Control operation method and device of automatic driving vehicle and unmanned vehicle

Abstract

The present application discloses a control operation method, device and unmanned vehicle for an autonomous driving vehicle. A specific implementation of the method includes: during the operation of the autonomous driving vehicle, the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment are subjected to constraint decomposition by a constraint decomposition algorithm, wherein the objective function and constraint conditions are used to determine the control sequence of the autonomous driving vehicle at a future moment; an acceleration algorithm is used to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence; and the operation of the autonomous driving vehicle at the next moment is controlled by the first control instruction in the control sequence. The present application improves the information processing efficiency of obtaining a control sequence based on the objective function and constraint conditions, and meets the online real-time requirements of the autonomous driving vehicle.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, specifically to autonomous driving technology, and more particularly to a control and operation method, device, computer-readable medium, electronic device and unmanned vehicle for an autonomous driving vehicle, which can be applied in autonomous driving scenarios. Background Art

[0002] With the support of Internet of Vehicles and artificial intelligence technologies, autonomous driving technology has developed rapidly. However, the existing methods for controlling the operation of autonomous vehicles are difficult to meet the real-time solution requirements of the on-board controller due to the heavy computational burden of online optimization. Summary of the invention

[0003] The embodiments of the present application provide a control operation method, device, computer-readable medium, electronic device and unmanned vehicle for an autonomous driving vehicle.

[0004] In a first aspect, an embodiment of the present application provides a method for controlling and operating an autonomous driving vehicle, comprising: during the operation of the autonomous driving vehicle, constraining the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment through a constraint decomposition algorithm, wherein the objective function and constraint conditions are used to determine the control sequence of the autonomous driving vehicle at future moments; using an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence; and controlling the operation of the autonomous driving vehicle at the next moment through the first control instruction in the control sequence.

[0005] In some embodiments, the above-mentioned iterative calculation process of the objective function and constraint conditions after constraint decomposition using the acceleration algorithm to obtain a control sequence includes: converting the process of solving the objective function and constraint conditions after constraint decomposition using the alternating direction multiplier method into a fixed point iterative form; using the acceleration algorithm to process the iterative calculation process corresponding to the fixed point iterative form to obtain a control sequence.

[0006] In some embodiments, the objective function and constraints corresponding to the autonomous driving vehicle at the current moment are constrained and decomposed through the constraint decomposition algorithm, including: taking the variables in the constraints as local variables, and adding the global variables corresponding to the local variables to the constraints; based on the global variables and local variables, decomposing the constraints into equality constraints corresponding to the local variables and inequality constraints corresponding to the global variables; and characterizing the consensus constraints between the local variables and the global variables through dual variables.

[0007] In some embodiments, the constraints include: constraints on control variables that control the operation of the autonomous driving vehicle, and constraints on state variables that characterize state information after the autonomous driving vehicle is running at a future time.

[0008] In some embodiments, the constraints include: constraints on control variables that control the operation of the autonomous vehicle.

[0009] In some embodiments, the above method also includes: predicting the state information sequence of the autonomous driving vehicle at a future moment according to the actual state information and control sequence of the autonomous driving vehicle at the current moment through a prediction model; and the above-mentioned iterative calculation process of the objective function and constraint conditions after constraint decomposition using an acceleration algorithm to obtain a control sequence, including: based on the state information sequence at the current moment, using an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence.

[0010] In some embodiments, the above method also includes: determining the actual state information of the autonomous driving vehicle at the next moment after the autonomous driving vehicle runs according to the first control instruction; and updating the prediction model based on the actual state information at the next moment and the state information sequence.

[0011] In the second aspect, an embodiment of the present application provides a control operation device for an autonomous driving vehicle, comprising: a constraint decomposition unit, configured to, during the operation of the autonomous driving vehicle, perform constraint decomposition on the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment through a constraint decomposition algorithm, wherein the objective function and constraint conditions are used to determine the control sequence of the autonomous driving vehicle at future moments; an acceleration unit, configured to use an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence; and a control unit, configured to control the operation of the autonomous driving vehicle at the next moment through the first control instruction in the control sequence.

[0012] In some embodiments, the acceleration unit is further configured to: convert the process of solving the objective function and constraint conditions after constraint decomposition using the alternating direction multiplier method into a fixed point iteration form; use the acceleration algorithm to process the iterative calculation process corresponding to the fixed point iteration form to obtain a control sequence.

[0013] In some embodiments, the constraint decomposition unit is further configured to: treat the variables in the constraint conditions as local variables, and add the global variables corresponding to the local variables to the constraint conditions; based on the global variables and the local variables, decompose the constraint conditions into equality constraints corresponding to the local variables and inequality constraints corresponding to the global variables; and characterize the consensus constraints between the local variables and the global variables through dual variables.

[0014] In some embodiments, the constraints include: constraints on control variables that control the operation of the autonomous driving vehicle, and constraints on state variables that characterize state information after the autonomous driving vehicle is running at a future time.

[0015] In some embodiments, the constraints include: constraints on control variables that control the operation of the autonomous vehicle.

[0016] In some embodiments, the above-mentioned device also includes: a prediction unit, configured to predict the state information sequence of the autonomous driving vehicle at a future moment through a prediction model based on the actual state information and control sequence of the autonomous driving vehicle at the current moment; and an acceleration unit, further configured to: based on the state information sequence at the current moment, use an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence.

[0017] In some embodiments, the above-mentioned device also includes: an updating unit, configured to: determine the actual state information of the autonomous driving vehicle at the next moment after the autonomous driving vehicle runs according to the first control instruction; and update the prediction model according to the actual state information at the next moment and the state information sequence.

[0018] In a third aspect, an embodiment of the present application provides a computer-readable medium having a computer program stored thereon, wherein when the program is executed by a processor, the method described in any implementation manner of the first aspect is implemented.

[0019] In a fourth aspect, an embodiment of the present application provides an electronic device, comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by one or more processors, the one or more processors implement the method described in any implementation manner of the first aspect.

[0020] In the fifth aspect, an embodiment of the present application provides an unmanned vehicle, comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation method of the first aspect.

[0021] The control operation method, device and unmanned vehicle of the autonomous driving vehicle provided in the embodiments of the present application, during the operation of the autonomous driving vehicle, use a constraint decomposition algorithm to decompose the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment, wherein the objective function and constraint conditions are used to determine the control sequence of the autonomous driving vehicle at the future moment; use an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after the constraint decomposition to obtain a control sequence; and control the operation of the autonomous driving vehicle at the next moment through the first control instruction in the control sequence, thereby improving the information processing efficiency of obtaining the control sequence based on the objective function and constraint conditions, and meeting the online real-time requirements of the autonomous driving vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0023] Figure 1 is an exemplary system architecture diagram in which an embodiment of the present application may be applied;

[0024] Figure 2 is a flow chart of an embodiment of a method for controlling and operating an autonomous driving vehicle according to the present application;

[0025] Figure 3 is a schematic diagram of a constraint decomposition algorithm including control variable constraints and state variable constraints;

[0026] Figure 4 is a schematic diagram of a constraint decomposition algorithm including control variable constraints;

[0027] Figure 5 is a schematic diagram of an application scenario of the control operation method of an autonomous driving vehicle according to this embodiment;

[0028] Figure 6 is a flowchart of another embodiment of a method for controlling an operation of an autonomous driving vehicle according to the present application;

[0029] Figure 7 is a structural diagram of an embodiment of a control operation device for an autonomous driving vehicle according to the present application;

[0030] Figure 8 is a schematic diagram of the structure of a computer system suitable for implementing the embodiments of the present application;

[0031] Fig. 9 is a schematic diagram of the system architecture of an unmanned vehicle according to the present application;

[0032] Fig.10 It is a schematic diagram of the structure of the unmanned vehicle according to the present application. DETAILED DESCRIPTION

[0033] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It should also be noted that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.

[0034] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0035] Figure 1An exemplary architecture 100 is shown to which the control operation method and apparatus of the autonomous driving vehicle of the present application can be applied.

[0036] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The communication connection between the terminal devices 101, 102, 103 constitutes a topological network, and the network 104 is used to provide a medium for the communication link between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0037] The terminal devices 101, 102, 103 interact with the server 105 via the network 104 to receive or send messages, etc. The terminal devices 101, 102, 103 may be hardware devices or software that support network connection for data interaction and data processing. When the terminal devices 101, 102, 103 are hardware, they may be various electronic devices that support network connection, information acquisition, interaction, display, processing and other functions, including but not limited to vehicle-mounted computers, smart phones, tablet computers, e-book readers, laptop portable computers and desktop computers, etc. When the terminal devices 101, 102, 103 are software, they may be installed in the electronic devices listed above. It may be implemented as multiple software or software modules, for example, for providing distributed services, or it may be implemented as a single software or software module. No specific limitation is made here.

[0038] Server 105 may be a server that provides various services, such as a background processing server that determines the control instructions of the autonomous driving vehicle based on the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment according to the state information of the autonomous driving vehicle obtained by terminal devices 101, 102, and 103. Optionally, the server may send the control instructions to the autonomous driving vehicle to control its operation. As an example, server 105 may be a cloud server.

[0039] It should be noted that the server can be hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or it can be implemented as a single server. When the server is software, it can be implemented as multiple software or software modules (for example, software or software modules used to provide distributed services), or it can be implemented as a single software or software module. No specific limitation is made here.

[0040] It should also be noted that the control operation method of the autonomous driving vehicle provided in the embodiment of the present application can be executed by a server, or by a terminal device, or by a server and a terminal device in cooperation with each other. Accordingly, the various parts (such as various units) included in the control operation device of the autonomous driving vehicle can be all set in the server, or all set in the terminal device, or can be set in the server and the terminal device respectively.

[0041] It should be understood that Figure 1 The number of terminal devices, networks, and servers in the system architecture is only illustrative. Any number of terminal devices, networks, and servers may be provided as required. When the electronic device on which the control operation method of the autonomous driving vehicle is running does not need to transmit data with other electronic devices, the system architecture may only include the electronic device (e.g., server or terminal device) on which the control operation method of the autonomous driving vehicle is running.

[0042] Continue to refer Figure 2 , a process 200 of an embodiment of a method for controlling an automatic driving vehicle is shown, comprising the following steps:

[0043] Step 201, during the operation of the autonomous driving vehicle, the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment are constrained and decomposed by a constraint decomposition algorithm.

[0044] In this embodiment, the execution subject of the control operation method of the automatic driving vehicle (for example Figure 1 The terminal device or server in the autonomous driving vehicle can determine the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment in real time during the operation of the autonomous driving vehicle, and decompose the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment through a constraint decomposition algorithm.

[0045] The objective function and the constraint conditions are used to determine the control sequence of the autonomous driving vehicle at a future time. Specifically, under the constraints of the constraint conditions, the optimal solution of the objective function is determined to determine the optimal solution as the control sequence of the autonomous driving vehicle.

[0046] As an example, the constraints may be constraints determined based on actual conditions such as the maneuverability of the autonomous driving vehicle, status information of the autonomous driving vehicle and the surrounding environment, driving safety, driving comfort, etc.

[0047] In some optional implementations of this embodiment, the constraints include: constraints on control variables that control the operation of the autonomous driving vehicle, and constraints on state variables that characterize state information of the autonomous driving vehicle after operation at a future moment.

[0048] In some optional implementations of this embodiment, the constraints include: constraints on control variables that control the operation of the autonomous driving vehicle.

[0049] In the field of autonomous driving, constraints generally include constraints corresponding to control variables and constraints corresponding to state variables; however, for some control optimization problems, only the constraints corresponding to the control variables need to be considered to control the autonomous driving vehicle.

[0050] For autonomous vehicles, constraints are often complicated and complex, which makes the objective function subject to the constraints more computationally intensive during the iterative calculation process and the convergence process of determining the optimal solution slower.

[0051] By decomposing the constraints, different types of constraints can be decomposed, so that the objective function determines the constraints according to the actual situation during the iterative calculation process, which helps to improve the calculation speed of the iterative calculation process of the objective function. As an example, the above-mentioned execution subject can decompose the constraints into equality constraints and inequality constraints.

[0052] In some optional implementations of this embodiment, the above-mentioned execution entity can perform the above-mentioned step 201 in the following manner: first, taking the variables in the constraint conditions as local variables, and adding the global variables corresponding to the local variables to the constraint conditions; then, based on the global variables and the local variables, decomposing the constraint conditions into equality constraints corresponding to the local variables and inequality constraints corresponding to the global variables; finally, characterizing the consensus constraints between the local variables and the global variables through dual variables.

[0053] As an example, consider a typical model predictive control problem under a linear discrete system, which is expressed as follows:

[0054]

[0055] stx t+1 =A t x t +B t u t , t=1,…,N (2)

[0056] x1=x init (3)

[0057]

[0058] Among them, formula (1) represents the objective function, and formulas (2), (3), and (4) represent the constraints. Specifically, N is the prediction time domain, and the state variable Control variables Weight Matrix The system state coefficient matrix and control coefficient matrix are Constraint Set

[0059] The core idea of ​​constraint decomposition model predictive control is to decompose the original quadratic programming problem into two parts, namely, the quadratic programming problem that only considers the objective function and equality constraints, and the small-scale problem that only considers the inequality constraints of the state variables and control variables in a single time domain. After decomposing the constraints, the quadratic programming problem containing the initial state constraints (3) and the equality constraints (2) caused by system dynamics can be solved by KKT (Karush-Kuhn-Tucker) conditions. The inequality constraints of the state variables and control variables in a single time domain can be calculated by projection or multi-parameter quadratic programming methods. In addition, since there is no coupling relationship between the inequality constraints in different time domains, they can be calculated in parallel to reduce the computational overhead.

[0060] In order to decouple the original problem represented by formula (1)-(4) in terms of constraints, this embodiment introduces global variables. The global variables are actually just copies of local variables (state variables and control variables). As a result, the local variables are responsible for solving sub-problems that consider the state constraints of the objective function and the initial moment and the equality constraints introduced by the system dynamics, and the global variables are responsible for solving the inequality constraints that need to be satisfied. Therefore, the above formulas (1)-(4) can be rewritten as follows:

[0061]

[0062] stx t+1 =A t x t +B t u t , t=1,…,N (6)

[0063] x1=x init (7)

[0064]

[0065]

[0066]

[0067] Among them, formula (5) represents the objective function, and formulas (6)-(10) represent the constraints. Specifically, the subscript t represents time. By introducing the global variable Decoupling the original problem is achieved, so that the local variables (x, u) are responsible for solving the sub-problems including the objective function (5) and the equality constraints (6) and (7). Responsible for solving the inequality constraint (8). Since x1 is a constant, and u N+1 The coefficient R N+1 is 0, so although the mathematical description of the problem has changed, it is still equivalent to the mathematical form (1)-(4) of the original problem. The above formulas (9) and (10) represent the consensus constraint relationship between local variables and global variables.

[0068] Continue to refer Figure 3 , showing the core idea of ​​the constraint decomposition algorithm including control variable constraints and state variable constraints.

[0069] For some control optimization problems in the field of autonomous driving, only the control variable constraints need to be considered to meet the control objectives. In this case, since the number of consensus constraints is reduced (i.e., the previous state variables and control variables need to reach consensus, it is compressed to only control variables need to reach consensus), the number of iterations required to achieve convergence is also reduced accordingly.

[0070] At this point, the original model predictive control problem (1)-(4) that only considers control constraints can be rewritten as follows:

[0071]

[0072] stx t+1 =A t x t +B t u t , t=1,…,N (6')

[0073] x1=x init (7')

[0074]

[0075]

[0076] The subscript t represents time. By introducing the variable Decoupling of the control problem that only considers the control variable constraints is achieved, so that is responsible for solving the sub-problems including the objective function and the equality constraints, and Responsible for solving inequality constraints. The above formula (9') represents the consensus constraint relationship between local variables and global variables.

[0077] Continue to refer Figure 4 , showing the core idea of ​​the constraint decomposition algorithm that only includes control variable constraints.

[0078] Step 202, using an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence.

[0079] In this embodiment, the execution subject may use an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence.

[0080] After the objective function and the constraint conditions are decomposed, the execution subject adopts an acceleration algorithm to accelerate the iterative calculation process based on the objective function after the constraint decomposition, so as to further improve the convergence speed of the iterative calculation process.

[0081] The acceleration algorithm may be any acceleration algorithm that can speed up the convergence speed of the iterative calculation process. As an example, the acceleration algorithm may be a z-acceleration method or a quasi-Newton method.

[0082] The control sequence includes multiple control instructions, and the number of the control instructions can be determined according to the prediction time domain of the objective function during the solution process.

[0083] In some optional implementations of this embodiment, the execution subject may perform step 202 in the following manner:

[0084] Firstly, the process of solving the objective function and constraint conditions after constraint decomposition by using the alternating direction multiplier method is converted into a fixed point iteration form. Then, an acceleration algorithm is used to process the iterative calculation process corresponding to the fixed point iteration form to obtain a control sequence.

[0085] The update process of the constraint decomposition algorithm is as follows:

[0086] After decomposing the constraints of the original problem by introducing global variables, the consensus optimization problem can be solved by the alternating direction multiplier method. To this end, the set is defined as follows:

[0087]

[0088] This set represents the state variable and control variable pairs that satisfy the system dynamics equation constraint (6) and the initial state constraint (7). Representing a collection The indicator function of the state variable control variable is in the set If the indicator function is within , the indicator function is 0, otherwise the value of the indicator function is +∞.

[0089] In addition, the objective function (5) is expressed as φ(x, u), and the inequality constraint (8) is expressed as With the help of the above defined formula, the mathematical expressions (5)-(10) of the above problem after the introduction of the operator splitting method can be written as follows:

[0090]

[0091]

[0092] In addition, in order to solve the consensus constraints between local variables and global variables, dual variables ω and λ are introduced:

[0093] ω t The relevant consensus constraints are: t=1,…,N+1

[0094] λ t The relevant consensus constraints are: t=1,…,N+1

[0095] At the same time, in order to keep the constraint decomposition model predictive control symbol system simple, the local variable X, the global variable z and the dual variable y are defined as follows:

[0096]

[0097]

[0098]

[0099] The above formulas (11)-(12) have been rewritten into a structure suitable for the alternating direction multiplier method to optimize the solution. Therefore, the augmented Lagrangian function form of the alternating direction multiplier method is given:

[0100]

[0101] After obtaining the augmented Lagrangian function, the local variable X is first optimized, which can be expressed as the following formula using a linear equality constrained quadratic programming problem:

[0102]

[0103] sA eq X=b eq (15)

[0104] Among them, the quadratic programming problem is expressed as follows:

[0105]

[0106]

[0107]

[0108] Among them, Q n is the weight in the weight matrix Q, R n is the weight in the weight matrix R, and I is the identity matrix.

[0109] Since the quadratic optimization problem only has equality constraints, the analytical solution can be directly obtained with the help of the matrix structure. For the above quadratic programming problem, the sufficient and necessary conditions for optimality can be expressed by the following formula:

[0110]

[0111] where v is the dual variable associated with the equality constraint. Since ρ>0, H is a positive definite matrix, and since A eq Full rank, so the KKT coefficient matrix is ​​invertible. Since each iteration needs to solve the quadratic programming problem, in which only f changes, its analytical solution with respect to f can be obtained. For the above formula (16), LDL can be used T The decomposition method is used to solve the problem, and the KKT coefficient matrix is ​​decomposed as follows:

[0112]

[0113] Among them, the matrix P is a permutation matrix, the matrix L is a lower triangular unit matrix (i.e., all elements on the main diagonal are 1), and the matrix D is a block diagonal matrix (1×1 or 2×2 diagonal blocks, and all elements except the diagonal blocks are 0). The choice of the permutation matrix P depends on the structure of the KKT matrix, in order to make the matrix L have fewer non-zero factors and make the decomposition more stable. Therefore, the above formula (16) can be solved by the following formula:

[0114]

[0115] In solving the linear equality constrained quadratic programming problem, the KKT coefficient matrix remains unchanged during each iterative optimization process, and only the first term f in the entire optimization function changes. Therefore, the matrices P, L, and D can be calculated and stored offline in advance. -1 , and then each step only needs to calculate the above formula (17).

[0116] After completing the update of the local variable X, the constraint decomposition algorithm needs to update the global variable z. The update formula is as follows:

[0117]

[0118] Since there is no coupling relationship between the constraints at different times, the N+1 sub-problems can be solved in parallel; in addition, if there is no coupling relationship between the state variable constraints and the control variable constraints, the state and control variables can be further decoupled to reduce the complexity of solving the optimization problem. The optimization of the N+1 sub-problems can be solved using a multi-parameter quadratic programming method because its dimension n+m is small, thereby reducing the computing time required for online optimization.

[0119] After completing the update optimization for the local variable X and the global variable z, the dual variable y is updated using the following formula:

[0120] y k+1 =y k +ρ(X k+1 -z k+1 ) (19)

[0121] After completing the update of each variable, the constraint decomposition algorithm needs to set a suitable iterative convergence criterion so that the variable stops iterating after meeting a certain accuracy requirement. The specific formula for the iterative termination condition of this algorithm is as follows:

[0122]

[0123]

[0124] Among them, ∈ pri ,∈ abs ,∈ rel ,∈ dual They represent the previous error variable, the absolute value of the error variable, the dual error variable, and the true error variable respectively.

[0125] Based on the update process of the above constraint decomposition, an iterative process acceleration strategy can be implemented.

[0126] Specifically, the constraint decomposition model predictive control algorithm still uses the first-order optimization method for iterative solution. Since only the gradient information of the function to be optimized can be used during the update, the algorithm converges and iterates slowly. In this implementation, an acceleration method (for example, a quasi-Newton method) will be used to accelerate the iterative convergence speed of the constraint decomposition model predictive control algorithm by extracting the information saved in the past iteration points, thereby solving the problem of slow convergence of the constraint decomposition algorithm.

[0127] First, the alternating direction multiplier method used to solve the constraint decomposition algorithm is converted into a fixed point iteration form, and then the acceleration method is applied to the fixed point iteration form to achieve the purpose of accelerating the algorithm iteration convergence speed. Since the global variable z needs to be updated to ensure that it satisfies the inequality constraint, the weighted values ​​of the first M iterative variables that satisfy the inequality constraint cannot be guaranteed to satisfy the constraint. Therefore, in order to successfully apply the acceleration algorithm, the update optimization order of the alternating direction multiplier method needs to be changed.

[0128] It is known that the standard alternating direction multiplier method first updates the local variable X, then updates the global variable z, and finally updates the dual variable y. For the constraint decomposition algorithm, it is necessary to first update the global variable z, then update the local variable X, and finally update the dual variable y. It should be noted that changing the update optimization order of the alternating direction multiplier method can also achieve the effect of accelerated convergence on some problems. Although the local variable X contains an equality constraint, since the sum of the weighted coefficients of the first M iterative variables in the accelerated algorithm is 1, the weighted value still satisfies the equality constraint. Therefore, the local variable and the dual variable can be used as the optimization variables for the fixed point iteration, which can be written in the following form using the iterative form:

[0129]

[0130] y k+1 :=y k +ρ(g(X k ,y k )-X k+1 ) (twenty two)

[0131] The function g(·) is used to calculate the global variable z, which is considered as an intermediate variable. Therefore, when the update of variables X and y does not contain constraints, the update of the alternating direction multiplier method can be expressed in the form of fixed point iteration:

[0132] (X k+1 ,y k+1 )=F ADMM (X k ,y k )

[0133] After completing the conversion of the fixed point iteration form of the alternating direction multiplier method, the accelerated constraint decomposition model predictive control algorithm is completed. Although the update of the local variable X in the constraint decomposition algorithm is obtained by centrally solving the equality constraint quadratic programming problem, since the local variable can still be separated into N+1 state control variable pairs in a physical sense, when using the accelerated algorithm, the new weighted value of each state control variable can still be obtained in parallel.

[0134] Step 203, controlling the operation of the automatic driving vehicle at the next moment through the first control instruction in the control sequence.

[0135] In this embodiment, the above-mentioned execution entity can control the operation of the automatic driving vehicle at the next moment through the first control instruction in the control sequence.

[0136] The control sequence includes multiple control instructions arranged in chronological order, and each control instruction corresponds to a future moment of the autonomous driving vehicle. In this embodiment, only the first control instruction in the control sequence is used to control the operation of the autonomous driving vehicle at the next moment, and after completing the control at the next moment, the information processing process of steps 201-203 is continued at the next moment to cyclically obtain the control instructions of the autonomous driving vehicle at each moment.

[0137] Continue to see Figure 5 , Figure 5 FIG. 5 is a schematic diagram 500 of an application scenario of the control operation method of an automatic driving vehicle according to this embodiment. Figure 5 In the application scenario, the autonomous driving vehicle 501 sends its own motion control information and surrounding environment information to the server 502 in real time. During the operation of the autonomous driving vehicle, the server first decomposes the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment through the constraint decomposition algorithm. Among them, the objective function and constraint conditions are used to determine the control sequence of the autonomous driving vehicle at the future moment. Then, the acceleration algorithm is used to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain the control sequence. Finally, the operation of the autonomous driving vehicle at the next moment is controlled by the first control instruction in the control sequence. The above control process is executed through training to complete the control of the entire operation process of the autonomous driving vehicle.

[0138] The method provided by the above-mentioned embodiments of the present application, during the operation of the autonomous driving vehicle, uses a constraint decomposition algorithm to decompose the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment, wherein the objective function and constraint conditions are used to determine the control sequence of the autonomous driving vehicle at the future moment; uses an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence; and controls the operation of the autonomous driving vehicle at the next moment through the first control instruction in the control sequence, thereby improving the information processing efficiency of obtaining the control sequence based on the objective function and constraint conditions, and meeting the online real-time requirements of the autonomous driving vehicle.

[0139] In some optional implementations of this embodiment, the above-mentioned execution entity may perform the following operations: through a prediction model, based on the actual state information and control sequence of the autonomous driving vehicle at the current moment, predict the state information sequence of the autonomous driving vehicle at a future moment.

[0140] The prediction model may adopt a neural network model with prediction function to characterize the correspondence between the actual state information of the autonomous driving vehicle at the current moment, the control sequence and the state information sequence of the autonomous driving vehicle at the future moment. For example, the prediction model may adopt a recurrent convolutional network.

[0141] In this implementation, the execution subject may perform step 202 in the following manner: based on the state information sequence at the current moment, an acceleration algorithm is used to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence.

[0142] When the constraint conditions are based on the state information sequence obtained by the prediction model, the solution process of the objective function has more accurate constraint conditions, which improves the accuracy of the control sequence obtained based on the objective function and improves the safety of the operation process of the self-driving vehicle.

[0143] In some optional implementations of this embodiment, the above-mentioned execution entity may also perform the following operations: first, determine the actual state information of the autonomous driving vehicle at the next moment after the autonomous driving vehicle runs according to the first control instruction; then, update the prediction model according to the actual state information at the next moment and the state information sequence.

[0144] As time goes by, the above-mentioned execution entity can determine the actual state information of the autonomous driving vehicle after running the first control instruction in the control sequence corresponding to the current moment when the next moment of the current moment arrives.

[0145] As an example, the state information sequence corresponding to the current moment includes the prediction results of the state information at the future moment. After the actual state information is determined at the next moment, the loss between the two can be determined, and then the prediction model can be updated according to the loss information.

[0146] In this implementation, based on the update of the prediction model, the accuracy of the prediction model is improved, thereby further improving the operating safety of the autonomous driving vehicle.

[0147] Continue to refer Figure 6 , shows a schematic process 600 of an embodiment of a control operation method of an autonomous driving vehicle according to the present application, comprising the following steps:

[0148] Step 601, during the operation of the autonomous driving vehicle, based on the state information sequence at the current moment, the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment are constrained and decomposed by a constraint decomposition algorithm.

[0149] Among them, the objective function and constraints are used to determine the control sequence of the autonomous driving vehicle at future moments.

[0150] Step 602, converting the process of solving the objective function and constraint conditions after constraint decomposition by using the alternating direction multiplier method into a fixed point iteration form.

[0151] Step 603, using an acceleration algorithm to process the iterative calculation process corresponding to the fixed point iteration form to obtain a control sequence.

[0152] Step 604, controlling the operation of the automatic driving vehicle at the next moment through the first control instruction in the control sequence.

[0153] Step 605, determining actual state information of the automatic driving vehicle at the next moment after the automatic driving vehicle runs according to the first control instruction;

[0154] Step 606, using the prediction model, based on the actual state information and control sequence of the autonomous driving vehicle at the current moment, predict the state information sequence of the autonomous driving vehicle at a future moment.

[0155] Step 607: Update the prediction model according to the actual state information and the state information sequence at the next moment.

[0156] By looping through steps 601-607, the control instructions of the autonomous driving vehicle at each moment can be determined to control the operation of the autonomous driving vehicle.

[0157] It can be seen from this embodiment that Figure 2 Compared with the corresponding embodiments, process 600 of the control operation method of the autonomous driving vehicle in this embodiment specifically illustrates the information processing process of the model predictive control algorithm and the acceleration of the iterative calculation process. On the basis of ensuring driving safety, it improves the information processing efficiency of obtaining the control sequence based on the objective function and constraints, and meets the online real-time requirements of the autonomous driving vehicle.

[0158] Continue to refer Figure 7 As an implementation of the methods shown in the above figures, the present application provides an embodiment of a control operation device for an autonomous driving vehicle, and the device embodiment is similar to Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0159] like Figure 7As shown, the control operation device of the autonomous driving vehicle includes: a constraint decomposition unit 701, which is configured to decompose the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment through a constraint decomposition algorithm during the operation of the autonomous driving vehicle, wherein the objective function and constraint conditions are used to determine the control sequence of the autonomous driving vehicle at a future moment; an acceleration unit 702, which is configured to use an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after the constraint decomposition to obtain a control sequence; and a control unit 703, which is configured to control the operation of the autonomous driving vehicle at the next moment through the first control instruction in the control sequence.

[0160] In some optional implementations of this embodiment, the acceleration unit 702 is further configured to: convert the process of solving the objective function and constraint conditions after constraint decomposition using the alternating direction multiplier method into a fixed point iteration form; use an acceleration algorithm to process the iterative calculation process corresponding to the fixed point iteration form to obtain a control sequence.

[0161] In some optional implementations of this embodiment, the constraint decomposition unit 701 is further configured to: take the variables in the constraint conditions as local variables, and add the global variables corresponding to the local variables to the constraint conditions; based on the global variables and the local variables, decompose the constraint conditions into equality constraints corresponding to the local variables and inequality constraints corresponding to the global variables; and characterize the consensus constraints between the local variables and the global variables through dual variables.

[0162] In some optional implementations of this embodiment, the constraints include: constraints on control variables that control the operation of the autonomous driving vehicle, and constraints on state variables that characterize state information of the autonomous driving vehicle after operation at a future moment.

[0163] In some optional implementations of this embodiment, the constraints include: constraints on control variables that control the operation of the autonomous driving vehicle.

[0164] In some optional implementations of this embodiment, the above-mentioned device also includes: a prediction unit (not shown in the figure), which is configured to predict the state information sequence of the autonomous driving vehicle at a future moment through a prediction model based on the actual state information and control sequence of the autonomous driving vehicle at the current moment; and an acceleration unit, which is further configured to: based on the state information sequence at the current moment, use an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence.

[0165] In some optional implementations of this embodiment, the above-mentioned device also includes: an updating unit (not shown in the figure), which is configured to: determine the actual state information of the autonomous driving vehicle at the next moment after the operation according to the first control instruction; and update the prediction model according to the actual state information at the next moment and the state information sequence.

[0166] In this embodiment, the constraint decomposition unit in the control operation device of the autonomous driving vehicle decomposes the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment through a constraint decomposition algorithm during the operation of the autonomous driving vehicle, wherein the objective function and constraint conditions are used to determine the control sequence of the autonomous driving vehicle at future moments; the acceleration unit uses an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence; the control unit controls the operation of the autonomous driving vehicle at the next moment through the first control instruction in the control sequence, thereby improving the information processing efficiency of obtaining the control sequence based on the objective function and constraint conditions, and meeting the online real-time requirements of the autonomous driving vehicle.

[0167] Reference below Figure 8 , which shows a device suitable for implementing the embodiments of the present application (eg Figure 1 Schematic diagram of the structure of a computer system 800 of the devices 101, 102, 103, 105 shown. Figure 8 The device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0168] like Figure 8 As shown, the computer system 800 includes a processor (e.g., CPU, central processing unit) 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage part 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the system 800 are also stored. The processor 801, ROM 802, and RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0169] The following components are connected to the I / O interface 805: an input section 806 including a keyboard, a mouse, etc.; an output section 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, a modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the I / O interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed, so that a computer program read therefrom is installed into the storage section 808 as needed.

[0170] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part 809, and / or installed from a removable medium 811. When the computer program is executed by the processor 801, the above-mentioned functions defined in the method of the present application are executed.

[0171] It should be noted that the computer-readable medium of the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0172] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the client computer, partially on the client computer, as a separate software package, partially on the client computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the client computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0173] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the device, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order 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 and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0174] The units involved in the embodiments described in the present application may be implemented by software or by hardware. The described units may also be arranged in a processor, for example, may be described as: a processor comprising a constraint decomposition unit, an acceleration unit and a control unit. Among them, the names of these units do not constitute a limitation on the unit itself in some cases. For example, the acceleration unit may also be described as "a unit that uses an acceleration algorithm to process the iterative calculation process of the objective function and constraint conditions after constraint decomposition to obtain a control sequence".

[0175] As another aspect, the present application also provides a computer-readable medium, which may be included in the device described in the above embodiment; or it may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by the device, the computer device: during the operation of the autonomous driving vehicle, the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment are constrained by the constraint decomposition algorithm, wherein the objective function and constraint conditions are used to determine the control sequence of the autonomous driving vehicle at the future moment; the iterative calculation process of the objective function and constraint conditions after the constraint decomposition is processed by the acceleration algorithm to obtain the control sequence; and the operation of the autonomous driving vehicle at the next moment is controlled by the first control instruction in the control sequence.

[0176] Continue to refer Fig. 9 , showing a schematic diagram 900 of the system architecture of the unmanned vehicle of the present application. In this example, the unmanned vehicle is an unmanned delivery vehicle for schematic illustration, but the type of unmanned vehicle is not limited. For example, the unmanned vehicle can also be an unmanned taxi, an unmanned bus, an unmanned inspection vehicle, or an unmanned work vehicle.

[0177] As attached Fig. 9 As shown, the unmanned vehicle 900 mainly includes four parts: chassis module 901, automatic driving module 902, cargo box module 903 and remote monitoring streaming module 904. The automatic driving module includes a core processing unit (Orin or Xavier module), a traffic light recognition camera, front and rear surround cameras, a multi-line laser radar, a positioning module (such as Beidou, GPS, etc.), and an inertial navigation unit. The camera can communicate with the automatic driving module. In order to increase the transmission speed and reduce the wiring harness, GMSL link communication can be used. The chassis module mainly includes a battery, a power management device, a chassis controller, a motor driver, and a power motor. The battery provides power for the entire unmanned vehicle system. The power management device converts the battery output into different voltage levels that can be used by each functional module and controls the power on and off. The chassis controller receives the motion instructions issued by the automatic driving module to control the unmanned vehicle to turn, move forward, move backward, brake, etc. The remote monitoring streaming module consists of a front monitoring camera, a rear monitoring camera, a left monitoring camera, a right monitoring camera and a streaming module. The module transmits the video data collected by the monitoring camera to the background server for viewing by the background operator. The wireless communication module communicates with the backend server through the antenna, and can realize the remote control of the unmanned vehicle by the backend operator. The cargo box module is the cargo carrying device of the unmanned vehicle. The cargo box module is also provided with a display interaction module, which is used for the interaction between the unmanned vehicle and the user. The user can perform operations such as picking up, storing, and purchasing goods through the display interaction module. The type of cargo box can be changed according to actual needs. For example, in a logistics scenario, the cargo box can include multiple sub-boxes of different sizes, which can be used to load goods for distribution. In a retail scenario, the cargo box can be set as a transparent box so that users can intuitively see the products for sale.

[0178] Continue to refer Fig.10 , showing a schematic diagram 1000 of the structure of the unmanned vehicle of the present application. Specifically, the core processing unit suite in the automatic driving module, such as various processors, can execute the methods shown in the above embodiments 200 and 600.

[0179] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above invention concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A method for controlling and operating an autonomous driving vehicle, comprising: During the operation of the autonomous driving vehicle, the objective function and constraint conditions corresponding to the autonomous driving vehicle at the current moment are subjected to constraint decomposition by a constraint decomposition algorithm, including: taking the variables in the constraint conditions as local variables, and adding the global variables corresponding to the local variables to the constraint conditions; based on the global variables and the local variables, decomposing the constraint conditions into equality constraints corresponding to the local variables and inequality constraints corresponding to the global variables; characterizing the consensus constraints between the local variables and the global variables by dual variables, wherein the objective function and the constraint conditions are used to determine the control sequence of the autonomous driving vehicle at a future moment; Using an acceleration algorithm to process the iterative calculation process of the objective function and the constraint conditions after the constraint decomposition to obtain the control sequence; The operation of the automatic driving vehicle at the next moment is controlled by the first control instruction in the control sequence.

2. The method according to claim 1, wherein: The iterative calculation process of the objective function and the constraint conditions after the constraint decomposition is processed by the acceleration algorithm to obtain the control sequence, including: The process of solving the objective function and constraint conditions after constraint decomposition by using the alternating direction multiplier method is converted into a fixed point iteration form; The acceleration algorithm is used to process the iterative calculation process corresponding to the fixed point iteration form to obtain the control sequence.

3. The method according to claim 1, wherein: The constraints include: constraints on control variables that control the operation of the autonomous driving vehicle, and constraints on state variables that characterize state information of the autonomous driving vehicle after operation at a future moment.

4. The method according to claim 1, wherein: The constraints include: constraints on control variables that control the operation of the autonomous driving vehicle.

5. The method according to claim 3, wherein: Also includes: Predicting, by means of a prediction model, a state information sequence of the autonomous driving vehicle at a future moment according to the actual state information of the autonomous driving vehicle at a current moment and the control sequence; as well as The iterative calculation process of the objective function and the constraint conditions after the constraint decomposition is processed by the acceleration algorithm to obtain the control sequence, including: Based on the state information sequence at the current moment, the acceleration algorithm is used to process the iterative calculation process of the objective function and the constraint conditions after the constraint decomposition to obtain the control sequence.

6. The method according to claim 5, wherein: Also includes: Determining actual state information of the automatic driving vehicle at a next moment after the automatic driving vehicle operates according to the first control instruction; The prediction model is updated according to the actual state information at the next moment and the state information sequence.

7. A control and operation device for an autonomous driving vehicle, comprising: A constraint decomposition unit is configured to perform constraint decomposition on an objective function and constraint conditions corresponding to the autonomous driving vehicle at a current moment by means of a constraint decomposition algorithm during operation of the autonomous driving vehicle, including: taking variables in the constraint conditions as local variables, and adding global variables corresponding to the local variables to the constraint conditions; decomposing the constraint conditions into equality constraints corresponding to the local variables and inequality constraints corresponding to the global variables based on the global variables and the local variables; and characterizing the consensus constraints between the local variables and the global variables by means of dual variables, wherein the objective function and the constraint conditions are used to determine a control sequence of the autonomous driving vehicle at a future moment; An acceleration unit is configured to use an acceleration algorithm to process the iterative calculation process of the objective function and the constraint conditions after the constraint decomposition to obtain the control sequence; A control unit is configured to control the operation of the autonomous driving vehicle at the next moment through the first control instruction in the control sequence.

8. A computer readable medium having a computer program stored thereon, wherein: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. An electronic device, comprising: one or more processors; a storage device having one or more programs stored thereon, 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 6.

10. An unmanned vehicle, comprising: one or more processors; a storage device having one or more programs stored thereon, 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 6.

Images