediction points as the first pose information, wherein the prediction points are determined from the predicted path corresponding to the vehicle.

[0173] A second determination module 32 is configured to determine a plurality of reference points on the reference driving path corresponding to the vehicle, and determine the pose information corresponding to each of the reference points as the second pose information, wherein the plurality of reference points include the first reference point corresponding to the target time and the second reference points of a preset number after the target time, and each of the reference points has a corresponding prediction point on the predicted path.

[0174] The constructing module 33 is configured to construct an LQR optimization model according to the first pose information and the second pose information.

[0175] The third determining module 34 is configured to determine a target control parameter of a next time point of the target time point according to the LQR optimization model.

[0176] The control module 35 is configured to control the vehicle according to the target control parameter at the next time point of the target time point.

[0177] Optionally, the pose information of the vehicle at the target time point includes a coordinate of the vehicle at the target time point.

[0178] The second determining module 32 includes:

[0179] The first determining sub-module is configured to determine a projection point of the coordinate of the vehicle at the target time point on the reference driving path as the first reference point.

[0180] The second determining sub-module is configured to determine the preset number of second reference points on the reference driving path one by one, starting from a position where the first reference point is located, and taking a preset distance as a spacing between adjacent reference points.

[0181] Optionally, the reference driving path corresponds to a path equation, and the pose information includes a coordinate and a heading angle.

[0182] The second determining module 32 includes:

[0183] The third determining sub-module is configured to determine a respective coordinate corresponding to each reference point according to the reference driving path.

[0184] The fourth determining sub-module is configured to determine, for each reference point, a heading angle corresponding to the reference point according to the coordinate corresponding to the reference point and the path equation.

[0185] Optionally, the heading angle θi corresponding to the i-th reference point Ri is determined by the following formula: Ri :

[0186] θ Ri = arctan(f'(x Ri ))

[0187] wherein 0≤i≤preset number, R0 is the first reference point, f(x Ri ) is the path equation, and f'(x Ri ) is a derivative of the path equation.

[0188] Optionally, the LQR optimization model includes a target cost function.

[0189] The target cost function is determined by the following manner:

[0190] Obtaining a pose cost coefficient and a control cost coefficient;

[0191] Generating a first cost function according to the first pose information, the second pose information and the pose cost coefficient;

[0192] Generating a second cost function according to a preset control parameter and the control cost coefficient;

[0193] Determining a sum of the first cost function and the second cost function as the target cost function.

[0194] Optionally, the target cost function F is generated according to the following formula:

[0195]

[0196] wherein, is the first cost function, is the second cost function; X i is the predicted pose information of the i th prediction point, X Ri is the second pose information of the i th reference point, X R0 is the second pose information of the first reference point, N is the preset quantity, Q is a diagonal matrix generated according to the pose cost coefficient; r is the control cost coefficient, κ i is the control parameter of the i th prediction point.

[0197] Optionally, the pose information includes coordinates and a heading angle, and the pose cost coefficient includes a position cost coefficient and a heading cost coefficient.

[0198] Q is generated according to the following formula:

[0199] Q = diag [q xy , q xy , q θ ]

[0200] wherein, q xy is the position cost coefficient, q θ is the heading cost coefficient.

[0201] Optionally, the LQR optimization model further includes a target constraint condition, and the target constraint condition includes a first constraint condition for constraining a generation manner of the predicted path and a second constraint condition for constraining X0 to be the first pose information.

[0202] Optionally, the pose information includes coordinates and a heading angle.

[0203] The first constraint condition is generated by the following manner:

[0204] constructing a kinematics model taking the pose information as a state variable and taking a preset control parameter as a control variable;

[0205] performing discretization processing on the kinematics model according to a preset distance to obtain a system equation for generating pose information of a second prediction point according to pose information of a first prediction point, the second prediction point being a next prediction point spaced apart from the first prediction point by a preset distance;

[0206] performing linearization processing on the system equation to obtain a processed target equation as the first constraint condition.

[0207] Optionally, the system equation is:

[0208]

[0209] wherein X i is the predicted pose information for the i th prediction point, Δs is the preset distance, and θ i is a heading angle corresponding to the i th prediction point.

[0210] Optionally, the linearization processing is Taylor expansion.

[0211] The target equation is:

[0212] X i+1 = A i X i +B i κ i +C i

[0213] wherein A i , B i , and C i are determined by the following manner:

[0214]

[0215]

[0216]

[0217] wherein θ Ri is a heading angle corresponding to the i th reference point.

[0218] Optionally, the third determination module 34 comprises:

[0219] a fifth determination sub-module configured to determine a feedback gain matrix capable of minimizing the target cost function under the condition of satisfying the target constraint condition.

[0220] A sixth determining sub-module is configured to determine a target control parameter of a next time point of the target time point according to the first pose information and the feedback gain matrix.

[0221] Optionally, the fifth determining sub-module is configured to solve the target cost function by using a dynamic programming solving method to obtain the feedback gain matrix.

[0222] Optionally, the sixth determining sub-module is configured to determine the target control parameter κ according to the following formula:

[0223] κ = -K i X i -L i

[0224] wherein K i , L i is the feedback gain matrix, and:

[0225]

[0226]

[0227]

[0228]

[0229] q i = -QX Ri

[0230] wherein P N = Q, F N = q N .

[0231] Optionally, the target control parameter is a steering curvature.

[0232] The control module 35 comprises:

[0233] A seventh determining sub-module is configured to determine a front wheel steering angle corresponding to the steering curvature.

[0234] A control sub-module is configured to control the vehicle according to the front wheel steering angle.

[0235] As to the apparatus in the above-mentioned embodiments, the specific manners in which various modules perform operations have been described in details in the embodiments about the method, and thus will not be described in details here.

[0236] The present disclosure also provides a non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor, implements the steps of the vehicle control method provided by any of the embodiments of the present disclosure.

[0237] The present disclosure also provides an electronic device, comprising:

[0238] a memory having stored thereon a computer program;

[0239] a processor configured to execute the computer program in the memory to implement the steps of the vehicle control method provided by any of the embodiments of the present disclosure.

[0240] The present disclosure also provides a vehicle comprising the electronic device provided by any of the embodiments of the present disclosure.

[0241] Figure 4 is a block diagram of an electronic device 700 according to an exemplary embodiment. As shown in Figure 4 the electronic device 700 can include a processor 701 and a memory 702. The electronic device 700 can also include one or more of a multimedia component 703, an input / output (I / O) interface 704, and a communication component 705.

[0242] The processor 701 is configured to control overall operations of the electronic device 700 to complete all or part of the steps of the vehicle control method described above. The memory 702 is configured to store various types of data to support operations of the electronic device 700, which can include, for example, instructions for operating any application or method on the electronic device 700, and application-related data, such as contact data, transmitted and received messages, pictures, audio, video, and the like. The memory 702 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic storage, a flash memory, a magnetic disk, or an optical disk. The multimedia component 703 can include a screen and an audio component. The screen can be, for example, a touch screen, and the audio component is configured to output and / or input audio signals. For example, the audio component can include a microphone configured to receive external audio signals. The received audio signals can be further stored in the memory 702 or transmitted through the communication component 705. The audio component further includes at least one speaker configured to output audio signals. The I / O interface 704 provides an interface between the processor 701 and other interface modules, which can be a keyboard, a mouse, a button, and the like. The buttons can be virtual buttons or physical buttons. The communication component 705 is configured to perform wired or wireless communication between the electronic device 700 and other devices. The wireless communication, such as Wi-Fi, Bluetooth, near field communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC, or other 5G, and the like, or a combination of one or more of them, is not limited herein. Therefore, the communication component 705 can include, for example, a Wi-Fi module, a Bluetooth module, an NFC module, and the like.

[0243] In an exemplary embodiment, the electronic device 700 can be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, micro-controllers, microprocessors, or other electronic elements for performing the vehicle control method described above.

[0244] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, which, when executed by a processor, implement the steps of the vehicle control method described above. For example, the computer-readable storage medium can be the memory 702 described above including program instructions executable by the processor 701 of the electronic device 700 to complete the vehicle control method described above.

[0245] In another exemplary embodiment, a computer program product is also provided, which contains a computer program executable by a programmable device, the computer program having code portions for performing the vehicle control method described above when executed by the programmable device.

[0246] The preferred embodiments of the present disclosure are described in detail above with reference to the accompanying drawings, but the present disclosure is not limited to the specific details of the above-described embodiments. Various simple modifications can be made to the technical solutions of the present disclosure within the scope of the technical concept of the present disclosure, and these simple modifications all belong to the protection scope of the present disclosure.

[0247] In addition, it should be noted that each specific technical feature described in the above specific embodiments can be combined in any appropriate manner without contradiction. In order to avoid unnecessary repetition, various possible combinations are not described again in the present disclosure.

[0248] Furthermore, any combination of the various different embodiments of the present disclosure can also be made, as long as it does not deviate from the idea of the present disclosure, and it should also be considered as disclosed by the present disclosure.

Claims

1. A vehicle control method, characterized in that, The method includes: The vehicle's pose information at the target time and the pose information corresponding to each of the multiple prediction points are determined as the first pose information. The prediction points are determined from the prediction path corresponding to the vehicle. On the reference driving path corresponding to the vehicle, multiple reference points are determined, and the pose information corresponding to each reference point is determined as the second pose information. The multiple reference points include a first reference point corresponding to the target time and a preset number of second reference points after the target time, and each reference point has a corresponding prediction point on the prediction path. An LQR optimization model is constructed based on the first pose information and the second pose information; the LQR optimization model includes an objective cost function. Based on the LQR optimization model, determine the target control parameters for the next time step after the target time step; At the next time point after the target time, the vehicle is controlled according to the target control parameters; The objective cost function F is generated according to the following formula: in, For the first cost function, X is the second cost function; i In order to target the i The predicted pose information of each prediction point, X Ri For the first i The second pose information of the reference point, X R0 The second pose information of the first reference point, where N is the preset number. Q This is a diagonal matrix generated based on the pose cost coefficients; r To control the cost coefficient, For the first i Control parameters for each prediction point; the pose information includes coordinates and heading angle; The LQR optimization model further includes objective constraints, which include a first constraint for constraining the generation method of the predicted path and a second constraint for constraining X0 to be the first pose information. The first constraint is generated in the following way: Construct a kinematic model with pose information as the state variable and preset control parameters as the control variable; According to a preset distance, the kinematic model is discretized to obtain a system equation for generating the pose information of the second prediction point based on the pose information of the first prediction point. The second prediction point is the next prediction point that is separated from the first prediction point by a preset distance. The system equations are linearized to obtain the processed objective equation, which serves as the first constraint condition. The system equations are as follows: Among them, X i In order to target the i The pose information predicted by each prediction point The preset distance, θ i For the first i The heading angles corresponding to each predicted point; The linearization process is a Taylor expansion; The objective equation is: Among them, A i B i C i Determined in the following ways: in, θ Ri For the first i The heading angle corresponding to each reference point.

2. The method according to claim 1, characterized in that, The vehicle's pose information at the target time includes the vehicle's coordinates at the target time; The determination of multiple reference points on the reference driving path of the vehicle includes: The projection point of the vehicle's coordinates at the target time onto the reference driving path is determined as the first reference point; Starting from the location of the first reference point, and using a preset distance as the spacing between adjacent reference points, the preset number of second reference points are determined one by one on the reference driving path.

3. The method according to claim 1, characterized in that, The reference driving path corresponds to a path equation, and the pose information includes coordinates and heading angle; Determining the pose information corresponding to each reference point includes: Based on the reference driving path, determine the coordinates corresponding to each of the reference points; For each reference point, the heading angle corresponding to the reference point is determined based on the coordinates of the reference point and the path equation.

4. The method according to claim 3, characterized in that, The number is determined by the following formula. i Reference point R i Corresponding heading angle θ Ri : Where, 0≤ i ≤Preset quantity, R0 is the first reference point, f ( x Ri ) represents the path equation. is the derivative of the path equation.

5. The method according to claim 1, characterized in that, The pose information includes coordinates and heading angle, and the pose cost coefficient includes position cost coefficient and heading cost coefficient; Generate according to the following formula Q : Q = diag[ q xy , q xy , q θ ] in, q xy The location cost coefficient is... q θ The heading cost coefficient is denoted as .

6. The method according to claim 1, characterized in that, The step of determining the target control parameters for the next time step based on the LQR optimization model includes: Determine the feedback gain matrix that minimizes the objective cost function while satisfying the objective constraints; Based on the first pose information and the feedback gain matrix, the target control parameters for the next time step at the target time step are determined.

7. The method according to claim 6, characterized in that, Determining the feedback gain matrix that minimizes the objective cost function while satisfying the objective constraints includes: The objective cost function is solved using dynamic programming to obtain the feedback gain matrix.

8. The method according to claim 6, characterized in that, The step of determining the target control parameters for the next time step based on the first pose information and the feedback gain matrix includes: The target control parameters are determined according to the following formula. : Among them, K i L i Let be the feedback gain matrix, and: in, P N = Q , F N = q N .

9. The method according to claim 1, characterized in that, The target control parameter is the steering curvature; The step of controlling the vehicle according to the target control parameters includes: Determine the front wheel steering angle corresponding to the steering curvature; The vehicle is controlled based on the front wheel steering angle.

10. A vehicle control device, characterized in that, The device includes: The first determining module is used to determine the vehicle's pose information at the target time and the pose information corresponding to each of the multiple prediction points, which are used as the first pose information. The prediction points are determined from the prediction path corresponding to the vehicle. The second determining module is used to determine multiple reference points on the reference driving path corresponding to the vehicle, and determine the pose information corresponding to each reference point as the second pose information. The multiple reference points include a first reference point corresponding to the target time and a preset number of second reference points after the target time, and each reference point has a corresponding prediction point on the prediction path. A construction module is used to construct an LQR optimization model based on the first pose information and the second pose information; the LQR optimization model includes an objective cost function; The third determining module is used to determine the target control parameters for the next time step after the target time step based on the LQR optimization model. A control module is used to control the vehicle according to the target control parameters at the next time after the target time. The objective cost function F is generated according to the following formula: in, For the first cost function, X is the second cost function; i In order to target the i The predicted pose information of each prediction point, X Ri For the first i The second pose information of the reference point, X R0 The second pose information of the first reference point, where N is the preset number. Q This is a diagonal matrix generated based on the pose cost coefficients; r To control the cost coefficient, For the first i Control parameters for each prediction point; the pose information includes coordinates and heading angle; The LQR optimization model further includes objective constraints, which include a first constraint for constraining the generation method of the predicted path and a second constraint for constraining X0 to be the first pose information. The first constraint is generated in the following way: Construct a kinematic model with pose information as the state variable and preset control parameters as the control variable; According to a preset distance, the kinematic model is discretized to obtain a system equation for generating the pose information of the second prediction point based on the pose information of the first prediction point. The second prediction point is the next prediction point that is separated from the first prediction point by a preset distance. The system equations are linearized to obtain the processed objective equation, which serves as the first constraint condition. The system equations are as follows: Among them, X i In order to target the i The pose information predicted by each prediction point The preset distance, θ i For the first i The heading angles corresponding to each predicted point; The linearization process is a Taylor expansion; The objective equation is: Among them, A i B i C i Determined in the following ways: in, θ Ri For the first i The heading angle corresponding to each reference point.

11. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the vehicle control method according to any one of claims 1-9.

12. An electronic device, characterized in that, include: A memory on which computer programs are stored; A processor for executing the computer program in the memory to implement the steps of the vehicle control method according to any one of claims 1-9.

13. A vehicle, characterized in that, Includes the electronic device as described in claim 12.

Images