Automatic driving vehicle green wave passing vehicle speed planning method and vehicle control method

Abstract

The invention relates to an automatic driving vehicle green wave passing vehicle speed planning method and a vehicle control method, and the method comprises the steps: obtaining traffic information; based on the traffic information, constructing traffic light time sequence information under distance domain discretization; obtaining a traffic light window probability passing result based on the traffic light window prediction model; screening a potential passage route based on the traffic light window probability passage result; and in the potential passing routes, solving an optimal passing route and an optimal acceleration route and an optimal speed route corresponding to the optimal passing route. Compared with the prior art, the method has the advantages of low time complexity, high reliability, high online deployment speed, high safety and the like.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a green wave speed planning method and a vehicle control method for an autonomous driving vehicle integrated with deep learning. Background Art

[0002] With the rapid development of road traffic, the huge number of vehicles commuting has posed severe challenges to energy consumption and road traffic safety. The acquisition of road traffic environment information by autonomous driving vehicles that integrate perception and communication provides a new solution for energy conservation, emission reduction and efficient passage. Specifically, in the traffic scene at a traffic light intersection, the autonomous driving vehicle can obtain the current traffic light timing information (Signal Phase and Timing, SPaT) through the traffic light roadside unit (RSU) to plan a green and efficient traffic speed trajectory.

[0003] However, the existing vehicle speed planning methods based on road condition selection have great limitations when dealing with the green pass problem at traffic light communication intersections, and it is difficult to achieve the optimal effect. This is because these methods can usually only adjust the speed based on the current road condition information, and cannot fully consider the traffic light signals and other dynamic constraints. In addition, optimization-based methods also face many challenges in practical applications. These methods require complex speed planning optimization under multiple constraints, including traffic light constraints, road speed limits, physical constraints of vehicle actuators, and vehicle state constraints. The existence of these multiple constraints makes the speed planning optimization problem extremely complex and difficult to solve. Especially in the context of high-load computing power requirements of the perception module, how to complete efficient speed planning under limited computing power resources of the on-board controller has become a major problem in the application of autonomous driving technology. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a green wave speed planning method and vehicle control method for autonomous driving vehicles with low time complexity, high reliability, fast online deployment speed and high safety.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] According to a first aspect of the present invention, a method for planning the speed of an autonomous driving vehicle for green wave passage is provided, the method comprising:

[0007] Get traffic information;

[0008] Based on the traffic information, constructing traffic light timing information in a discrete distance domain;

[0009] Based on the traffic light window prediction model, obtain the traffic light window probability passing result;

[0010] Filter potential travel routes based on the traffic light window probability travel results;

[0011] Among the potential travel routes, an optimal travel route and its corresponding optimal acceleration route and optimal speed route are solved.

[0012] As a preferred technical solution, the method further includes:

[0013] Build a model training dataset;

[0014] The traffic light window prediction model is trained using the training data set with the goal of minimizing the loss function.

[0015] As a preferred technical solution, the step of constructing a model training data set includes:

[0016] Based on the historical traffic information, an optimization problem is constructed, and the optimization problem is solved to obtain traffic light window selection information corresponding to the historical traffic information; wherein the historical traffic information includes: historical traffic light position information, historical traffic light timing SPaT and historical vehicle initial speed information;

[0017] The optimization problem is:

[0018]

[0019] Wherein, J1 is the first cost function; v and a are the vehicle speed and vehicle acceleration respectively; is the vehicle energy consumption index; s0 and s f are the starting point and the end point respectively; t(s f ) is the travel time for the vehicle to reach the destination; a min and a max are the minimum and maximum acceleration of the vehicle respectively; t max is the maximum travel time for the vehicle to reach the destination; v max is the road speed limit; T i,j and are the upper and lower time limits of the jth green light window of the ith traffic light; t(s i ) is the time when the vehicle reaches the i-th traffic light position; w c and w t are the weights of energy consumption item and travel time respectively.

[0020] As a preferred technical solution, the construction of the model training data set also includes:

[0021] Using 0-1 logical variables To select a suitable green light window for passing, the optimization problem is converted into a mixed integer programming problem, and the mixed integer programming problem is relaxedly solved using the big M method, and the traffic light window passing constraint in the optimization problem is expressed as:

[0022]

[0023] Among them, M i,j is a larger integer under the j-th green light window constraint of the i-th traffic light to help relax the constraint; b i,j Indicates whether to pass through the j-th green light window of the ith traffic light, 1 means passing, 0 means not passing.

[0024] As a preferred technical solution, the construction of the model training data set also includes:

[0025] The mixed integer programming problem is expressed in discrete form:

[0026]

[0027] Among them, J2 is the second cost function; are the trajectories of the state quantity and the control quantity respectively; Select the matrix for the traffic light window; Δs = (s f -s0) / N is the discrete distance interval; N is the total discrete steps; k is the ordinal number of discrete steps; t(N) is the time it takes for the vehicle to reach the destination; a min and a max are the minimum and maximum acceleration of the vehicle respectively; v max is the road speed limit; P(k) indicates whether there is a traffic light at the kth discrete position; n t is the number of traffic lights, n w is the number of traffic light windows.

[0028] As a preferred technical solution, the loss function is:

[0029]

[0030] in, The traffic light window prediction model predicts the passing probability of the jth window of the i-th traffic light during the training process; y i,j is the true value of the sample.

[0031] As a preferred technical solution, the screening of potential travel routes based on the traffic light window probability travel result includes:

[0032] The traffic light window probability passing result for:

[0033]

[0034] For the first traffic light intersection, the following screening criteria are used to screen potential traffic light windows:

[0035]

[0036] Among them, s1 is the location of the first traffic light intersection, t acc is the actual acceleration time, which is calculated as:

[0037]

[0038] in, is the nominal acceleration time, calculated as:

[0039]

[0040] For subsequent traffic light intersections Use the following screening criteria to screen potential traffic light windows:

[0041]

[0042] in, Indicates the flag selected in the green light window of the previous traffic light intersection i-1,j =True's green light window upper limit time;

[0043] Obtain potential routes based on the following filling criteria

[0044]

[0045] As a preferred technical solution, in the potential travel routes, solving the optimal travel route and its corresponding optimal acceleration route and optimal speed route includes:

[0046] Select any route from the potential route set δ The number of integers in the mixed integer optimization problem of green wave traffic light is fixed to solve the continuous optimization problem;

[0047] Iterate and solve to obtain the optimal cost function value for all candidate routes, and compare them to obtain the optimal solution corresponding to the mixed integer optimization problem

[0048] According to the optimal solution Determine the optimal travel route and its corresponding optimal acceleration route and optimal speed route.

[0049] As a preferred technical solution, the optimization problem after fixing the number of integers in the mixed integer optimization problem is expressed as:

[0050]

[0051] in, For the selected route nonzero entries in the matrix; p1,...,p t is the discrete grid number corresponding to the location of the traffic light after discretization in the distance domain, where the subscript t represents the number of traffic lights in the predicted distance domain.

[0052] According to a second aspect of the present invention, there is provided a vehicle control method, the method comprising:

[0053] Controlling the vehicle using the optimal acceleration route and the optimal speed route corresponding to the optimal travel route;

[0054] Among them, the optimal acceleration route and the optimal speed route corresponding to the optimal passage route are obtained based on the green wave passage speed planning method for an autonomous driving vehicle provided by the first aspect or any one of the implementation methods of the first aspect.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] 1. Low time complexity: The green wave speed planning method for autonomous driving vehicles provided by the present invention introduces a traffic light window passing probability prediction model based on supervised learning, which effectively solves the NP problem caused by the non-continuous optimization characteristics in the traffic light green wave speed planning, and significantly reduces the complexity of solving the mixed integer programming problem.

[0057] 2. High reliability: The green wave speed planning method for autonomous driving vehicles provided by the present invention proposes green light route screening criteria based on the predicted probability value and actual working conditions, avoiding the problems of inaccurate neural network prediction and infeasible speed planning, and ensuring the practicality and reliability of the solution.

[0058] 3. Fast online deployment speed: The green wave speed planning method for autonomous driving vehicles provided by the present invention combines model prediction with a fast solution method for nonlinear programming. This solution mainly relies on a nonlinear programming algorithm, which significantly improves the speed of online deployment.

[0059] 4. High safety: Compared with the traditional mixed integer programming method, the green wave speed planning method for autonomous driving vehicles provided by the present invention combines model prediction with the fast solution method of nonlinear programming. It is not only efficient, but also ensures that the planned speed and acceleration meet the actual physical constraints, thereby ensuring the safety and standardization of vehicle driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A schematic diagram of a flow chart of a method for planning green wave speed for an autonomous driving vehicle integrating deep learning provided in an embodiment of the present invention;

[0061] Figure 2 A schematic diagram of the implementation process of the traffic light window prediction model provided by an embodiment of the present invention;

[0062] Figure 3 A schematic diagram of a green light window selection model for a traffic light intersection provided by an embodiment of the present invention;

[0063] Figure 4 A flowchart of an algorithm for integrating network prediction and fast solution of nonlinear programming provided in an embodiment of the present application;

[0064] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0065] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present invention. The "one embodiment" or "embodiment" referred to herein refers to specific features, structures or characteristics that can be included in at least one implementation of the present application.

[0066] Figure 1 This is a flowchart of a method for planning the speed of an autonomous vehicle for green wave traffic provided in an embodiment of the present application. This application provides method operation steps as described in the embodiment or flowchart, but may include more or fewer operation steps based on conventional or non-creative labor. The order of steps listed in the embodiment is only one of the many execution orders of the method for planning the speed of an autonomous vehicle for green wave traffic and the vehicle control method, and does not represent the only execution order. The method should be implementable by software and / or hardware. Please refer to Figure 1 , the green wave speed planning method for autonomous driving vehicles can include:

[0067] Step S110: Obtaining traffic information;

[0068] Step S120: constructing traffic light timing information in a distance domain discrete manner based on traffic information;

[0069] Step S130: obtaining a traffic light window probability passing result based on the traffic light window prediction model;

[0070] Step S140: Screening potential travel routes based on the traffic light window probability travel results;

[0071] Step S150: among the potential travel routes, find the optimal travel route and its corresponding optimal acceleration route and optimal speed route.

[0072] The traffic information in step S110 includes: location information of several future traffic light intersections and traffic light timing SPaT information. Traffic information can use vehicle Internet devices to establish communication with RSU to obtain location information and SPaT information of several future traffic light intersections in real time.

[0073] Step S120 can normalize and standardize the data according to the traffic information obtained in step S110 according to the output of the constructed traffic light window prediction model and the requirements of subsequent algorithms to construct traffic light timing information under distance domain discretization.

[0074] The following describes the training method of the above traffic light window prediction model:

[0075] Optionally, the above-mentioned green wave speed planning method for the autonomous driving vehicle further includes:

[0076] Step S210: construct a model training data set;

[0077] Step S220: Using the training data set, the traffic light window prediction model is trained with the goal of minimizing the loss function.

[0078] It can be understood that the above-mentioned traffic light window prediction model is constructed based on a neural network, and a high-quality traffic light data set can be constructed by learning using a data-driven method to realize the probability prediction mapping of future traffic light intersections. In addition to the above-mentioned steps S210 and S220, its implementation process can also include other steps, such as Figure 2 As shown in the figure, the implementation process of the above traffic light window prediction model can be divided into the steps of constructing a data set, segmenting traffic light traffic strategies, model design, model training and model evaluation. The following is a detailed introduction to each step:

[0079] Step 1: Construct a data set;

[0080] Prepare a traffic light intersection data set, take the traffic light intersection location, traffic light timing SPaT information, vehicle initial speed, intersection selection information and other data as the data set, and divide it into training set, validation set and test set.

[0081] First, the scheme for constructing the model training data set in step S210 is introduced:

[0082] Optionally, the above step S210 includes:

[0083] Based on the historical traffic information, an optimization problem is constructed and solved to obtain the traffic light window selection information corresponding to the historical traffic information; wherein the historical traffic information includes: historical traffic light position information, historical traffic light timing SPaT and historical vehicle initial speed information;

[0084] The optimization problem is:

[0085]

[0086] The above formula only considers the point model of the vehicle at the macro level as the controlled object passing through the traffic light intersection, the travel time t(s), speed v(s) and acceleration a(s) as the state quantity and control quantity of the vehicle at the macro level, and selects, where J1 is the first cost function; v and a are the vehicle speed and vehicle acceleration respectively; is the vehicle energy consumption index; s0 and s f are the starting point and the end point respectively; t(s f ) is the travel time for the vehicle to reach the destination; a min and a max are the minimum and maximum acceleration of the vehicle respectively; t max is the maximum travel time for the vehicle to reach the destination; v max is the road speed limit; T i,j and are the upper and lower time limits of the jth green light window of the ith traffic light; t(s i ) is the time when the vehicle reaches the i-th traffic light position; w c and w t are the weights of energy consumption item and travel time respectively.

[0087] Optionally, the above step S210 further includes:

[0088] Using 0-1 logical variables To select a suitable green light window for passing, the optimization problem is converted into a mixed integer programming problem, and the mixed integer programming problem is relaxedly solved using the big M method, and the traffic light window passing constraint in the optimization problem is expressed as:

[0089]

[0090] Among them, M i,j is a larger integer under the j-th green light window constraint of the i-th traffic light to help relax the constraint; b i,j Indicates whether to pass through the j-th green light window of the ith traffic light, 1 means passing, 0 means not passing.

[0091] Optionally, the above step S210 further includes:

[0092] Considering that the nonlinear mixed integer programming algorithm and the traffic light window prediction model are solved and reasoned based on discrete input and output in the actual solution process, the above mixed integer programming problem is expressed in discrete form:

[0093]

[0094] Where J2 is the second cost function; V: = [v(0), ..., v(N)] T And A:=[a(0),...,a(N-1)] T are the trajectories of the state quantity and the control quantity respectively; Select the matrix for the traffic light window; Δs = (s f -s0) / N is the discrete distance interval; N is the total discrete steps; k is the ordinal number of discrete steps; t(N) is the time it takes for the vehicle to reach the destination; a min and a max are the minimum and maximum acceleration of the vehicle respectively; v max is the road speed limit; P(k) indicates whether there is a traffic light at the kth discrete position; n t is the number of traffic lights, n w is the number of traffic light windows.

[0095] It can be understood that the dynamic equation of the above vehicle mass model uses forward Euler integration to achieve state update.

[0096] like Figure 3 As shown, a schematic diagram of the vehicle's speed trajectory selection at multiple consecutive traffic light intersections in a smooth traffic scenario is shown. The vehicle chooses the second green light window of the first traffic light, the third green light window of the second traffic light, and the third green light window of the third traffic light to pass. The above symbol explanations can be combined with the diagram to correspond to each other.

[0097] It is understandable that the existing mature mixed integer solver can be used to solve the above optimization problem to obtain the true value label of the traffic light intersection. After constructing the data set, the input feature values ​​of the data set are further determined: traffic light position, SPaT, vehicle initial speed, and output label: traffic light window selection.

[0098] Step 2: Segment traffic light intersection traffic strategy and model design;

[0099] Design continuous n t Traffic lights wThe method of passing through a traffic light window is used to predict the traffic flow at a certain traffic light window. A traffic light window prediction model is constructed based on a neural network to realize end-to-end prediction of data. The probability prediction of traffic light intersections is realized by using the outermost output Softmax activation function. A loss function based on binary cross entropy is constructed to make the predicted value approach the true value label obtained by the solver. The neural network traffic light intersection prediction module constructed in the present invention is trained based on the above strategy.

[0100] Step 3: Model training;

[0101] Use the training set to train the model, and design the binary cross entropy as the loss function of the model training. The loss function expression can be:

[0102]

[0103] in, The traffic light window prediction model predicts the passing probability of the jth window of the i-th traffic light during the training process; y i,j is the true value of the sample, a successful behavior is 1, and an unsuccessful behavior is 0.

[0104] Step 4: Model evaluation;

[0105] The trained model weight bias is saved and verified on the test set split from the prepared data set to measure the accuracy of traffic light intersections. The specific evaluation indicators can be: prediction accuracy, inference speed, and model parameter quantity. Based on the above comprehensive evaluation indicators, a prediction model that takes into account accuracy, lightweight online deployment, and real-time inference can be obtained.

[0106] The following is an introduction to the solution of screening potential travel routes in step S140:

[0107] Optionally, the above step S140 includes:

[0108] Traffic light window probability passing results under transient vehicle driving conditions for:

[0109]

[0110] Considering the problem of prediction accuracy under different working conditions, a screening criterion for probabilistic pass results can be designed to obtain potential integer feasible solutions;

[0111] For the first traffic light intersection, the following screening criteria are used to screen potential traffic light windows:

[0112]

[0113] Among them, s1 is the location of the first traffic light intersection, t accis the actual acceleration time, which is calculated as:

[0114]

[0115] in, is the nominal acceleration time, calculated as:

[0116]

[0117] For subsequent traffic light intersections Use the following screening criteria to screen potential traffic light windows:

[0118]

[0119] in, Indicates the flag selected in the green light window of the previous traffic light intersection i-1,j =True's green light window upper limit time;

[0120] Based on this, we can construct a set of candidate strategies for traffic light window timing selection First, the set is set to an empty set, and then the potential feasible traffic light window routes are screened according to the above screening criteria, so as to fill the set in a loop. Where n r Represents the number of potentially feasible window routes screened out.

[0121] Optionally, you can obtain potential routes based on the following filling criteria:

[0122]

[0123] In the formula, The i-th row (i.e. the i-th signal light) has only one green light window, and the i+1-th row / matrix element follows the above screening criteria flag i,j Fill with unique elements.

[0124] The following is an introduction to the solution of solving the optimal route in step S150:

[0125] Optionally, the above step S150 includes:

[0126] Select any route from the potential route set δ The number of integers in the mixed integer optimization problem of green wave traffic light is fixed to solve the continuous optimization problem and avoid NP hard problems.

[0127] According to the above optimization problem, the optimal cost function value is obtained Iterative solution can be performed to obtain the traffic light route to be selected The optimal cost function value corresponding to each route in the mixed integer optimization problem is obtained by comparing the optimal solution.

[0128] According to the optimal solution Determine the optimal travel route and its corresponding optimal acceleration route and optimal speed route.

[0129] Optionally, the optimization problem after fixing the number of integers in the mixed integer optimization problem is expressed as:

[0130]

[0131]

[0132] in, For the selected route nonzero entries in the matrix; p1,...,p t is the discrete grid number corresponding to the location of the traffic light after discretization in the distance domain, where the subscript t represents the number of traffic lights in the predicted distance domain.

[0133] So far, the embodiment of the present application has established an algorithm flow that integrates model prediction and nonlinear programming fast solution for traffic light green wave speed planning, such as Figure 4 As shown, the algorithm flow may include:

[0134] (1) Initially, initialize the optimal solution of (3) Define the trained neural network f Θ (θ), the traffic light timing information, traffic light position, initial vehicle speed and other information are normalized into network input problem parameters according to the standard normalization method of the training process At the same time, the problem parameters such as maximum / minimum acceleration, maximum vehicle speed, upper limit of driving time, discrete interval, and discrete step number in formula (3) are defined as α.

[0135] (2) Bring in the trained neural network f Θ Get the traffic light window probability passing result According to the candidate route screening criteria of formula (6), (7), (8), (9), the potential traffic light route set δ is obtained, and the number of elements n in the set is calculated. r , set the number of loop solutions iter←0.

[0136] (3) Repeat the solution for the number of times iter←iter+1, and select the pass window route from the set δ in order Combination The non-zero elements and problems in Traffic light location in Simplify the definition of the optimization problem (10) Solving the nonlinear programming problem (10), we obtain

[0137] (4) If the solution to problem (10) is feasible, jump to the following (5); otherwise, return to the above (3) and continue iterating.

[0138] (5) The optimization result of the window route and the optimal solution For comparison, if Update And record the optimization results If the number of solutions exceeds the number of elements in the set δ, iter ≥ n r , then stop the iteration and input the optimal acceleration and speed trajectory to the vehicle; otherwise, return to the above (3) to continue the iteration.

[0139] Based on the same inventive concept, the embodiment of the present application further provides a vehicle control method, the method comprising:

[0140] Controlling the vehicle using the optimal acceleration route and the optimal speed route corresponding to the optimal travel route;

[0141] Among them, the optimal acceleration route and the optimal speed route corresponding to the optimal passage route are obtained based on any of the above-mentioned green wave passage speed planning methods for autonomous driving vehicles.

[0142] Figure 5 Schematic block diagram of an electronic device that can be used to implement an embodiment of the present disclosure is shown. Figure 5 As shown, the electronic device of the present invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) or loaded from a storage unit into a random access memory (RAM). In the RAM, various programs and data required for device operation can also be stored. The CPU, ROM and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0143] Multiple components in the device are connected to the I / O interface, including: input units, such as keyboards, mice, etc.; output units, such as various types of displays, speakers, etc.; storage units, such as disks, optical disks, etc.; and communication units, such as network cards, modems, wireless communication transceivers, etc. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunication networks.

[0144] The processing unit performs the various methods and processes described above, such as steps S110 to S150 of the method of the present invention. For example, in some embodiments, steps S110 to S150 of the method of the present invention may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via a ROM and / or a communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more of steps 101 to 105 of the method of the present invention described above may be executed. Alternatively, in other embodiments, the CPU may be configured to execute steps S110 to S150 of the method of the present invention by any other appropriate means (e.g., by means of firmware).

[0145] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0146] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0147] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, 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 foregoing.

[0148] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for planning the speed of an autonomous driving vehicle for green wave passage, characterized in that: The method comprises: Get traffic information; Based on the traffic information, constructing traffic light timing information in a discrete distance domain; Based on the traffic light window prediction model, obtain the traffic light window probability passing result; Filter potential travel routes based on the traffic light window probability travel results; Among the potential travel routes, an optimal travel route and its corresponding optimal acceleration route and optimal speed route are solved.

2. The green wave speed planning method for an autonomous driving vehicle according to claim 1 is characterized in that: The method further comprises: Build a model training dataset; The traffic light window prediction model is trained using the training data set with the goal of minimizing the loss function.

3. The green wave speed planning method for an autonomous driving vehicle according to claim 2 is characterized in that: The constructing of the model training data set includes: Based on the historical traffic information, an optimization problem is constructed, and the optimization problem is solved to obtain traffic light window selection information corresponding to the historical traffic information; wherein the historical traffic information includes: historical traffic light position information, historical traffic light timing SPaT and historical vehicle initial speed information; The optimization problem is: Wherein, J1 is the first cost function; v and a are the vehicle speed and vehicle acceleration respectively; is the vehicle energy consumption index; s0 and s f are the starting point and the end point respectively; t(s f ) is the travel time for the vehicle to reach the destination; a min and a max are the minimum and maximum acceleration of the vehicle respectively; t max is the maximum travel time for the vehicle to reach the destination; v max is the road speed limit; T i,j and are the upper and lower time limits of the jth green light window of the ith traffic light; t(s i ) is the time when the vehicle reaches the i-th traffic light position; w c and w t are the weights of energy consumption item and travel time respectively.

4. The green wave speed planning method for an autonomous driving vehicle according to claim 3 is characterized in that: The constructing of the model training data set also includes: Using 0-1 logical variables To select a suitable green light window for passing, the optimization problem is converted into a mixed integer programming problem, and the mixed integer programming problem is relaxedly solved using the big M method, and the traffic light window passing constraint in the optimization problem is expressed as: Among them, M i,j is a larger integer under the j-th green light window constraint of the i-th traffic light to help relax the constraint; b i,j Indicates whether to pass through the j-th green light window of the ith traffic light, 1 means passing, 0 means not passing.

5. The green wave speed planning method for an autonomous driving vehicle according to claim 4 is characterized in that: The constructing of the model training data set also includes: The mixed integer programming problem is expressed in discrete form: Among them, J2 is the second cost function; and are the trajectories of the state quantity and the control quantity respectively; Select the matrix for the traffic light window; Δs = (s f -s0) / N is the discrete distance interval; N is the total discrete steps; k is the ordinal number of discrete steps; t(N) is the time it takes for the vehicle to reach the destination; a min and a max are the minimum and maximum acceleration of the vehicle respectively; v max is the road speed limit; P(k) indicates whether there is a traffic light at the kth discrete position; n t is the number of traffic lights, n w is the number of traffic light windows.

6. The green wave speed planning method for an autonomous driving vehicle according to claim 2, characterized in that: The loss function is: in, The traffic light window prediction model predicts the passing probability of the jth window of the i-th traffic light during the training process; y i,j is the true value of the sample.

7. The green wave speed planning method for an autonomous driving vehicle according to claim 1, characterized in that: The screening of potential travel routes based on the traffic light window probability travel result includes: The traffic light window probability passing result for: For the first traffic light intersection, the following screening criteria are used to screen potential traffic light windows: Among them, s1 is the location of the first traffic light intersection, t acc is the actual acceleration time, which is calculated as: in, is the nominal acceleration time, calculated as: For subsequent traffic light intersections Use the following screening criteria to screen potential traffic light windows: in, Indicates the flag selected in the green light window of the previous traffic light intersection i-1,j =True's green light window upper limit time; Obtain potential routes based on the following filling criteria 8. The green wave speed planning method for an autonomous driving vehicle according to claim 1, characterized in that: The step of solving the optimal route and its corresponding optimal acceleration route and optimal speed route in the potential routes includes: Select any route from the potential route set δ The number of integers in the mixed integer optimization problem of green wave traffic light is fixed to solve the continuous optimization problem; Iterate and solve to obtain the optimal cost function value for all candidate routes, and compare them to obtain the optimal solution corresponding to the mixed integer optimization problem According to the optimal solution Determine the optimal travel route and its corresponding optimal acceleration route and optimal speed route.

9. The green wave speed planning method for an autonomous driving vehicle according to claim 8, characterized in that: The optimization problem after fixing the number of integers in the mixed integer optimization problem is expressed as: in, For the selected route nonzero entries in the matrix; p1,…,p t is the discrete grid number corresponding to the location of the traffic light after discretization in the distance domain, where the subscript t represents the number of traffic lights in the predicted distance domain.

10. A vehicle control method, characterized in that: The method comprises: Controlling the vehicle using the optimal acceleration route and the optimal speed route corresponding to the optimal travel route; The optimal acceleration route and the optimal speed route corresponding to the optimal passage route are obtained based on the green wave passage speed planning method for an autonomous driving vehicle according to any one of claims 1 to 9.

Images