CN117227759A

CN117227759A - Automatic driving transverse control method, device, equipment and storage medium

Info

Publication number: CN117227759A
Application number: CN202311353684.9A
Authority: CN
Inventors: 刘世达; 黄威; 吉鸿海; 王力; 刘鹏
Original assignee: North China University of Technology
Current assignee: North China University of Technology
Priority date: 2023-10-19
Filing date: 2023-10-19
Publication date: 2023-12-15

Abstract

The invention provides an automatic driving transverse control method, an automatic driving transverse control device, automatic driving transverse control equipment and a storage medium, wherein the automatic driving transverse control method comprises the following steps: acquiring a transverse corner and an expected corner of a vehicle at the current moment, calculating a transverse corner error at the current moment according to the transverse corner and the expected corner at the current moment, designing an R-MFAILC controller, and constructing a control expression of the R-MFAILC controller; and controlling the transverse rotation angle of the vehicle by using the transverse rotation angle error of the next moment in the last iteration as the input quantity of the R-MFAILC controller and using the steering wheel rotation angle input quantity of the current moment in the current iteration output by the R-MFAILC controller. The R-MFAILC controller is enabled to make corresponding self-adaptive change along with the change of error feedback information, and when the influence of measurable disturbance on the controller is processed, the advantages that no model self-adaptive iterative learning control algorithm does not need an accurate mathematical model of a controlled object and only input and output information are needed can be simultaneously reserved.

Description

Automatic driving transverse control method, device, equipment and storage medium

Technical Field

The invention belongs to the field of automatic control, and particularly relates to an automatic driving transverse control method, an automatic driving transverse control device, automatic driving transverse control equipment and a storage medium.

Background

At present, with the continuous development of social economy, the general improvement of the living standard of people and the rapid growth of motor vehicles, the saturation of urban traffic road networks is continuously increased, the problem of traffic jam is increasingly serious, the great development of public traffic is an important means for relieving the traffic jam, and the application of the automatic driving technology of vehicles and the upgrading of an intelligent system play a vital role in improving the service quality of a public transport system and the running efficiency of the public transport.

The control execution technology is used for controlling the driving behavior of the vehicle according to the result output by the intelligent decision of the vehicle and guaranteeing the safety of the vehicle to reach the process of a preset target point. The control execution technology of the vehicle includes lateral control that mainly controls the steering of the vehicle under the condition of ensuring safety and comfort, and longitudinal control that mainly controls the lateral turning angle (i.e., running turning angle) of the vehicle so that the vehicle can safely reach the destination in accordance with a predetermined trajectory. The coupling relation exists between the transverse movement and the longitudinal movement of the vehicle, and an accurate mathematical model is required by adopting a transverse and longitudinal comprehensive control method, so that the control problem of the unmanned vehicle is complicated.

Under the low-speed driving condition, the controllers are independently designed for transverse control and longitudinal control respectively, and a good tracking effect can be achieved. When a vehicle travels in a specific traveling environment, such as a special lane, it is usually neglected that pedestrians, other vehicles and sudden obstacles cause interference to the normal traveling of the vehicle, and compared with the longitudinal control and other influencing conditions of the vehicle, the transverse corner motion control becomes a primary factor for whether the vehicle can normally run.

In most vehicle operation studies, the establishment of a vehicle lateral dynamics process involving a lateral corner change is typically dependent on time, position and lateral corner. The vehicle is typically a four-wheeled vehicle having two rear wheels and two front wheels, typically the rear wheels being the drive wheels and the front wheels being the steering wheels.

Thus, there is a need for an autopilot lateral control method, apparatus, device and storage medium.

Disclosure of Invention

The invention provides an automatic driving transverse control method, an automatic driving transverse control device, automatic driving transverse control equipment and a storage medium, and solves the problem that the accurate control of transverse rotation angles cannot be realized under the condition that an accurate mathematical model of transverse dynamics of a vehicle is difficult to obtain in the prior art.

The technical scheme of the invention is realized as follows: an autopilot lateral control method, the method comprising the steps of:

acquiring a transverse corner and an expected corner of a vehicle at the current moment, calculating a transverse corner error at the current moment according to the transverse corner and the expected corner at the current moment, designing an R-MFAILC controller, and constructing a control expression of the R-MFAILC controller;

and controlling the transverse rotation angle of the vehicle by using the transverse rotation angle error of the next moment in the last iteration as the input quantity of the R-MFAILC controller and using the steering wheel rotation angle input quantity of the current moment in the current iteration output by the R-MFAILC controller.

The control expression of the R-MFAILC controller is

Wherein u (i, k) is the steering wheel angle input quantity of the vehicle at the current moment in the current iteration, u (i-1, k) is the steering wheel angle input quantity of the vehicle at the current moment in the last iteration,as the estimated value of PG vector, beta _d (k+1) is the desired rotation angle at the next time of the vehicle,β _p (i-1, k+1) is the lateral rotation angle of the vehicle at the next moment in the last iteration, e _p (i-1，k+1)＝β _d (k+1)-β _p (i-1, k+1) is the error of the lateral angle error of the vehicle at the next moment in the last iteration, ρ ₂ ∈(0，1]So that the control algorithm has better adaptability, d (k) is a measurable disturbance that is only related to the change in time k.

As a preferred embodiment, the desired rotation angle β at the next time is _d The specific expression of (k+1) is

Where k is the vehicle running time, the desired angle of rotation β _d (k+1) is a sine function of a constant 0 at 0 to 100 time instants and a sine function at 100 to 600 time instants.

As a preferred embodiment, the specific design steps of the R-MFAILC controller are as follows:

introducing an iteration axis i and defining a nonlinear discrete time dynamics equation of the transverse motion of the vehicle:

where T is the sampling time length, i=1, 2, … represents the number of iterations, k e {0,1, …, T } is the sampling time, β (i, k) is the lateral rotation angle of the vehicle at the current time in the present iteration, β (i, k+1) is the lateral rotation angle of the vehicle at the next time in the present iteration (i.e., the predicted lateral rotation angle), u (i, k) represents the steering wheel rotation angle input amount of the vehicle at the current time in the present iteration, and L represents the sampling time;

As a preferred embodiment, the R-MFAILC controller is controlled by a robust model-free adaptive iterative learning method, and the R-MFAILC controller is learned by constructing a robust model-free adaptive iterative learning method, and the specific expression is:

based on the R-MFAILC controller, by introducing attenuation factorsAnd constructing an R-MFAILC controller by using the piecewise dynamic linearization iterative learning law, so that the R-MFAILC controller can accurately track a required transverse corner curve, thereby improving the control accuracy of the transverse corner of the automatic driving vehicle.

In actual operation, as a preferred embodiment, the specific implementation process of controlling the transverse rotation angle of the vehicle at the current moment in the iteration of the vehicle by using the R-MFAILC controller comprises,

if the transverse measurement rotation angle beta of the next moment in the previous iteration _Measuring (i-1, k+1) and the desired rotation angle beta at the next time _d (k+1) is not equal, transmitting a transverse rotation angle error e (i-1, k+1) of the next moment in the last iteration to an R-MFAILC controller, transmitting a steering wheel rotation angle input quantity u (i-1, k+1) of the next moment in the last iteration output by the R-MFAILC controller to a steering wheel rotation angle execution module, controlling the rotation angle adjustment times of the steering wheel, enabling a steering system of an automatic driving vehicle to obtain electricity to generate a transverse rotation angle output signal beta (i, k+1) of the next moment in the current iteration, and measuring beta through an on-vehicle sensor _Measuring (i, k+1) from the expected rotation angle beta at the next time _q And (k+1) comparing to obtain a transverse rotation angle error e (i, k+1) at the next moment in the iteration, wherein the transverse rotation angle error e is used for the next iteration learning of the R-MFAILC controller so as to realize the adjustment of the transverse rotation angle.

As a preferred embodiment, the specific implementation process of controlling the transverse rotation angle of the next moment in the last iteration of the vehicle by using the R-MFAILC controller further includes:

if beta is _d (k+1)＝β _Measuring (i-1, k+1), sending a normal message to the steering wheel angle execution module;

if beta is _d (k+1)＞β _Measuring (i-1, k+1), sending a steering wheel angle increasing message to the steering wheel angle executing module if the actual angle of rotation is too small;

if beta is _d (k+1)＜β _Measuring (i-1, k+1), the actual steering angle radian is too large, and a steering wheel steering angle reducing message is sent to a steering wheel steering angle executing module.

An autopilot lateral control apparatus, the apparatus comprising:

the acquisition and calculation unit is configured to acquire the transverse rotation angle and the expected rotation angle of the vehicle at the current moment, and calculate the rotation angle error at the current moment according to the transverse rotation angle and the expected rotation angle at the current moment;

the system comprises an S-MFAILC controller, a controller and a controller, wherein the S-MFAILC controller is configured to calculate steering wheel angle input quantity at the next moment in the last iteration according to the transverse angle error at the next moment in the last iteration of the vehicle;

And the control unit is configured to control the transverse rotation angle of the vehicle at the next moment in the current iteration according to the steering wheel rotation angle input quantity of the next moment in the last iteration and the transverse rotation angle error of the next moment in the last iteration, which are output by the R-MFAILC controller.

A computer-readable storage medium having stored thereon a computer program, wherein the program when executed by a processor implements the automatic lateral driving control method according to any one of claims 1 to 7.

An electronic device, the electronic device comprising: a processor and a memory storing a computer program, the processor being configured to perform the autopilot lateral control method of any one of claims 1 to 7 when the computer program is run.

After the technical scheme is adopted, the invention has the beneficial effects that: in the original model-free self-adaptive iterative learning controller (MFAIL)C controller) by introducing an attenuation factor of 1/(k) ² +1) to construct an R-MFAILC controller and by setting an error threshold e _m The sectional attenuation factor control method of the transverse rotation angles u (i, k) is carried out, so that the R-MFAILC controller makes corresponding self-adaptive changes along with the change of error feedback information, when the influence of measurable disturbance on the controller is processed, the model-free self-adaptive iterative learning control algorithm can be kept at the same time, the accurate mathematical model of a controlled object is not needed, only the advantages of inputting and outputting information are needed, the R-MFAILC controller can realize the accurate tracking of a needed transverse rotation angle curve, and the control precision of the transverse rotation angle of the bus is improved.

Drawings

In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.

FIG. 1 is a prior art vehicle transverse dynamics analysis chart;

FIG. 2 is an exemplary flow chart of a method for controlling the lateral direction of bus autopilot in accordance with an embodiment of the present invention;

FIG. 3 is a block diagram of a control system according to an embodiment of the present invention;

FIG. 4 shows different error thresholds e according to an embodiment of the present invention _m The lower R-MFAILC output curve;

FIG. 5 shows different error thresholds e according to an embodiment of the present invention _m The lower R-MFAILC error curve;

FIG. 6 is a graph showing a comparison of the tracking effect of three methods R-MFAILC, MFAILC, PD-ILC in accordance with an embodiment of the present invention;

FIG. 7 is a graph comparing error curves of three methods R-MFAILC, MFAILC, PD-ILC in accordance with an embodiment of the present invention;

FIG. 8 is a graph showing the comparison of the input effects of three methods R-MFAILC, MFAILC, PD-ILC according to an embodiment of the present invention;

FIG. 9 is a graph showing a comparison of curve tracking effects of three methods R-MFAILC, MFAILC, PD-ILC according to an embodiment of the present invention;

FIG. 10 is a graph showing the comparison of the tracking effect of three methods R-MFAILC, MFAILC, PD-ILC in the X and Y dimensions according to an embodiment of the present invention;

FIG. 11 is a graph showing the error contrast of three methods R-MFAILC, MFAILC, PD-ILC under a curved track in accordance with an embodiment of the present invention;

FIG. 12 is a graph showing the comparison of the input effects of three R-MFAILC, MFAILC, PD-ILC methods according to an embodiment of the present invention under a curved track;

FIG. 13 is an external view of a BJ6105EVCA-49-Type automated driving bus according to an embodiment of the present invention;

FIG. 14 is a joint simulation diagram of a Truck-sim and Simulink platform according to an embodiment of the present invention;

FIG. 15 is a diagram of a vehicle traffic information setting interface under the Truck-sim platform according to an embodiment of the present invention;

FIG. 16 is a diagram of a vehicle dynamics parameter information setup interface under a Truck-sim platform according to an embodiment of the present invention;

FIG. 17 is a diagram of a vehicle physical structure information setup interface under the Truck-sim platform according to an embodiment of the present invention;

FIG. 18 is a X, Y-dimensional tracking effect graph of three algorithms R-MFAILC, MFAILC and PD-ILC under a Truck-sim platform according to an embodiment of the present invention;

FIG. 19 is a graph of the lateral position error versus the R-MFAILC, MFAILC and PD-ILC algorithms for a Truck-sim platform in accordance with an embodiment of the present invention;

FIG. 20 is a graph of the transverse angle error versus the R-MFAILC, MFAILC and PD-ILC algorithms for a Truck-sim platform in accordance with an embodiment of the present application;

FIG. 21 is an exemplary block diagram of a bus autopilot lateral control arrangement in accordance with an embodiment of the present application;

fig. 22 is an exemplary structural diagram of an electronic device capable of implementing the method of embodiments of the present application.

Detailed Description

The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.

Examples:

the method adopted by the file is an improved robust model-free self-adaptive iterative learning control method with attenuation factors, and firstly, a data model PG with time-varying parameters is established by adopting an iterative domain dynamic linearization method based on the characteristic of periodic repeated operation of an autonomous vehicle; then, an R-MFAILC controller with an adaptive attenuation factor was designed. One of the advantages of the method is that the controller not only needs to use the input and output data of the controlled system, but also has stronger robustness and the capability of processing nonlinear interference of the measuring system.

Based on the analysis of the background art, the transverse motion process of the unmanned vehicle is expressed by adopting a general nonlinear discrete time dynamics equation:

β(i，k+1)＝f(u(i，k)，β(i，k)，k)

where K e {0,1, …, K } represents a time series, i represents the number of iterations, u (i, K) and β (i, K) represent the steering wheel angle (i.e., the angle input) and the lateral angle (i.e., the angle output) at this time in the iteration of the vehicle, respectively, and f (·) is an unknown nonlinear function.

The control object of the invention is an unmanned vehicle, and compared with a small unmanned vehicle, the invention has the following differences: in the actual running of the small unmanned vehicle, the establishment of the route mainly follows the own or navigation will of the driver, and has larger uncertainty; for unmanned vehicles, due to the social requirement of the vehicles, the vehicles need to run strictly according to a fixed route every day, and the unmanned vehicles have certain characteristic of repeated running, and the difference provides a precondition for the use of the iterative learning control method.

Aiming at the problem that the traditional vehicle transverse dynamics mathematical model cannot realize the accurate control of the transverse rotation angle, the invention designs an R-MFAILC controller (Robust-Model Free Adaptive Iterative Learning Control, robust model-free self-adaptive iterative learning control) so as to realize the accurate tracking control of the transverse rotation angle of the vehicle. The specific design steps of the R-MFAILC controller are as follows:

Step a: description of lateral dynamics: introducing an iteration axis i and defining a nonlinear discrete time dynamics equation of the transverse motion of the vehicle as follows:

where β (i, k) is the lateral rotation angle of the vehicle at the current time in the current iteration, β (i+1, k) is the lateral rotation angle of the vehicle at the current time in the next iteration (i.e., the predicted lateral rotation angle), and k represents the sampling time.

For tracking the track of a bus under a circular route, given the desired track β, zero tracking error is reached as the number of iterations approaches infinity, the control target being to locate the appropriate front wheel steering angle u. To obtain the vehicle control signal, the controller uses the difference between the actual angle and the expected angle.

Step b: full-format dynamic linearization design under iterative domain:

for track tracking under a fixed cycle path with repeated tasks, the control input and tracking error information of the past tracking process are stored in a memory device, and the current control input is corrected to improve the control accuracy of the current track tracking cycle.

In the iterative domain, the dynamic linearization process of a nonlinear system can be summarized as:

lemma 1: assumption 1-4 for a nonlinear system, whenThere is an iteratively dependent Pseudo-Gradient parameter vector Pseudo-Gradient (PG), i.e. >So that the system can be turnedThe data model is dynamically linearized in the following iterative domain full form:

wherein,is a pseudo-gradient parameter vector of this system, define +.>An integer vector for all system output signals comprising all control inputs [ k-L ] _u +1，k]Output [ k-L ] _β +1，k]Related input/output signals, integer L _β ，L _u (0≤L _β ≤k _β ，1≤L _u ≤k _u ) Named control output linearization length constant and control input linearization length constant, and there are:

in this context, L is based on the actual operating requirements of the vehicle _β ＝L _u =1, the above formula can be expressed as follows, specifically developed:

Δβ(i，k+1)＝φ _p ^T (i，k)ΔH _p (i，k)

wherein phi is _p (i，k)＝[φ ₁ (i，k)φ ₂ (i，k)] ^T Phi (i, k) is bounded, deltaH _p (i, k) as a whole as an input signal in the control design.

In the real world, measurement disturbances often affect the output of the system. Thus, the actual output is:

Δβ _p (i，k+1)＝φ _p ^T (i，k)ΔH _p (i，k)+d(k)

then, the process is carried out,

Δβ _p (i，k+1)＝φ ₁ (i，k)Δβ(i，k)+φ ₂ (i，k)Δu(i，k)+d(k)

in phi ₁ (i, k) and φ ₂ (i, k) is a pseudo-gradient vector, and d (k) represents a measurable disturbance of the vehicle system.

R-MFAILC controller design:

for vehicle trajectory tracking under a circular path, a given number of iterations tends to an expected trajectory β with zero tracking error at infinity _d (k+1), the control objective is to find the appropriate control input. Will be the desired body angle beta _d (k+1) and the actual body angle beta _p The difference of (i, k+1) is used as an input to the controller to obtain a vehicle control signal.

The tracking error is defined as follows:

e _p (i，k+1)＝β _d (k+1)-β _p (i，k+1)

the control input criteria function is:

J(u(i，k))＝|e _p (i，k+1)| ² +λ|u(i，k)-u(i-1，k)| ²

substituting the linearization model into a criterion function, using the derivative of the control quantity to make the linearization model equal to zero, and introducing an attenuation factor of 1/(k) ² +1), a linearization dynamic linearization iterative control law with an attenuation factor can be obtained:

wherein the step size factor ρ ₁ ，ρ ₂ ∈(0，1]So that the control algorithm has greater flexibility.

And (2) injection: compared with the traditional FFDL-MFAC algorithm, an attenuation factor of 1/(k) is introduced into the control algorithm equation ² +1) the attenuation factor has the effect of outputting an error e with increasing time k _p (i, k) has less and less impact on control volume update. Since the measured perturbation of the system is due to the output error e _p (i, k) the influence of the measurement disturbance will gradually decrease with increasing time k.

In practice, from Δβ (i, k+1))＝φ _p ^T (i，k)ΔH _p (i, k) we can know when

The measurement error signal causes the measurement disturbance to be equal to zero, and the disturbance is then completely suppressed. However, the algorithm may stop because the correction term becomes zero, and it will then not react to changes in the desired output, system parameters, and structure. To overcome this disadvantage, an R-MFAILC improvement algorithm based on fading gain is proposed herein:

Wherein the error threshold is e _m ，k _p Is e _p (i, k) from |e _p (i，k)|≥e _m To |e _p (i，k)|＜e _m Is a time of (a) to be used.

And (3) injection: the control algorithm is divided into two steps according to the actual output of the system often affected by the measured disturbance. The stability of the two-stage algorithm makes it robust. In the R-MFAILC algorithm, which does not take the decay factor as the first stage, the current control input is updated by the previous control input and current measurement error. According to the analysis of the third chapter, the controlled system can be stabilized because of the measurement disturbance, the tracking error is not equal to zero at k→infinity. In the second phase, the R-MFAILC algorithm has an attenuation factor when the measurement error is less than e _m When it is put in place to stop measuring disturbances. When stage 2 suddenly increases, if e _m The algorithm equation may again shift to phase 1 and loop the adjustment process. It is therefore able to sense and adjust for system structure, parameters and output variations.

The PG vector in the control law is an unknown parameter, which needs to be estimated since it is unknown. An improved projection algorithm is used herein to estimate it. First, an index function is defined as follows:

J(φ(i，k))＝|Δβ _p (i-1，k+1)-d(k-1)-φ _p ^T (i，k)ΔH _p (i-1，k)| ² +μ||φ _p (i，k)-φ _p (i-1，k)|| ²

for phi in defining tracking error _p (i, k) taking extremum, and using the 'inverse matrix' lemma, the estimated value of the PG vector is:

wherein mu E (0, 2)]Is a step factor which increases the flexibility of the control algorithmIs phi _p An estimate of (i, k).

The control law shows that the parameter phi _p (i, k) plays an important role in the system, while the parameter phi _p (i, k) is a time-varying parameter vector. To enhance the stability and time-varying parameter tracking capabilities of the system we provide a parameter reset algorithm if

And combining the R-MFAILC improved algorithm of the attenuation gain to obtain a PG estimation algorithm equation and a reset algorithm. The robust model-free adaptive iterative learning control scheme of the vehicle track tracking control system is as follows:

the specific implementation process for controlling the transverse rotation angle of the vehicle at the current moment in the current iteration by using the R-MFAILC controller comprises the following steps:

if the transverse measured rotation angle at the next moment in the previous iteration is not equal to the expected rotation angle at the next moment, transmitting a transverse rotation angle error at the next moment in the previous iteration to an R-MFAILC controller, transmitting the steering wheel rotation angle input quantity at the next moment in the previous iteration output by the R-MFAILC controller to a steering wheel rotation angle execution module, controlling the rotation angle adjustment times of the steering wheel, enabling a steering system of an automatic driving vehicle to obtain electricity to generate a transverse rotation angle output signal at the next moment in the current iteration, and comparing the transverse rotation angle error at the next moment in the current iteration obtained through measurement of a vehicle-mounted sensor with the expected rotation angle at the next moment to obtain the transverse rotation angle error at the next moment in the current iteration for the next iteration learning of the R-MFAILC controller so as to realize adjustment of the transverse rotation angle.

The specific implementation process for controlling the transverse rotation angle of the vehicle at the current moment in the current iteration by using the R-MFAILC controller further comprises the following steps:

if beta is _d (k+1)＝β _Measuring (i-1, k+1) are equal, then send the normal message to the steering wheel angle execution module;

In a second aspect, an embodiment of the present invention provides a vehicle autopilot lateral control apparatus, the apparatus comprising:

an R-MFAILC controller configured to calculate a vehicle corner input u (i, k) at a next time in a previous iteration of the vehicle based on a lateral corner error at the next time in the previous iteration of the vehicle, the specific control expression being

Where u (i, k) is the input amount of the transverse rotation angle of the vehicle at the current moment in the current iteration, u (i-1, k) is the input amount of the transverse rotation angle of the vehicle at the current moment in the previous iteration, e _p (i-1, k+1) is the lateral angle error, Δβ, of the vehicle at the next time in the previous iteration _p (i, k) is the difference of the transverse rotation angles at the current moment in the iteration, e _p (i-1，k+1)＝β _d (k+1)-β _p (i-1，k+1)，β _d (k) Beta is the desired angle of rotation at the current moment of the vehicle _p (i-1, k+1) is the lateral rotation angle of the vehicle at the next moment in the last iteration, e _m Is an error threshold;

based on the designed R-MFAILC controller, the embodiment of the invention provides a vehicle automatic driving transverse control method, as shown in FIG. 2, which comprises the following steps:

step 1100: obtaining the transverse rotation angle measurement value beta of the next moment in the last iteration of the vehicle _Measuring (i-1, k+1) and the desired rotation angle beta _d (k+1) thereby calculating the lateral rotation angle error e at the next time _p (i-1，k+1)；

Step 1200: designing an R-MFAILC controller, wherein the R-MFAILC controller is shown as a formula (47);

step 1300: and using the transverse angle error at the next moment in the previous iteration as the input quantity of the R-MFAILC controller, and controlling the transverse angle at the next moment in the current iteration of the vehicle by using the steering wheel angle input quantity at the next moment in the current iteration output by the R-MFAILC controller.

In step 1300, if the lateral rotation angle measurement β at the next time in the previous iteration _Measuring (i-1, k+1) and the desired rotation angle beta at the next time _d (k+1) are not equal, the lateral rotation angle error e at the next moment in the previous iteration _p (i-1, k+1) is transmitted to the R-MFAILC controller, the steering wheel angle input quantity u (i, k) of the next moment in the iteration output by the R-MFAILC controller is transmitted to the steering wheel angle execution module, the angle adjustment times of the vehicle body angle module are controlled, the engine is electrified to generate an actual transverse angle output signal beta (i, k+1), and the angle measurement value beta is transmitted to the steering wheel angle execution module _Measuring (i, k+1) from the desired rotation angle beta _d (k+1) comparing to obtain the transverse angle error e of the next moment in the iteration _p (i, k+1) for the next iteration of learning of the R-MFAILC controller to achieve adjustment of the lateral rotation angle.

Since in actual operation the lateral angle value β (i, k+1) transmitted via the engine is to be measured by the vehicle-mounted sensor to form β _Measuring (i, k+1) can be equal to the desired rotation angle beta _d (k+1) comparing to form a transverse angle error value e _p (i, k+1), so e here _p (i，k+1)＝β _d (k+1)-β _Measuring (i, k+1), which is e in the design of the controller _p (i，k+1)＝β _d (k+1)-β _p (i, k+1) is different, and β (i, k+1) is a theoretical lateral rotation angle value of the R-MFAILC controller, which is not present in practical application.

if beta is _d (k+1)＜β _Measuring (i-1, k+1), the actual turning angle radian is too large, and the actual turning angle radian is sent to the steering wheel turning angle execution moduleAnd reducing the steering wheel angle message.

As shown in fig. 3, the R-MFAILC controller, the steering wheel angle execution module, the steering wheel angle control module, and the steering system (including the engine) of the automatic driving vehicle are sequentially connected, the steering wheel angle input information u (i, k) generated by the R-MFAILC controller is used for starting the steering wheel angle control module to control the number of times of transverse angle adjustment of the steering system of the automatic driving vehicle, and the generator is enabled to generate an actual angle output signal beta (i, k+1) to drive the expected angle beta _d (k+1) and the rotation angle measurement value beta _Measuring (i, k+1) to obtain a transverse angle error e at the current moment in the iteration _p (i, k+1) for the next iteration of learning of the R-MFAILC controller to achieve adjustment of the lateral rotation angle.

In the actual running process of the vehicle, the R-MFAILC controller designed by the invention takes the steering wheel angle information of the vehicle as a control input and is equivalent to an actual steering wheel of the vehicle, takes the transverse steering angle of the vehicle as an output, compares the actual steering angle value generated by each iteration of the vehicle with an expected steering angle value, generates transverse steering angle error information which is used in the next iteration learning process, and the attenuation factor 1/(k) in the R-MFAILC controller ² +1) output error e with increasing time k _p (i, k+1) has a smaller and smaller effect in the control quantity update, since the measured perturbation of the system is due to the output error e _p (i, k+1) introducing that the influence of the measured perturbation gradually weakens with the increase of the output error, for realizing the accurate tracking of the required transverse angle curve; the R-MFAILC controller, the steering wheel angle execution module, the control car body angle module and the steering system (comprising an engine) of the automatic driving vehicle are sequentially connected, steering wheel angle input information u (i, k) generated by the R-MFAILC controller is started to the steering wheel angle execution module to control the car body angle module, so as to control the transverse angle adjustment times of the steering system of the automatic driving vehicle, and the generator is enabled to generate an actual angle output signal beta (i, k+1) electrically, so that the expected angle beta is obtained _d (k+1) and the rotation angle measurement value beta _Measuring (i, k+1) to obtain the current timeThe transverse rotation angle error e (i, k+1) at the current moment in iteration is used for the next iteration learning of the R-MFAILC controller so as to realize the adjustment of the transverse rotation angle.

In the actual running process, the R-MFAILC controller designed by the invention takes steering wheel angle information of a vehicle as control input and is equivalent to an actual vehicle steering wheel, and takes a transverse steering angle of a bus as output, the R-MFAILC controller compares an actual steering angle value generated by each iteration of the bus with an expected steering angle value, the generated transverse steering angle error information is used in the next iteration learning process, attenuation factors in the R-MFAILC controller are increased along with time, the effect of output errors in control quantity updating is smaller and smaller, and because the measurement perturbation of the system is introduced by the output errors, the influence of the measurement perturbation is gradually weakened along with the increase of the output errors, so that the accurate tracking of a required transverse steering angle curve is realized; the steering system (comprising an engine) of the automatic driving vehicle is sequentially connected with the R-MFAILC controller, the steering wheel angle execution module, the vehicle body angle control module and the steering system (comprising the engine) of the automatic driving vehicle, steering wheel angle input information generated by the R-MFAILC controller is started to control the vehicle body angle control module for controlling the transverse angle adjustment times of the steering system of the automatic driving vehicle, the generator is enabled to generate an actual angle output signal, an expected angle is compared with an angle measured value, and the transverse angle error at the current moment under the iteration is obtained for the next iteration learning of the R-MFAILC controller so as to realize the adjustment of the transverse angle.

In practice, a bus generally travels along a road in the form of a curve when changing directions according to the characteristics of road construction. To verify the validity of the control method herein. This section is divided into two parts.

In the first part, we use a nonlinear non-minimum phase system to verify the validity of the element attenuation factor in the R-MFAILC algorithm presented herein. The second section uses the proposed lateral dynamic description system to verify the applicability of the R-MFAILC algorithm. The third section verifies the effectiveness of the proposed R-MFAILC method in an actual curvilinear driving trajectory.

Two iterative learning algorithms are used herein for comparison experiments.

The traditional MFAILC control algorithm is:

/>

in addition, the PD-type iterative learning control (PD-ILC) algorithm is:

u(i，k)＝u(i-1，k)+α _d Δe _p (i，k-1)+α _p e _p (i-1，k)

wherein alpha is _d The gain coefficient as a differential term is differentiated from the previous error by Δe _p (i, k-1), and e _p (i，k-1)＝β _d (k-1)-β _p (i，k-1)。α _p Is the gain coefficient of the proportional term. The parameter settings for both comparison algorithms are given in table 1.

Table 1: parameter setting of two control methods

Algorithm	Parameter setting
		MFAILC	λ＝1，μ＝2，ρ＝0.2，η＝1，φ(i，1)＝0
PD-ILC	α _d ＝0.6，α _p ＝0.6

In order to verify the effectiveness of the method provided by the invention, the method of the invention is subjected to numerical simulation verification and semi-physical simulation verification and comprises the following steps of

a) Numerical simulation verification

(1) Simulation analysis of R-MFAILC algorithm

In this section, the control effect of the R-MFAILC algorithm will be examined. An initial value (x (0, k), y (0, k)) = (0, 0) is set and a non-repetitive disturbance d (k) = 0.04 sin (k/200+ (kpi)/100) +2 is added. The total number of simulation iterations is 100. The controller parameters are set as follows: λ=1, μ=2, ρ=0.2, η=1. The effect of the attenuation factor on the control is evident in the R-MFAILC algorithm. Therefore, the scheme focuses on the influence of different attenuation factors on the R-MFAILC control algorithm. The effect is shown in fig. 4 and 5.

The desired trajectory can be described as:

β _d (k)＝0.5×(-1) ^round(k/100)

the system for which the simulation is directed can be described as follows:

error threshold e _m The effect on control is also evident in the R-MFAILC algorithm. Therefore, the scheme focuses on the influence of different parameters on the R-MFAILC control algorithm. The results of the R-MFAILC simulation for different error thresholds are shown in fig. 4 and 5 below. The tracking errors for the different cases are shown in table 2.

Different error threshold e _m The R-MFAILC output curve of (c) is shown in fig. 4. And output curve e _m When=0.9, e _m Curve e compared to when=1.1 _m Tracking was better when=1. For example, when k=300s, e _m =0.9 and e _m The overshoot values of the curves corresponding to =1.1 are 0.4rad and 0.3rad, respectively. The origin of this phenomenonThe method comprises the following steps: if the selection value is greater than 1, there will be an output static error in the system. If the selected value is less than 1, the purpose of suppressing the measurement disturbance cannot be achieved. Thus, for the system, e _m =1 has a good angle tracking performance.

The R-MFAILC error curves at different angular error thresholds are shown in fig. 5, the error evaluation index is defined as the mean absolute error (Mean Absolute Error, MAE), and the MAE function can be expressed and used by the following formula:

in e _p (i, k) represents an absolute error obtained by the number of iterations, and n is the maximum number of iterations.

From fig. 5 different error threshold e _m Although e is seen from the R-MFAILC error curve _m =1 initial error convergence speed is not as good as e _m =1.1 (0 to 20 s), but the initial error value is significantly smaller than (e _m Maximum error of about 0.6 when=1.1, e _m Maximum error at=1 is about 0.5). On the other hand, due to the error threshold e _m The selected value of (2) is smaller than 1, and the purpose of suppressing measurement disturbance cannot be achieved. It can be observed that when the number of iterations is greater than 8, the error curve fluctuates greatly, and normal operation of the system is difficult to meet. Again verify e _m Validity of =1.

Table 2 shows the calculated values of the error threshold at different iteration times: e according to the error statistics in Table 2 _m The error value at=1 is smaller than the other two values, whether the error convergence speed (in the tenth iteration, e _m MAE error at=0.9 is 0.274, e _m MAE error at=1 is 0.076) and the final error stable value (at iteration 100, e _m MAE error when=0.9 is 0.183,2.97 ×10 ^-5 The error value at the time was 2.97X10 ^-5 ，e _m The error value when=1 is 1.55×10 ^-15 ). It is easy to find e _m Is critical to the operation of the algorithm, if the selection value is large, there will be output static differences in the system. If the selected value is small, there is noThe method achieves the purpose of suppressing measurement disturbance. Therefore, the error threshold e is reasonably selected for different control systems _m Is very important.

Table 2 different values e _m R-MFAILC algorithm error table of (C)

It can also be seen from the error analysis of fig. 2-3 and table 2 that the tracking effect is different for the three different values. e, e _m The error threshold of=1 is verified to have significant advantages. Thus, e is employed in subsequent experimental simulations _m ＝1。

Simulation analysis of R-MFAILC and different control algorithms

To verify the effectiveness of the proposed R-MFAILC method, the R-MFAILC algorithm is compared herein with both the PD-ILC and the conventional MFAILC existing ILC algorithms. The control targets and controller parameter settings of this section are the same as those of section a), and the corresponding tracking performance curves and error tracking curves are shown in fig. 6 and 7.

The R-MFAILC tracking effect is shown in FIG. 6, compared to the MFAILC and PD-ILC algorithms. Compared with the other two ILC algorithms, the R-MFAILC algorithm has better tracking effect, and is particularly characterized in that the MFAILC algorithm oscillates up and down on a tracking curve in about 450 seconds, and the PD-ILC algorithm cannot be completely combined with a required angle after overshooting at the initial moment. Therefore, the self-adaption capability and the anti-interference capability of the piecewise attenuation factors are obviously superior to those of the comparison algorithm.

The error effect of R-MFAILC is shown in FIG. 7, and the error evaluation index is defined as the mean absolute error (Mean Absolute Error, MAE) compared to the MFAILC and PD-ILC algorithms. Compared with the other two ILC algorithms, the R-MFAILC algorithm shows a faster error convergence speed and a smaller initial error value, and particularly shows that the final error value is difficult to stabilize after 100 iterations of the MFAILC algorithm, and the error convergence speed of the MFAILC algorithm is obviously behind the R-MFAILC algorithm. On the other hand, the PD-ILC algorithm, although having a faster error convergence rate than the algorithm presented herein, has an excessive initial error value. And the self-adaptability and the anti-interference capability of two evaluation indexes of the comprehensive error convergence rate and the initial error value of the R-MFAILC algorithm are further verified.

Table 3 gives the error calculation of the error effect for three methods (R-MFAILC, MFAILC and PD-ILC algorithm) at different iterations: according to the error statistics, in Table 3, the error value of R-MFAILC is smaller than the other two methods, regardless of the error convergence speed (at iteration 6, the MAE error of R-MFAILC is 0.268, the MAE error of PD-ILC is 0.355, the MAE error of MFAILC is 0.489) and the initial error stability value (at iteration 3, the MAE error of R-MFAILC is 0.513, the MAE error of PD-ILC is 0.518, and the MAE error of MFAILC is 0.518). The R-MFAILC algorithm has good anti-interference effect and self-adaptation capability, which are easily found from two evaluation indexes of error convergence rate and initial error value.

TABLE 3 Algorithm tracking error table for different body position angles

The controller input pair for the R-MFAILC algorithm and the other two algorithms at different times is shown in FIG. 8. It can be seen that the MFAILC algorithm exhibits different degrees of overshoot at the moment of angle change (t=50s, 150s, 460 s,4500 s), whereas the PD-ILC algorithm has difficulty tracking the desired angle. The control inputs of the two ILC algorithms described above are less effective than the R-MFAILC algorithm. Therefore, based on the simulation results, the R-MFAILC algorithm has the advantage of precisely controlling the input angle. After the R-MFAILC algorithm is adopted, the overshoot of the angle of the vehicle body is obviously reduced.

From the error analysis of fig. 6-8 and table 3, it can also be seen that the tracking effect of the three methods is different, and the adaptive capacity and anti-interference capacity of the R-MFAILC algorithm are further verified compared to the conventional ILC algorithm.

Curve track following

To verify the effectiveness of the proposed R-MFAILC method in an actual curve driving trajectory, the R-MFAILC algorithm is compared herein with both the existing PD-ILC and the conventional MFAILC ILC algorithm. The control object of the section adopts a transverse model, and the concrete form is as follows:

where t=0.01 s, l=20m, and furthermore, in an ideal case, the constant tracking speed is 10 km/h. In the case of curve tracking, the desired body angle of a bus can be described as:

in a simulation environment, the initial state is set to β in each iteration _p (0, k) =0.2×sin (k/20); a non-repetitive perturbation d (k) =0.04 sin (k/200+ (kpi)/100) +2 is added; the initial parking position of the bus is the same as the desired track, and is set to (x (0, k), y (0, k))= (0, 0), and the control input of the first iteration is u (i, 0) =0. The corresponding tracking performance curves and error tracking curves are shown in fig. 9 and 10.

The R-MFAILC tracking effect curves are shown in fig. 9, compared to the MFAILC and PD-ILC algorithms. Compared with other two ILC algorithms, the R-MFAILC algorithm has better tracking effect, and is specifically characterized in that the MFAILC algorithm oscillates up and down on a tracking curve after 400 seconds, and the MFAILC algorithm cannot completely combine the required angle after overshooting for 100-400 seconds; also, the PD-ILC algorithm cannot fully combine the required angle after 100 seconds of overshoot, and the tracking curve oscillates up and down in about 400-550 seconds. Therefore, the self-adaption capability of the piecewise attenuation factors and the anti-interference capability of the curve paths are obviously superior to those of the comparison algorithm.

FIG. 10 better illustrates the curve tracking effect of the R-MFAILC algorithm in the X and Y dimensions. It can be seen that the PD-ILC algorithm has difficulty tracking the desired trajectory after 100 meters in the X dimension, which in reality corresponds to the difficulty steering the vehicle through a suitable angle, thereby deviating from the desired path. Whereas the MFAILC algorithm tracks better before 400m in the X dimension, and tracks worse after 400 m. Therefore, based on the simulation results, the curve path tracking performance of the R-MFAILC algorithm is verified.

Error effects of the R-MFAILC algorithm on the curve path versus MFAILC and PD-ILC algorithms as shown in fig. 11, the error evaluation index is defined as the mean absolute error (Mean Absolute Error, MAE). Compared with the other two ILC algorithms, the error convergence speed of the R-MFAILC algorithm is faster, the initial error value is smaller, the error curve of the final error value PD-ILC algorithm of the PD-ILC algorithm is unstable between the 3 rd iteration and the 20 th iteration after 100 iterations of the MFAILC algorithm, and the initial error value is much larger than that of the R-MFAILC algorithm. On the other hand, although the initial error value of the MFAILC algorithm is smaller than the algorithm presented herein, the error convergence rate of the MFAILC algorithm is significantly lower than the R-MFAILC algorithm after about 10 th iteration. And the adaptability and the anti-interference capability of two evaluation indexes of the comprehensive error convergence rate and the initial error value of the R-MFAILC algorithm on a curve path are further verified.

The error calculation of the error effect curve path for the three methods (R-MFAILC, MFAILC and PD-ILC algorithms) at different iterations is given in table 4: as can be seen from the error statistics of Table 4, the error value of R-MFAILC is smaller than the other two methods at the error convergence rate (0.232 for the R-MFAILC, 0.961 for the PD-ILC, 0.137 for the MFAILC) and the initial error stability value (0.300 for the R-MFAILC algorithm and 0.193 for the MFAILC algorithm) at the 6 th iteration. The R-MFAILC algorithm has better anti-interference effect and curve path self-adaptive capacity, which are easily found from two evaluation indexes of error convergence rate and initial error value.

Table 4 algorithm tracking error table under curve track

The R-MFAILC algorithm is shown in FIG. 12 with the other two algorithms at different time pairs of controller inputs. It can be seen that MFAILC and PD-ILC exhibit different levels of overshoot at the instants of angular change (100 seconds and 300 seconds). Compared with the R-MFAILC algorithm, the two ILC algorithms have poorer control input effects, can cause the vehicle to shake in actual running, and seriously affect the running stability and safety. Therefore, based on the simulation result, the excellent effect of the R-MFAILC algorithm in accurately controlling the input angle is reflected, and the body angle overshoot of the R-MFAILC algorithm is obviously smaller than that of the R-MFAILC algorithm.

From the error analysis of fig. 10-12 and table 4, it can also be seen that the tracking effect of the three methods is different, and the self-adaptation capability and anti-interference capability of the R-MFAILC algorithm on the curve path are further verified compared with the conventional ILC algorithm.

Semi-physical simulation verification

In order to further verify the practicability of the proposed method, a semi-physical simulation experiment was performed in this chapter. The simulation environment is set as a conventional two-lane road with a certain curvature. And a simulation model is established by taking a BJ6105EVCA-49 bus manufactured by North automobile Futian automobile Co., ltd as an example. Further, fig. 13 is an external view of a vehicle BJ6105EVCA-49 type.

The corresponding specific vehicle configuration is shown in table 5.

Table 5 BJ6105EVCA-49 bus related data

The simulation platform is a joint simulation platform based on Truck-Sim and MATLAB Simulink, and as shown in FIG. 14, the joint simulation model of Truck-Sim and Simulink is shown.

According to Table 5, the vehicle path and parameter settings are shown in FIGS. 15-17. In addition, in the joint simulation, external interference is added in the running process of the vehicle, and the method specifically comprises the following steps:

d(k)＝0.04*sin(k/200+(kπ)/100)+2

it should be noted that the relevant parameter settings in Simulink are the same as in section a).

Fig. 18-20 are simulation results of lateral trajectory tracking after 100 iterations in three control modes. For the R-MFAILC control method, the lateral position of the vehicle has a certain deviation in the range of 17s and 47 s. Within the curve range of 3-22s, there is a certain deviation in the yaw angle of the aircraft. The maximum absolute value of the lateral position error is not more than 0.117m, and the maximum absolute yaw angle error is not more than 9.632 degrees. The maximum lateral error and turning error of the PD-ILC algorithm are 0.166m and 10.285 °, respectively, and the maximum lateral error and turning error of the MFAILC algorithm are 0.132m and 21.771 °, respectively. The R-MFAILC is better able to meet the requirements of track following than the PD-ILC and MFAILC, which can also be concluded from the position error and the angular error of fig. 19 and 20.

To demonstrate the control effect of the controller in more detail, table 6 gives the lateral motion control error, in which RMSE error evaluation index is as follows:

wherein e _p (i, k) represents an absolute error obtained by the number of iterations, and n is the number of maximum iterations. The MAE error evaluation index is the same as section a).

As is evident from table 6, the control algorithm presented herein is able to track lateral position more effectively among the results of the three control methods.

TABLE 6 motion control error

Fig. 21 shows a block diagram of a transverse control device for automatic driving of a bus according to an embodiment of the present invention.

As shown in fig. 21, an embodiment of the present invention provides a bus autopilot lateral control apparatus 2000 including an acquisition and calculation unit 2100, an R-MFAILC controller 2200, and a control unit 2300.

An acquiring and calculating unit 2100 configured to acquire a lateral rotation angle and an expected rotation angle of the vehicle at a next time in the current iteration, and calculate a rotation angle error of the vehicle at the next time in the current iteration;

the R-MFAILC controller 2200 is configured to calculate the steering wheel angle input quantity at the next moment in the iteration according to the angle error at the next moment in the iteration of the vehicle, and the specific control expression is shown in the formula (24);

And the control unit 2300 is configured to control the transverse rotation angle of the next time in the next iteration of the vehicle according to the steering wheel rotation angle input quantity of the next time in the current iteration and the transverse rotation angle error of the next time in the current iteration, which are output by the R-MFAILC controller.

In the automatic driving transverse control device of the bus, which is used in the embodiment of the invention, in the actual running process of the bus, the R-MFAILC controller designed by the invention takes the steering wheel angle information of the vehicle as control input, takes the equivalent of the actual steering wheel of the vehicle, takes the transverse steering angle of the bus as output, and the R-MFAILC controller compares the actual steering angle value generated by each iteration of the bus with the expected steering angle value, so that the generated transverse steering angle error information is used in the next iteration learning process, and the attenuation factor 1/(k) in the R-MFAILC controller ² +1) output error e with increasing time k _p (i, k+1) has a smaller and smaller effect in the control quantity update, since the measured perturbation of the system is due to the output error e _p (i, k+1) introducing that the influence of the measured perturbation gradually weakens with the increase of the output error, for realizing the accurate tracking of the required transverse angle curve; the R-MFAILC controller, the steering wheel angle execution module, the control car body angle module and the steering system (comprising an engine) of the automatic driving vehicle are sequentially connected, steering wheel angle input information u (i, k) generated by the R-MFAILC controller is started to the steering wheel angle execution module to control the car body angle module, so as to control the transverse angle adjustment times of the steering system of the automatic driving vehicle, and the generator is enabled to generate an actual angle output signal beta (i, k+1) electrically, so that the expected angle beta is obtained _d (i, k+1) and the rotation angle measurement value beta _Measuring (i, k+1) to obtain a transverse angle error e at the current moment in the iteration _p (i, k+1) for the next iteration of learning of the R-MFAILC controller to achieve adjustment of the lateral rotation angle.

In some embodiments, the device for controlling the lateral direction of bus autopilot may incorporate the method features of the method for controlling the lateral direction of bus autopilot of any embodiment, and vice versa, which is not described herein.

In an embodiment of the present invention, there is provided an electronic device including: a processor and a memory storing a computer program, the processor being configured to perform any of the bus autopilot lateral control methods of the embodiments of the present invention when the computer program is run.

Fig. 22 shows a schematic diagram of an electronic device 3000 in which embodiments of the present invention may be implemented or implemented, and in some embodiments may include more or fewer electronic devices than shown. In some embodiments, it may be implemented with a single or multiple electronic devices. In some embodiments, implementation may be with cloud or distributed electronic devices.

As shown in fig. 22, the electronic device 3000 includes a processor 3001 that can perform various appropriate operations and processes in accordance with programs and/or data stored in a Read Only Memory (ROM) 3002 or programs and/or data loaded from the storage portion 1008 into a Random Access Memory (RAM) 3003. Processor 3001 may be a multi-core processor or may include a plurality of processors. In some embodiments, the processor 3001 may comprise a general-purpose main processor and one or more special coprocessors such as, for example, a Central Processing Unit (CPU), a Graphics Processor (GPU), a neural Network Processor (NPU), a Digital Signal Processor (DSP), and so forth. In the RAM 3003, various programs and data necessary for the operation of the electronic device 3000 are also stored. The processor 3001, the ROM 3002, and the RAM 3003 are connected to each other by a bus 3004. An input/output (I/O) interface 3005 is also connected to bus 3004.

The above-described processor is used in combination with a memory to execute a program stored in the memory, which when executed by a computer is capable of implementing the methods, steps or functions described in the above-described embodiments.

The following components are connected to the I/O interface 3005: an input portion 3006 including a keyboard, a mouse, a touch screen, and the like; an output portion 3007 including a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, a speaker, and the like; a storage portion 3008 including a hard disk or the like; and a communication section 3009 including a network interface card such as a LAN card, a modem, or the like. The communication section 3009 performs communication processing via a network such as the internet. The drive 3010 is also connected to the I/O interface 3005 as needed. A removable medium 3011, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like, is installed on the drive 3010 as needed, so that a computer program read out therefrom is installed into the storage section 3008 as needed. Only some of the components are schematically illustrated in fig. 10, which does not mean that the computer system 3000 includes only the components shown in fig. 22.

The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer or its associated components. The computer may be, for example, a mobile terminal, a smart phone, a personal computer, a laptop computer, a car-mounted human-computer interaction device, a personal digital assistant, a media player, a navigation device, a game console, a tablet, a wearable device, a smart television, an internet of things system, a smart home, an industrial computer, a server, or a combination thereof.

Although not shown, in an embodiment of the present invention, there is provided a storage medium storing a computer program configured to, when executed, perform any of the file difference-based compiling methods of the embodiment of the present invention.

Storage media in embodiments of the invention include both permanent and non-permanent, removable and non-removable items that may be used to implement information storage by any method or technology. Examples of storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, read only compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium which can be used to store information that can be accessed by a computing device.

Methods, programs, systems, apparatus, etc. in accordance with embodiments of the invention may be implemented or realized in single or multiple networked computers, or in distributed computing environments. In the present description embodiments, tasks may be performed by remote processing devices that are linked through a communications network in such a distributed computing environment.

It will be appreciated by those skilled in the art that embodiments of the present description may be provided as a method, system, or computer program product. Thus, it will be apparent to those skilled in the art that the functional modules/units or controllers and associated method steps set forth in the above embodiments may be implemented in software, hardware, and a combination of software/hardware.

The acts of the methods, procedures, or steps described in accordance with the embodiments of the present invention do not have to be performed in a specific order and still achieve desirable results unless explicitly stated. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Various embodiments of the invention are described herein, but for brevity, description of each embodiment is not exhaustive and features or parts of the same or similar between each embodiment may be omitted. Herein, "one embodiment," "some embodiments," "example," "specific example," or "some examples" means that it is applicable to at least one embodiment or example, but not all embodiments, according to the present invention. The above terms are not necessarily meant to refer to the same embodiment or example. Those skilled in the art may combine and combine the features of the different embodiments or examples described in this specification and of the different embodiments or examples without contradiction.

The exemplary systems and methods of the present invention have been particularly shown and described with reference to the foregoing embodiments, which are merely examples of the best modes for carrying out the systems and methods. It will be appreciated by those skilled in the art that various changes may be made to the embodiments of the systems and methods described herein in practicing the systems and/or methods without departing from the spirit and scope of the invention as defined in the following claims.

Claims

1. An automatic driving lateral control method, characterized in that the method comprises the steps of:

2. The automatic driving lateral control method according to claim 1, characterized in that: the control expression of the R-MFAILC controller is

Wherein u (i, k) is the steering wheel angle input quantity of the vehicle at the current moment in the current iteration, u (i-1, k) is the steering wheel angle input quantity of the vehicle at the current moment in the last iteration,as the estimated value of PG vector, beta _d (k+1) is the desired rotation angle at the next time of the vehicle, β _p (i-1, k+1) is the lateral rotation angle of the vehicle at the next moment in the last iteration, e _p (i-1,k+1)＝β _d (k+1)-β _p (i-1, k+1) is the error of the lateral angle error of the vehicle at the next moment in the last iteration, ρ ₂ ∈(0,1]So that the control algorithm has better adaptability, d (k) is a measurable disturbance that is only related to the change in time k.

3. The automatic driving lateral control method according to claim 1, characterized in that: the expected rotation angle beta at the next time _d The specific expression of (k+1) is

4. The automatic driving lateral control method according to claim 1, characterized in that: the specific design steps of the R-MFAILC controller are as follows:

where T is the sampling time length, i=1, 2, … represents the number of iterations, k e {0,1, …, T } is the sampling time, β (i, k) is the lateral rotation angle of the vehicle at the current time in the present iteration, β (i, k+1) is the lateral rotation angle of the vehicle at the next time in the present iteration (i.e., the predicted lateral rotation angle), u (i, k) represents the steering wheel rotation angle input amount of the vehicle at the current time in the present iteration, and L represents the sampling time.

5. The automatic driving lateral control method according to claim 1, characterized in that: the R-MFAILC controller is controlled through robust model-free self-adaptive iterative learning, and the R-MFAILC controller is learned through constructing robust model-free self-adaptive iterative learning, and the specific expression is as follows:

6. The automated driving lateral control method of claim 1, wherein in actual operation, the specific implementation of controlling the lateral angle of the vehicle at the current time in the current iteration of the vehicle using the R-MFAILC controller comprises,

if go upTransverse measurement rotation angle beta of next time in one iteration _Measuring (i-1, k+1) and the desired rotation angle beta at the next time _d (k+1) is not equal, transmitting a transverse rotation angle error e (i-1, k+1) of the next moment in the last iteration to an R-MFAILC controller, transmitting a steering wheel rotation angle input quantity u (i-1, k+1) of the next moment in the last iteration output by the R-MFAILC controller to a steering wheel rotation angle execution module, controlling the rotation angle adjustment times of the steering wheel, enabling a steering system of an automatic driving vehicle to obtain electricity to generate a transverse rotation angle output signal beta (i, k+1) of the next moment in the current iteration, and measuring beta through an on-vehicle sensor _Measuring (i, k+1) from the expected rotation angle beta at the next time _q And (k+1) comparing to obtain a transverse rotation angle error e (i, k+1) at the next moment in the iteration, wherein the transverse rotation angle error e is used for the next iteration learning of the R-MFAILC controller so as to realize the adjustment of the transverse rotation angle.

7. The automatic driving lateral control method according to claim 6, wherein the specific implementation process of controlling the lateral rotation angle of the vehicle at the next moment in the last iteration by using the R-MFAILC controller further comprises:

8. Automatic driving lateral control device, characterized in that it comprises:

an R-MFAILC controller configured to calculate a steering wheel angle input at a next time in a previous iteration from a lateral angle error at the next time in the previous iteration of the vehicle;

9. A computer-readable storage medium, characterized by: a computer program stored thereon, wherein the program when executed by a processor implements the autopilot lateral control method of any one of claims 1 to 7.

10. An electronic device, the electronic device comprising: a processor and a memory storing a computer program, the processor being configured to perform the autopilot lateral control method of any one of claims 1 to 7 when the computer program is run.