CN108791289A

CN108791289A - A kind of control method for vehicle and device

Info

Publication number: CN108791289A
Application number: CN201810404991.8A
Authority: CN
Inventors: 李小凤; 赵翾; 宋永刚
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2018-04-28
Filing date: 2018-04-28
Publication date: 2018-11-13
Anticipated expiration: 2038-04-28
Also published as: CN108791289B

Abstract

A kind of control method for vehicle and device, this method includes, according to the information of road surface of road residing for the status information of target vehicle and target vehicle, determine current lateral displacement deviation and lateral angles deviation, according to current lateral displacement deviation, lateral angles deviation and preview distance, determine the lateral displacement deviation for taking aim at position in advance, hereafter, according to current lateral displacement deviation, lateral angles deviation, the pre- lateral displacement deviation for taking aim at position, sliding formwork diverter surface equation and sliding formwork control parameter, determine sliding formwork control rate, finally by sliding-mode control, using sliding formwork control rate as input, determine the steering angle of output and according to steering angle control targe vehicle, the steering angle is travelled for adjusting target vehicle to along destination path.Since the above control method for vehicle uses nonlinear sliding-mode control so that the effect of adjustment target vehicle driving trace is more ideal.

Description

Vehicle control method and device

Technical Field

The application relates to the technical field of unmanned vehicles, in particular to a vehicle control method and device.

Background

The unmanned vehicle is an intelligent vehicle which makes a corresponding global or local path plan according to all or part of known and real-time acquired environmental condition information and automatically makes a control decision so as to ensure that the vehicle safely and reliably runs to a preset destination.

The vehicle control technology is the core of the unmanned vehicle technology, mainly comprises a plurality of parts such as speed and direction control of a vehicle, and can realize the control of a vehicle running route through the speed and direction control. The existing vehicle control method calculates a control quantity through a robust PID (proportional, integral, differential) method, and can control the actual traveling route of the vehicle to continuously approach the expected traveling route of the vehicle according to the control quantity until the actual traveling route is overlapped with the expected traveling route. However, the robust PID method is a linear control method, and therefore, the method can only determine the control amount of the vehicle in a linear method, so that when the vehicle is controlled according to the control amount, the convergence speed of the vehicle toward the desired traveling route is the same regardless of the degree of deviation of the current actual traveling route of the vehicle from the desired traveling route, and the control effect is not ideal.

Disclosure of Invention

The application provides a vehicle control method and device, which are used for solving the technical problem that the effect is not ideal when the vehicle travelling route is controlled by the conventional vehicle control method.

In a first aspect, an embodiment of the present application provides a vehicle control method, first determining a current lateral displacement deviation and a lateral angle deviation according to state information of a target vehicle and road surface information of a road where the target vehicle is located, where the current lateral displacement deviation is used to represent a distance between a target position where the target vehicle is located at a current time and a first lateral position on a target path corresponding to the road, the lateral angle deviation is used to represent an angle deviation between a traveling direction of the target vehicle at the current time and a tangential direction of the target path at the first lateral position, the first lateral position is a position on the target path where a distance from the target position is smallest, the lateral angle deviation is less than or equal to 90 degrees, and the target path is a path expected to be traveled by the target vehicle; then, determining the transverse displacement deviation of the pre-aiming position according to the current transverse displacement deviation, the transverse angle deviation and the pre-aiming distance, wherein the transverse displacement deviation of the pre-aiming position is used for indicating the distance between the pre-aiming position and a second transverse position on the target path when the target vehicle runs to the pre-aiming position, the distance between the projection position of the pre-aiming position on the central axis of the target vehicle and the target position is the pre-aiming distance, the included angle between the vector direction pointing from the target position to the pre-aiming position and the current traveling direction of the target vehicle is less than ninety degrees, and the second transverse position is the position with the minimum distance between the target position and the pre-aiming position on the target path; and finally, determining a sliding mode control rate according to the current transverse displacement deviation, the transverse angle deviation, the transverse displacement deviation of the pre-aiming position, a sliding mode switching surface equation and sliding mode control parameters, and finally determining an output steering angle and controlling the target vehicle according to the steering angle by taking the sliding mode control rate as input through a sliding mode control method, wherein the steering angle is used for adjusting the target vehicle to run along a target path.

By adopting the method, the control on the running route of the target vehicle is realized by the sliding mode control method, so that the running route of the target vehicle approaches to the target route until the running route is overlapped with the target route.

In practice, the current lateral displacement deviation y may be determined according to a first formula_LFirst, aThe formula is as follows:

wherein,represents a pair y_LDerivative, v_yIs the lateral speed, y, of the target vehicle in the direction of travel perpendicular to the current time of the target vehicle_LDetermined from a three-dimensional system state representation of the target vehicle.

In addition, the lateral angle deviation ε may be determined according to a second formula_LThe second formula is:

wherein,represents a pair of ∈_LDerivation, p being the curvature of the target path, v_xis the longitudinal speed of the target vehicle in the traveling direction of the target vehicle at the current moment, β is the included angle between the traveling direction of the target vehicle at the current moment and the projection of the traveling direction on the horizontal plane at the current moment, y_LIs determined according to the three-dimensional system state expression of the target vehicle, wherein w is the yaw rate of the target vehicle, and w is determined according to the three-dimensional system state expression of the target vehicle. Due to the determination of the lateral angle deviation epsilon_Lthe inclined angle beta between the road and the horizontal plane is considered, so that the deviation epsilon is determined according to the transverse angle_LThe result obtained by fitting the driving route is closer to the actual road condition, and the control effect on the driving track of the target vehicle is more ideal.

In implementation, the three-dimensional system state expression of the target vehicle is:

wherein u is a steering angle of the target vehicle,represents a pair v_yThe derivation is carried out by the derivation,denotes derivation of w, a₁₁、a₁₂、a₂₁、a₂₂、b₁₁And b₂₁Are coefficients determined according to the state information of the target vehicle at the present time and the road surface information of the road.

In an implementation, a may be determined from the third indication₁₁The third formula is:

wherein, C_rCornering stiffness of rear tires of a target vehicle, C_fIs the cornering stiffness of the front tyre of the target vehicle, m is the mass of the target vehicle, v_xthe longitudinal speed of the target vehicle in the traveling direction of the target vehicle at the current moment is taken as beta, and beta is an included angle between the traveling direction of the target vehicle at the current moment and the horizontal plane at the current moment;

a may be determined from the fourth publication₁₂The fourth formula is:

wherein a is the front wheel base of the target vehicle, and b is the rear wheel base of the target vehicle;

a can be determined according to a fifth formula₂₁The fifth formula is:

wherein J is the moment of inertia of the target vehicle;

a can be determined according to a sixth formula₂₂The sixth formula is:

b can be determined according to the seventh formula₁₁The seventh formula is:

b can be determined according to the seventh formula₂₁The eighth formula is:

in implementation, the preview distance D may be determined according to a ninth formula, where the ninth formula is:

where d0 is the set distance value, v is the traveling speed of the target vehicle, ρ_frontIs of a first curvature, p_nextIs the second curvature. Wherein the first curvature is the curvature of the target path at the first lateral position, the curvature of the target path at the first pre-aim point, and the mean curvature of the target path between the first lateral position and the first pre-aim point, and the second curvature is the curvature of the target path at the first lateral position, the curvature of the target path at the second pre-aim point, and the curvature of the target path at the first lateral position and the second pre-aim pointThe first pre-aiming point is the point with the smallest distance to the target position among the points on the target path, the distance between which and the target position is larger than the set distance value, the included angle between the vector which is sent out from the target position and points to the first pre-aiming point and the traveling direction of the target vehicle at the current moment is smaller than ninety degrees, the second pre-aiming point is the point with the smallest distance to the target position among the points on the target path, the distance between which and the target position is larger than the set distance value, and the included angle between the vector which is sent out from the target position and points to the second pre-aiming point and the traveling direction of the target vehicle at the current moment is not smaller than ninety degrees.

In practice, the lateral displacement deviation y of the home position may be determined according to the tenth formula_e：

Wherein R is the turning radius of the target path, D is the pre-aiming distance, y_LFor the current lateral displacement deviation, ε_LIs the lateral angular deviation.

In implementation, the following sliding-mode switching surface equation can be constructed:

wherein, y_efor deviation of the horizontal displacement of the preview position, alpha, beta_sand (q/p) is a sliding mode control parameter, α, beta_sAnd (q/p) is a positive odd number;

order above S₁The sliding mode control parameter u may be determined according to an eleventh formula, where:

wherein,g(x)＝1/((R+y_L-Dsinε_L)v_y)，

f(x)＝2(R-Dsinε_L)(a₁₁v_y+a₁₂w)+2D(ρv_xcosβ-w)cosε_L，

r is the turning radius of the target path, D is the pre-aiming distance, y_LFor the current lateral displacement deviation, ε_LIs a lateral angular deviation, v_xIs the longitudinal speed, v, of the target vehicle in the direction of travel of the target vehicle at the current moment in time_yIs the lateral velocity of the target vehicle in the direction of travel perpendicular to the target vehicle's current time, w is the target vehicle's yaw rate, a₁₁、a₁₂Is a coefficient determined based on the state information of the target vehicle at the present time and the road surface information of the road.

In implementation, after the control target vehicle runs, the adjustment value of the sliding mode control parameter can be determined by a deep learning method; specifically, a loss function representing an actual loss between the predicted travel path and the actual travel path of the target vehicle may be determined, and the adjustment value of the sliding-mode control parameter may be determined based on the loss function. The method for determining the loss function comprises the following steps: determining a loss function according to the deviation amount between the predicted lateral displacement deviation and the actual lateral displacement deviation of the target vehicle at least one sampling moment; or determining a loss function according to the deviation amount between the predicted lateral angle deviation and the actual lateral angle deviation of the target vehicle at least one sampling moment. The predicted transverse displacement deviation is used for representing the distance between a first predicted position where the target vehicle is located and a third transverse position on the target path at any sampling time, and the third transverse position is the position where the distance between the target path and the first predicted position is the minimum; the actual transverse displacement deviation is used for representing the distance between the actual position of the target vehicle and a fourth transverse position on the target path at the sampling moment, and the fourth transverse position is the position with the minimum distance between the actual position and the target path; the predicted transverse angle deviation is used for representing the angle deviation between the predicted travelling direction of the target vehicle and the tangent direction of the target path at a fifth transverse position at any sampling moment, and the fifth transverse position is the position with the minimum distance between the target path and a second predicted position where the target vehicle is located; the actual lateral angle deviation is used to indicate an angle deviation between the actual traveling direction of the target vehicle and a tangential direction of the target path at a fifth lateral position at the sampling time, where the fifth lateral position is a position on the target path where a distance from the second predicted position where the target vehicle is located is the smallest.

In practice, prior to determining the loss function, a convolved signature C of the target vehicle at the sampling time may be determined according to a twelfth formula_xThe twelfth formula is:

C_x＝f(∑IG_x+b_x)，

wherein I is an input array for representing state information of the target vehicle and road surface information of a road where the target vehicle is located at the sampling time, G_xFor trainable convolution kernels, b_xFor trainable biasing, G_xIs a first random value, b_xIs a second random value;

from the convolution feature map C_xAnd determining the predicted lateral displacement deviation of the target vehicle at the sampling moment. In addition, in the implementation, the feature graph C can also be obtained according to convolution_xAnd determining the predicted transverse angle deviation of the target vehicle at the sampling moment.

In practice, G can also be adjusted by the gradient descent method_xAnd/or b_x. Specifically, the sensitivity Ω can be determined according to the thirteenth formula_xThe thirteenth formula is:

Ω_x＝w_x+1(σ'(w_x+1C_x+b_x+1)Ψ_i-1)，

wherein, w_x+1Represents the value of the multiplicative bias in the convolution kernel, σ' (w)_x+1C_x+b_x+1) Representing the sensitivity function, Ψ_i-1Up-sampling the input array;

the sensitivity function is:

determining G according to the fourteenth formula_xWeight value update of (Δ w)_xThe fourteenth formula is:

Δw_x＝-η∑(Ω_xΨ_i+1)，

where η represents the learning rate, Ψ_i+1Down-sampling the input array;

according to G_xWeight value update of (Δ w)_xAnd adjusting the value of Gx.

in implementation, the sliding mode control parameters α, β may be determined according to the fifteenth formula_sAnd (q/p), the fifteenth formula being:

wherein,

f (Q, k) represents a weight factor of alpha, f (Q, epsilon) represents β_sf (Q, Q/p) represents a weight factor of (Q/p), n represents a variable range of α, and m represents β_sG represents the variable range of q/p, γ_aIs the gain factor, s, z, c₁，c₂respectively, a constant greater than zero, Delta alpha is the adjustment value of alpha, Delta beta_sis beta_sΔ (q/p) is the adjustment value of (q/p). According to sliding modecontrol parameters α, beta_sAnd (q/p) the sliding mode control rate can be adjusted by the adjustment value, the target vehicle is controlled according to the steering angle determined by the adjusted sliding mode control rate, and the running track of the target vehicle can further quickly approach to the target path.

In a second aspect, an embodiment of the present application provides a vehicle control apparatus, which includes a memory and a processor, where the memory stores a computer program or instructions, and the processor calls the computer program or instructions stored in the memory to implement the method performed by the cache policy server in any one of the possible designs of the first aspect and the first aspect.

In a third aspect, the present application provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a computer, the computer can be caused to implement the vehicle control method in any one of the above first aspect and possible designs of the first aspect.

In a fourth aspect, the present application provides a computer program product, which when executed by a computer, can enable the computer to implement the vehicle control method in the first aspect and any one of the possible designs of the first aspect.

In a fifth aspect, the present application provides a chip, which is coupled to a transceiver, and is used to implement the vehicle control method in any one of the possible designs of the first aspect and the first aspect.

Drawings

Fig. 1 is a schematic system architecture diagram of a vehicle control method according to an embodiment of the present application;

fig. 2 is a schematic structural diagram of a control device according to an embodiment of the present disclosure;

fig. 3 is a schematic view of a driving state of a target vehicle according to an embodiment of the present application;

FIG. 4 is a flow chart of a vehicle control method provided by an embodiment of the present application;

FIG. 5 is a flow chart of a method for adjusting a synovial control parameter provided in an embodiment of the present application;

FIG. 6 is a flow chart of another vehicle control method provided by an embodiment of the present application;

fig. 7 is a schematic structural diagram of another control device provided in an embodiment of the present application.

Detailed Description

The embodiment of the invention provides a media content caching method and device, which are used for solving the technical problem of unsatisfactory control effect when a robust PID method is used for controlling a vehicle running track.

The following describes several concepts related to embodiments of the present application:

the state information of the vehicle refers to the parameters of the vehicle and the running parameters during the running of the vehicle, wherein the parameters of the vehicle include but are not limited to the following parameters: mass of the vehicle, moment of inertia of the vehicle, wheelbase of the vehicle, front wheel cornering stiffness of the vehicle, rear wheel cornering stiffness of the vehicle; the driving parameters in the driving of the vehicle include, but are not limited to, the following parameters: the vehicle position information, the vehicle running speed, the vehicle traveling direction, the vehicle yaw rate, the vehicle front wheel rotation angle and the like, wherein the direction pointed by the vector which is emitted by the vehicle center of mass and points to the vehicle head middle position can be taken as the vehicle traveling direction, and the position of the vehicle center of mass can be taken as the vehicle position.

The road surface information of the road refers to information of the road on which the vehicle runs, and includes but is not limited to the following parameters: position information of a road (e.g., coordinates of the road), width information of the road, a traveling direction of the road, a curvature of the road, a turning radius of the road, and a tilt angle of the road (an angle between the traveling direction of the road and a projection of the traveling direction of the road on a horizontal plane), and the like. The position information of a target path located in a road can be used as the position information of the road, the advancing direction of the target path can be used as the advancing direction of the road, the curvature of the target path can be used as the curvature of the road, the turning radius of the target path can be used as the turning radius of the road, and the included angle between the advancing direction of the target path and the projection of the advancing direction on a horizontal plane can be used as the inclined angle of the road; the target path may be a path expected to be traveled by the vehicle on a road on which the vehicle is traveling, for example, the target path may coincide with a center line of the road, and a traveling direction of the target path is the same as the traveling direction of the road.

In order to make the objects, technical solutions and advantages of the present application more clear, the present application will be further described in detail with reference to the accompanying drawings.

As shown in fig. 1, a system 100 of a vehicle control method provided in an embodiment of the present application includes a controlled target vehicle 101 and a control device 102, where the control device 102 may be a terminal device located on the controlled target vehicle 101, or a device capable of acquiring state information of the controlled target vehicle 101 and road surface information of a road where the target vehicle 101 is located and performing remote control on the target vehicle 101, such as a cloud server device.

In implementation, the control device 102 may directly acquire the information of the target vehicle 101 from the target vehicle 101 and directly control the traveling of the target vehicle 101, or may acquire the information of the target vehicle 101 through another relay device and control the traveling of the target vehicle 101 through another relay device. Note that the travel of the control target vehicle 101 here may be the travel of the control target vehicle 101 by a control signal, which may be generated directly by the control device 102 or generated by the control device 102 instructing another device.

As shown in fig. 2, the control device 102 of a vehicle control method provided in an embodiment of the present application may include a memory 201 and a processor 202, where, for example, through the control device 102, the memory 201 is configured to store a computer program or instructions, and the processor 202 is configured to execute the computer program or instructions stored in the memory, so that the control device 102 implements the steps involved in the control device 102 in the method for controlling data transmission provided in an embodiment of the present application.

It should be understood that, in an embodiment, the control device 102 may be located on a target vehicle to be controlled, and the control device 102 implements, through the memory 201 and the processor 202, a calculation capability for supporting the control device 102 to determine a steering angle of the target vehicle according to the vehicle control method provided in the embodiment of the present application, and a control capability for supporting the control device 102 to control the target vehicle to travel according to the steering angle. In another embodiment, if the control device 102 is a remote device, the control device 102 further needs to have a communication capability, specifically, as shown in fig. 2, the control device 102 may further have a transceiver 203, where the transceiver 203 is used for the control device 102 to interact, and in this case, the transceiver 203 may be a wireless transceiver for supporting the control device 102 to remotely acquire information of the target vehicle and remotely control the target vehicle to travel according to the steering angle after the steering angle is determined.

Fig. 3 is a schematic diagram of a state of the target vehicle 101 in driving, in which the centroid of the target vehicle 101 is located at point M, that is, the target vehicle 101 is located at point M, the central axis of the target vehicle 101 is a straight line located at point M, N, point N is located at a middle position of the front wheel axle of the target vehicle 101, and the vector isThe same direction as the traveling direction of the target vehicle, in the embodiment of the present application, may be passedIndicating a direction of travel of the target vehicle; the center line of the road on which the target vehicle 101 is located is a curve S₀Wherein, curve S₀Is ρ and the point O is the steering center of the target vehicle 101. The vehicle control method provided by the embodiment of the application can be used for adjusting the track of the target vehicle 101 to approach the track of an expected target vehicle form, namely the road center line S₀Up to the target vehicle 101 trajectory and road centerline S₀And (4) overlapping.

Next, with the target position where the target vehicle 101 is located as M, the target path where the target vehicle 101 is expected to travel is set as the road center line S₀For example, a method of controlling the target vehicle 101 by the control apparatus 102 in the embodiment of the present application is explained, the method having the steps shown in fig. 4:

step S101: the control device 102 determines a current lateral displacement deviation and a lateral angle deviation according to the state information of the target vehicle 101 at the current time and the road surface information of the road where the target vehicle 101 is currently located, wherein the current lateral displacement deviation is used for representing the target position M where the target vehicle 101 is currently located and the road center line S₀First lateral position M of₁The lateral angle deviation is used to indicate the traveling direction of the target vehicle 101 at the present timeWith the road centre line S₀At a first transverse position M₁Angle deviation between the tangential directions of (1), M₁Is the road center line S₀A position where the distance between the upper and target positions is minimum; in practice, the lateral angular deviation is less than or equal to 90 degrees;

step S102: the control device 102 determines the lateral displacement deviation of the preview position according to the current lateral displacement deviation, the lateral angle deviation and the preview distance, and the lateral displacement deviation of the preview position is used for indicating that the target vehicle 101 runs to the preview position M' and is in contact with the road center line S₀Second lateral position M of₂Wherein the projected position M 'of the home position M' on the central axis of the current time of the target vehicle is at a distance equal to the home distance from the target position M, and a second lateral positionM₂The position with the minimum distance from the pre-aiming position M' in all the positions on the target path; in implementation, M "may be taken as the preview location M';

step S103: the control equipment 102 determines a sliding mode control rate according to the current transverse displacement deviation, the transverse angle deviation, the transverse displacement deviation of the pre-aiming position and the sliding mode control parameters; in implementation, when the sliding mode control rate is initially determined, the sliding mode control parameters can be configured as preset parameters;

step S104: the control equipment 102 determines a steering angle according to the sliding mode control rate; the steering angle is used to adjust the traveling route of the target vehicle 101 to travel along the target path;

step S105: the control apparatus 102 controls the target vehicle 101 to travel according to the determined steering angle.

In the implementation of step S101, the control device 102 may determine a three-dimensional system state expression according to the state information of the target vehicle 101 and the road surface information of the road, further determine the lateral speed of the target vehicle 101 in the y-axis direction (perpendicular to the traveling direction of the target vehicle 101 at the current moment), and the lateral speed of the target vehicle 101 in the y-axis direction may be used to determine the target position M where the target vehicle 101 is located, and the first lateral position M on the target path₁The distance between the two, i.e. the current lateral displacement deviation; and, the control apparatus 102 may determine the current yaw rate of the vehicle from the three-dimensional system state expression of the target vehicle 101, and the determined yaw rate may be used to determine the current direction of travel of the target vehicle 101An angular deviation from a tangent of the target path at the first transverse position, i.e., a transverse angular deviation.

In implementation, the control device 102 determines the state information of the target vehicle 101, including, but not limited to, the information shown in table 1.

Status information	Sign of status information/unit of status information	Numerical value of status information (example)
			Quality of	m/kg	3200
Moment of inertia	J/kg*m2	9500
			Front wheel base	a/m	1.2
Rear wheel base	b/m	1.7
			Front wheel cornering stiffness	C_f/N*rad-1	190000
Rear wheel cornering stiffness	C_r/N*rad-1	210000

TABLE 1 State information Table for target vehicle 101

Specifically, the control device 102 may determine the current lateral displacement deviation y according to a first formula_LThe first formula is:

wherein,represents a pair y_LDerivative, v_yFor the target vehicle 101 in the vertical directionTransverse velocity in the direction of (v)_yDetermined from a three-dimensional system state expression of the target vehicle 101;

the control device 102 may also determine the lateral angle deviation epsilon according to a second formula_LThe second formula is:

wherein,represents a pair of ∈_LTaking a derivative, rho is a road center line S₀Curvature of v_xIs a target vehicle 101longitudinal velocity in the direction, β isAndthe angle between the projections on the horizontal plane (i.e., the slope angle of the road on which the target vehicle 101 is currently located), y_LIs root ofAnd w is the yaw velocity of the target vehicle and is determined according to the three-dimensional system state expression of the target vehicle.

In implementation, the three-dimensional system state expression of the target vehicle 101 determined by the control device 102 (hereinafter, referred to as equation (1) for convenience of description) may be:

where u is the steering angle of the target vehicle 101 at the present moment,represents a pair v_yThe derivation is carried out by the derivation,denotes derivation of w, a₁₁、a₁₂、a₂₁、a₂₂、b₁₁And b₂₁Are coefficients determined from the state information of the target vehicle 101 at the present time and the road surface information of the road on which the target vehicle 101 is located.

In implementation, in the three-dimensional system state expression of the target vehicle 101, a₁₁Can be determined by the third formula:

wherein, C_rCornering stiffness of rear tires of the target vehicle 101, C_fThe cornering stiffness of the front tires of the target vehicle 101, m being the mass of the target vehicle 101;

a₁₂can be determined by the fourth equation:

wherein a is a front wheel base of the target vehicle, and b is a rear wheel base of the target vehicle;

a₂₁may be determined by the fifth equation:

wherein J is a moment of inertia of the target vehicle;

a₂₂may be determined by the sixth equation:

b₁₁can be determined by the seventh equation:

b₁₂may be determined by the eighth equation:

in practice, the third, fourth, fifth, sixth, seventh and eighth formulas are substituted into the formula (1), that is, the three-dimensional system state expression of the target vehicle 101, to obtain the respective formulasAre expressed byIs further according toCan determine v_yAnd according toCan determine w, v_ySubstituting into the first formula, y can be obtained_LAnd substituting w into the second formula, can yield ε_LIs described in (1).

In the implementation of step S102, the preview distance may be determined according to a set distance value, and specifically, the control device 102 may determine the preview distance according to a ninth formula:

where d0 is a set distance value, v is a running speed of the target vehicle, ρ_frontIs the first curvature, p_nextIs the second curvature.

In practice, the pre-line distance D is used to represent the driving track of the target vehicle 101 and the road center line S₀At the time of coincidence, the target vehicle 101 is expected to travel a distance ofA distance component in the direction. It can be understood that if the pre-aiming distance D is too small, the change rate of the steering angle of the target vehicle 101 may be too large, which may cause the target vehicle to shake violently in the adjustment of the driving track; the too large preview distance D may result in too slow adjustment of the driving track of the target vehicle 101, and thus the adjustment effect is not good. By adopting the method, the value of the pre-aiming distance D can be dynamically adjusted, and the influence on the control precision of the target vehicle 101 caused by the overlarge or undersize value of the pre-aiming distance D is prevented.

Taking fig. 3 as an example, in implementation, the control device 102 may determine the first curvature ρ according to the following method_front：

The first step is as follows: selecting a first preview point N₁The first preview point N₁Is the road center line S₀The point closest to the M point among the points with the distance between the upper M point and the M point larger than d0, and the vectorDirection of travel with target vehicleThe included angle between is less than ninety degrees; in particular, this step is performed in such a way that the control device 102 can select a first preliminary pointing point N_1xWherein the first preliminary preview point N_1xLocation vectorDirection of travel with target vehicleLess than ninety degrees, the control device 102 further determines N_1xThe distance S between the target vehicle and the current M point_xIf d0 is greater, if so, N is determined_1xIf the point is the first pre-aiming point, otherwise, continuously selecting the next first pre-aiming point and judging whether the distance between the first pre-aiming point and the point M is greater than d 0;

the second step is that: the center line S of the road₀At a first transverse position M₁Curvature of, road center line S₀At a first preview point N₁And the road center line S₀At a first transverse position M₁And a first preview point N₁The average value of the curvatures of the points in between is taken as the first curvature rho_front。

Further in implementation, the control device 102 may determine the second curvature ρ according to the following method_next：

The first step is as follows: selecting a second preview point N₂The second preview point N₂In the roadCore line S₀The point closest to the M point among the points with the distance between the upper M point and the M point larger than d0, and the vectorDirection of travel with target vehicleThe included angle between the two is not less than ninety degrees; in another embodiment, the first lateral position M₁To a first preview point N₁The number of points of (2) and a second preview point N₂To a first transverse position M₁The number of points is the same;

the second step is that: the center line S of the road₀At a first transverse position M₁Curvature of, road center line S₀At a second preview point N₂And the road center line S₀At a first transverse position M₁And a second preview point N₂The average value of the curvatures of the points in between is taken as the first curvature rho_front。

In the implementation of step S102, the lateral displacement deviation of the preview location may be determined according to the current lateral displacement deviation, the lateral angle deviation, and the preview distance according to the geometric relationship. Specifically, taking fig. 3 as an example, the control device 102 may determine the lateral displacement deviation y of the home position according to a tenth formula_e：

Wherein R is the turning radius of the target path, D is the pre-aiming distance, y_LFor the current lateral displacement deviation, ε_LIs the lateral angular deviation. In practice, the turning radius R of the target path X point_xAnd curvature rho_xThe following relations exist between the following components:wherein, the X point is any point on the target path.

In the implementation of step S102, the following sliding-mode switching surface equation (hereinafter, abbreviated as formula (2)) may be designed:

wherein, y_eIs the lateral displacement deviation of the home position,represents a pair y_ederivative, α, β_sand (q/p) is a sliding mode control parameter, α, beta_sAnd q/p is a positive odd number.

In practice, the sliding mode switching surface S₁When 0, the natural buffeting of the sliding mode is minimal, so that the sliding mode control rate u can be expressed as the eleventh formula:

wherein g (x) can be represented by formula (3):

g(x)＝1/((R+y_L-Dsinε_L)v_y)，(3)

f (x) can be represented by the formula (4):

f(x)＝2(R-Dsinε_L)(a₁₁v_y+a₁₂w)+2D(ρv_xcosβ-w)cosε_L，(4)

in the eleventh formula, the (3) and the (4) formulas, R is a steering radius, D may be a home distance determined according to a ninth formula, and y_LFor the current lateral displacement deviation, ε_LIs a lateral angular deviation, v_xIs the longitudinal speed, v, of the target vehicle 101 in the x-axis direction_yIs the lateral velocity of the target vehicle 101 in the y-axis direction, w is the yaw rate of the target vehicle 101, a₁₁、a₁₂Is a coefficient determined based on the state information of the target vehicle 101 and the road surface information of the road on which the target vehicle 101 is located. In practice a₁₁Can be determined according to a third formula, a₁₂May be determined according to a fourth formula.

In the implementation of step S104, the control apparatus 102 may determine the steering angle of the target vehicle 101 by the sliding mode control method using the determined sliding mode control rate as an input variable for the sliding mode control, and control the running of the target vehicle 101 according to the determined steering angle.

After step S105, the control apparatus 102 may also, according to the route on which the target vehicle 101 actually travels and the road center line S on which the target vehicle 101 is expected to travel₀Determining an adjustment value of the sliding mode control parameter through a deep learning method, and adjusting the sliding mode control parameter according to the adjustment value of the sliding mode control parameter, so that the driving path of the target vehicle 101 is controlled through the adjusted sliding mode control parameter to be closer to the road center line S₀。

Specifically, the control apparatus 102 may determine the adjustment value of the sliding-mode control parameter according to the following method:

determining the predicted lateral displacement deviation e of the target vehicle 101 at least one sampling instant by means of a deep learning method₁And the actual lateral displacement deviation e of the target vehicle 101 at the same sampling time₁'; thereafter, the predicted lateral displacement deviation e of the target vehicle 101 is based on at least one sampling instant₁Deviation from actual lateral displacement e₁' deviation amount therebetween, determining a loss function f (e)₁，e₁') for representing a loss between the predicted lateral displacement deviation and the actual lateral displacement deviation of the target vehicle 101 at least one sampling instant; according to a loss function f (e)₁，e₁') determining an adjustment value of the sliding mode control parameter, and adjusting the sliding mode control parameter according to the adjustment value of the sliding mode control parameter.

In an implementation, at least one sampling instant may also be determinedPredicted lateral angle deviation e of target vehicle 101₂And the actual lateral angle deviation e of the target vehicle 101 at the same sampling instant₂'; then, the predicted lateral angle deviation e of the target vehicle 101 is calculated from at least one sampling time₂Deviation from true transverse angle e₂' deviation amount therebetween, determining a loss function f (e)₂，e₂') for representing a loss between the predicted lateral displacement deviation and the actual lateral displacement deviation of the target vehicle 101 at least one sampling instant; according to a loss function f (e)₂，e₂') determining an adjustment value of the sliding mode control parameter, and adjusting the sliding mode control parameter according to the adjustment value of the sliding mode control parameter.

In addition, in an implementation, the loss function may be determined based on a deviation amount between the predicted lateral displacement deviation and the actual lateral displacement deviation of the target vehicle 101 at least one sampling time and a deviation amount between the predicted lateral angle deviation and the actual lateral angle deviation of the target vehicle 101 at the same sampling time.

In implementation, the lateral displacement deviation of the target vehicle 101 at the sampling time and/or the lateral angle deviation of the target vehicle 101 at the sampling time may be determined by a convolutional neural network, as follows:

constructing the twelfth formula by determining the convolution feature map C_x：

C_x＝f(∑IG_x+b_x) (twelfth formula)

Wherein I is an input matrix for representing state information of the target vehicle 101 and road surface information of a road on which the target vehicle 101 is located at least one sampling time, G_xFor trainable convolution kernels, b_xFor trainable biasing, G_xIs a first random value, b_xIs a second random value; in implementation, the state information of the target vehicle 101 and the road surface information of the road where the target vehicle 101 is located include, but are not limited to, the location of the target vehicle 101, the traveling direction of the target vehicle 101, the targetThe traveling speed of the vehicle 101, the curvature of a target path corresponding to a road where the target vehicle 101 is located, the shortest distance (i.e., the lateral displacement deviation at the sampling time) from the target vehicle 101 to the target path corresponding to the road where the target vehicle 101 is located, the pre-aiming distance, and the like, according to an input array at a certain time, the convolutional neural network can predict the lateral displacement deviation and/or the lateral angle deviation of the target vehicle 101;

from the convolution feature map C_xThe predicted lateral displacement deviation of the target vehicle 101 at the sampling instant can be determined and/or the predicted lateral angle deviation of the target vehicle at the sampling instant can be determined.

In implementation, training update of the convolution kernel can be performed for multiple times until the convolution kernel converges, and then a minimum empirical risk function En (f) is applied;

wherein,

f((x),y_i) Is a loss function for a certain sample.

To minimize the loss function, the convolution kernel can be trained by a gradient descent method. The method has the following ideas: if the function f (x) is differentiable and defined somewhere (e.g., point n), then the function is in the opposite direction along the gradient at that pointThe decrease is fastest. En (f) is obtained from the calculation every iteration.

Specific adjustment G_xThe method of (3), may comprise:

determining G according to the thirteenth formula_xSensitivity omega of_xThe thirteenth formula is:

Ω_x＝w_x+1(σ'(w_x+1C_x+b_x+1)Ψ_i-1) (thirteenth formula)

Wherein, w_x+1Represents the value of the multiplicative bias in the convolution kernel, σ' (w)_x+1C_x+b_x+1) Representing the sensitivity function, Ψ_i-1Up-sampling the input array I; the sensitivity function is:

Δw_x＝-η∑(Ω_xΨ_i+1) (fourteenth formula)

wherein η represents a learning factor, Ψ_i+1in implementation, η may be a percentage constant;

G_xweight value update of (Δ w)_xAdjusting G_xThe value of (a).

In practice, train G_xBefore, it may also be determined whether training is needed, for example, it may be determined whether training of the convolution kernel is needed according to a deviation degree between a convolution result and an actual measurement result.

in implementation, the sliding mode control parameters include alpha and beta_sand (q/p), wherein, α, β_sAnd (q/p) is a positive odd number, the adjustment value of the sliding mode control parameter may be determined according to a fifteenth formula:

wherein,

f (Q, k) represents a weight factor of alpha, f (Q, epsilon) represents β_sf (Q, Q/p) represents a weighting factor of (Q/p), n represents a value of alphain the variation range, m represents beta_sG represents the variable range of q/p, γ_afor the gain factor, s, z, c1, c2 are respectively a constant greater than zero, Δ α is the adjustment value of α, Δ β_sis beta_sΔ (q/p) is the adjustment value of (q/p).

by the above method, α beta can be obtained_sand the adjustment values of (q/p) Δ α, Δ β_sand Δ (q/p), Δ α, Δ β can be determined the next time the synovial control rate is determined_sAnd Δ (q/p) as a synovial membrane control parameter.

As shown in fig. 5, a method for adjusting a synovial membrane control parameter provided by an embodiment of the present application comprises the following steps:

step 501: performing convolution processing on the state information of the target vehicle 101 at the current moment and the road surface information of the road where the target vehicle 101 is located; in practice, the state information of the target vehicle 101 at the sampling time may be used to determine the predicted lateral displacement deviation and the predicted lateral angle deviation;

step 502: judging whether the convolution kernel needs to be updated according to the convolution processing result, if so, executing a step 503, otherwise, executing a step 504: in implementation, whether the convolution kernel needs to be trained may be determined according to a deviation amount between the predicted lateral displacement deviation and the actual lateral displacement deviation, and/or according to a deviation amount between the predicted lateral angle deviation and the actual lateral angle deviation, where the actual lateral displacement deviation and/or the actual lateral angle deviation may be determined according to an inertial navigation method;

step 503: the convolution kernel is updated according to the gradient descent method, and then step 501 is performed:

step 504: determining a predicted lateral displacement deviation and a predicted lateral angle deviation of the target vehicle 101 at least one sampling moment according to the result of the convolution processing;

step 505: determining a loss function according to the predicted lateral displacement deviation, the predicted lateral angle deviation, the actual lateral displacement deviation and the actual lateral angle deviation of the target vehicle 101 at least one sampling moment; in implementation, the actual lateral displacement deviation and the actual lateral angle deviation may be determined from inertial navigation;

step 502: based on the loss function, an adjustment value for the synovial membrane control parameter is determined.

As shown in fig. 6, if the control device 102 is located on the target vehicle 101 shown in fig. 3, a method for controlling a vehicle provided by an embodiment of the present application includes the following steps:

step 601: the control device 102 acquires state information of the target vehicle 101 and road surface information of a road on which the target vehicle 101 is located; in implementation, the road surface information of the road on which the target vehicle 101 is located includes information of a target path that the target vehicle 101 is expected to travel;

step 602: the control device 102 determines the current lateral displacement deviation y based on the state information of the target vehicle 101 and the road surface information of the road_LTransverse angle deviation epsilon_LAnd a pre-aiming distance D;

step 603: the control device 102 is dependent on the current lateral displacement deviation y_LTransverse angle deviation epsilon_LAnd a pre-aiming distance D, determining the transverse displacement deviation y of the pre-aiming position_e；

Step 604: the control device 102 deviates from the lateral displacement y of the home position according to the home position_eAnd sliding mode control parameters are constructed to form a sliding mode switching surface equation S₁Determining a sliding mode control rate u;

step 605: the control apparatus 102 determines an output amount of the sliding mode control method according to the sliding mode control method with the sliding mode control rate u as an input amount, and controls the steering angle of the target vehicle 101 according to the output amount of the sliding mode control method, so that the running track of the target vehicle 101 continuously approaches the target path.

Based on the same concept as that of the method embodiment, the embodiment of the present application further provides a control device, configured to implement the method related to the control device 102 in the embodiment of the present application, and in a specific implementation, the control device may be a terminal, a cloud device, or other device or hardware with a type of function, which is used to implement the method. The control device may have a structure as shown in fig. 2.

As shown in fig. 2, in a schematic diagram of a possible logical structure of the control device 102 involved in the foregoing embodiments provided for the embodiments of the present application, the control device 102 includes a processor 202. In the embodiment of the present application, the processor 202 is configured to control and manage the actions of the control device 102. The control device 102 may also include a memory 201 and a transceiver 203. The memory 202 is used, among other things, to store computer programs or instructions for controlling the device 102. The transceiver 201 is used to support the control device 102 for communication.

In the control device 102 shown in fig. 2, the processor 202 may be a central processing unit, a general purpose processor, a digital signal processor, an application specific integrated circuit, a field programmable gate array or other programmable logic device, a transistor logic device, a hardware component, or any combination thereof. Which may implement or perform the various illustrative logical blocks, modules, and circuits described in connection with the disclosure. Processor 202 may also be a combination that performs a computing function, including for example, one or more microprocessors, a digital signal processor and a combination of microprocessors, and the like.

Specifically, in the control device 102 shown in fig. 2, the transceiver 203 is used for the control device 102 to communicate;

a memory 201 for storing computer code or instructions;

a processor 202 for calling computer code or instructions in the memory 201 to perform the following steps:

a memory for storing computer code or instructions;

a processor for calling computer code or instructions in memory to perform the steps of:

determining a current transverse displacement deviation and a transverse angle deviation according to the state information of the target vehicle and the road surface information of a road where the target vehicle is located, wherein the current transverse displacement deviation is used for representing the distance between a target position where the target vehicle is located at the current moment and a first transverse position on a target path corresponding to the road, the transverse angle deviation is used for representing the angle deviation between the traveling direction of the target vehicle at the current moment and the tangential direction of the target path at the first transverse position, the first transverse position is a position where the distance between the target path and the target position is the minimum, the transverse angle deviation is smaller than or equal to 90 degrees, and the target path is a path expected to be traveled by the target vehicle;

determining the transverse displacement deviation of the pre-aiming position according to the current transverse displacement deviation, the transverse angle deviation and the pre-aiming distance, wherein the transverse displacement deviation of the pre-aiming position is used for indicating the distance between the pre-aiming position and a second transverse position on the target path when the target vehicle runs to the pre-aiming position, the distance between the projection position of the pre-aiming position on the central axis of the target vehicle and the target position is the pre-aiming distance, the included angle between the vector direction pointing from the target position to the pre-aiming position and the current traveling direction of the target vehicle is less than ninety degrees, and the second transverse position is the position with the minimum distance between the second transverse position and the pre-aiming position on the target path;

determining a sliding mode control rate according to the current transverse displacement deviation, the current transverse angle deviation, the transverse displacement deviation of the pre-aiming position, a sliding mode switching surface equation and sliding mode control parameters, wherein the sliding mode switching surface equation is used for indicating a sliding mode switching surface of a sliding mode control method, and the sliding mode control rate is used for controlling a target vehicle to run along a target path;

and controlling the running track of the target vehicle according to the sliding mode control rate.

Optionally, when determining the current lateral displacement deviation, the processor 202 is specifically configured to:

determining a current lateral position according to a first formulaShift deviation y_LThe second formula is:

wherein,represents a pair y_LDerivative, v_yIs the lateral speed, y, of the target vehicle in the direction of travel perpendicular to the current time of the target vehicle_LDetermined according to a three-dimensional system state expression of the target vehicle;

in determining the lateral angle deviation, the processor is specifically configured to:

determining the lateral angle deviation epsilon according to a second formula_LThe second formula is:

wherein,represents a pair of ∈_LDerivation, p being the curvature of the target path, v_xis the longitudinal speed of the target vehicle in the traveling direction of the target vehicle at the current moment, β is the included angle between the traveling direction of the target vehicle at the current moment and the projection of the traveling direction on the horizontal plane at the current moment, y_LIs determined according to the three-dimensional system state expression of the target vehicle, wherein w is the yaw rate of the target vehicle, and w is determined according to the three-dimensional system state expression of the target vehicle.

Optionally, the three-dimensional system state expression of the target vehicle is:

Alternatively, a₁₁The third formula, determined according to the third expression, is:

where Cr is the cornering stiffness of the rear tires of the target vehicle, Cf is the cornering stiffness of the front tires of the target vehicle, m is the mass of the target vehicle, v_xthe longitudinal speed of the target vehicle in the traveling direction of the target vehicle at the current moment is taken as beta, and beta is an included angle between the traveling direction of the target vehicle at the current moment and the horizontal plane at the current moment;

a₁₂as determined according to the fourth disclosure, the fourth formula is:

a₂₁determined according to the fifth disclosure, the fifth formula is:

wherein J is the moment of inertia of the target vehicle;

a₂₂as determined according to the sixth disclosure, the sixth formula is:

b₁₁a seventh formula, determined according to the seventh disclosure, is:

b₂₁an eighth formula, determined according to the eighth disclosure, is:

optionally, when determining the pre-aiming distance, the processor 202 is specifically configured to:

determining a first curvature and a second curvature, wherein the first curvature is a curvature of the target path at a first lateral position, a curvature of the target path at a first pre-aim point, and an average curvature of the target path between the first lateral position and the first pre-aim point, the second curvature is a curvature of the target path at the first lateral position, a curvature of the target path at a second pre-aim point, and an average curvature of the target path between the first lateral position and the second pre-aim point, the first pre-aim point is a point on the target path where a distance from the target position is greater than a set distance value, the point where the distance from the target position is the smallest, an angle between a vector emanating from the target position and pointing at the first pre-aim point and a direction of travel of the target vehicle at the current time is less than ninety degrees, and the second pre-aim point is a point on the target path where the distance from the target position is greater than the set distance value, the point with the minimum distance to the target position, and the included angle between the vector which is emitted from the target position and points to the second preview point and the traveling direction of the target vehicle at the current moment is not less than ninety degrees;

according to a ninth formula, the pre-aiming distance D is determined, and the ninth formula is as follows:

where d0 is the set distance value, v is the traveling speed of the target vehicle, ρ_frontIs of a first curvature, p_nextIs the second curvature.

Optionally, when determining the lateral displacement deviation of the preview position, the processor 202 is specifically configured to:

according to the tenth formula, the transverse displacement deviation y of the preview position is determined_e：

Optionally, the sliding-mode switching surface equation S1 is:

determining the sliding mode control parameter u according to an eleventh formula, wherein the eleventh formula is obtained after S1 is zero, and the eleventh formula is as follows:

wherein g (x) ═ 1/((R + y)_L-Dsinε_L)v_y)，

f(x)＝2(R-Dsinε_L)(a₁₁v_y+a₁₂w)+2D(ρv_xcosβ-w)cosε_L，

R is the turning radius of the target path, D is the pre-aiming distance, y_LFor the current lateral displacement deviation, ε_LIs a lateral angular deviation, v_xIs the longitudinal speed, v, of the target vehicle in the direction of travel of the target vehicle at the current moment in time_yIs the lateral velocity of the target vehicle in the traveling direction perpendicular to the current time of the target vehicle, w is the yaw rate of the target vehicle, and a11, a12 are coefficients determined from the state information of the target vehicle at the current time and the road surface information of the road.

Optionally, the processor 202 is further configured to:

determining a loss function representing an actual loss between a predicted travel path and an actual travel path of the target vehicle after controlling the target vehicle to travel;

and determining an adjusting value of the sliding mode control parameter according to the loss function.

Optionally, when determining the loss function, the processor 202 is specifically configured to:

determining a loss function according to the deviation amount between the predicted lateral displacement deviation and the actual lateral displacement deviation of the target vehicle at least one sampling moment;

the predicted transverse displacement deviation is used for representing the distance between a first predicted position where the target vehicle is located and a third transverse position on the target path at any sampling time, and the third transverse position is the position where the distance between the target path and the first predicted position is the minimum;

the actual lateral displacement deviation is used to indicate the distance between the actual position of the target vehicle and a fourth lateral position on the target path at the sampling time, where the distance between the actual position and the fourth lateral position on the target path is the smallest.

determining a loss function according to the deviation amount between the predicted transverse angle deviation and the actual transverse angle deviation of the target vehicle at least one sampling moment;

the predicted transverse angle deviation is used for representing the angle deviation between the predicted travelling direction of the target vehicle and the tangent direction of the target path at a fifth transverse position at any sampling moment, and the fifth transverse position is the position with the minimum distance between the target path and a second predicted position where the target vehicle is located;

the actual lateral angle deviation is used to indicate an angle deviation between the actual traveling direction of the target vehicle and a tangential direction of the target path at a fifth lateral position at the sampling time, where the fifth lateral position is a position on the target path where a distance from the second predicted position where the target vehicle is located is the smallest.

Optionally, the processor 202 is further configured to:

determining a convolution characteristic diagram C of the target vehicle at the sampling moment according to a twelfth formula_xThe twelfth formula is:

C_x＝f(∑IG_x+b_x)；

from the convolution feature map C_xAnd determining the predicted lateral displacement deviation of the target vehicle at the sampling moment.

Optionally, the processor 202 is further configured to:

C_x＝f(∑IG_x+b_x)；

from the convolution feature map C_xAnd determining the predicted transverse angle deviation of the target vehicle at the sampling moment.

Optionally, the processor 202 is further configured to:

using a gradient descent method, adjusting G_xAnd/or b_x。

Optionally, adjusting G_xThe processor 202 is specifically configured to:

determining the sensitivity Ω according to the thirteenth formula_xThe thirteenth formula is:

Ω_x＝w_x+1(σ'(w_x+1C_x+b_x+1)Ψ_i-1)，

the sensitivity function is:

Δw_x＝-η∑(Ω_xΨ_i+1)，

where η represents the learning rate, Ψ_i+1Down-sampling the input array;

according to G_xWeight value update of (Δ w)_xAdjusting G_xThe value of (a).

optionally, the sliding mode control parameters comprise alpha and beta_sand (q/p), α, β_sAnd (q/p) is a positive odd number, and when the adjustment value of the sliding mode control parameter is determined according to the loss function, the processor 202 is specifically configured to:

determining an adjustment value of the sliding mode control parameter according to a fifteenth formula, wherein the fifteenth formula is as follows:

wherein,

f (Q, k) represents a weight factor of alpha, f (Q, epsilon) represents β_sf (Q, Q/p) represents a weight factor of (Q/p), n represents a variable range of α, and m represents β_sG represents the variable range of q/p, γ_aIs the gain factor, s, z, c₁，c₂respectively, a constant greater than zero, Delta alpha is the adjustment value of alpha, Delta beta_sis beta_sΔ (q/p) is the adjustment value of (q/p).

In the case of dividing the functional modules by corresponding functions, fig. 7 shows a schematic structural diagram of a possible functional module of the control device 102 involved in the above embodiment, where the control device 700 includes: the system comprises a preview processing module 701, a sliding mode control module 702, a neural network learning controller module 703, a vehicle control module 704 and a map matching module 705.

The preview processing module 701 is specifically configured to determine the current lateral displacement deviation y according to the obtained state information of the target vehicle 101 and the road surface information of the road where the target vehicle 101 is located_LTransverse angle deviation epsilon_LThe horizontal displacement deviation y of the pre-aiming distance D and the pre-aiming position_eEqual parameters and variables; the sliding mode control module 702 is specifically configured to vary the lateral displacement y according to the preview location_eEstablishing a sliding mode switching surface equation by using the sliding mode control parameters, and determining the sliding mode control rate; the neural network learning controller module 703 is configured to determine, by using a deep learning method, predicted state information of the target vehicle 101 during traveling (for example, determine a predicted lateral displacement deviation and a predicted lateral angle deviation of the target vehicle 101 at a sampling time), determine a loss function, determine an adjustment value of a slip-mode control parameter according to the loss function, output the adjusted slip-mode control parameter to the slip-mode control module 702, and enable the slip-mode control module 702 to update the slip-mode control parameter; the vehicle control module 704 is specifically configured to control the steering angle of the target vehicle 101 according to the sliding mode control rate output by the sliding mode control module 702; the map matching module 705 is configured to obtain information such as an actual position and a traveling speed of the target vehicle 101 (for example, determine an actual lateral displacement deviation and an actual lateral angle deviation of the target vehicle 101 at a sampling time), and output the information to the neural network learning controller module 703, where the information is used by the neural network learning controller module 703 to determine a loss function.

Based on the same concept as the method embodiment, the embodiment of the present application further provides a computer-readable storage medium, on which some instructions are stored, and when the instructions are called and executed, the instructions may cause the control device to perform the functions related to the control device in any one of the possible designs of the method embodiment and the method embodiment. In the embodiment of the present application, the readable storage medium is not limited, and may be, for example, a RAM (random-access memory), a ROM (read-only memory), and the like.

Based on the same concept as the method embodiments, the embodiments of the present application further provide a computer program product, which, when executed by a computer, can enable a control device to perform the functions related to the control device in any one of the possible designs of the method embodiments and the method embodiments.

Based on the same concept as the method embodiment, the embodiment of the present application further provides a chip, which can be coupled to a transceiver, and is used to implement the functions related to the control device in any one of the possible designs for implementing the method embodiment and the method embodiment described above.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

While some possible embodiments of the present application have been described, additional variations and modifications of these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including the embodiments of the application and all alterations and modifications as fall within the scope of the application.

It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.

Claims

1. A vehicle control method characterized by comprising:

determining a current transverse displacement deviation and a transverse angle deviation according to state information of a target vehicle and road surface information of a road where the target vehicle is located, wherein the current transverse displacement deviation is used for representing a distance between a target position where the target vehicle is located at the current moment and a first transverse position on a target path corresponding to the road, the transverse angle deviation is used for representing an angle deviation between a traveling direction of the target vehicle at the current moment and a tangential direction of the target path at the first transverse position, the first transverse position is a position on the target path where the distance between the target path and the target position is minimum, the transverse angle deviation is smaller than or equal to ninety degrees, and the target path is a path expected to be traveled by the target vehicle;

determining a lateral displacement deviation of a pre-aiming position according to the current lateral displacement deviation, the lateral angle deviation and a pre-aiming distance, wherein the lateral displacement deviation of the pre-aiming position is used for representing the distance between the pre-aiming position and a second lateral position located on the target path when the target vehicle runs to the pre-aiming position, the distance between the projection position of the pre-aiming position on the central axis of the target vehicle and the target position is the pre-aiming distance, an included angle between a vector direction pointing from the target position to the pre-aiming position and the current traveling direction of the target vehicle is less than ninety degrees, and the second lateral position is a position on the target path with the minimum distance to the pre-aiming position;

determining a sliding mode control rate according to the current transverse displacement deviation, the transverse angle deviation, the transverse displacement deviation of the pre-aiming position, a sliding mode switching surface equation and sliding mode control parameters, wherein the sliding mode switching surface equation is used for indicating a sliding mode switching surface of a sliding mode control method, and the sliding mode control rate is used for controlling the target vehicle to run along the target path;

determining a steering angle according to the sliding mode control rate, wherein the steering angle is used for adjusting the running route of the target vehicle to the target route;

and controlling the target vehicle to run according to the steering angle.

2. The method of claim 1, wherein determining a current lateral displacement bias comprises:

determining the current lateral displacement deviation y according to a first formula_LThe first formula is:

wherein,represents a pair y_LDerivative, v_yFor the transverse speed of the target vehicle in the direction of travel perpendicular to the target vehicle's current time, y_LDetermined according to a three-dimensional system state expression of the target vehicle;

determining a lateral angle deviation, comprising:

wherein,represents a pair of ∈_LTaking a derivative, ρ being the curvature of the target path, v_xthe longitudinal speed of the target vehicle in the traveling direction of the target vehicle at the current moment, β is an included angle between the traveling direction of the target vehicle at the current moment and the projection of the traveling direction on the horizontal plane at the current moment, and y_LW is the yaw rate of the target vehicle, and w is determined according to the three-dimensional system state expression of the target vehicle.

3. The method of claim 2, wherein the target vehicle's three-dimensional system state expression is:

wherein u is a steering angle of the target vehicle,represents a pair v_yThe derivation is carried out by the derivation,denotes derivation of w, a₁₁、a₁₂、a₂₁、a₂₂、b₁₁And b₂₁The coefficients are determined according to the state information of the target vehicle at the current moment and the road surface information of the road.

4. The method of claim 3, wherein a is₁₁Determined according to a third expression, the third formula is:

wherein, C_rIs the cornering stiffness of the rear tires of said target vehicle, C_fIs the cornering stiffness of the front tyre of the target vehicle, m is the mass of the target vehicle, v_xthe longitudinal speed of the target vehicle in the traveling direction of the target vehicle at the current moment is taken as beta, and beta is an included angle between the traveling direction of the target vehicle at the current moment and a horizontal plane at the current moment;

a is a₁₂Determined according to a fourth public, the fourth formula is:

a is a₂₁Determined according to a fifth disclosure, the fifth formula is:

wherein J is a moment of inertia of the target vehicle;

a is a₂₂Determined according to a sixth expression, the sixth formula is:

b is₁₁Determined according to a seventh disclosure, the seventh formula is:

b is₂₁Determined according to an eighth disclosure, the eighth formula is:

5. the method of claim 1, wherein the pre-range is determined according to the following method:

determining a first curvature and a second curvature, wherein the first curvature is a curvature of the target path at the first lateral position, a curvature of the target path at a first preaction point, and an average curvature of the target path between the first lateral position and the first preaction point, the second curvature is a curvature of the target path at the first lateral position, a curvature of the target path at a second preaction point, and an average curvature of the target path between the first lateral position and the second preaction point, the first preaction point is a point on the target path at which a distance from the target position is greater than a set distance value, the point at which the distance from the target position is the smallest, and an angle between a vector emanating from the target position and pointing at the first preaction point and a direction of travel of the target vehicle at the current time is less than ninety degrees, the second pre-aiming point is the point with the smallest distance to the target position in the points with the distance to the target position larger than the set distance value on the target path, and the included angle between the vector which is emitted from the target position and points to the second pre-aiming point and the traveling direction of the target vehicle at the current moment is not less than ninety degrees;

determining the pre-aiming distance D according to a ninth formula, wherein the ninth formula is as follows:

6. The method of claim 1, wherein determining a lateral displacement bias for the home position comprises:

according to a tenth formula, determining the transverse displacement deviation y of the preview position_e：

7. The method of claim 1, wherein the sliding-mode switching surface equation S₁Comprises the following steps:

wherein, y_eas a deviation of the lateral displacement of the preview position, alpha, β_sand (q/p) is the sliding mode control parameter, the α, the beta_sAnd said (q/p) is a positive odd number;

determining the sliding mode control parameter according to an eleventh formulaU, the eleventh formula is that S₁Zero, the eleventh formula is:

wherein g (x) ═ 1/((R + y)_L-Dsinε_L)v_y)，

f(x)＝2(R-Dsinε_L)(a₁₁v_y+a₁₂w)+2D(ρv_xcosβ-w)cosε_L，

R is the turning radius of the target path, D is the pre-aiming distance, y_LFor the current lateral displacement deviation, ε_LIs the transverse angular deviation, v_xIs the longitudinal speed, v, of the target vehicle in the direction of travel of the target vehicle at the current moment in time_yIs the lateral velocity of the target vehicle in the direction of travel perpendicular to the target vehicle's current time, w is the target vehicle's yaw rate, a₁₁、a₁₂Is a coefficient determined according to the state information of the target vehicle at the present time and the road surface information of the road.

8. The method of claim 1, wherein after controlling the target vehicle to travel, the method further comprises:

determining a loss function representing an actual loss between a predicted travel path and the actual travel path of the target vehicle;

9. The method of claim 8, wherein determining a loss function comprises:

determining the loss function according to the deviation amount between the predicted lateral displacement deviation and the actual lateral displacement deviation of the target vehicle at least one sampling moment;

the predicted lateral displacement deviation is used for representing the distance between a first predicted position where the target vehicle is located and a third lateral position on the target path at any sampling moment, wherein the third lateral position is the position where the distance between the target path and the first predicted position is the smallest;

the actual lateral displacement deviation is used for representing the distance between the actual position of the target vehicle and a fourth lateral position on the target path at the sampling moment, wherein the fourth lateral position is the position on the target path where the distance between the actual position and the actual position is minimum.

10. The method of claim 8, wherein determining a loss function comprises:

determining the loss function according to the deviation amount between the predicted transverse angle deviation and the actual transverse angle deviation of the target vehicle at least one sampling moment;

the predicted lateral angle deviation is used for representing an angle deviation between a predicted travelling direction of the target vehicle and a tangential direction of the target path at a fifth lateral position at any sampling moment, and the fifth lateral position is a position on the target path where a distance between the target vehicle and a second predicted position where the target vehicle is located is the smallest;

the actual lateral angle deviation is used for representing the angle deviation between the actual traveling direction of the target vehicle and the tangential direction of the target path at a fifth lateral position at the sampling moment, wherein the fifth lateral position is a position on the target path where the distance between the position and the second predicted position where the target vehicle is located is the smallest.

11. The method of claim 9, wherein determining the loss function is preceded by:

determining the convolution characteristic graph C of the target vehicle at the sampling moment according to a twelfth formula_xThe twelfth formula is:

C_x＝f(∑IG_x+b_x)；

wherein I is an input array for representing the state information of the target vehicle and the road surface information of the road where the target vehicle is located at the sampling time, G_xFor trainable convolution kernels, b_xFor trainable biasing, G_xIs a first random value, b_xIs a second random value;

according to the convolution characteristic diagram C_xAnd determining the predicted lateral displacement deviation of the target vehicle at the sampling moment.

12. The method of claim 10, wherein determining the loss function is preceded by:

C_x＝f(∑IG_x+b_x)；

according to the convolution characteristic diagram C_xAnd determining the predicted transverse angle deviation of the target vehicle at the sampling moment.

13. The method of claim 11 or 12, further comprising:

adjusting said G by gradient descent_xAnd/or b_x。

14. The method of claim 13, wherein adjusting the G_xThe method comprises the following steps:

Ω_x＝w_x+1(σ'(w_x+1C_x+b_x+1)Ψ_i-1)；

wherein, w_x+1Represents the value of the multiplicative bias in the convolution kernel, σ' (w)_x+1C_x+b_x+1) Representing the sensitivity function, Ψ_i-1Upsampling the input array;

the sensitivity function is:

Δw_x＝-η∑(Ω_xΨ_i+1)；

where η represents the learning rate, Ψ_i+1Down-sampling of the input array;

according to G_xWeight value update of (Δ w)_xAdjusting G_xThe value of (a).

15. the method of claim 8, wherein the sliding-mode control parameters include α, β_sand (q/p), the alpha, the beta_sAnd the (q/p) is a positive odd number, determining an adjustment value of the sliding mode control parameter, and including:

wherein,

f (Q, k) tableweight factor denoted alpha, f (Q, epsilon) denotes beta_sf (Q, Q/p) represents a weight factor of (Q/p), n represents a variable range of α, and m represents β_sG represents the variable range of q/p, γ_aIs the gain factor, s, z, c₁，c₂respectively, a constant greater than zero, Delta alpha is the adjustment value of alpha, Delta beta_sis beta_sΔ (q/p) is the adjustment value of (q/p).

16. A control apparatus for vehicle control, characterized by comprising a memory and a processor:

the memory for storing computer code or instructions;

the processor is used for calling the computer code or the instructions in the memory and executing the following steps:

determining a current transverse displacement deviation and a transverse angle deviation according to state information of a target vehicle and road surface information of a road where the target vehicle is located, wherein the current transverse displacement deviation is used for representing a distance between a target position where the target vehicle is located at the current moment and a first transverse position on a target path corresponding to the road, the transverse angle deviation is used for representing an angle deviation between a traveling direction of the target vehicle at the current moment and a tangential direction of the target path at the first transverse position, the first transverse position is a position on the target path where the distance between the target path and the target position is the smallest, the transverse angle deviation is smaller than or equal to 90 degrees, and the target path is a path expected to be traveled by the target vehicle;

17. The control device of claim 16, wherein in determining the current lateral displacement bias, the processor is specifically configured to:

determining the current lateral displacement deviation y according to a first formula_LThe second formula is:

18. The control apparatus according to claim 17, wherein the three-dimensional system state expression of the target vehicle is:

19. The control apparatus according to claim 18, wherein a is the same as₁₁Determined according to a third expression, the third formula is:

a is a₁₂Determined according to a fourth public, the fourth formula is:

a is a₂₁Determined according to a fifth disclosure, the fifth formula is:

wherein J is a moment of inertia of the target vehicle;

a is a₂₂Determined according to a sixth expression, the sixth formula is:

b is₁₁Determined according to a seventh disclosure, the seventh formula is:

b is₂₁Determined according to an eighth disclosure, the eighth formula is:

20. the control device of claim 16, wherein in determining the pre-range, the processor is specifically configured to:

21. The control device of claim 16, wherein in determining the lateral displacement deviation of the preview location, the processor is specifically configured to:

22. The control apparatus of claim 16, wherein the sliding-mode shift surface equation S₁Comprises the following steps:

determining the sliding mode control parameter u according to an eleventh formula, wherein the eleventh formula is that the S is₁Zero, the eleventh formula is:

wherein g (x) ═ 1/((R + y)_L-Dsinε_L)v_y)，

f(x)＝2(R-Dsinε_L)(a₁₁v_y+a₁₂w)+2D(ρv_xcosβ-w)cosε_L，

R is the turning radius of the target path, D is the pre-aiming distance, y_LFor the current lateral displacement deviation, ε_LIs the transverse angular deviation, v_xFor the target vehicle to be at the targetLongitudinal speed in the direction of travel of the vehicle at the present moment, v_yIs the lateral velocity of the target vehicle in the direction of travel perpendicular to the target vehicle's current time, w is the target vehicle's yaw rate, a₁₁、a₁₂Is a coefficient determined according to the state information of the target vehicle at the present time and the road surface information of the road.

23. The control device of claim 16, wherein the processor is further configured to:

determining a loss function representing an actual loss between a predicted travel path and the actual travel path of the target vehicle after controlling the target vehicle to travel;

24. The control device of claim 23, wherein in determining the loss function, the processor is specifically configured to:

25. The control device of claim 23, wherein in determining the loss function, the processor is specifically configured to:

26. The control device of claim 24, wherein the processor is further configured to:

C_x＝f(∑IG_x+b_x)；

27. The control device of claim 25, wherein the processor is further configured to:

C_x＝f(∑IG_x+b_x)；

28. The control device of claim 26 or 27, wherein the processor is further configured to:

adjusting said G by gradient descent_xAnd/or b_x。

29. The control apparatus of claim 28, wherein adjusting the G_xThe processor is specifically configured to:

Ω_x＝w_x+1(σ'(w_x+1C_x+b_x+1)Ψ_i-1)；

the sensitivity function is:

Δw_x＝-η∑(Ω_xΨ_i+1)；

where η represents the learning rate, Ψ_i+1Down-sampling of the input array;

according to G_xWeight value update of (Δ w)_xAdjusting G_xThe value of (a).

30. the control apparatus of claim 23, wherein the sliding mode control parameters include α, β_sand (q/p), the alpha, the beta_sAnd the (q/p) is a positive odd number, and when the adjustment value of the sliding mode control parameter is determined according to the loss function, the processor is specifically configured to:

wherein,