A prediction control method and device for an autonomous vehicle

Abstract

The application discloses a kind of prediction control method and device of automatic driving vehicle, it is related to intelligent driving technical field.The specific embodiment of the method includes: based on vehicle dynamics model, the control system model considering control input delay and state delay is established;The control variable and state variable of the control system model are augmented, and the augmented system model is obtained;Based on model predictive control, the augmented system model is used as prediction model, the objective function of the vehicle to be controlled and the constraint condition of the objective function are established;According to the prediction model, the objective function is solved under the constraint condition, and the control sequence in the prediction domain is obtained, to control the operation of the vehicle to be controlled using the control sequence.The embodiment can compensate control input delay and state delay, improve driving safety, and the robustness and stability of control system.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving technology, and in particular to a predictive control method and device for an autonomous driving vehicle. Background Art

[0002] In the control system of an autonomous vehicle, there are control input delays and state delays. Control input delays typically include the computation time required for the controller to generate control instructions and the communication time required on the CAN bus. CAN stands for Controller Area Network (CAN).

[0003] State delay is caused by the perception system. Due to the complexity of the real traffic environment and the computer's insufficient ability to understand the traffic environment, the perception system consumes a lot of computing time in the process of obtaining traffic environment information and the state information of the controlled vehicle, which introduces a very obvious state delay.

[0004] However, current control system designs often ignore control input delay and state delay. That is, they don't consider control input delay, and assume that traffic environment information and the state of the controlled vehicle can be obtained in real time.

[0005] In the process of implementing the present invention, there are at least the following problems in the prior art:

[0006] Ignoring the existence of control input delay and state delay when designing a control system will seriously reduce driving safety, robustness and stability of the control system. Summary of the Invention

[0007] In view of this, an embodiment of the present invention provides a predictive control method and device for an autonomous driving vehicle. The method establishes a control system model that takes into account control input delay and state delay, and expands the control system model. Then, based on model predictive control theory, a predictive model, objective function and constraints of model predictive control are established to solve the objective function to obtain a control sequence, which can compensate for control input delay and state delay, improve driving safety, and improve the robustness and stability of the control system.

[0008] To achieve the above objectives, according to one aspect of an embodiment of the present invention, a predictive control method for an autonomous driving vehicle is provided.

[0009] A predictive control method for an autonomous driving vehicle according to an embodiment of the present invention includes: establishing a control system model that takes into account control input delay and state delay based on a vehicle dynamics model; augmenting the control variables and state variables of the control system model to obtain an augmented system model; based on model predictive control, using the augmented system model as a prediction model to establish an objective function of the controlled vehicle and constraints of the objective function; and solving the objective function under the constraints according to the prediction model to obtain a control sequence within a prediction domain, so as to use the control sequence to control the operation of the controlled vehicle.

[0010] Optionally, based on the vehicle dynamics model, a control system model that takes into account control input delay and state delay is established, including: linearizing and discretizing the vehicle dynamics model to obtain a discretized linear model; introducing control input delay and state delay into the linear model to obtain a corresponding control system model.

[0011] Optionally, the linear model is expressed as:

[0012] x k+1 =A k x k +B k u k-d

[0013] Where x k+1 represents the state variable at time k+1; x k represents the state variable at time k; u k-d represents the control variable at time kd, d represents the control input delay; A k 、B k Respectively represent the parameter matrices of the linear model at time k;

[0014] The expression of the control system model is:

[0015] x k+1 =A k x k-b +B k u k

[0016] Where x k-b represents the state variable at time kb, b represents the state delay; u k represents the control variable at time k.

[0017] Optionally, the objective function is expressed as:

[0018]

[0019] Where, represents the value of the objective function; U(X k+1 ,Y k+1 ) indicates that the controlled vehicle is at position (X k+1 ,Y k+1 ) comprehensive potential energy value at u k represents the augmented control variable at time k; S and R are semi-positive definite matrices; r k+1 Represents the reference data at time k+1; C∈R 1×(m+n×d) , m and n are the dimensions of the space R, and d represents the control input delay;

[0020] The solving the objective function under the constraint conditions includes: solving the minimum value of the objective function in the prediction domain under the constraint conditions.

[0021] Optionally, the constraints include any one or more of the following: the augmented system, state variable constraints, control variable constraints, control variable increment constraints, safety distance constraints between the controlled vehicle and obstacles, and real-time capture of the state of the controlled vehicle at the current moment in each control cycle.

[0022] Optionally, augmenting the control variables and state variables of the control system model to obtain an augmented system model includes: augmenting the control variables of the control system model from the present to the future to obtain augmented control variables; augmenting the state variables of the control system model from the past to the present to obtain augmented state variables; and using the augmented control variables and the augmented state variables to replace the control variables and state variables of the control system model respectively to obtain an augmented system model.

[0023] Optionally, the augmented system model is expressed as:

[0024] x k+1 =A k x k +B k u k

[0025] Where x k+1 represents the augmented state variable at time k+1; x k represents the augmented state variable at time k; u k represents the augmented control variable at time k; A k 、B k represent the parameter matrices of the augmented system model at time k respectively.

[0026] Optionally, the method further includes: the controlled vehicle operates a perception system, a controller, and an actuator in parallel; wherein the relationship between the acquisition period of the perception system, the control period of the controller, and the execution period of the actuator is:

[0027] T b =T c =aT a

[0028] Where, T b is the acquisition period of the perception system; T c is the control period of the controller; T a is the execution cycle of the actuator, and a is a positive number;

[0029] The discretization process includes: performing discretization process using the execution period of the actuator as a sampling period.

[0030] Optionally, the use of the control sequence to control the operation of the controlled vehicle includes: the controller sending the first a elements of the control sequence to the actuator, and the actuator executing corresponding control instructions in sequence according to the order of the elements.

[0031] To achieve the above objectives, according to another aspect of an embodiment of the present invention, a predictive control device for an autonomous driving vehicle is provided.

[0032] A predictive control device for an autonomous driving vehicle according to an embodiment of the present invention includes: a first model building module, used to establish a control system model that takes into account control input delay and state delay based on a vehicle dynamics model; a second model building module, used to augment the control variables and state variables of the control system model to obtain an augmented system model; an objective function building module, based on model predictive control, uses the augmented system model as a prediction model to establish an objective function of the controlled vehicle and constraints of the objective function; and an objective function solving module, used to solve the objective function under the constraints according to the prediction model to obtain a control sequence within the prediction domain, so as to use the control sequence to control the operation of the controlled vehicle.

[0033] Optionally, the first model building module is further used to linearize and discretize the vehicle dynamics model to obtain a discretized linear model; and introduce control input delay and state delay into the linear model to obtain a corresponding control system model.

[0034] Optionally, the linear model is expressed as:

[0035] x k+1 =A k x k +B k u k-d

[0036] Where x k+1 represents the state variable at time k+1; x krepresents the state variable at time k; u k-d represents the control variable at time kd, d represents the control input delay; A k 、B k Respectively represent the parameter matrices of the linear model at time k;

[0037] The expression of the control system model is:

[0038] x k+1 =A k x k-b +B k u k

[0039] Where x k-b represents the state variable at time kb, b represents the state delay; u k represents the control variable at time k.

[0040] Optionally, the objective function is expressed as:

[0041]

[0042] Where, represents the value of the objective function; U(X k+1 ,Y k+1 ) indicates that the controlled vehicle is at position (X k+1 ,Y k+1 ) comprehensive potential energy value at u k represents the augmented control variable at time k; S and R are semi-positive definite matrices; r k+1 Represents the reference data at time k+1; C∈R 1×(m+n×d) , m and n are the dimensions of the space R, and d represents the control input delay;

[0043] The objective function solving module is further configured to solve the minimum value of the objective function in the prediction domain under the constraint conditions.

[0044] Optionally, the constraints include any one or more of the following: the augmented system, state variable constraints, control variable constraints, control variable increment constraints, safety distance constraints between the controlled vehicle and obstacles, and real-time capture of the state of the controlled vehicle at the current moment in each control cycle.

[0045] Optionally, the second model building module is also used to augment the control variables of the control system model from the present to the future to obtain augmented control variables; augment the state variables of the control system model from the past to the present to obtain augmented state variables; and use the augmented control variables and the augmented state variables to replace the control variables and state variables of the control system model respectively to obtain an augmented system model.

[0046] Optionally, the augmented system model is expressed as:

[0047] x k+1 =A k x k +B k u k

[0048] Where x k+1 represents the augmented state variable at time k+1; x k represents the augmented state variable at time k; u k represents the augmented control variable at time k; A k 、B k represent the parameter matrices of the augmented system model at time k respectively.

[0049] Optionally, the device further comprises: a parallel operation module configured to cause the controlled vehicle to operate a perception system, a controller, and an actuator in parallel; wherein the relationship between the acquisition period of the perception system, the control period of the controller, and the execution period of the actuator is as follows:

[0050] T b =T c =aT a

[0051] Where, T b is the acquisition period of the perception system; T c is the control period of the controller; T a is the execution cycle of the actuator, and a is a positive number;

[0052] The first model building module is further configured to perform discretization processing using the execution period of the actuator as a sampling period.

[0053] Optionally, the objective function solving module is further configured for the controller to send the first a elements of the control sequence to the executor, and for the executor to sequentially execute corresponding control instructions in the order of the elements.

[0054] To achieve the above objective, according to another aspect of the embodiments of the present invention, an electronic device is provided.

[0055] An electronic device according to an embodiment of the present invention includes: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement a predictive control method for an autonomous driving vehicle according to an embodiment of the present invention.

[0056] To achieve the above objective, according to another aspect of the embodiments of the present invention, a computer-readable medium is provided.

[0057] A computer-readable medium according to an embodiment of the present invention stores a computer program, which, when executed by a processor, implements a predictive control method for an autonomous driving vehicle according to an embodiment of the present invention.

[0058] One embodiment of the above invention has the following advantages or beneficial effects: by establishing a control system model that takes into account control input delay and state delay, and expanding the control system model, and then based on model predictive control theory, establishing a prediction model, objective function and constraints of model predictive control to solve the objective function to obtain a control sequence, it is possible to compensate for control input delay and state delay, improve driving safety, and improve the robustness and stability of the control system.

[0059] After linearizing and discretizing the vehicle dynamics model, control input delay and state delay are introduced to convert the nonlinear vehicle dynamics model into a linear, discrete model, reducing computational complexity and solution time. The controlled vehicle's integrated potential energy, augmented control variables, and reference data are incorporated into the objective function. The objective function is then minimized within the prediction domain under constraints to obtain the optimal control sequence and ensure ideal control results.

[0060] By augmenting both the control variables and the state variables, the control system model is expanded. Parallel strategies can reduce delays within the control cycle and the sampling time of the discretization process, enabling higher-frequency and smoother control of the controlled vehicle. The first a elements of the control sequence are sent to the actuator for execution, further optimizing the control effect.

[0061] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] The accompanying drawings are provided for a better understanding of the present invention and are not intended to limit the present invention.

[0063] Figure 1 is a schematic diagram of the main steps of a predictive control method for an autonomous driving vehicle according to an embodiment of the present invention;

[0064] Figure 2 1 is a schematic diagram of the main flow of a predictive control method for an autonomous driving vehicle according to an embodiment of the present invention;

[0065] Figure 3 This is a schematic diagram of the update of control instructions and system states in a discrete control system when there is a control input delay;

[0066] Figure 4is a parallel strategy timing diagram of a predictive control method for an autonomous driving vehicle according to an embodiment of the present invention;

[0067] Figure 5 It is a transfer and update trajectory diagram of the control sequence output of the controller and the control instructions executed by the actuator in the embodiment of the present invention;

[0068] Figure 6 is a schematic diagram of a traffic scene according to an embodiment of the present invention;

[0069] Figure 7 It is a schematic diagram of the control effect of the control system under different control input delays;

[0070] Figure 8 is a schematic diagram of main modules of a predictive control device for an autonomous driving vehicle according to an embodiment of the present invention;

[0071] Figure 9 It is a schematic structural diagram of a computer device suitable for implementing an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0072] The following description of exemplary embodiments of the present invention is made in conjunction with the accompanying drawings, in which various details of the embodiments of the present invention are included to facilitate understanding. These details should be considered as merely exemplary. Therefore, it should be appreciated by those skilled in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0073] Figure 1 FIG. 1 is a schematic diagram of the main steps of the predictive control method for an autonomous driving vehicle according to an embodiment of the present invention. Figure 1 As shown, the predictive control method for an autonomous driving vehicle according to an embodiment of the present invention mainly includes the following steps:

[0074] Step S101: Based on the vehicle dynamics model, a control system model is established that accounts for control input delay and state delay. Control input delay and state delay are introduced into the vehicle dynamics model to generate a corresponding control system model. The control system model has control variables and state variables. Control variables include, for example, the front wheel steering angle and longitudinal force in the vehicle body coordinate system; state variables include, for example, vehicle speed, yaw rate, and vehicle position.

[0075] Step S102: Augmenting the control variables and state variables of the control system model to obtain an augmented system model. The control variables and state variables of the control system model are augmented starting from the current moment to obtain corresponding augmented control variables and augmented state variables. The augmented control variables and augmented state variables are then used to replace the control variables and state variables of the control system model, respectively, to obtain an augmented system model.

[0076] Step S103: Based on model predictive control, the augmented system model is used as a prediction model to establish an objective function of the controlled vehicle and constraints of the objective function. The augmented system model is used as a prediction model for model predictive control (MPC), and the objective function of the model predictive control and constraints of the objective function are established.

[0077] Model predictive control (MPC) is a special type of control. Its current control action is obtained at each sampling instant by solving a finite-horizon open-loop optimal control problem. The current state of the process serves as the initial state of the optimal control problem, and the resulting optimal control sequence only implements the first control action. Essentially, MPC solves an open-loop optimal control problem.

[0078] In this embodiment, the objective function includes three influencing factors: the comprehensive potential energy of the controlled vehicle, the augmented control variables, and the reference data. Comprehensive potential energy refers to the superposition of the potential energy fields of different obstacles (such as road boundaries, lane lines, obstructing vehicles, etc.) to express the combined impact of these obstacles on the controlled vehicle. Among them, potential energy refers to the energy possessed by an object or system due to its position or state. Potential energy can be generated by gravity, electrostatic potential, mechanical force, and chemical force. The location of multiple potential energy bodies forms a potential energy field.

[0079] The augmented control variable is the result of augmenting the control variable. The reference data can be the speed limit set by the road or the control instructions input by the operator, such as lane change instructions, follow instructions, overtaking instructions, etc.

[0080] The constraints may include any one or more of the following: augmented system, state variable constraints, control variable constraints, control variable increment constraints, safe distance constraints between the controlled vehicle and obstacles, and real-time capture of the state of the controlled vehicle at the current moment in each control cycle.

[0081] It should be noted that the constraints can be any one or more of the above-mentioned combinations. When the constraints include all of the above-mentioned constraints, the optimal control sequence can be obtained by solving the minimum value of the objective function in the prediction domain. Compared with other single constraints or constraint combinations, the best control effect can be obtained.

[0082] Step S104: Based on the prediction model, the objective function is solved under the constraints to obtain a control sequence within the prediction domain, and the control sequence is used to control the operation of the controlled vehicle. Based on the prediction model, an optimization solver, such as fmincon, is used to solve the objective function under the constraints to obtain a control sequence within the prediction domain. The first element or the first part of the elements of the obtained control sequence is then sent as a control instruction to the actuator. After a control input delay, the actuator receives the control instruction and executes it to control the operation of the controlled vehicle.

[0083] Figure 2 FIG. 1 is a schematic diagram of the main flow chart of the predictive control method for an autonomous driving vehicle according to an embodiment of the present invention. Figure 2 As shown, the predictive control method for an autonomous driving vehicle according to an embodiment of the present invention, taking into account the case of time-invariant delay (i.e., delay that does not change with time), mainly includes the following steps:

[0084] Step S201: Linearize and discretize the vehicle dynamics model to obtain a discretized linear model. Since the vehicle dynamics model is a nonlinear model, solving it online consumes a large amount of computing resources and takes a long time. To reduce computational complexity and save computing time, the vehicle dynamics model can be linearized and discretized to obtain a discretized linear model. The linear model is expressed as:

[0085] x k+1 =A k x k +B k u k-d

[0086] Formula 1

[0087] Where x k+1 represents the state variable at time k+1; x k represents the state variable at time k; u k-d represents the control variable at time kd; d represents the control input delay; A k 、B k They represent the parameter matrices of the linear model at time k.

[0088] in, τ in is the maximum input delay time, Δt is the sampling period of the digital system; A k =e A(t )Δt ,

[0089] The estimated state information used by the control system is often outdated. In discrete control systems, the state delay can be expressed as τ st is the maximum state delay time. Figure 3 This is a schematic diagram of the update of control instructions and system states in a discrete control system when there is a control input delay. Figure 3 As shown in Figure 1, when there is a control input delay, the controller sends a control instruction u, and the system state x will be updated after a period of time.

[0090] In light of the background technology, it can be seen that ignoring control input delays and state delays when controlling a controlled vehicle will reduce driving safety, the robustness and stability of the control system, and even lead to traffic accidents. Therefore, this embodiment improves the existing predictive control method to compensate for the adverse effects of delays on the control system. It should be noted that this example has a prerequisite, specifically that the control system is aware of the traffic environment within a certain field of view in front of the controlled vehicle. This traffic environment is obtained by the perception system capturing environmental information in the field of view in front of the controlled vehicle during each sampling period.

[0091] Step S202: Introduce control input delay and state delay into the linear model to obtain the corresponding control system model. Introduce control input delay into the linear model and state delay The control system model that takes into account control input delay and state delay can be obtained. Represents a space. The expression of the control system model is:

[0092] x k+1 =A k x k-b +B k u k

[0093] Formula 2

[0094] Where x k-b represents the state variable at time kb, b represents the state delay; u k represents the control variable at time k.

[0095] Step S203: Augment the control variables and state variables of the control system model to obtain an augmented system model. In this step, the control variables of the control system model are augmented (or expanded) from the present (i.e., the current moment) to the future, and the state variables of the control system model are augmented (or expanded) from the past to the present, thereby obtaining augmented control variables and augmented state variables. The augmented control variables and augmented state variables are then used to replace the control variables and state variables of the control system model, respectively, to obtain an augmented system model. The augmented system model is expressed as follows:

[0096] x k+1 =Ak x k +B k u k

[0097] Formula 3

[0098] Where x k+1 represents the augmented state variable at time k+1; x k represents the augmented state variable at time k; u k represents the augmented control variable at time k; A k 、B k They represent the parameter matrices of the augmented system model at time k. k 、A k 、B k and u k The expression is as follows:

[0099]

[0100] Where, Representatives and Another different space; m and n are spaces dimension.

[0101]

[0102] Where I is the unit matrix; A k 、B k They represent the parameter matrices of the linear model at time k.

[0103]

[0104] u k =u k+d

[0105] Formula 7

[0106] Step S204: Based on model predictive control, the augmented system model is used as the prediction model to establish the objective function of the controlled vehicle and the constraints of the objective function. By using the augmented system as the prediction model in model predictive control, the predictive control method with delay compensation in this embodiment can be obtained. The objective function is expressed as:

[0107]

[0108] Where, represents the value of the objective function; U(X k+1 ,Y k+1 ) indicates that the controlled vehicle is at position (X k+1 ,Y k+1 ) comprehensive potential energy value at uk represents the augmented control variable at time k; S and R are semi-positive definite matrices; r k+1 represents the reference data at time k+1; (·) T It should be noted that, in this embodiment, no additional method is needed to plan the speed in advance.

[0109] definition is a vector whose i-th element is 1 and the other elements are 0, then we can get:

[0110]

[0111] In the above formula, the value of i is an empirical value, which can achieve better prediction results.

[0112] Constraints include augmented systems, state variable constraints, control variable constraints, control variable increment constraints, safe distance constraints between the controlled vehicle and obstacles, and one or more combinations of real-time capture of the current state of the controlled vehicle within each control cycle. The corresponding expressions for each constraint are as follows:

[0113] (1) Augmented system constraint expression:

[0114] x k+1 =A k x k +B k u k ,k=0,1,…,N p -1

[0115] Formula 10

[0116] Where, represents the prediction domain; Representatives and A different space.

[0117] (2) State variable constraint expression:

[0118] x min ≤x k ≤x max ,k=1,2,…,N p

[0119] Formula 11

[0120] (3) Control variable constraint expression:

[0121] u min ≤u k ≤u max ,k=0,1,…,N p -1

[0122] Formula 12

[0123] (4) Control variable increment constraint expression:

[0124]

[0125] (5) Safety distance constraint expression between the controlled vehicle and the obstacle:

[0126]

[0127] Where, X j,k is the horizontal coordinate of the j-th obstacle at time k; Y j,k is the ordinate of the j-th obstacle at time k; N o is the total number of obstacles within the considered range; P and Q are unit matrices.

[0128] (6) The constraint expression for capturing the state of the controlled vehicle at the current moment in real time during each control cycle:

[0129] x0=x(t)

[0130] Formula 15 Step S205: Solve the objective function under the constraints according to the prediction model to obtain the control sequence within the prediction domain. Solving the objective function is to find the minimum value of the objective function within the prediction domain, that is, to solve the following formula:

[0131]

[0132] Where,

[0133] When solving Equation 16, the constraints need to be abbreviated for ease of processing. This is explained in detail below.

[0134] State variable constraint expression x min ≤x k ≤x max ,k=1,2,…,N p It can be expressed as:

[0135]

[0136] Substituting the augmented system constraint (i.e., Equation 10) into Equation 17, the state variable constraint can be converted into a constraint on the control input, as follows:

[0137] A e U≤B e

[0138] Formula 18

[0139] in,

[0140]

[0141]

[0142] Control variable increment constraint expression Δu min ≤u -1 -u0≤Δu max , is a constraint on the first element of the control variable, representing the control instruction executed in the previous control cycle. Therefore, for the control variable constraint expression (i.e., Formula 12), when k = 0, it should satisfy:

[0143]

[0144] in,

[0145] Therefore, Δu from Equation 12 and Equation 13 is min ≤u -1 -u0≤Δu max , the constraints on the control input can be derived as:

[0146]

[0147] Similarly, Δu in Formula 13 min ≤u k -u k+1 ≤Δu max ,k=0,1,…,N p -2 can be converted to:

[0148]

[0149] Finally, all linear inequality constraints can be simplified as follows:

[0150]

[0151] Formula 14 is a nonlinear inequality constraint, which can be simplified as follows:

[0152] f non (X,Y)≤0

[0153] Formula 23

[0154] Where,

[0155]

[0156]

[0157] as well as

[0158]

[0159] Therefore, the nonlinear inequality constraint can be written as:

[0160]

[0161] Finally, the predictive control problem with delay compensation can be transformed into the following optimization problem with nonlinear constraints:

[0162]

[0163] sA in U≤B in

[0164]

[0165] Among them, st represents the constraint condition. The above optimization problem can be solved using an optimization solver, such as fmincon. After solving, when the system cycle is , the controller generates a control sequence.

[0166] Step S206: The control sequence is sent to the actuator via the CAN bus. After a control input delay, the actuator receives the control sequence, allowing it to control the operation of the controlled vehicle according to the control sequence. In this embodiment, the first element, or the first part of the elements, of the control sequence are sent to the actuator as a control instruction. After a control input delay d, when the system period reaches k+d, the actuator receives the control instruction and executes it.

[0167] In practical applications, for discrete systems, the smaller the sampling time, the smaller the discrete error. Assume that the maximum time delays of the perception system, controller, and communication are τ st , τ c and τ d If these events are executed in a serial manner, the control cycle Δt of the control system should be no less than max(τ st ,τ c +τ d ), resulting in a longer control cycle and larger discrete error.

[0168] In order to solve the above problems, in a preferred embodiment, the advantages of model predictive control theory can be utilized to use a parallel strategy to implement a higher frequency control system. Figure 4 This is a parallel strategy timing diagram of the predictive control method for an autonomous driving vehicle according to an embodiment of the present invention. Figure 4The controller's operating frequency is different from the actuator's, and the actuator's frequency is twice that of the controller. Therefore, the control system can be discretized with a smaller sampling period to achieve higher-frequency and smoother control of the controlled vehicle.

[0169] Exemplarily, the controlled vehicle operates the perception system, controller, and actuator in parallel; wherein the relationship between the acquisition cycle of the perception system, the control cycle of the controller, and the execution cycle of the actuator is:

[0170] T b =T c =aT a

[0171] Formula 25

[0172] Where, T b ≥τ st , is the acquisition cycle of the perception system; T c ≥τ c +τ d , is the control period of the controller; T a is the execution cycle of the actuator; Is a positive number.

[0173] Based on the above periodic relationship, when the linearized vehicle dynamics model is discretized in step S201, the execution period of the actuator can be used as the sampling period for discretization, that is, the control period Δt=T a Subsequently, the objective function is solved according to steps S202 to S205 to obtain an optimized control sequence U. The first a elements of the control sequence U are sent as control instructions to the actuator. The actuator temporarily stores these control instructions in a cache and executes the corresponding control instructions in sequence according to the order of the elements to complete the control of the controlled vehicle.

[0174] Among them, the update rule of the control instructions in the cache can be described as the cache rule, which defines the control instruction cache of the actuator as The indicator variable for receiving a new control sequence is ω k The last time a command was received was The cache update rule can be expressed as:

[0175]

[0176] In the formula, when a new control sequence is received at time k, ω k is equal to 1, otherwise it is equal to 0; is the received control sequence; is the transformation matrix, which is in the form:

[0177]

[0178] Therefore, the control instruction to be executed at time k can be expressed as:

[0179]

[0180] Figure 5 It is a diagram of the control sequence output of the controller of the embodiment of the present invention and the transmission and update trajectory of the control instructions executed by the actuator. Figure 5 It can be seen that the control sequence output at time k-3 is passed to the executor for execution at time k-1; the control sequence output at time k-1 is passed to the executor for execution at time k+1.

[0181] The following is a simulation experiment to demonstrate the compensation effect of the predictive control method considering delay compensation in this embodiment. Figure 6 Schematic diagram of a traffic scene according to an embodiment of the present invention. Figure 6 During the simulation, a two-lane road in the same direction extends along the positive direction of the horizontal axis of the world coordinate system. The width of the lane is 3.5m. The sampling time of the discrete system is Δt = 0.05 seconds. The prediction domain length of the model predictive control is N p = 20, upper and lower limits of state vector constraints x min and x max It can be derived from the following constraints:

[0182]

[0183] Where γ is the angular velocity of the actuator, is the angle of the actuator output.

[0184] For some of the unconstrained variables, in order to meet the feasibility of numerical solution using computers, some relatively large values ​​can be used to constrain their upper and lower bounds in the simulation. Considering the control input delay, it is assumed that the controlled vehicle starts from the center of the second lane on the right at a speed of 8m / s and moves in the positive direction of the X axis. There are no other vehicles or obstacles on the road within the considered range (see Figure 6 ).

[0185] Based on the above traffic scenario, the predictive control method considering delay compensation of this embodiment is used to control the controlled vehicle. The specific control effect is shown in Figure 7 . Figure 7 The horizontal axis represents time, and the vertical axis represents longitudinal velocity. Figure 7 As can be seen, as the control input delay increases, the longitudinal speed adjustment begins to fluctuate; and the greater the control input delay, the longer the longitudinal speed adjustment takes. Based on this figure, it can be seen that the method of this embodiment can significantly reduce the adverse effects of delay on the control system.

[0186] The predictive control method for an autonomous driving vehicle in an embodiment of the present invention analyzes the control input delay and state delay existing in the autonomous driving system, expands the prediction model, and proposes a predictive control method with delay compensation for autonomous driving vehicles, which can reduce the adverse effects of control input delay and state delay in the autonomous driving system; and designs a parallel control strategy for the perception system, controller and actuator, which can directly reduce the delay time within the control cycle and reduce the sampling frequency of the discrete system.

[0187] Figure 8 Schematic diagram of the main modules of the predictive control device for an autonomous driving vehicle according to an embodiment of the present invention. Figure 8 As shown, the predictive control device 800 for an autonomous driving vehicle according to an embodiment of the present invention mainly includes:

[0188] The first model building module 801 is used to build a control system model based on the vehicle dynamics model, taking into account control input delay and state delay. Control input delay and state delay are introduced into the vehicle dynamics model to obtain a corresponding control system model. The control system model has control variables and state variables. Control variables include, for example, the front wheel steering angle and longitudinal force in the vehicle body coordinate system; state variables include, for example, vehicle speed, yaw rate, and vehicle position.

[0189] The second model building module 802 is configured to augment the control variables and state variables of the control system model to obtain an augmented system model. The control variables and state variables of the control system model are augmented starting from the current moment to obtain corresponding augmented control variables and augmented state variables. The augmented control variables and augmented state variables are then used to replace the control variables and state variables of the control system model, respectively, to obtain an augmented system model.

[0190] The objective function establishment module 803 establishes the objective function of the controlled vehicle and the constraints of the objective function based on the model predictive control, using the augmented system model as the prediction model. The augmented system model is used as the prediction model of the MPC, and the objective function of the model predictive control and the constraints of the objective function are established.

[0191] In this embodiment, the objective function includes three influencing factors: the comprehensive potential energy of the controlled vehicle, the augmented control variables, and the reference data. Comprehensive potential energy refers to the superposition of the potential energy fields of different obstacles (such as road boundaries, lane lines, obstructing vehicles, etc.) to express the combined impact of these obstacles on the controlled vehicle. Among them, potential energy refers to the energy possessed by an object or system due to its position or state. Potential energy can be generated by gravity, electrostatic potential, mechanical force, and chemical force. The location of multiple potential energy bodies forms a potential energy field.

[0192] The augmented control variable is the result of augmenting the control variable. The reference data can be the speed limit set by the road or the control instructions input by the operator, such as lane change instructions, follow instructions, overtaking instructions, etc.

[0193] Constraints can include any one or more of the following: augmented systems, state variable constraints, control variable constraints, control variable increment constraints, safe distance constraints between the controlled vehicle and obstacles, and real-time capture of the controlled vehicle's state at each control cycle. It should be noted that the constraints can be any one or a combination of the above. When the constraints include all of the above, the optimal control sequence can be obtained by minimizing the objective function within the prediction domain. This can achieve the best control effect compared to other single constraints or constraint combinations.

[0194] The objective function solving module 804 is configured to solve the objective function under the constraints based on the prediction model to obtain a control sequence within the prediction domain, thereby controlling the operation of the controlled vehicle using the control sequence. Based on the prediction model, an optimization solver, such as fmincon, is employed to solve the objective function under the constraints to obtain a control sequence within the prediction domain. The first element or the first portion of the elements of the obtained control sequence is then sent as a control instruction to an actuator. After a control input delay, the actuator receives and executes the control instruction to control the operation of the controlled vehicle.

[0195] In addition, the prediction control device 800 of the autonomous driving vehicle according to the embodiment of the present invention may further include: a parallel operation module ( Figure 8 (not shown). This module is used for the controlled vehicle to run the perception system, controller, and actuator in parallel; wherein the relationship between the acquisition period of the perception system, the control period of the controller, and the execution period of the actuator is:

[0196] T b =T c =aT a

[0197] Where, T b is the acquisition period of the perception system; T c is the control period of the controller; T a is the execution period of the actuator, and a is a positive number.

[0198] From the above description, it can be seen that by establishing a control system model that takes into account control input delay and state delay, and expanding the control system model, and then based on the model predictive control theory, establishing a prediction model, objective function and constraints of model predictive control to solve the objective function to obtain a control sequence, it is possible to compensate for control input delay and state delay, improve driving safety, and enhance the robustness and stability of the control system.

[0199] Reference below Figure 9 , which shows a structural diagram of a computer system 900 suitable for implementing an electronic device according to an embodiment of the present invention. Figure 9 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0200] like Figure 9 As shown, the computer system 900 includes a central processing unit (CPU) 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage unit 908 into a random access memory (RAM) 903. Various programs and data required for the operation of the computer system 900 are also stored in the RAM 903. The CPU 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0201] The following components are connected to the I / O interface 905: an input section 906 including a keyboard, a mouse, and the like; an output section 907 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 908 including a hard disk and the like; and a communication section 909 including a network interface card such as a LAN card or a modem. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the I / O interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 910 as needed, so that computer programs read therefrom can be installed into the storage section 908 as needed.

[0202] In particular, according to an embodiment disclosed in the present invention, the process described in the main step diagram above can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program code for executing the method shown in the main step diagram. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 909, and / or installed from a removable medium 911. When the computer program is executed by the central processing unit (CPU) 901, the above-mentioned functions defined in the system of the present invention are performed.

[0203] It should be noted that the computer-readable medium described in the present invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media can include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. This propagated data signal can take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. Program code embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wireline, optical fiber cable, RF, or any suitable combination thereof.

[0204] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the above-mentioned module, program segment, or a part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0205] The modules described in the embodiments of the present invention may be implemented in software or hardware. The modules described may also be provided in a processor. For example, they may be described as follows: a processor comprising a first model building module, a second model building module, an objective function building module, and an objective function solving module. The names of these modules do not, in some cases, limit the modules themselves. For example, the first model building module may also be described as a "module for building a control system model that takes into account control input delays and state delays based on a vehicle dynamics model."

[0206] As another aspect, the present invention further provides a computer-readable medium, which may be included in the device described in the above embodiment, or may exist independently and not be incorporated into the device. The computer-readable medium carries one or more programs, and when executed by the device, the device includes the following steps: establishing a control system model that takes into account control input delay and state delay based on a vehicle dynamics model; augmenting the control variables and state variables of the control system model to obtain an augmented system model; establishing an objective function of a controlled vehicle and constraints of the objective function based on model predictive control using the augmented system model as a prediction model; solving the objective function under the constraints according to the prediction model to obtain a control sequence within a prediction domain, and using the control sequence to control the operation of the controlled vehicle.

[0207] According to the technical solution of the embodiment of the present invention, by establishing a control system model that takes into account control input delay and state delay, and expanding the control system model, and then based on the model predictive control theory, establishing a prediction model, objective function and constraints of model predictive control to solve the objective function to obtain a control sequence, it is possible to compensate for control input delay and state delay, improve driving safety, and improve the robustness and stability of the control system.

[0208] The above-mentioned product can execute the method provided by the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not fully described in this embodiment, please refer to the method provided by the embodiment of the present invention.

[0209] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A predictive control method for an autonomous vehicle, characterized in that: include: Based on the vehicle dynamics model, a control system model is established that takes into account control input delay and state delay; Augmenting the control variables and state variables of the control system model to obtain an augmented system model; Based on model predictive control, the augmented system model is used as a prediction model to establish an objective function for the controlled vehicle and constraints for the objective function. The objective function includes the controlled vehicle's comprehensive potential energy, augmented control variables, and reference data. The comprehensive potential energy is the sum of the potential energy fields of different obstacles. The augmented control variables are the result of augmenting the control variables. The reference data is the speed limit set by the road or the control command input by the operator. The objective function is solved under the constraint conditions according to the prediction model to obtain a control sequence within the prediction domain, so as to control the operation of the controlled vehicle using the control sequence.

2. The method according to claim 1, characterized in that The control system model considering control input delay and state delay is established based on the vehicle dynamics model, including: Perform linearization and discretization on the vehicle dynamics model to obtain a discretized linear model; Control input delay and state delay are introduced into the linear model to obtain a corresponding control system model.

3. The method according to claim 2, characterized in that The linear model is expressed as: x k+1 =A k x k +B k u k-d Where x k+1 represents the state variable at time k+1; x k represents the state variable at time k; u k-d represents the control variable at time kd, d represents the control input delay; A k 、B k Respectively represent the parameter matrices of the linear model at time k; The expression of the control system model is: x k+1 =A k x k-b +B k u k Where x k-b represents the state variable at time kb, b represents the state delay; u k represents the control variable at time k.

4. The method according to claim 1, wherein The expression of the objective function is: Where, represents the value of the objective function; U(X k+1 ,Y k+1 ) indicates that the controlled vehicle is at position (X k+1 ,Y k+1 ) comprehensive potential energy value at u k represents the augmented control variable at time k; S and R are semi-positive definite matrices; r k+1 represents the reference data at time k+1; m and n are spaces The dimension of , d represents the control input delay; Solving the objective function under the constraint conditions includes: Under the constraints, the minimum value of the objective function in the prediction domain is solved.

5. The method according to claim 1, wherein The constraint conditions include any one or more of the following: the augmented system, state variable constraints, control variable constraints, control variable increment constraints, safety distance constraints between the controlled vehicle and obstacles, and real-time capture of the state of the controlled vehicle at the current moment in each control cycle.

6. The method according to any one of claims 1 to 5, characterized in that The step of augmenting the control variables and state variables of the control system model to obtain an augmented system model includes: Augmenting the control variables of the control system model from the present to the future to obtain augmented control variables; Augmenting the state variables of the control system model from the past to the present to obtain augmented state variables; The augmented control variables and the augmented state variables are used to replace the control variables and the state variables of the control system model respectively, so as to obtain an augmented system model.

7. The method according to claim 6, characterized in that The expression of the augmented system model is: x k+1 =A k x k +B k u k Where x k+1 represents the augmented state variable at time k+1; x k represents the augmented state variable at time k; u k represents the augmented control variable at time k; A k 、B k represent the parameter matrices of the augmented system model at time k respectively.

8. The method according to claim 2, characterized in that The method further comprises: The controlled vehicle operates a perception system, a controller, and an actuator in parallel; wherein the relationship between the acquisition period of the perception system, the control period of the controller, and the execution period of the actuator is: T b =T c =aT a Where, T b is the acquisition period of the perception system; T c is the control period of the controller; T a is the execution cycle of the actuator, and a is a positive number; The discretization process includes: performing discretization process using the execution period of the actuator as a sampling period.

9. The method according to claim 8, characterized in that The using the control sequence to control the operation of the controlled vehicle includes: The controller sends the first a elements of the control sequence to the executor, and the executor executes the corresponding control instructions in sequence according to the order of the elements.

10. A predictive control device for an autonomous driving vehicle, characterized in that: include: A first model building module is used to build a control system model taking into account control input delay and state delay based on the vehicle dynamics model; A second model building module is used to augment the control variables and state variables of the control system model to obtain an augmented system model; an objective function establishment module, based on model predictive control, using the augmented system model as a prediction model to establish an objective function for the controlled vehicle and constraints for the objective function; the objective function includes the controlled vehicle's comprehensive potential energy, augmented control variables, and reference data; the comprehensive potential energy refers to the superposition of potential energy fields of different obstacles; the augmented control variables are the results of augmenting the control variables; and the reference data are speed limits set along the road or control instructions input by the operator; The objective function solving module is used to solve the objective function under the constraint conditions according to the prediction model to obtain a control sequence within the prediction domain, so as to use the control sequence to control the operation of the controlled vehicle.

11. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 9.

12. A computer-readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images