Intelligent networked vehicle cloud path tracking control method and system

Abstract

The application relates to an intelligent networked vehicle cloud path tracking control method and system. The method iteratively executes the following steps, including: constructing a vehicle tracking error model based on a vehicle state; obtaining a historical time delay data sequence and signal strength, generating a feature sequence, using a time delay prediction model based on the feature sequence, and outputting a time delay prediction value and uncertainty; performing QoS mapping based on a service guarantee level parameter, a time delay prediction value and uncertainty to obtain a time delay confidence upper bound and a preview distance; using the vehicle tracking error model to calculate a preview tracking error based on the preview distance, designing a steering control law based on the preview tracking error; performing intelligent networked vehicle cloud path tracking control based on the steering control law, and obtaining a vehicle state after the vehicle executes the steering control law. Compared with the prior art, the application introduces time delay prediction uncertainty into vehicle control strategy generation, thereby improving path tracking performance and operation safety under random communication time delay conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent networked vehicles and vehicle-cloud collaborative control, and in particular to an intelligent networked vehicle cloud path tracking control method and system. BACKGROUND

[0002] With the development of vehicle networking and intelligent networked vehicle technology, vehicle-cloud collaborative control has gradually become an important form of high-level automatic driving. By deploying sensors, actuators and basic controllers on the vehicle end and deploying high-performance computing and decision units on the cloud end, the vehicle end can upload the vehicle state to the cloud control center in real time through a wireless communication network (such as a 4G / 5G cellular network), the cloud control center generates control instructions based on global environmental information and planned trajectories, and then the control instructions are sent to the vehicle end through the network for execution, realizing vehicle-cloud collaborative path tracking and decision control. This architecture can utilize the powerful computing and perception capabilities of the cloud to improve the global nature of path planning and the optimization level of control strategies. However, vehicle-cloud collaboration inevitably introduces a wireless communication link with time-varying delay, and the delay is affected by network load, link state and computing queue, etc., showing randomness and uncertainty. For time-sensitive path tracking control, if this type of communication delay and its fluctuations are not reasonably handled, it can easily lead to control instruction lag, increased path tracking error, and even cause lateral oscillation and decreased comfort and stability.

[0003] The existing technology mainly focuses on the following methods for handling communication delay in vehicle-cloud collaborative path tracking: 1) fixed delay compensation: treating the delay as a constant and designing a fixed compensation amount, which is difficult to adapt to dynamic network environments; 2) worst-case design: designing a controller based on the maximum delay, which is robust but overly conservative, affecting performance. To solve the above problems, Chinese Patent Application CN120808624A provides a cloud-controlled autonomous vehicle control method considering vehicle-cloud communication delay, which predicts the number of steps of the delay, combines historical vehicle states and control inputs, and deduces future vehicle states. Although it achieves dynamic compensation for the impact of delay, the method used is single-point prediction, which only obtains a deterministic estimate value without considering the uncertainty of the prediction result. In addition, it uses a fixed structure and fixed weight state feedback control mechanism, i.e., based on the deterministic delay obtained by prediction, the state is forward compensated through the vehicle lateral dynamics model, and a linear quadratic regulator (LQR) is designed based on the compensated state. The prediction delay only affects the state compensation link and does not affect the final control strategy, thereby affecting the path tracking accuracy and the stability of vehicle driving.

[0004] Therefore, it is a technical problem to provide a method that can guide the vehicle control strategy generation based on delay prediction uncertainty. SUMMARY

[0005] The present application aims to overcome the defects of the prior art and provides an intelligent connected vehicle cloud path tracking control method and system. The method predicts the time delay and uncertainty through a Bayesian Long Short-Term Memory Neural Network (BLSTM), constructs a mapping relationship between the Quality of Service (QoS) and the vehicle tracking control parameters based on the predicted time delay and uncertainty, and thereby achieves control.

[0006] The object of the present application can be achieved by the following technical solutions:

[0007] According to a first aspect of the present application, an intelligent connected vehicle cloud path tracking control method is provided. The method iteratively performs the following steps to achieve vehicle cloud path tracking control, including:

[0008] A vehicle tracking error model is constructed based on the vehicle state. The error model includes vehicle lateral error and vehicle heading error.

[0009] A historical time delay data sequence and corresponding signal strength are obtained, and a feature sequence is generated. Based on the feature sequence, a trained time delay prediction model is used to output a time delay prediction value at the next time and the uncertainty of the time delay prediction value.

[0010] A service guarantee level parameter is constructed. Based on the service guarantee level parameter, the time delay prediction value, and the uncertainty, a QoS mapping is performed to obtain a time delay confidence upper bound of the communication time delay. Based on the time delay confidence upper bound, a preview distance is obtained.

[0011] Based on the preview distance, the vehicle tracking error model is used to calculate a preview tracking error. Based on the preview tracking error, a steering control law is designed. The preview tracking error includes a preview heading error and a preview lateral error.

[0012] Based on the steering control law, intelligent connected vehicle cloud path tracking control is performed, and the vehicle state after the vehicle executes the steering control law is obtained.

[0013] As a preferred technical solution, the method for constructing the vehicle tracking error model is:

[0014] The vehicle state vector is defined as wherein, represents the vehicle coordinates, represents the vehicle heading angle, represents the vehicle longitudinal speed; represents the transpose operation.

[0015] reference point of the vehicle on the reference trajectory The reference point coordinates of the vehicle on the reference trajectory are , and the reference tangent heading angle and the reference curvature at the reference point are and respectively;

[0016] The vehicle lateral error and the vehicle heading error are calculated based on the vehicle state vector, the reference point coordinates, the reference tangent heading angle and the reference curvature, and are

[0017] ,

[0018] wherein represents the vehicle lateral error; represents the vehicle heading error; represents an angle normalization operation.

[0019] As a preferred technical solution, in the time delay prediction model, the feature sequence is processed by a forget gate, an input gate and an output gate in the time delay prediction model to generate a hidden state at the current time;

[0020] The signal strength at the current time is obtained, the importance score of the hidden state is calculated, and the attention weight is calculated based on the importance score, and is

[0021] ,

[0022] ,

[0023] wherein represents the importance score of the hidden state at the current time t; represents an attention vector; represents a transposition operation; represents a weight matrix of the hidden state; represents a weight matrix of the signal strength; represents a signal strength sequence at time t; represents a bias term; represents the attention weight; represents the total time length.

[0024] As a preferred technical solution, the method for obtaining the time delay confidence upper bound is

[0025] The reliability coefficient is calculated based on the service guarantee level parameter , and is , represents the reliability coefficient; denotes a service assurance level coefficient;

[0026] denotes a time delay confidence upper bound, calculated based on the reliability coefficient, as: , denotes a time delay confidence upper bound; denotes a time delay prediction value; denotes an uncertainty.

[0027] As a preferred technical solution, the method for obtaining the preview distance is:

[0028] ,

[0029] denotes a preview distance; denotes a nominal preview distance, which is adopted without considering the communication time delay; denotes a vehicle longitudinal speed; denotes a time delay confidence upper bound; denotes a time delay prediction value; denotes an uncertainty.

[0030] As a preferred technical solution, the method for constructing the steering control law is:

[0031]

[0032] denotes a steering control law; denotes a preview lateral error, and , denotes a vehicle heading angle at a reference point corresponding to the preview distance, denotes a vehicle coordinate of a reference point corresponding to the preview distance, denotes a vehicle coordinate; denotes a preview distance; denotes a lateral error gain; denotes a preview heading error, and , denotes a vehicle heading angle, denotes an angle normalization operation; denotes a vehicle longitudinal speed.

[0033] As a preferred technical solution, the method for calculating the lateral error gain is:

[0034] ,

[0035] wherein, denotes a nominal preview distance, which is adopted without considering the communication time delay; denotes an adjustment parameter.

[0036] According to a second aspect of the present application, a cloud path tracking control system for intelligent connected vehicles is provided, which comprises a vehicle end controller, a wireless communication module and a cloud control center, wherein the cloud control center comprises:

[0037] a communication time delay collection module configured to collect a historical time delay data sequence and corresponding signal strength;

[0038] an error model construction module configured to construct a vehicle tracking error model based on vehicle states collected by the vehicle end controller, wherein the error model comprises vehicle lateral error and vehicle heading error;

[0039] a time delay prediction network module configured to generate a feature sequence based on the historical time delay data sequence and corresponding signal strength, and output a time delay prediction value at the next time and an uncertainty of the time delay prediction value based on the feature sequence using a trained time delay prediction model;

[0040] a QoS indicator mapping module configured to construct a service guarantee level parameter, perform QoS mapping to obtain a time delay confidence upper bound of the communication time delay based on the service guarantee level parameter, the time delay prediction value and the uncertainty, and obtain a preview distance based on the time delay confidence upper bound;

[0041] a path tracking control module configured to calculate a preview tracking error based on the preview distance using the vehicle tracking error model, design a steering control law based on the preview tracking error, and perform intelligent connected vehicle cloud path tracking control by issuing the steering control law to the vehicle end controller through the wireless communication module; the preview tracking error comprises a preview heading error and a preview lateral error.

[0042] As a preferred technical solution, the vehicle end controller is arranged at the vehicle end, configured to collect vehicle states and add local time stamps to the collected data; and the vehicle end controller is further configured to receive the steering control law issued by the cloud control center and convert the steering control law into execution signals of a vehicle target steering execution mechanism and a target driving execution mechanism; the vehicle states comprise pose information and running state information.

[0043] As a preferred technical solution, the wireless communication module is located at the vehicle end, configured to realize communication between the vehicle end controller and the cloud control center; and in the communication process, the module records or carries network side information and sends the network side information to the communication time delay collection module; the network side information comprises a sending time stamp, a receiving time stamp and a current communication network signal strength.

[0044] Compared with the prior art, the present application has the following beneficial effects:

[0045] 1) In the latency data acquisition and prediction network module of this invention, in addition to acquiring vehicle-to-network-to-vehicle (V2N2V) latency data, network quality information such as the signal strength of the current communication network is also acquired and recorded. This information, together with the latency, is used to construct a time-series feature sequence as the input to the latency prediction network. Furthermore, the current signal strength is introduced into the latency prediction model to guide the generation of attention weights, which can more accurately capture the correspondence between the communication link status and latency fluctuations. This effectively reduces the probability of missing large latency and prediction distortion, improves the accuracy and robustness of latency QoS estimation, and provides a more reliable input for subsequent control strategies.

[0046] 2) Based on existing technologies, this invention introduces the concept of latency prediction uncertainty. This uncertainty is combined with the latency prediction value to construct an upper bound for latency confidence under a given service reliability. First, the predicted latency is used to compensate for the traditional fixed aiming distance, in the form of " This item corrects the problem that traditional fixed aiming distance cannot reflect the impact of time delay, reducing path tracking lag caused by communication delay; it utilizes the uncertainty of time delay to construct a safety margin, in the form of " This feature ensures that the vehicle still has sufficient aiming distance, making steering commands smoother and thus enhancing the robustness of the control system under communication delay fluctuations and prediction errors. Compared with control methods that only use fixed aiming distance or simple gain adjustment, it can simultaneously take into account path tracking accuracy and driving stability in complex and changing network environments. Attached Figure Description

[0047] Figure 1 This is a flowchart of the method of the present invention;

[0048] Figure 2 This is a schematic diagram of the delay prediction model structure in an embodiment of the present invention;

[0049] Figure 3 This is a schematic diagram of the system framework of the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0051] The prior art urgently needs a cloud path tracking control method capable of directly mapping the uncertainty of time delay prediction as a QoS indicator into vehicle control parameters, making full use of time delay prediction and its uncertainty information to adaptively adjust the path tracking control strategy to improve the path tracking performance and operation safety under the condition of random communication time delay.

[0052] To solve the above technical problems, the present application provides a smart connected vehicle cloud path tracking control method, the flowchart of which is shown in Figure 1 The method iteratively performs the following steps until the path tracking task is completed, including:

[0053] S1, a vehicle tracking error model is constructed based on the vehicle state, and the error model includes vehicle lateral error and vehicle heading error.

[0054] In this embodiment, the vehicle state vector is defined as , wherein represents the vehicle coordinates, represents the vehicle heading angle, represents the vehicle longitudinal speed, represents the transposition operation; and the kinematic model can be constructed as:

[0055] ,

[0056] wherein , , are respectively x the speed component in the x-axis direction, y the speed component in the y-axis direction, and the heading angle speed, is the vehicle front wheel steering angle, which is the controlled variable; L is the vehicle wheelbase.

[0057] The reference point of the vehicle on the reference trajectory is the reference point coordinates , and the reference tangent heading angle and the reference curvature at the reference point are respectively: and ; the main purpose of vehicle tracking control is to minimize the deviation between the actual position of the vehicle and the reference trajectory, including vehicle lateral error and vehicle heading error , the vehicle lateral error and the vehicle heading error are calculated based on the vehicle state vector, the reference point coordinates, the reference tangent heading angle and the reference curvature, which are:

[0058] ,

[0059] wherein represents the vehicle lateral error; represents the vehicle heading error; denotes an angle normalization operation, in particular normalizing the angle to .

[0060] Based on the foregoing, the error dynamics can be obtained as:

[0061] ,

[0062] where, denotes the lateral error dynamics; denotes the vehicle longitudinal speed; denotes the heading error dynamics; denotes the equivalent disturbance term, which is used to describe the influence of unmodeled dynamics, external disturbances, and parameter uncertainties on the heading error dynamics; denotes the curvature of the reference trajectory at the reference point.

[0063] Under the condition that , , there is , so the above error dynamics can be approximated as:

[0064] ,

[0065] The above formula is the state equation of the controlled object (error), which reflects how the steering control law acts on the change rate of the error, thereby ensuring that the vehicle tracks the control requirement that the lateral error and the heading error tend to 0.

[0066] S2, acquire a historical time delay data sequence and corresponding signal strength and generate a feature sequence, and based on the feature sequence, use a trained time delay prediction model to output a time delay prediction value at the next moment and an uncertainty of the time delay prediction value.

[0067] According to actual needs, the application focuses on the V2N2V time delay of vehicle motion after vehicle state sending to control command execution, in particular, the time stamp of the vehicle end sending data is denoted as , the time stamp of the cloud receiving signal is denoted as ; the time stamp of the cloud sending data is denoted as , and the time stamp of the vehicle receiving cloud instructions is denoted as , so the V2N2V time delay can be represented as .

[0068] A large number of time delay data sets are collected under different speeds and network environments as a training set. Each set of time delay data contains a time stamp, vehicle pose information (including position, speed, and heading angle), V2N2V time delay data, and communication network signal strength.

[0069] After the collection is completed, the data is organized into time sequence sample segments according to time sequence, so as to be input into a time sequence prediction model, and for each time segment, an input feature sequence is constructed: wherein, represents data at time t, represents V2N2V delay at time t, represents signal strength at time t; and the feature sequence comprises a delay sequence and a signal strength sequence.

[0070] In the embodiment, the delay prediction model adopted is set to be BSLTM, the BLSTM network takes the feature sequence as input, the input sequence length is set to be 32, the network hidden unit number is set to be 64, and the learning rate is set to be 1e-4; the network comprises a batch normalization layer, an LSTM network layer, an attention layer, a residual network layer and a distribution parameter layer, and the number of LSTM network layers is at least 4 layers, and the specific structure is as shown in Figure 2 Because the average value of 5G communication delay is about 19ms, and the delay has obvious time-varying and gamma distribution characteristics, Monte Carlo sampling is further embedded before the network output layer in the present application, the gamma distribution is taken as a prior distribution to constrain the probability characteristics of the delay prediction value, Bayesian inference is realized through random Dropout, and finally the delay prediction value is output.

[0071] In addition, during training, based on the training set, the network parameter w is regarded as a random variable through the Bayesian inference idea, and the parameter posterior distribution is estimated through the variational inference or Monte Carlo sampling method:

[0072] ,

[0073] wherein, D is a training data set, is a parameterized approximate posterior distribution; and finally, model optimization is realized by maximizing the log-likelihood lower bound.

[0074] During model deduction, in the delay prediction model, the feature sequence is processed through the forget gate, the input gate and the output gate in the delay prediction model to generate the hidden state at the current time, and the process is as follows:

[0075] ,

[0076] wherein, , , are output vectors of the forget gate, the input gate and the output gate respectively; is a candidate unit state vector, which represents the candidate memory content generated by the current input and the historical hidden state; W * is a weight matrix input to each gate / candidate state; U *is the recurrent weight matrix of hidden state to each gate / candidate state; b * is the bias vector of each gate / candidate state; s (·) is a sigmoid activation function; tanh(·) is a hyperbolic tangent activation function; denotes element-wise multiplication of vectors; denotes the hidden state at time t . denotes the cell state at time t .

[0077] The signal strength at the current time is obtained, the importance score of the hidden state is calculated, the sensitivity of the model to the change of the communication signal is enhanced, the attention weight is calculated based on the importance score, and the attention weight is:

[0078] ,

[0079] ,

[0080] wherein, denotes the importance score of the hidden state at the current time t; denotes the attention vector; denotes the transposition operation; denotes the weight matrix of the hidden state; denotes the weight matrix of the signal strength; denotes the signal strength sequence at time t; denotes the bias term; denotes the attention weight; denotes the total time length.

[0081] The context vector is obtained by weighted summation of the hidden state based on the attention weight: , and then the context vector is processed through the residual network layer and the distribution parameter layer to output the predicted time delay value and the corresponding uncertainty.

[0082] S3, a service guarantee level parameter is constructed, a time delay confidence upper bound of the communication time delay is obtained based on the service guarantee level parameter, the time delay prediction value and the uncertainty, and a preview distance is obtained based on the time delay confidence upper bound.

[0083] The application further constructs a QoS mapping module for cloud control path tracking control, which is used to map the communication time delay and its uncertainty into a preview distance that can be directly used for path tracking control.

[0084] In order to depict the requirement for communication service quality, a service guarantee level parameter is set (wherein ).

[0085] The reliability coefficient is calculated based on the service guarantee level parameter, and is: , denotes the reliability coefficient; denotes the service guarantee level coefficient.

[0086] Based on the reliability coefficient, combined with the one-sided probability inequality, under the condition that the service guarantee level parameter is not less than , the delay confidence upper limit is: , denotes the delay confidence upper limit; denotes the delay prediction value; denotes the uncertainty.

[0087] The present application maps the delay confidence upper limit to the preview distance, thereby explicitly considering the influence of the communication delay on the original fixed preview distance design in the control law, and is:

[0088] ,

[0089] denotes the preview distance; denotes the nominal preview distance adopted without considering the communication delay; denotes the vehicle longitudinal speed; denotes the delay confidence upper limit; denotes the delay prediction value; denotes the uncertainty.

[0090] S4, based on the preview distance, a preview tracking error is calculated using a vehicle tracking error model, and a steering control law is designed based on the preview tracking error; the preview tracking error includes a preview heading error and a preview lateral error.

[0091] Specifically, on the basis of the classic Stanley path tracking control method, a dynamic preview mechanism considering the communication delay and uncertainty is introduced, and a cloud path tracking control law considering delay prediction uncertainty is constructed to improve the path tracking accuracy and stability of the vehicle under the condition of communication delay and uncertainty.

[0092] The preview distance L d The position selection preview reference point is denoted as , and the lateral error of the vehicle relative to the preview reference point is defined as :

[0093] ,

[0094] The preview heading error is defined as :

[0095] .

[0096] On the basis of the above, then:

[0097]

[0098] denotes a steering control law; denotes a preview lateral error, and , denotes a preview distance corresponding to the vehicle heading angle at the reference point, denotes a preview distance corresponding to the vehicle coordinates of the reference point, denotes vehicle coordinates; denotes a preview distance; denotes a preview heading error, and , denotes a vehicle heading angle, denotes an angle normalization operation; denotes a vehicle longitudinal speed; denotes a lateral error gain, and:

[0099] ,

[0100] wherein, denotes an adjustment parameter.

[0101] S5, based on the steering control law, intelligent connected vehicle cloud path tracking control is performed, and the vehicle state after the vehicle executes the steering control law is obtained.

[0102] In this step, if the vehicle path tracking control task has not ended, then based on the newly collected vehicle state, the execution step S1 is returned to form a closed-loop control; otherwise, it is ended.

[0103] Embodiment 2

[0104] The present application provides an intelligent connected vehicle cloud path tracking control system, which has the structure as shown in Figure 3 The present application provides an intelligent connected vehicle cloud path tracking control system, which has the structure as shown in

[0105] Specifically, the vehicle end controller: used for collecting vehicle state, and adding local timestamp to the collected data; and, the vehicle end controller is also used for receiving the steering control law issued by the cloud control center, and converting the steering control law into the execution signal of the vehicle target steering execution mechanism and the target driving execution mechanism; the vehicle state includes pose information and running state information.

[0106] Wireless communication module: used to realize the communication between the vehicle controller and the cloud control center; and in the communication process, the module records or carries the network side information, and sends the network side information to the communication delay acquisition module; the network side information includes sending timestamp, receiving timestamp and current communication network signal strength.

[0107] The cloud control center comprises:

[0108] a) A communication delay acquisition module, which is used to acquire a historical delay data sequence and corresponding signal strength, through which communication delay information in a real dynamic network environment can be obtained, providing a data basis for subsequent prediction and control.

[0109] b) An error model construction module, which is used to construct a vehicle tracking error model based on the vehicle state collected by the vehicle controller, the error model including vehicle lateral error and vehicle heading error, and the model constructed by the module is used to guide the path tracking control module to generate a control law with smaller error.

[0110] c) A delay prediction network module, which is used to generate a feature sequence based on the historical delay data sequence and the corresponding signal strength, and output a delay prediction value and an uncertainty of the delay prediction value at the next moment based on the trained delay prediction model, which not only ensures the prediction accuracy but also provides a quantitative evaluation of the prediction reliability, providing a basis for introducing a confidence margin in the design of the control layer.

[0111] d) A QoS index mapping module, in which a service guarantee level parameter is constructed, a delay confidence upper bound of the communication delay is obtained based on the service guarantee level parameter, the delay prediction value and the uncertainty through QoS mapping, and a preview distance is obtained based on the delay confidence upper bound; the module supports flexible adjustment under different service levels, enabling the system to achieve a controllable compromise between performance and safety.

[0112] e) A path tracking control module, which is used to calculate a preview tracking error based on the preview distance and the vehicle tracking error model, and design a steering control law based on the preview tracking error; the module is used to perform intelligent connected vehicle cloud path tracking control by issuing the steering control law to the vehicle controller through the wireless communication module, realizing robust path tracking of the vehicle under the condition of communication delay and prediction uncertainty, wherein the preview tracking error includes preview heading error and preview lateral error.

[0113] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described modules can refer to the corresponding process in the foregoing method embodiments, which will not be described herein.

[0114] The present application also provides an electronic device including a central processing unit (CPU) that can perform various appropriate actions and processes in accordance with computer program instructions stored in a read-only memory (ROM) or computer program instructions loaded from a storage unit into a random access memory (RAM). Various programs and data required for device operation can also be stored in the RAM. The CPU, the ROM, and the RAM are connected to each other through a bus. An input / output (I / O) interface is also connected to the bus.

[0115] A plurality of components in the device are connected to the I / O interface, including: an input unit such as a keyboard, a mouse, and the like; an output unit such as various types of displays, a speaker, and the like; a storage unit such as a magnetic disk, an optical disk, and the like; and a communication unit such as a network card, a modem, a wireless communication transceiver, and the like. The communication unit allows the device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0116] The processing unit performs the various methods and processes described above, such as the methods S1-S5. For example, in some embodiments, the methods S1-S5 can be implemented as a computer software program that is tangibly embodied in a machine-readable medium, such as the storage unit. In some embodiments, part or all of the computer program can be loaded and / or installed on the device via the ROM and / or the communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more steps of the methods S1-S5 described above can be performed. Alternatively, in other embodiments, the CPU can be configured to perform the methods S1-S5 by any other appropriate means, such as by means of firmware.

[0117] The functionality described above in this document can be performed, at least in part, by one or more hardware logic components. For example, and without limitation, example types of hardware logic components that can be used include field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip systems (SOCs), complex programmable logic devices (CPLDs), etc.

[0118] Program code for carrying out the methods of the present application can be written in any combination of one or more programming languages. The program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, causes the machine to perform the functions / acts specified in the flow diagrams and / or block diagrams. The program code can execute entirely on a machine, partly on a machine, as a stand-alone software package, partly on a machine and partly on a remote machine or entirely on a remote machine or server.

[0119] In the context of the present application, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable storage medium can include, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage medium would include one or more lines of electrical wire, portable computer diskette, hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the foregoing.

[0120] The above description is only specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present application, and these modifications or replacements should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for intelligent connected vehicle cloud path tracking control, characterized in that, The method iteratively performs the following steps to realize the cloud path tracking control of the vehicle, comprising: A vehicle tracking error model is constructed based on the vehicle state, which includes vehicle lateral error and vehicle heading error; A historical time delay data sequence and corresponding signal strength are obtained, and a feature sequence is generated. Based on the feature sequence, a trained time delay prediction model is used to output a time delay prediction value at the next time and an uncertainty of the time delay prediction value; A service guarantee level parameter is constructed. Based on the service guarantee level parameter, the time delay prediction value, and the uncertainty, a QoS mapping is performed to obtain a time delay confidence upper bound of the communication time delay, and a preview distance is obtained based on the time delay confidence upper bound; Based on the preview distance, a preview tracking error is calculated using the vehicle tracking error model, and a steering control law is designed based on the preview tracking error; the preview tracking error includes a preview heading error and a preview lateral error; Based on the steering control law, intelligent connected vehicle cloud path tracking control is performed, and the vehicle state after the vehicle executes the steering control law is obtained. 2.The intelligent connected vehicle cloud path tracking control method of claim 1, wherein, The method for constructing the vehicle tracking error model is: Define the vehicle state vector as wherein denotes the vehicle coordinates, denotes the vehicle heading angle, denotes the vehicle longitudinal velocity; denotes the transposition operation; Reference point of vehicle on reference trajectory Reference point coordinates of , and at the reference point, there are reference tangent heading angle and reference curvature respectively and ; Based on the vehicle state vector, the reference point coordinates, the reference tangent heading angle, and the reference curvature, the vehicle lateral error and the vehicle heading error are calculated, which are: ， wherein, represents a vehicle lateral error; represents a vehicle heading error; represents an angle normalization operation. 3.The intelligent connected vehicle cloud path tracking control method of claim 1, wherein, In the time delay prediction model, the feature sequence is processed through a forget gate, an input gate, and an output gate in the time delay prediction model to generate a hidden state at the current time; The signal strength at the current time is obtained, the importance score of the hidden state is calculated, and the attention weight is calculated based on the importance score, which is: ， ， wherein, denotes the hidden state at the current time t denotes the importance score of the hidden state denotes the attention vector denotes the transpose operation denotes the weight matrix of the hidden state denotes the weight matrix of the signal strength denotes the signal strength sequence at time t denotes the bias term denotes the attention weight denotes the total time length 4.The intelligent connected vehicle cloud path tracking control method of claim 1, wherein, The method for obtaining the time delay confidence upper bound is: based on the service assurance level parameter The reliability coefficient is calculated as: , represents the reliability coefficient; represents the service assurance level coefficient; Based on the reliability coefficient, a time delay confidence upper bound is calculated as: , denotes the time delay confidence upper bound; denotes the time delay prediction value; denotes the uncertainty. 5.The intelligent networked vehicle cloud path tracking control method of claim 1, wherein, The method for obtaining the preview distance is: ， denotes a preview distance; denotes a nominal preview distance taken without considering communication latency; denotes a vehicle longitudinal speed; denotes a latency confidence upper bound; denotes a latency prediction value; denotes an uncertainty; denotes a reliability coefficient. 6.The intelligent networked vehicle cloud path tracking control method of claim 1, wherein, The method for constructing the steering control law is: represents a steering control law; represents a preview lateral error, and , represents a preview distance corresponding vehicle heading angle at a reference point, represents a preview distance corresponding vehicle coordinates at a reference point, represents vehicle coordinates; represents a preview distance; represents a lateral error gain; represents a preview heading error, and , represents a vehicle heading angle, represents an angle normalization operation; represents a vehicle longitudinal speed. 7.The intelligent vehicle cloud path tracking control method of claim 6, wherein, The calculation method of the lateral error gain is: ， wherein, represents a nominal preview distance employed without considering communication latency; represents an adjustment parameter.

8. An intelligent connected vehicle cloud path tracking control system, characterized in that, The system includes a vehicle end controller, a wireless communication module, and a cloud control center. The cloud control center includes: A communication time delay acquisition module is used to acquire a historical time delay data sequence and corresponding signal strength; An error model construction module is used to construct a vehicle tracking error model based on the vehicle state acquired by the vehicle end controller, which includes vehicle lateral error and vehicle heading error; A time delay prediction network module is used to generate a feature sequence based on the historical time delay data sequence and corresponding signal strength, and to output a time delay prediction value at the next time and an uncertainty of the time delay prediction value using a trained time delay prediction model based on the feature sequence; A QoS index mapping module is used to construct a service guarantee level parameter, perform QoS mapping based on the service guarantee level parameter, the time delay prediction value, and the uncertainty to obtain a time delay confidence upper bound of the communication time delay, and obtain a preview distance based on the time delay confidence upper bound; A path tracking control module, which calculates a preview tracking error based on the preview distance using the vehicle tracking error model, designs a steering control law based on the preview tracking error, and issues the steering control law to a vehicle-side controller through the wireless communication module to perform cloud path tracking control of the intelligent connected vehicle based on the steering control law; the preview tracking error includes a preview heading error and a preview lateral error. 9.The intelligent vehicle cloud path tracking control system of claim 8, wherein, The vehicle-side controller is arranged at the vehicle side to collect vehicle states and add local time stamps to the collected data; the vehicle-side controller is also used to receive the steering control law issued by the cloud control center and convert the steering control law into execution signals of a vehicle target steering execution mechanism and a target driving execution mechanism; the vehicle states include pose information and running state information. 10.The intelligent vehicle cloud path tracking control system of claim 8, wherein, The wireless communication module is located at the vehicle side to realize communication between the vehicle-side controller and the cloud control center; during communication, the module records or carries network side information and sends the network side information to the communication time delay collection module; the network side information includes a sending time stamp, a receiving time stamp, and a current communication network signal strength.

Images