A continuous space driving interactive decision-making method, device, equipment and medium

Abstract

This application relates to the field of autonomous driving technology. It discloses a continuous spatial driving interaction decision-making method, apparatus, device, and medium. The method includes: acquiring other vehicle operation data and the vehicle's own operation data; determining the other vehicle's acceleration based on the other vehicle's operation data; using the other vehicle's acceleration as a feature quantity; constructing a driving style evaluation rule based on the Naive Bayes principle and likelihood function to determine the other vehicle's driving style; and calculating the vehicle's longitudinal acceleration based on the other vehicle's driving style, its operation data, and the vehicle's own operation data. This application uses the other vehicle's acceleration as a feature quantity, constructs a driving style evaluation rule based on the Naive Bayes principle and likelihood function to determine the other vehicle's driving style, and calculates the vehicle's longitudinal acceleration based on the other vehicle's driving style, its operation data, and the vehicle's own operation data. This allows for accurate prediction of the other vehicle's driving style within a local planning window, making the driving style during the driving process deterministic, reducing the solution dimensionality, and facilitating engineering implementation.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and more specifically, to a continuous spatial driving interaction decision-making method, device, equipment, and medium. Background Technology

[0002] Fully automated driving systems are an effective means of solving the problem of driving difficulties. At present, domestic and foreign OEMs and parts suppliers focus on the comfort and safety of fully automated driving systems, while intelligence, which is the core element for fully automated driving systems to move from assisted driving to highly automated driving, has not yet received enough attention from domestic and foreign OEMs and parts suppliers. Furthermore, the problem of interactive decision-making in uncertain scenarios such as other vehicles crossing the road, which fully embodies intelligence, has not yet been well resolved.

[0003] Fully automated driving systems are a crucial component in solving the last-mile problem of autonomous driving, and driving interaction decision-making strategies are the core technology for fully automated driving systems to move from automation to intelligence. The driving style of a vehicle during the driving interaction decision-making process is uncertain due to environmental traversal. Driving decision-making strategies that consider this uncertainty typically require solving high-dimensional optimization problems, resulting in a high computational load and hindering their engineering implementation. Summary of the Invention

[0004] In view of the above situation, embodiments of this application provide a continuous space driving interaction decision-making method, device, equipment and medium, which aims to solve the above problems or at least partially solve the above problems.

[0005] In a first aspect, this application provides a continuous spatial driving interaction decision-making method, including:

[0006] Obtain operational data from other vehicles and your own vehicle;

[0007] The acceleration of another vehicle is determined based on its operating data. Using the acceleration of another vehicle as a feature, a driving style evaluation rule is constructed based on the Naive Bayes principle and the likelihood function to determine the driving style of another vehicle.

[0008] Calculate the longitudinal acceleration of the vehicle based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the vehicle itself.

[0009] Based on the other vehicle's operating data and the vehicle's operating data, the vehicle's lateral path curve is calculated.

[0010] The longitudinal velocity curve of the vehicle is calculated based on the longitudinal acceleration of the vehicle and the lateral path curve.

[0011] The vehicle's driving trajectory is calculated based on the vehicle's longitudinal speed curve and the vehicle's lateral path curve.

[0012] Secondly, this application provides a continuous space driving interactive decision-making device, comprising:

[0013] The acquisition module is used to acquire operating data from other vehicles and the vehicle itself.

[0014] The driving style prediction module is used to determine the acceleration of other vehicles based on the other vehicle's operating data, and to construct driving style evaluation rules based on the Naive Bayes principle and likelihood function using the other vehicle's acceleration as a feature quantity to judge the driving style of other vehicles.

[0015] The longitudinal interaction decision module is used to calculate the longitudinal acceleration of the vehicle based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the vehicle itself.

[0016] The lateral path planning module is used to calculate the lateral path curve of the vehicle based on the operation data of other vehicles and the operation data of the vehicle itself.

[0017] The longitudinal velocity planning module is used to calculate the longitudinal velocity curve of the vehicle based on the longitudinal acceleration of the vehicle and the lateral path curve.

[0018] The driving trajectory planning module is used to calculate the driving trajectory of the vehicle based on the vehicle's longitudinal speed curve and the vehicle's lateral path curve.

[0019] Thirdly, this application provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the continuous space driving interaction decision-making method as described above.

[0020] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the continuous space driving interaction decision-making method described above.

[0021] The above-described technical solutions adopted in the embodiments of this application can achieve the following beneficial effects:

[0022] This application uses the acceleration of other vehicles as a feature quantity, constructs driving style evaluation rules based on the Naive Bayes principle and likelihood function, judges the driving style of other vehicles, and calculates the longitudinal acceleration of the own vehicle based on the driving style of other vehicles, the driving data of other vehicles and the driving data of the own vehicle. It can accurately predict the driving style of other vehicles within the local planning window, making the driving style in the driving process deterministic, reducing the solution dimension, and facilitating engineering implementation. Attached Figure Description

[0023] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0024] Figure 1 This is a schematic diagram of an application environment for the continuous space driving interaction decision-making method in one embodiment of the present invention;

[0025] Figure 2 This is a flowchart illustrating a continuous spatial driving interaction decision-making method according to an embodiment of the present invention;

[0026] Figure 3 This is a likelihood function of different driving styles of other vehicles in one embodiment of the present invention;

[0027] Figure 4 yes Figure 2 A schematic diagram of a specific implementation method for step S30;

[0028] Figure 5 This is a schematic diagram of the driving interaction decision-making process in one embodiment of the present invention;

[0029] Figure 6 This is a schematic diagram of the driving interaction decision abstraction process in one embodiment of the present invention;

[0030] Figure 7 This is a schematic diagram of a continuous space driving interaction decision-making device according to an embodiment of the present invention;

[0031] Figure 8 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention;

[0032] Figure 9 This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0034] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such use can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the term "comprising" and its variations should be interpreted as open-ended terms meaning "including but not limited to."

[0035] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0036] As mentioned earlier, the uncertainty of driving style often necessitates solving high-dimensional optimization problems for driving decision-making strategies, resulting in a high computational load and hindering their engineering implementation. To address this technical issue, embodiments of this application provide a continuous spatial driving interaction decision-making method.

[0037] The continuous spatial driving interaction decision-making method provided in this embodiment of the invention can be applied to, for example... Figure 1 In this application environment, the device communicates with the server via a network. The server can obtain the driving data of other vehicles and its own vehicle from the device. Based on the driving data of other vehicles, it determines the acceleration of other vehicles. Using the acceleration of other vehicles as a feature, it constructs driving style evaluation rules based on the Naive Bayes principle and likelihood function to determine the driving style of other vehicles. Based on the driving style of other vehicles, the driving data of other vehicles, and the driving data of its own vehicle, it calculates the longitudinal acceleration of its own vehicle. Based on the driving data of other vehicles and its own vehicle, it calculates the lateral path curve of its own vehicle. Based on the longitudinal acceleration and the lateral path curve of its own vehicle, it calculates the longitudinal velocity curve of its own vehicle. Based on the longitudinal velocity curve and the lateral path curve of its own vehicle, it calculates the driving trajectory of its own vehicle. This application uses the acceleration of other vehicles as a feature quantity, constructs driving style evaluation rules based on the Naive Bayes principle and likelihood function, determines the driving style of other vehicles, and calculates the longitudinal acceleration of the own vehicle based on the driving style of other vehicles, their operating data, and the driving data of the own vehicle. This enables accurate prediction of the driving style of other vehicles within a local planning window, making the driving style during the driving process deterministic, reducing the dimensionality of the solution, and facilitating engineering implementation. The continuous space driving interaction decision-making method of this application has the characteristics of "easy to adapt, easy to maintain, easy to expand, and easy to upgrade," and can be used to create a fully automated driving system that balances safety, comfort, and intelligence.

[0038] The device side can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server side can be implemented using a standalone server or a server cluster consisting of multiple servers. The invention will now be described in detail through specific embodiments.

[0039] Please see Figure 2 As shown, Figure 2 A flowchart illustrating the continuous space driving interaction decision-making method provided in this embodiment of the invention includes the following steps:

[0040] S10: Obtain the operating data of other vehicles and the operating data of your own vehicle.

[0041] It is understandable that, such as Figure 2 As shown, the vehicles include other vehicles and the vehicle itself. Various sensors on the vehicle acquire operational data from both other vehicles and the vehicle itself, taking environmental factors into account. This includes the vehicle's longitudinal position. Longitudinal velocity and longitudinal acceleration The longitudinal position of his car Longitudinal velocity and longitudinal acceleration longitudinal acceleration of the vehicle And his car's longitudinal acceleration wait.

[0042] S20: Determine the acceleration of the other vehicle based on the other vehicle's operating data, use the other vehicle's acceleration as a feature quantity, construct a driving style evaluation rule based on the Naive Bayes principle and the likelihood function, and judge the driving style of the other vehicle.

[0043] Specifically, the Naive Bayes principle can fully integrate driving experience information and real-time observation information to achieve accurate prediction of the driving style of other vehicles.

[0044] Specifically, step S20 includes:

[0045] S21: Pre-set the prior probabilities of various driving styles of other vehicles and the likelihood functions of various driving styles of other vehicles.

[0046] Specifically, based on past driving experience information, a prior probability p(R) of another vehicle exhibiting an aggressive driving style and a prior probability p(C) of a normal driving style are pre-set within the local planning window. For example... Figure 3 As shown, the likelihood function for setting the driving style of another car as an aggressive driving style is p(f v |R), the likelihood function for his driving style being a normal driving style is p(f v |C), representing the known relationship between another driver's driving style and a certain inherent attribute, assuming p(f) v |R)=1-p(f v |C) follows a Gaussian distribution, and the Gaussian distribution parameters can be obtained using maximum likelihood estimation.

[0047] S22: Based on the Naive Bayes principle, using the acceleration of the other vehicle as a feature, and according to the prior probabilities of the other vehicle's various driving styles and the likelihood functions of the other vehicle's various driving styles, the probability of the other vehicle's driving style within the local planning window is calculated.

[0048] Specifically, the acceleration of the other vehicle is used as the characteristic quantity f. v =[a o Based on the Naive Bayes principle, the probability of another car's driving style within the local programming window is:

[0049]

[0050] In the formula, p(R|f v p(R) is the probability of another car's driving style within the local planning window; p(C) is the prior probability of another car's aggressive driving style within the local planning window; p(F) is the prior probability of another car's normal driving style within the local planning window. In this implementation, it is assumed that p(R) = p(C) = 0.5; p(f) = p(C) = 0.5. v |R) is the likelihood function for a driver with an aggressive driving style; p(f) v |C) is the likelihood function for a driver whose driving style is normal; p(f v ) represents the total probability.

[0051] S23: Based on the probability of the other vehicle's driving style within the local planning window and the likelihood function of each of the other vehicle's driving styles, determine whether the other vehicle's driving style is an aggressive driving style or a normal driving style.

[0052] If the probability value of his car's driving style within the local planning window is located in the likelihood function of the aggressive driving style, then his car's driving style is determined to be the aggressive driving style; if the probability value of his car's driving style within the local planning window is located in the likelihood function of the normal driving style, then his car's driving style is determined to be the normal driving style.

[0053] S30: Calculate the longitudinal acceleration of the vehicle based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the vehicle itself.

[0054] This application utilizes the Naive Bayes principle to fully integrate driving experience information and real-time observation information to accurately predict the driving style of other vehicles within a local planning window. This allows for the reasonable allocation of acceleration thresholds during the driving interaction decision-making process. Specifically, vehicles with aggressive driving styles are configured with larger acceleration thresholds during the driving interaction decision-making process, while vehicles with normal driving styles are configured with relatively smaller acceleration thresholds.

[0055] Specifically, when it is determined that the driving style of another vehicle is aggressive, a first threshold is determined when calculating the longitudinal acceleration of the vehicle; when it is determined that the driving style of another vehicle is normal, a second threshold is determined when calculating the longitudinal acceleration of the vehicle; the first threshold is greater than the second threshold.

[0056] Specifically, such as Figure 4 As shown, step S30 calculates the longitudinal acceleration of the vehicle based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the vehicle itself, including:

[0057] S31: Based on the driving style of the other vehicle, the operation data of the other vehicle and the operation data of the own vehicle, establish an interactive decision-making problem on uncertainties in the driving process when the other vehicle crosses the scene, which includes uncertainties in system state variables, uncertainties in vehicle obstacle avoidance and chance constraints in system state variables.

[0058] Using driving path length to analyze driving interaction decision-making processes (such as...) Figure 5 (as shown) is transformed into a driving interaction decision abstraction process (such as...) Figure 6 As shown in the figure, only the longitudinal solution space of driving needs to be considered in the driving interaction decision process, which simplifies the modeling and solving of the driving interaction decision problem.

[0059] Based on the longitudinal position of the vehicle longitudinal speed of the vehicle longitudinal acceleration of the vehicle The longitudinal position of his car His car's longitudinal speed His car's longitudinal acceleration The system state vector is defined as follows: With longitudinal acceleration of the vehicle And his car's longitudinal acceleration The control vector is defined as follows: Based on the longitudinal position of the vehicle longitudinal speed of the vehicle The longitudinal position of his car His car's longitudinal speed The observation vector is defined as follows: The discretized system state equation and observation equation can then be expressed as:

[0060]

[0061] In the formula, x k Let Ax be the system state vector at time k. k-1 Let u be the system state vector at time k-1; k-1 z is the control vector at time k-1; k Let ω be the observation vector at time k;k-1 The system process noise at time k-1 follows a Gaussian distribution; υ k Let Q be the observation noise at time k that follows a Gaussian distribution; let Q be the system process noise covariance matrix; let R be the system observation noise covariance matrix; the system state matrix A, control matrix B, and observation matrix C can be expressed as follows:

[0062]

[0063] Define the optimization objective as

[0064]

[0065] In the formula, W x W is the system state vector weight matrix; u The control vector weight matrix; x k Let x be the system state vector at time k; k,ref Let x be the system reference state vector at time k; N x is the system state vector in the Nth step of the calculation. N,ref u is the system reference state vector in the Nth step of the calculation. k l is the control vector at time k; N (b N ) represents the terminal state weight item; l k (b k ,u k () represents the process weighting item.

[0066] Considering the system state variables, the longitudinal position of the vehicle And the longitudinal position of his car The uncertainty of vehicle obstacle avoidance is as follows: The uncertainty conditions of vehicle obstacle avoidance are transformed into deterministic conditions using the expectation model, i.e.

[0067]

[0068] In the formula, x k Let k be the system state vector at time k; P is the mean of the system state vector at time k; k Let ρ be the covariance matrix at time k; m The safe distance between your vehicle and other vehicles; ρ min W represents the minimum safe distance between the vehicle and other vehicles; W is the obstacle avoidance weight matrix; the weight matrix W... b It can be represented as:

[0069]

[0070] Considering the uncertainty of the system state variables, the system state variable constraints are constructed using a chance constraint model.

[0071]

[0072] In the formula, x k,max (i) represents the upper boundary of the system state vector; x k,min (i) represents the lower boundary of the system state vector; p x (i) represents the probability that the system state vector satisfies the boundary constraints. i is the index of an element in the state vector.

[0073] It should be noted that x k,max (i) represents the upper boundary of the longitudinal acceleration threshold determined based on the driving style of other vehicles; x k,min (i) represents the lower boundary of the longitudinal acceleration threshold determined based on the driving style of other vehicles.

[0074] Considering system control constraints,

[0075]

[0076] In the formula, u k,max (i) represents the upper boundary of the system control vector; u k,min (i) represents the lower boundary of the system control vector. Here, i is the element index of the control vector.

[0077] In summary, based on equations (2), (3), (4), (5), and (6), the uncertain interactive decision-making problem in the scenario of another vehicle crossing the road during the driving process is established as follows:

[0078]

[0079] S32: The derivation process of the uncertainty of the system state variables is described by the Kalman filter method; the uncertainty condition of vehicle obstacle avoidance is transformed into the deterministic condition of vehicle obstacle avoidance by the expectation model; the chance constraint condition of the uncertainty of the system state variables is transformed into the deterministic constraint condition of the system state variables by the one-sided Chebyshev inequality; and the uncertainty interaction decision problem of other vehicles crossing the scene during the driving process is transformed into the deterministic interaction decision problem of other vehicles crossing the scene during the driving process.

[0080] Assuming the initial confidence level b(x0) is Gaussian distributed, any linear transformation of the Gaussian random variable will result in another Gaussian random variable. Therefore, the confidence level b(x0) at any time k-1 is... k-1 It follows a Gaussian distribution and can be expressed as: Furthermore, the confidence level at time k can be recursively solved based on the Bayesian criterion. Then the mean of the system state vector at time k is obtained. The covariance matrix P at time k k for:

[0081]

[0082] In the formula, Bu is the mean of the system state vector at time k-1; k-1 z is the control vector at time k-1; k P is the observation vector at time k; k-1 Let K be the covariance matrix at time k-1; k I is the gain matrix; A is the identity matrix; B is the system state matrix; C is the observation matrix; Q is the system process noise covariance matrix; and R is the system observation noise covariance matrix.

[0083] Considering the unknowns in the observation information during the deduction process, the observation information is approximated using the maximum likelihood of the observation equation, i.e.: Therefore, equation (8) can be simplified to

[0084]

[0085] Consider E[x] T W[x] = E[x] T WE[x]+tr[WVar[x]], then the optimization objective described by equation (3) can be transformed into

[0086]

[0087] In the formula, x is the system state vector; This is the mean of the system state vector in the Nth step of the calculation; W is the mean of the system state vector at time k; x P is the system state vector weight matrix; N To calculate the covariance matrix in the Nth step; P k Let x be the covariance matrix at time k; N,ref x is the system reference state vector in the Nth step of the calculation; k,ref Let u be the system reference state vector at time k; k W is the control vector at time k; W is the obstacle avoidance weight matrix; W u This is the control vector weight matrix.

[0088] Consider tr[W x P N ] and tr[W x P k If the optimization objective described by equation (10) is unrelated to the system control quantity, then it can be simplified to:

[0089]

[0090] The system state vector x at time k k (i) Decompose into definite parts and the uncertain part ek (i), that is: Combining the chance constraint inequality (5), we can obtain

[0091]

[0092] In the formula, x k,max (i) represents the upper boundary of the system state vector; x k,min (i) represents the lower boundary of the system state vector.

[0093] Because of e k (i)~N(0,P) k (ii)), -e k (i)~N(0,P) k (ii) Using the one-sided Chebyshev inequality X~N(0,σ 2 If a > 0, then the chance constraint inequality can be obtained.

[0094]

[0095] In the formula, p x (i) represents the probability that the system state variables satisfy the boundary constraints; α k (i) and β k (i) can be expressed as

[0096]

[0097] From the chance constraint inequality (13), we can obtain

[0098]

[0099] In the formula, Let x be the system state vector at time k. k (i) is a definite part; e k (i) is the system state vector x at time k. k The uncertain part of (i); p x (i) represents the probability that the system state variables satisfy the boundary constraints.

[0100] Combining the chance constraint inequality (13) and the deterministic constraint inequality (16), it can be seen that the necessary and sufficient condition for the chance constraint inequality (5) to hold is that the deterministic constraint inequality (16) holds. Therefore, the chance constraint conditions of the system state variables are transformed into deterministic constraint conditions of the system state variables:

[0101]

[0102] In the formula, Let x be the system state vector at time k. k (i) is a definite part; xk,max (i) represents the upper boundary of the system state vector; x k,min (i) represents the lower boundary of the system state vector.

[0103] Further rearrangement of the constraint inequality (16) yields:

[0104]

[0105] Using equations (9), (11), (4), (17), and (6), the uncertain interactive decision-making problem of the driving process in which another vehicle crosses the scene, as described in equation (7), is transformed into a deterministic interactive decision-making problem of the driving process in which another vehicle crosses the scene:

[0106]

[0107] Further abstracting the deterministic interactive decision-making problem of the driving process described by equation (18) involving other vehicles crossing the scene, we can obtain the following nonlinear constrained optimization problem:

[0108]

[0109] In the formula, , is the mean of the optimal system state vector; u * The optimal control vector; + represents the penalty term for the terminal state vector; + represents the penalty term for the process state vector; ) represents the penalty term for the control vector; The mean of the system state vector at time k+1; , is the system equation; For system state vector constraints; h k (u k ) represents the control vector constraint; The mean of the system state vector at the initial moment; The initial value of the state vector.

[0110] Using the inequality constraints in the following obstacle function expression (19), we have

[0111]

[0112] In the formula, The obstacle function is a state vector constraint. This is a system state vector constraint.

[0113]

[0114] In the formula, d k (u k ) represents the obstacle function for the control vector constraint; h k(u k ) represents the control vector constraint.

[0115] When the inequalities in equation (19) satisfy the constraints, the function values ​​of equations (20) and (21) are 0; when the inequalities in equation (19) do not satisfy the constraints, the function values ​​of equations (20) and (21) are positive infinity. Therefore, by using equations (20) and (21), the nonlinear constrained optimization problem described by equation (19) can be transformed into...

[0116]

[0117] In the formula: t is the first parameter of the outer loop, which gradually increases with the number of iterations of the outer loop; , is the mean of the optimal system state vector; u * The optimal control vector; + represents the penalty term for the terminal state vector; + represents the penalty term for the process state vector; ) represents the penalty term for the control vector; The mean of the system state vector at time k+1; , is the system equation; The mean of the system state vector at the initial moment; The initial values ​​for the state vector; The obstacle function is the state vector constraint; d k (u k ) represents the obstacle function for the control vector constraint; + represents the obstacle function for the terminal state vector constraint.

[0118] Furthermore, the nonlinear constrained optimization problem described by equation (22) can be simplified as follows:

[0119]

[0120] In the formula, and M k (u k ) represents

[0121]

[0122]

[0123]

[0124] S33: Solve the deterministic interactive decision-making problem of the other vehicle crossing the scene during the driving process, and calculate the longitudinal acceleration of the vehicle.

[0125] Specifically, including:

[0126] S331: Initialize solver parameters t = 1;

[0127] S332: Update the deterministic interactive decision problem of the driving process other vehicles crossing the scene according to the initialized solver parameters, and transform the updated deterministic interactive decision problem of the driving process other vehicles crossing the scene using Taylor expansion in a preset domain.

[0128] In this embodiment, the deterministic interactive decision-making problem of the driving process other vehicle crossing scenario (the nonlinear constrained optimization problem described by equation (23)) is updated according to the initialized solver parameters, based on... and For equation (23) and M k (u k )exist Performing a second-order Taylor expansion within the domain yields the following results:

[0129]

[0130] In the formula, δx N This is the increment of the terminal state vector; This is the mean of the system state vector in the Nth step of the calculation; For terminal state vector reference; It is the first derivative of the function; It is the second derivative of the function.

[0131]

[0132] In the formula, x Let δx be the mean of the system state vector at time k; k Let k be the change in the system state vector at time k; For process state vector reference; It is the first derivative of the function; δ is the second derivative of the function.

[0133]

[0134] In the formula, u k = is the control vector at time k; δu k ) represents the change in the control vector; For control vector reference; It is the first derivative of the function; δ is the second derivative of the function.

[0135] based on For the system equation in equation (23) exist A first-order Taylor expansion within the domain yields:

[0136]

[0137] Substituting equations (27) to (29) into equation (23), the nonlinear constrained optimization problem described by equation (23) can be transformed into:

[0138]

[0139] In the formula, δx k+1 This represents the change in the system control vector at time k+1. The mean of the system control vector at time k+1; The process state vector reference is at time k+1.

[0140] The value function at step N is:

[0141]

[0142] In the formula, () is the first derivative of the function with respect to the state vector; Let be the first derivative of the function with respect to the control vector.

[0143] The action-value function for step N-1 is:

[0144]

[0145] use Transform the action-value function of step N-1 into

[0146]

[0147] Minimizing the action-value function described by equation (34) yields the control law for the (N-1)th step.

[0148]

[0149] Substituting equation (35) into equation (34), we obtain the value function for the (N-1)th step as V(δx). N-1 Meanwhile, the action-value function Q(δx) for the N-2th step is constructed using the patterns described by equations (33) and (34). N-2 ,δu N-2 Minimize Q(δx) N-2 ,δu N-2 The control law for step N-2 is obtained. This process continues iteratively from back to front, obtaining the control law and value function for each step; this process is called the Backward process. Subsequently, starting from the initial state of the system, the control law and derivation process obtained from the Backward process are used to derive each step. The process of iteratively calculating the system state variables at each step from front to back is called the Forward process.

[0150] S333: Based on the preset Backward process, solve the deterministic interactive decision-making problem of the transformed driving process in the scenario where another vehicle crosses the road, and calculate the change in control vector δu. k ;

[0151] S334: Based on the preset Forward process and the control vector change δu k The deterministic interactive decision-making problem of the transformed driving process involving other vehicles crossing the scene is solved, and the change in the system state vector δx is calculated. k ;

[0152] S335: Based on the system state vector change δx k The longitudinal acceleration of the vehicle is calculated.

[0153] The method further includes:

[0154] The change in the system state vector is compared with a preset first parameter to determine whether the change in the system state vector satisfies the inner-loop stopping criterion. Specifically, the first parameter is the maximum value δx of the change in the system state vector. max The inner-loop stopping criterion is that the absolute value of the change in the system state vector is no greater than the first parameter, i.e., |δx|. k |≤δx max .

[0155] When the change in the system state vector satisfies the inner loop stopping criterion, the first parameter of the outer loop is obtained, and the first parameter of the outer loop is compared with a preset second parameter to determine whether the first parameter of the outer loop satisfies the outer loop stopping criterion. Specifically, the first parameter of the outer loop is t, and the second parameter is the maximum value t of the first parameter of the outer loop. max The outer loop stopping criterion is that the first parameter of the outer loop is not less than the second parameter, i.e., t ≥ t. max .

[0156] When the first parameter of the outer loop satisfies the outer loop stopping criterion, the longitudinal acceleration of the vehicle is calculated based on the change in the system state vector. Specifically, when the change in the system state vector satisfies both the inner loop stopping criterion and the first parameter of the outer loop, the longitudinal acceleration of the vehicle is calculated based on the change in the system state vector. and Calculate the mean of the system state vector. Further based on Obtain the longitudinal acceleration of the vehicle

[0157] The method further includes:

[0158] When determining the change δx of the system state vector k If the inner loop stopping criterion is not met, or if the first parameter t of the outer loop does not meet the outer loop stopping criterion, update the preset domain. Within the updated preset domain, Taylor expansion is used to transform the updated deterministic interactive decision-making problem of the driving process and the other vehicle crossing the scene into a deterministic interactive problem until the change in the system state vector satisfies the inner loop stopping criterion and the first parameter of the outer loop satisfies the outer loop stopping criterion. The longitudinal acceleration of the vehicle is then calculated based on the change in the system state vector.

[0159] Specifically, S33: Solve the deterministic interactive decision-making problem of the other vehicle crossing the scene during the driving process, and calculate the longitudinal acceleration of the vehicle, including:

[0160] Step 1: Initialize the solver parameters;

[0161] Step 2 updates the nonlinear constraint optimization problem using equation (23), and in Within the field, Taylor expansion is used to transform the updated nonlinear constrained optimization problem into the form described by equation (31);

[0162] Step 3: Calculate the system state vector change δu iteratively from back to front using the Backward process. k ;

[0163] Step 4: Calculate the system state vector change δx iteratively from front to back using the forward process. k ,renew If the system state vector change δx k Satisfying the inner loop stopping criterion |δx k |≤δx max If not, proceed to Step 5; otherwise, proceed to the inner loop entry point Step 2.

[0164] Step 5: Extract the first parameter t of the outer loop. If the first parameter t of the outer loop satisfies the outer loop stopping criterion t≥t max Exit and calculate the longitudinal acceleration of the vehicle; otherwise, increase t and jump to the outer circulation entry Step 2.

[0165] S40: Calculate the lateral path curve of the vehicle based on the other vehicle's operating data and the vehicle's operating data.

[0166] Specifically, in this embodiment, the lateral path curve of the vehicle is calculated based on the other vehicle's operating data, the vehicle's operating data, and environmental data.

[0167] S50: Calculate the longitudinal velocity curve of the vehicle based on the longitudinal acceleration of the vehicle and the lateral path curve.

[0168] Specifically, in this embodiment, the longitudinal acceleration of the vehicle is used as one of the constraints, and the longitudinal velocity curve of the vehicle is calculated by combining it with the lateral path curve.

[0169] S60: Calculate the driving trajectory of the vehicle based on the longitudinal speed curve and the lateral path curve of the vehicle.

[0170] Specifically, in this embodiment, the vehicle chassis system is controlled to accurately track the longitudinal speed curve and lateral path curve of the vehicle based on the vehicle's driving trajectory, thereby achieving autonomous driving.

[0171] This application uses the acceleration of other vehicles as a feature quantity, constructs driving style evaluation rules based on the Naive Bayes principle and likelihood function, judges the driving style of other vehicles, and calculates the longitudinal acceleration of the own vehicle based on the driving style of other vehicles, the driving data of other vehicles and the driving data of the own vehicle. It can accurately predict the driving style of other vehicles within the local planning window, making the driving style in the driving process deterministic, reducing the solution dimension, and facilitating engineering implementation.

[0172] It is understood that the continuous spatial driving interaction decision-making method of this application can be directly applied to fully automatic parking systems, promoting the transition of fully automatic parking systems from automation to intelligence. Furthermore, the continuous spatial driving interaction decision-making method of this application has universality; with appropriate parameter adjustments, it can solve the interactive decision-making problems of autonomous valet parking and urban navigation-assisted driving. Therefore, the fully automatic parking interactive decision-making strategy, the autonomous valet parking interactive decision-making strategy, and the urban navigation-assisted driving interactive decision-making strategy implemented using this method all fall within the scope of protection of this invention patent.

[0173] As can be seen, in the above-mentioned scheme, the continuous spatial driving interaction decision-making method of this application uses the acceleration of other vehicles to predict their driving style, sets different acceleration thresholds when calculating the longitudinal acceleration of the vehicle based on the driving style of other vehicles, calculates the longitudinal velocity curve of the vehicle based on the longitudinal acceleration and the lateral path curve of the vehicle, calculates the driving trajectory of the vehicle based on the lateral path curve and the longitudinal velocity curve of the vehicle, and controls the vehicle chassis system to accurately track the longitudinal velocity curve and the lateral path curve based on the driving trajectory of the vehicle, thereby achieving autonomous driving. Therefore, the continuous spatial parking interaction decision-making method of this application has the characteristics of "easy to adapt, easy to maintain, easy to expand, and easy to upgrade", and can be used to create a fully automatic parking system that takes into account safety, comfort, and intelligence.

[0174] Furthermore, the continuous space parking interaction decision-making method of this application establishes an uncertain interaction decision-making problem in the driving process involving other vehicles crossing the scene, which includes uncertainties in system state variables, uncertainties in vehicle obstacle avoidance, and chance constraints in system state variables. It uses Kalman filtering to describe the derivation process of uncertainties in system state variables, uses the expectation model to transform uncertainties in vehicle obstacle avoidance into deterministic conditions in vehicle obstacle avoidance, and uses the one-sided Chebyshev inequality to transform chance constraints in system state variables into deterministic constraints in system state variables. This transforms the uncertain interaction decision-making problem in the driving process involving other vehicles crossing the scene into a nonlinear constraint optimization problem for solution, thereby suppressing the impact of uncertainties in system state information and driving style on the driving interaction decision-making process. Moreover, this method can quickly solve for the optimal decision strategy in continuous space, which is convenient for engineering implementation.

[0175] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0176] In one embodiment, a continuous space driving interaction decision-making device for uncertainty is provided, which corresponds one-to-one with the continuous space driving interaction decision-making method in the above embodiments. For example... Figure 7 As shown, the continuous space driving interaction decision-making device includes an acquisition module 101, a driving style prediction module 102, a longitudinal interaction decision-making module 103, a lateral path planning module 104, a longitudinal speed planning module 105, and a driving trajectory planning module 106. Detailed descriptions of each functional module are as follows:

[0177] Module 101 is used to acquire the operating data of other vehicles and the operating data of the vehicle itself;

[0178] The driving style prediction module 102 is used to determine the acceleration of another vehicle based on the other vehicle's operating data, and to construct driving style evaluation rules based on the Naive Bayes principle and likelihood function using the other vehicle's acceleration as a feature quantity to judge the driving style of the other vehicle.

[0179] The longitudinal interaction decision module 103 is used to calculate the longitudinal acceleration of the vehicle based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the vehicle itself.

[0180] The lateral path planning module 104 is used to calculate and obtain the lateral path curve of the vehicle based on the other vehicle's running data and the vehicle's running data.

[0181] The longitudinal speed planning module 105 is used to calculate the longitudinal speed curve of the vehicle based on the longitudinal acceleration of the vehicle and the lateral path curve.

[0182] The driving trajectory planning module 106 is used to calculate the driving trajectory of the vehicle based on the vehicle's longitudinal speed curve and the vehicle's lateral path curve.

[0183] Specifically, module 101 is also used for:

[0184] Obtain environmental data.

[0185] Specifically, the driving style prediction module 102 is used for:

[0186] Pre-set the prior probabilities of various driving styles of other vehicles and the likelihood functions of various driving styles of other vehicles;

[0187] Based on the Naive Bayes principle, using the acceleration of the other vehicle as a feature, the probability of the other vehicle's driving style within the local planning window is calculated according to the prior probability of the other vehicle's various driving styles and the likelihood function of the other vehicle's various driving styles.

[0188] Based on the probability of the other vehicle's driving style within the local planning window and the likelihood function of each of the other vehicle's driving styles, determine whether the other vehicle's driving style is an aggressive driving style or a normal driving style.

[0189] Specifically, the vertical interactive decision-making module 103 is used for:

[0190] When it is determined that the driving style of another vehicle is an aggressive driving style, the longitudinal acceleration threshold is determined to be the first threshold when calculating the longitudinal acceleration of the own vehicle.

[0191] When it is determined that the driving style of another vehicle is a normal driving style, the longitudinal acceleration threshold is determined to be the second threshold when calculating the longitudinal acceleration of the own vehicle.

[0192] The first threshold is greater than the second threshold.

[0193] Specifically, the vertical interactive decision-making module 103 is also used for:

[0194] Based on the driving style of the other vehicle, the operation data of the other vehicle and the operation data of the own vehicle, a driving process uncertain interactive decision-making problem in the scenario of the other vehicle crossing the traverse is established, which includes uncertain conditions of system state variables, uncertain conditions of vehicle obstacle avoidance and uncertain conditions of system state variables.

[0195] The derivation process of the uncertainty of the system state variables is described by Kalman filtering method; the uncertainty condition of vehicle obstacle avoidance is transformed into the deterministic condition of vehicle obstacle avoidance by expectation model; the chance constraint condition of the uncertainty of the system state variables is transformed into the deterministic constraint condition of the system state variables by one-sided Chebyshev inequality; and the uncertainty interaction decision problem of other vehicles crossing the scene during the driving process is transformed into the deterministic interaction decision problem of other vehicles crossing the scene during the driving process.

[0196] The deterministic interactive decision-making problem of the driving process involving other vehicles crossing the scene is solved, and the longitudinal acceleration of the vehicle is calculated.

[0197] Specifically, the vertical interactive decision-making module 103 is also used for:

[0198] Initialize solver parameters;

[0199] The deterministic interactive decision problem of the driving process other vehicles crossing the scene is updated according to the initialized solver parameters, and the updated deterministic interactive decision problem of the driving process other vehicles crossing the scene is transformed by Taylor expansion in the preset domain.

[0200] Based on the preset Backward process, the deterministic interactive decision-making problem of the transformed driving process and other vehicles crossing the scene is solved, and the change in control vector is calculated.

[0201] Based on the preset Forward process and the change in the control vector, the deterministic interactive decision-making problem of the transformed driving process in the scenario of another vehicle crossing is solved, and the change in the system state vector is calculated.

[0202] The longitudinal acceleration of the vehicle is calculated based on the change in the system state vector.

[0203] Specifically, the vertical interactive decision-making module 103 is also used for:

[0204] The change in the system state vector is compared with a preset first parameter to determine whether the change in the system state vector satisfies the inner loop stopping criterion.

[0205] When it is determined that the change in the system state vector satisfies the inner loop stopping criterion, the outer loop first parameter is obtained, and the outer loop first parameter is compared with the preset second parameter to determine whether the outer loop first parameter satisfies the outer loop stopping criterion.

[0206] When it is determined that the first parameter of the outer loop meets the outer loop stopping criterion, the longitudinal acceleration of the vehicle is calculated based on the change in the system state vector.

[0207] Specifically, the vertical interactive decision-making module 103 is also used for:

[0208] When it is determined that the change in the system state vector does not meet the inner loop stopping criterion, or when it is determined that the first parameter of the outer loop does not meet the outer loop stopping criterion, the preset domain is updated, and Taylor expansion is used within the updated preset domain to transform the updated deterministic interactive decision problem of the driving process other vehicle crossing the scene.

[0209] This application uses the acceleration of other vehicles as a feature quantity, constructs driving style evaluation rules based on the Naive Bayes principle and likelihood function, judges the driving style of other vehicles, and calculates the longitudinal acceleration of the own vehicle based on the driving style of other vehicles, the driving data of other vehicles and the driving data of the own vehicle. It can accurately predict the driving style of other vehicles within the local planning window, making the driving style in the driving process deterministic, reducing the solution dimension, and facilitating engineering implementation.

[0210] The continuous spatial driving interaction decision-making device of this application uses a driving style prediction module to predict the driving style of other vehicles based on their acceleration. Vehicles with aggressive driving styles have a larger acceleration threshold configured in the longitudinal interaction decision-making module, while vehicles with moderate driving styles have a relatively smaller acceleration threshold configured in the same module. The longitudinal velocity planning module uses the output of the longitudinal interaction decision-making module as one of the constraints to plan the desired longitudinal velocity curve of the vehicle. The driving trajectory planning and control module controls the vehicle chassis system to accurately track the outputs of the longitudinal velocity planning module and the lateral path planning module, thereby achieving autonomous driving. Therefore, the continuous spatial driving interaction decision-making framework of this application has the characteristics of being "easy to adapt, easy to maintain, easy to expand, and easy to upgrade," and can be used to create a fully autonomous driving system that balances safety, comfort, and intelligence.

[0211] Furthermore, the continuous space parking interactive decision-making device of this application establishes an uncertain interactive decision-making problem in the driving process involving other vehicles crossing the scene, which includes uncertainties in system state variables, uncertainties in vehicle obstacle avoidance, and chance constraints in system state variables. It uses the Kalman filter method to describe the derivation process of uncertainties in system state variables, uses the expectation model to transform uncertainties in vehicle obstacle avoidance into deterministic conditions in vehicle obstacle avoidance, and uses the one-sided Chebyshev inequality to transform chance constraints in system state variables into deterministic constraints in system state variables. This transforms the uncertain interactive decision-making problem in the driving process involving other vehicles crossing the scene into a nonlinear constraint optimization problem for solution, thereby suppressing the influence of uncertainties in system state information and driving style on the driving interactive decision-making process. Moreover, this method can quickly solve for the optimal decision strategy in continuous space, which is convenient for engineering implementation.

[0212] Specific limitations regarding the continuous space driving interaction decision-making device can be found in the limitations of the continuous space driving interaction decision-making method described above, and will not be repeated here. Each module in the aforementioned continuous space driving interaction decision-making device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0213] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used for communication with external devices via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a continuous spatial driving interactive decision-making method on the server side.

[0214] In one embodiment, a computer device is provided, which may be a device terminal, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a continuous spatial driving interactive decision-making method on the device side.

[0215] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps:

[0216] Obtain operational data from other vehicles and your own vehicle;

[0217] The acceleration of another vehicle is determined based on its operating data. Using the acceleration of another vehicle as a feature, a driving style evaluation rule is constructed based on the Naive Bayes principle and the likelihood function to determine the driving style of another vehicle.

[0218] Calculate the longitudinal acceleration of the vehicle based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the vehicle itself.

[0219] Based on the other vehicle's operating data and the vehicle's operating data, the vehicle's lateral path curve is calculated.

[0220] The longitudinal velocity curve of the vehicle is calculated based on the longitudinal acceleration of the vehicle and the lateral path curve.

[0221] The vehicle's driving trajectory is calculated based on the vehicle's longitudinal speed curve and the vehicle's lateral path curve.

[0222] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0223] Obtain operational data from other vehicles and your own vehicle;

[0224] The acceleration of another vehicle is determined based on its operating data. Using the acceleration of another vehicle as a feature, a driving style evaluation rule is constructed based on the Naive Bayes principle and the likelihood function to determine the driving style of another vehicle.

[0225] Calculate the longitudinal acceleration of the vehicle based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the vehicle itself.

[0226] Based on the other vehicle's operating data and the vehicle's operating data, the vehicle's lateral path curve is calculated.

[0227] The longitudinal velocity curve of the vehicle is calculated based on the longitudinal acceleration of the vehicle and the lateral path curve.

[0228] The vehicle's driving trajectory is calculated based on the vehicle's longitudinal speed curve and the vehicle's lateral path curve.

[0229] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and device side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0230] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0231] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0232] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A continuous spatial driving interactive decision-making method, characterized in that, include: Obtain operational data from other vehicles and your own vehicle; The acceleration of another vehicle is determined based on its operating data. Using the acceleration of another vehicle as a feature, a driving style evaluation rule is constructed based on the Naive Bayes principle and the likelihood function to determine the driving style of another vehicle. Calculate the longitudinal acceleration of the vehicle based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the vehicle itself. Based on the other vehicle's operating data and the vehicle's operating data, the vehicle's lateral path curve is calculated. The longitudinal velocity curve of the vehicle is calculated based on the longitudinal acceleration of the vehicle and the lateral path curve. The vehicle's driving trajectory is calculated based on the vehicle's longitudinal speed curve and the vehicle's lateral path curve. Based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the own vehicle, the longitudinal acceleration of the own vehicle is calculated, including: Based on the driving style of the other vehicle, the operation data of the other vehicle and the operation data of the own vehicle, a driving process uncertain interactive decision-making problem in the scenario of the other vehicle crossing the traverse is established, which includes uncertain conditions of system state variables, uncertain conditions of vehicle obstacle avoidance and uncertain conditions of system state variables. The derivation process of the uncertainty of the system state variables is described by Kalman filtering method; the uncertainty condition of vehicle obstacle avoidance is transformed into the deterministic condition of vehicle obstacle avoidance by expectation model; the chance constraint condition of the uncertainty of the system state variables is transformed into the deterministic constraint condition of the system state variables by one-sided Chebyshev inequality; and the uncertainty interaction decision problem of other vehicles crossing the scene during the driving process is transformed into the deterministic interaction decision problem of other vehicles crossing the scene during the driving process. The deterministic interactive decision-making problem of the driving process involving other vehicles crossing the scene is solved, and the longitudinal acceleration of the vehicle is calculated.

2. The method according to claim 1, characterized in that, The step of determining the acceleration of another vehicle based on its operating data, using the acceleration as a feature quantity, and constructing a driving style evaluation rule based on the Naive Bayes principle and likelihood function to determine the driving style of another vehicle includes: Pre-set the prior probabilities of various driving styles of other vehicles and the likelihood functions of various driving styles of other vehicles; Based on the Naive Bayes principle, using the acceleration of the other vehicle as a feature, the probability of the other vehicle's driving style within the local planning window is calculated according to the prior probability of the other vehicle's various driving styles and the likelihood function of the other vehicle's various driving styles. Based on the probability of the other vehicle's driving style within the local planning window and the likelihood function of each of the other vehicle's driving styles, determine whether the other vehicle's driving style is an aggressive driving style or a normal driving style.

3. The method according to claim 1, characterized in that, The step of calculating the longitudinal acceleration of the vehicle based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the vehicle itself includes: When it is determined that the driving style of another vehicle is an aggressive driving style, the longitudinal acceleration threshold is determined to be the first threshold when calculating the longitudinal acceleration of the own vehicle. When it is determined that the driving style of another vehicle is a normal driving style, the longitudinal acceleration threshold is determined to be the second threshold when calculating the longitudinal acceleration of the own vehicle. The first threshold is greater than the second threshold.

4. The method according to claim 1, characterized in that, Solving the deterministic interactive decision-making problem of other vehicles crossing the scene during the driving process, and calculating the longitudinal acceleration of the vehicle, includes: Initialize solver parameters; The deterministic interactive decision problem of the driving process other vehicles crossing the scene is updated according to the initialized solver parameters, and the updated deterministic interactive decision problem of the driving process other vehicles crossing the scene is transformed by Taylor expansion in the preset domain. Based on the preset Backward process, the deterministic interactive decision-making problem of the transformed driving process and other vehicles crossing the scene is solved, and the change in control vector is calculated. Based on the preset Forward process and the change in the control vector, the deterministic interactive decision-making problem of the transformed driving process in the scenario of another vehicle crossing is solved, and the change in the system state vector is calculated. The longitudinal acceleration of the vehicle is calculated based on the change in the system state vector.

5. The method according to claim 4, characterized in that, The method further includes: The change in the system state vector is compared with a preset first parameter to determine whether the change in the system state vector satisfies the inner loop stopping criterion. When it is determined that the change in the system state vector satisfies the inner loop stopping criterion, the outer loop first parameter is obtained, and the outer loop first parameter is compared with the preset second parameter to determine whether the outer loop first parameter satisfies the outer loop stopping criterion. When it is determined that the first parameter of the outer loop meets the outer loop stopping criterion, the longitudinal acceleration of the vehicle is calculated based on the change in the system state vector.

6. The method according to claim 5, characterized in that, The method further includes: When it is determined that the change in the system state vector does not meet the inner loop stopping criterion, or when it is determined that the first parameter of the outer loop does not meet the outer loop stopping criterion, the preset domain is updated, and Taylor expansion is used within the updated preset domain to transform the updated deterministic interactive decision problem of the driving process other vehicle crossing the scene.

7. A continuous space driving interactive decision-making device, characterized in that, include: The acquisition module is used to acquire operating data from other vehicles and the vehicle itself. The driving style prediction module is used to determine the acceleration of other vehicles based on the other vehicle's operating data, and to construct driving style evaluation rules based on the Naive Bayes principle and likelihood function using the other vehicle's acceleration as a feature quantity to judge the driving style of other vehicles. The longitudinal interaction decision module is used to calculate the longitudinal acceleration of the vehicle based on the driving style of the other vehicle, the operating data of the other vehicle, and the operating data of the vehicle itself. The longitudinal interactive decision-making module is further configured to, based on the driving style of the other vehicle, its operating data, and the vehicle's operating data, establish an uncertain interactive decision-making problem concerning the other vehicle's traversal scenario during the driving process, including uncertainties in system state variables, uncertainties in vehicle obstacle avoidance, and chance constraints in system state variables. It utilizes a Kalman filter to describe the derivation process of the uncertainties in system state variables, uses an expectation model to transform the uncertainties in vehicle obstacle avoidance into deterministic conditions, and uses a one-sided Chebyshev inequality to transform the chance constraints in system state variables into deterministic constraints, thus transforming the uncertain interactive decision-making problem concerning the other vehicle's traversal scenario during the driving process into a deterministic interactive decision-making problem. Finally, it solves the deterministic interactive decision-making problem concerning the other vehicle's traversal scenario during the driving process to calculate the longitudinal acceleration of the vehicle. The lateral path planning module is used to calculate the lateral path curve of the vehicle based on the operation data of other vehicles and the operation data of the vehicle itself. The longitudinal velocity planning module is used to calculate the longitudinal velocity curve of the vehicle based on the longitudinal acceleration of the vehicle and the lateral path curve. The driving trajectory planning module is used to calculate the driving trajectory of the vehicle based on the vehicle's longitudinal speed curve and the vehicle's lateral path curve.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the continuous space driving interaction decision-making method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the continuous space driving interaction decision-making method as described in any one of claims 1 to 6.

Images