Transverse and longitudinal coordinated control method and system for unmanned vehicle on large curvature curve

Abstract

The present invention relates to a method and system for coordinated lateral and longitudinal control of an unmanned vehicle on a curve with large curvature, which comprises: establishing a kinematic lateral and longitudinal coupling MPC model including centripetal acceleration constraints, and tracking the trajectory of the unmanned vehicle according to the MPC model; establishing an FNN model based on laser radar point cloud data, and performing real-time obstacle avoidance of the unmanned vehicle according to the FNN model; integrating the MPC model with the FNN model, and performing coordinated lateral and longitudinal control of the unmanned vehicle on a curve with large curvature to achieve safe driving. The present invention can achieve safe driving of the unmanned vehicle in the case of trajectory loss according to the planned trajectory state. The present invention can be applied in the field of unmanned vehicle control.

Description

Technical Field

[0001] The present invention relates to the field of unmanned vehicle control, and in particular to a lateral and longitudinal coordinated control method and system for an unmanned vehicle on a curve with a large curvature. Background Art

[0002] Ensuring the safety of driverless vehicles on curved roads is an urgent need for vehicle control. Driverless vehicles are non-complete motion constraint systems with multiple degrees of freedom, highly nonlinear characteristics, and model parameter uncertainty. Their underlying motion control problem is a complex nonlinear system control problem with multiple inputs and multiple outputs. At present, the motion control of driverless vehicles is mostly focused on traditional control algorithms with independent lateral and longitudinal control. Although it simplifies the complexity of a single problem and can achieve relatively good results on common roads, it has system nonlinear characteristics, time lags, and random uncertainties. It is difficult to achieve good control effects on certain specific roads, such as curves with large curvatures and sections with small traversable areas. The data-driven intelligent control algorithm can solve the problem of complex model establishment, make up for the disadvantages of modern control theory, and combine the intelligent control algorithm with the traditional control algorithm to better solve the problems of model establishment and control accuracy.

[0003] As the earliest proposed intelligent control algorithm, fuzzy control has been unanimously recognized by researchers for its robustness. However, due to the extremely high requirements of fuzzy controllers on input rule experience, there is no mature fuzzy rule base in many application scenarios in many fields, and researchers need to establish it themselves according to research and application scenarios. In order to meet the needs of real-time online optimization control and improve the vehicle stability control performance under multiple constraints, model predictive control, as a new control method with excellent performance, can effectively perform rolling optimization under dynamic constraint sets. Existing lateral control algorithms for unmanned vehicles mostly rely on the reliable trajectory given by the planning layer, and do not consider the safety control strategy when the trajectory is lost.

[0004] As an indispensable part of traffic roads, temporary roads are unpredictable and have uncertain curve curvatures, which easily lead to failure of trajectory planning at the planning level. Summary of the invention

[0005] In view of the above problems, the purpose of the present invention is to provide a lateral and longitudinal coordinated control method and system for an unmanned vehicle on a curve with large curvature, which can achieve safe driving of the unmanned vehicle in the event of trajectory loss based on the planned trajectory state.

[0006] To achieve the above-mentioned purpose, the present invention adopts the following technical scheme: a method for coordinated lateral and longitudinal control of an unmanned vehicle on a curve with large curvature, which comprises: establishing an MPC model of kinematic lateral and longitudinal coupling including centripetal acceleration constraints, and tracking the trajectory of the unmanned vehicle according to the MPC model; establishing an FNN model based on lidar point cloud data, and performing real-time obstacle avoidance of the unmanned vehicle according to the FNN model; integrating the MPC model with the FNN model, and performing coordinated lateral and longitudinal control of the unmanned vehicle on a curve with large curvature to achieve safe driving.

[0007] Furthermore, the establishment of a kinematic lateral and longitudinal coupled MPC model including a centripetal acceleration constraint includes:

[0008] Establish a vehicle kinematic model with lateral and longitudinal coupling, and add longitudinal speed control parameters to the kinematic model;

[0009] Linearizing the kinematic model to obtain a corresponding state space equation;

[0010] Set the centripetal acceleration constraints for lateral and longitudinal control to prevent the driverless vehicle from losing the feasible domain during the turning process;

[0011] Construct the horizontal and vertical comprehensive evaluation function indicators, set the objective function, and adjust the gradient minimum dynamic adjustment coefficient to make the evaluation function indicator less than the preset value, so as to minimize the error between the current control state and the target control state.

[0012] Furthermore, the centripetal acceleration constraint of the lateral and longitudinal control is:

[0013]

[0014] Among them, a max is the maximum centripetal acceleration, v x is the longitudinal velocity of the vehicle, L is the length of the vehicle, and δ is the target steering angle of the vehicle.

[0015] Furthermore, the construction of the horizontal and vertical comprehensive evaluation function index and setting of the objective function include:

[0016] Construct the horizontal and vertical comprehensive evaluation function index and set N p prediction horizon and N c After the control time domain, the objective function is set as:

[0017]

[0018] Where Y is the evaluation function index of the MPC model, Q represents the minimum dynamic adjustment coefficient of the gradient of the curve tracking error, R represents the minimum dynamic adjustment coefficient of the gradient of the control efficiency, X(k+i)-X ref(k+i) represents the trajectory tracking effect, represents the control efficiency, and k represents the time domain coefficient.

[0019] Further, the establishing of the FNN model and performing obstacle avoidance according to the FNN model include:

[0020] Collect data sets including obstacle location information, vehicle speed and corner feedback information as training sets and validation sets for the FNN model;

[0021] Using laser radar as the environment to perceive the road environment and determine the fuzzy domain of the target perceived obstacle information;

[0022] Combining the logical rationality and transparency of fuzzy logic control and the self-adaptation of artificial neural network parameters, a fuzzy neural network model is set up with the position information of the obstacle relative to the laser radar as input and the target speed and angle as output;

[0023] The obstacle position, speed and turning angle data collected in advance are input into the fuzzy neural network model, and the actual output value of each group of data network and the expected output value in the data are obtained through self-supervised learning. The two are compared to calculate the error, and the coefficients and parameters of the fuzzy neural network model are corrected based on the error. The real-time obstacle avoidance of the unmanned vehicle is achieved through the trained fuzzy neural network model.

[0024] Further, the setting of the fuzzy neural network model includes:

[0025] For the input of m-dimensional obstacle position information, the membership degree of each input variable is calculated according to the fuzzy rule;

[0026] Perform fuzzy calculation on each membership degree, and the fuzzy operator adopts the continuous multiplication operator;

[0027] Calculate the network output value according to the fuzzy calculation result;

[0028] Calculate the error between the network output and the expected output;

[0029] The coefficients and parameters are corrected according to the error obtained from each input to obtain the corrected fuzzy neural network model.

[0030] Furthermore, the MPC model is integrated with the FNN model to perform lateral and longitudinal coordinated control of the unmanned vehicle on a curve with large curvature, including:

[0031] Determine whether there is a successfully planned trajectory. If the time of trajectory planning failure is less than the preset time t set , it is considered as normal jitter of planning, and the motion state of the previous moment is maintained and continued to move; if the time of trajectory planning failure is greater than the preset time t set, then it is considered that the trajectory planning has failed. At this time, the trajectory tracking mode is converted to the obstacle avoidance mode, and the FNN model is used to avoid obstacles for the unmanned vehicle; if the trajectory planning is successful, the duration of the stable trajectory is determined. If the stable time parameter is greater than several times the preset time t set , the MPC model is used to track the trajectory of the unmanned vehicle.

[0032] A lateral and longitudinal coordinated control system for an unmanned vehicle on a curve with large curvature comprises: a trajectory tracking module, which establishes a kinematic lateral and longitudinal coupled MPC model including a centripetal acceleration constraint, and performs trajectory tracking of the unmanned vehicle according to the MPC model; an obstacle avoidance module, which establishes an FNN model based on laser radar point cloud data, and performs real-time obstacle avoidance of the unmanned vehicle according to the FNN model; and a fusion module, which fuses the MPC model with the FNN model, performs lateral and longitudinal coordinated control of the unmanned vehicle on a curve with large curvature, and realizes safe driving.

[0033] A computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any of the above methods.

[0034] A computing device comprises: one or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any of the above methods.

[0035] The present invention adopts the above technical solution, which has the following advantages:

[0036] 1. The present invention improves the traditional MPC algorithm by establishing a kinematic lateral and longitudinal coupling model, realizes lateral and longitudinal coordinated control of trajectory tracking, and ensures that the unmanned vehicle can travel normally with a trajectory.

[0037] 2. The present invention applies the FNN algorithm to the motion control of unmanned vehicles, improves the accuracy of the model through offline training of the model, and improves the obstacle avoidance performance of the unmanned vehicle.

[0038] 3. To address the problem of trajectory loss during curve driving, the present invention sets a safety control strategy for the mutual coordination and conversion of the trajectory tracking controller and the obstacle avoidance controller, so that the unmanned vehicle can drive safely even when the trajectory is lost on a temporary road. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is an overall flow chart of a horizontal and vertical coordinated control method in one embodiment of the present invention;

[0040] Figure 2 is a detailed flow chart of a horizontal and vertical coordinated control method in one embodiment of the present invention;

[0041] Figure 3 is the fuzzy domain of the laser radar sensing area in one embodiment of the present invention;

[0042] Figure 4 is a fuzzy neural network structure diagram in one embodiment of the present invention;

[0043] Figure 5 It is a schematic diagram of the structure of a computing device in one embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0045] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0046] The present invention provides a lateral and longitudinal coordinated control method and system for an unmanned vehicle on a curve with a large curvature, mainly aiming at the safe driving of the unmanned vehicle on a curve with a large curvature on a temporary road.

[0047] In view of the fact that the unpredictability of temporary roads and the uncertainty of curve curvature easily lead to the failure of trajectory planning at the planning layer, the present invention adds a safety redundancy method for obstacle avoidance at the control layer. Based on the model predictive control algorithm and the fuzzy neural network algorithm, the safe driving of the unmanned vehicle in the event of trajectory loss is achieved according to the planned trajectory state.

[0048] In one embodiment of the present invention, a method for coordinated lateral and longitudinal control of an unmanned vehicle on a large curvature curve is provided. This embodiment uses the method applied to a terminal as an example for illustration. It is understandable that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the present invention is used to achieve safe driving of an unmanned vehicle on a large curvature curve on a temporary road, such as Figure 1 As shown, the method comprises the following steps:

[0049] 1) Establish a kinematic lateral and longitudinal coupling MPC model including centripetal acceleration constraints, and track the trajectory of the unmanned vehicle based on the MPC model;

[0050] 2) Establish an FNN model based on the lidar point cloud data, and perform real-time obstacle avoidance for unmanned vehicles based on the FNN model;

[0051] 3) The MPC model is integrated with the FNN model to perform lateral and longitudinal coordinated control of the unmanned vehicle on large curvature curves to achieve safe driving.

[0052] In the above step 1), the MPC control algorithm is improved to achieve tracking of the curve trajectory. On the basis of the basic dynamic kinematic model of the vehicle, the lateral and longitudinal kinematic coupling conditions and the centripetal acceleration constraint conditions are added to establish a lateral and longitudinal coupled vehicle kinematic model. Since the vehicle is a typical complex nonlinear system, it is necessary to convert the established model into a linear model through Taylor expansion and then establish the state space equation. The MPC control algorithm is used to predict, feedback correct, and roll the established model, and then the optimal solution is obtained for the established objective function to achieve tracking of the curve trajectory by the unmanned vehicle and improve the tracking accuracy.

[0053] Specifically, Figure 2 As shown in the figure, a kinematic lateral and longitudinal coupled MPC model including centripetal acceleration constraint is established, including the following steps:

[0054] 1.1) Establish a vehicle kinematic model with lateral and longitudinal coupling, and add longitudinal speed control parameters to the kinematic model;

[0055] Since the original kinematic model only includes the lateral control of the vehicle, and the vehicle is a complex lateral and longitudinal coupling system, the longitudinal speed control parameter is added to the lateral kinematic model to achieve lateral and longitudinal coordinated control of the unmanned vehicle. In this embodiment, the lateral and longitudinal coupled vehicle kinematic model is:

[0056]

[0057] State quantity Where (x, y) is the coordinate position of the target path point relative to the unmanned vehicle. is the yaw angle of the target path point relative to the vehicle, v x is the longitudinal speed of the vehicle; the control quantity u = [δ, a], where δ is the target steering angle of the vehicle, L is the length of the vehicle, and a is the acceleration command, which will be mapped to the torque control command through a table lookup later.

[0058] 1.2) Linearize the kinematic model to obtain the corresponding state space equation;

[0059] Since the vehicle system of Ackerman steering is a nonlinear system, the kinematic model cannot be directly converted into the state space equation. It needs to be transformed by a series of mathematical methods to linearize it and then expressed as the state space equation. r ,u r ) is expanded as shown in equations (2) to (5):

[0060]

[0061]

[0062]

[0063]

[0064] Among them, X r Represents the state quantity at any point in the state space equation, u r Represents the control quantity corresponding to any point in the state space equation.

[0065] Combine equation (2) with the state variable of the random Taylor expansion point Subtracting them, we get equation (6) and its new linearized state space equation (7):

[0066]

[0067]

[0068] make Then, according to the forward Euler method, the predicted value at the next moment can be calculated. As shown in formula (8):

[0069]

[0070] Wherein, T represents the sampling period between two adjacent time domain states, A represents the state matrix of the system, B represents the control matrix of the system, and k represents the time domain coefficient.

[0071] 1.3) In order to ensure the coordination between speed and turning angle and not lose the real feasible domain, the centripetal acceleration constraints of lateral and longitudinal control are set to prevent the unmanned vehicle from losing the feasible domain during the turning process;

[0072] Among them, the centripetal acceleration constraints of the lateral and longitudinal control are:

[0073]

[0074] Among them, a max is the maximum centripetal acceleration, vx is the longitudinal velocity of the vehicle, L is the length of the vehicle, and δ is the target steering angle of the vehicle.

[0075] 1.4) Construct the horizontal and vertical comprehensive evaluation function index, set the objective function, and adjust the gradient minimum dynamic adjustment coefficient to make the evaluation function index less than the preset value, so that the error between the current control state and the target control state is minimized;

[0076] Among them, construct the horizontal and vertical comprehensive evaluation function indicators and set N p prediction horizon and N c After the control time domain, the objective function is set as:

[0077]

[0078] Where Y is the evaluation function index of the MPC model, Q represents the minimum dynamic adjustment coefficient of the gradient of the curve tracking error, R represents the minimum dynamic adjustment coefficient of the gradient of the control efficiency, X(k+i)-X ref (k+i) represents the trajectory tracking effect, represents the control efficiency, and k represents the time domain coefficient.

[0079] By dynamically adjusting the coefficients Q and R according to the minimum gradient, as shown in equations (11) and (12), the evaluation function index Y is kept relatively small (i.e., smaller than the preset value), so that the error between the current control state and the target control state is minimized.

[0080]

[0081]

[0082] Where γ represents the gradient adjustment parameter of coefficient Q, and δ represents the gradient adjustment parameter of coefficient R.

[0083] In the above step 2), for the case of trajectory loss, vehicle obstacle avoidance is studied based on the FNN model. Use LiDAR to perceive the road environment, and construct a fuzzy neural network controller with the position (angle and distance) information of the obstacle relative to the LiDAR as input and the target speed and angle as output. The neural network mainly includes an input layer, a fuzzification layer, a fuzzy rule layer, a defuzzification layer and an output layer. The weight coefficient and the parameters of the membership function are corrected through the error between the network output and the actual reference value during each training, so that it is finally close to the actual output of the data set. Finally, the real-time obstacle avoidance of unmanned vehicles is achieved through the trained model.

[0084] Specifically, Figure 2 As shown, an FNN model is established, and obstacle avoidance is performed according to the FNN model, including the following steps:

[0085] 2.1) Collecting data sets including obstacle location (distance and angle) information, vehicle speed and turning angle feedback information as training and validation sets for the FNN model;

[0086] In this embodiment, data collection is performed through manual driving.

[0087] 2.2) Using LiDAR as the environment to perceive the road environment and determine the fuzzy domain of the target perceived obstacle information;

[0088] The specific parameter values ​​of the input and output variables are normalized and mapped to appropriate fuzzy sets.

[0089] In this embodiment, the 180° area in front of the laser radar is taken as the region of interest and evenly divided into d1~d 18 The five regions represent the angles of obstacles relative to the vehicle, represented by the fuzzy set [L8, L7, …, L1, L, R, …, R7, R8]. The distance is divided into five regions, represented by the fuzzy set [DN, DNM, DM, DMF, DF]. Figure 3 shown.

[0090] 2.3) Combining the logical rationality and transparency of fuzzy logic control and the self-adaptation of artificial neural network parameters, a fuzzy neural network model is set up with the position information of the obstacle relative to the laser radar as input and the target speed and angle as output;

[0091] The fuzzy neural network model is a five-layer fuzzy neural network including an input layer, a fuzzification layer, a fuzzy rule layer, a defuzzification layer and an output layer.

[0092] Wherein, setting the fuzzy neural network model includes the following steps:

[0093] 2.3.1) For an m-dimensional input x = [x1, x2, ..., x m ], calculate each input variable x according to the fuzzy rules j The membership degree of , j∈m;

[0094] Among them, the membership function The Gaussian type is used, as shown in formula (13):

[0095]

[0096] in, are the center and width of the membership function respectively; m is the dimension of the input parameter (i.e. the number of eigenvectors); n is the number of fuzzy subsets.

[0097] 2.3.2) Perform fuzzy calculation on each membership degree, and the fuzzy operator ω i The continuous multiplication operator is used, as shown in formula (14):

[0098]

[0099] 2.3.3) Calculate the network output value y based on the fuzzy calculation result i ;

[0100]

[0101] in, Represents the weight coefficient of the mth input quantity in the i-th calculation.

[0102] 2.3.4) Calculate the error e between the network output and the expected output as shown in formula (16):

[0103]

[0104] Among them, y d is the expected output of the network, y c is the actual output of the network.

[0105] 2.3.5) Correct the coefficients and parameters according to the error obtained by each input to obtain the corrected fuzzy neural network model;

[0106] As shown in formulas (17) to (19):

[0107]

[0108]

[0109]

[0110] in, represents the weight coefficient of the jth input quantity in the i-th calculation, represents the center of the membership function of the jth input quantity in the i-th calculation, It represents the width of the membership function of the j-th input quantity in the i-th calculation.

[0111] 2.4) Input the pre-collected obstacle position, unmanned vehicle speed and turning angle data based on LiDAR into the fuzzy neural network model (such as Figure 4 As shown in the figure, the actual output value of each data network and the expected output value in the data are obtained through self-supervised learning, the two are compared to calculate the error, and the coefficients and parameters of the fuzzy neural network model are corrected based on the error. The real-time obstacle avoidance of the unmanned vehicle is achieved through the trained fuzzy neural network model.

[0112] In the above step 3), the MPC model is integrated with the FNN model to perform lateral and longitudinal coordinated control of the unmanned vehicle on a curve with large curvature, specifically:

[0113] Since large curvature curves can easily lead to trajectory planning failure, we first determine whether there is a successfully planned trajectory. If the time for trajectory planning failure is less than the preset time t set , it is considered as normal jitter of planning, and the motion state of the previous moment is maintained and continued to move; if the time of trajectory planning failure is greater than the preset time t set , it is considered that the trajectory planning has failed. At this time, the control mode will be switched, the trajectory tracking mode will be converted to the obstacle avoidance mode, and the FNN model will be used to avoid obstacles for the unmanned vehicle; if the trajectory planning is successful, the duration of the stable trajectory will be determined. If the stable time parameter is greater than several times the preset time t set , the MPC model is used to track the trajectory of the unmanned vehicle.

[0114] In one embodiment of the present invention, a lateral and longitudinal coordinated control system for an unmanned vehicle on a curve with a large curvature is provided, comprising:

[0115] The trajectory tracking module establishes a kinematic lateral and longitudinal coupled MPC model including centripetal acceleration constraints, and performs trajectory tracking of the unmanned vehicle based on the MPC model;

[0116] The obstacle avoidance module builds an FNN model based on the lidar point cloud data and performs real-time obstacle avoidance for unmanned vehicles according to the FNN model;

[0117] The fusion module integrates the MPC model with the FNN model to perform lateral and longitudinal coordinated control of the unmanned vehicle on large curvature curves to achieve safe driving.

[0118] The system provided in this embodiment is used to execute the above-mentioned method embodiments. Please refer to the above-mentioned embodiments for specific processes and detailed contents, which will not be repeated here.

[0119] like Figure 5As shown, it is a schematic diagram of the structure of a computing device provided in an embodiment of the present invention. The computing device can be a terminal, which can include: a processor, a communication interface, a memory, a display screen and an input device. Among them, the processor, the communication interface and the memory complete mutual communication through a communication bus. The processor is used to provide computing and control capabilities. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. When the computer program is executed by the processor, a control method is implemented; the internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be realized through WIFI, a management network, NFC (near field communication) or other technologies. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the housing of the computing device, or an external keyboard, touchpad or mouse. The processor can call the logic instructions in the memory to execute the following methods: establish an MPC model of kinematic lateral and longitudinal coupling including centripetal acceleration constraints, and track the trajectory of the unmanned vehicle according to the MPC model; establish an FNN model based on lidar point cloud data, and perform real-time obstacle avoidance for the unmanned vehicle according to the FNN model; integrate the MPC model with the FNN model, and perform lateral and longitudinal coordinated control of the unmanned vehicle on large curvature curves to achieve safe driving.

[0120] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0121] Those skilled in the art will understand that Figure 5The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the computing device to which the scheme of the present application is applied. The specific computing device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0122] In one embodiment of the present invention, a computer program product is provided, wherein the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, wherein the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the methods provided by the above-mentioned method embodiments, for example, including: establishing an MPC model of kinematic lateral and longitudinal coupling including centripetal acceleration constraints, and tracking the trajectory of an unmanned vehicle according to the MPC model; establishing an FNN model based on lidar point cloud data, and performing real-time obstacle avoidance for the unmanned vehicle according to the FNN model; integrating the MPC model with the FNN model, and performing lateral and longitudinal coordinated control of the unmanned vehicle on a curve with large curvature to achieve safe driving.

[0123] In one embodiment of the present invention, a non-transitory computer-readable storage medium is provided, which stores server instructions, and the computer instructions enable a computer to execute the methods provided by the above embodiments, for example, including: establishing an MPC model with kinematic lateral and longitudinal coupling including centripetal acceleration constraints, and tracking the trajectory of the unmanned vehicle according to the MPC model; establishing an FNN model based on lidar point cloud data, and performing real-time obstacle avoidance for the unmanned vehicle according to the FNN model; integrating the MPC model with the FNN model, and performing lateral and longitudinal coordinated control of the unmanned vehicle on a curve with large curvature to achieve safe driving.

[0124] The above embodiment provides a computer-readable storage medium, whose implementation principle and technical effect are similar to those of the above method embodiment, and will not be repeated here.

[0125] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0126] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0127] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for coordinated lateral and longitudinal control of an unmanned vehicle on a curve with large curvature, characterized in that: include: Establishing a kinematic lateral and longitudinal coupled MPC model including centripetal acceleration constraints, and tracking the trajectory of the unmanned vehicle according to the MPC model; Establishing an FNN model based on the laser radar point cloud data, and performing real-time obstacle avoidance for the unmanned vehicle according to the FNN model; The MPC model is integrated with the FNN model to perform lateral and longitudinal coordinated control of the unmanned vehicle on a curve with large curvature to achieve safe driving; The establishing of the FNN model and performing obstacle avoidance according to the FNN model include: Collect data sets including obstacle location information, vehicle speed and corner feedback information as training sets and validation sets for the FNN model; Using laser radar as the environment to perceive the road environment and determine the fuzzy domain of the target perceived obstacle information; Combining the logical rationality and transparency of fuzzy logic control and the self-adaptation of artificial neural network parameters, a fuzzy neural network model is set up with the position information of the obstacle relative to the laser radar as input and the target speed and angle as output; The obstacle position, speed and turning angle data collected in advance are input into the fuzzy neural network model, and the actual output value of each group of data network and the expected output value in the data are obtained through self-supervised learning. The two are compared to calculate the error, and the coefficients and parameters of the fuzzy neural network model are corrected based on the error. The real-time obstacle avoidance of the unmanned vehicle is achieved through the trained fuzzy neural network model.

2. The method for coordinated lateral and longitudinal control of an unmanned vehicle on a curve with a large curvature as claimed in claim 1, characterized in that: The establishment of a kinematic transverse and longitudinal coupled MPC model including a centripetal acceleration constraint comprises: Establish a vehicle kinematic model with lateral and longitudinal coupling, and add longitudinal speed control parameters to the kinematic model; Linearizing the kinematic model to obtain a corresponding state space equation; Set the centripetal acceleration constraints for lateral and longitudinal control to prevent the driverless vehicle from losing the feasible domain during the turning process; Construct the horizontal and vertical comprehensive evaluation function indicators, set the objective function, and adjust the gradient minimum dynamic adjustment coefficient to make the evaluation function indicator less than the preset value, so as to minimize the error between the current control state and the target control state.

3. The method for coordinated lateral and longitudinal control of an unmanned vehicle on a curve with a large curvature as claimed in claim 2, characterized in that: The centripetal acceleration constraints of the lateral and longitudinal control are: Among them, a max is the maximum centripetal acceleration, v x is the longitudinal velocity of the vehicle, L is the length of the vehicle, and δ is the target steering angle of the vehicle.

4. The method for coordinated lateral and longitudinal control of an unmanned vehicle on a curve with a large curvature as claimed in claim 2, characterized in that: The construction of the horizontal and vertical comprehensive evaluation function index and setting of the objective function include: Construct the horizontal and vertical comprehensive evaluation function index and set N p prediction horizon and N c After the control time domain, the objective function is set as: Where Y is the evaluation function index of the MPC model, Q represents the minimum dynamic adjustment coefficient of the gradient of the curve tracking error, R represents the minimum dynamic adjustment coefficient of the gradient of the control efficiency, X(k+i)-X ref (k+i) represents the trajectory tracking effect, represents the control efficiency, and k represents the time domain coefficient.

5. The method for coordinated lateral and longitudinal control of an unmanned vehicle on a curve with a large curvature as claimed in claim 1, characterized in that: The fuzzy neural network model is set, comprising: For the input of m-dimensional obstacle position information, the membership degree of each input variable is calculated according to the fuzzy rule; Perform fuzzy calculation on each membership degree, and the fuzzy operator adopts the continuous multiplication operator; Calculate the network output value according to the fuzzy calculation result; Calculate the error between the network output and the expected output; The coefficients and parameters are corrected according to the error obtained from each input to obtain the corrected fuzzy neural network model.

6. The method for coordinated lateral and longitudinal control of an unmanned vehicle on a curve with a large curvature as claimed in claim 1, characterized in that: The MPC model is integrated with the FNN model to perform lateral and longitudinal coordinated control of the unmanned vehicle on a curve with large curvature, including: Determine whether there is a successfully planned trajectory. If the time of trajectory planning failure is less than the preset time t set , it is considered as normal jitter of planning, and the motion state of the previous moment is maintained and continued to move; if the time of trajectory planning failure is greater than the preset time t set , then it is considered that the trajectory planning has failed. At this time, the trajectory tracking mode is converted to the obstacle avoidance mode, and the FNN model is used to avoid obstacles for the unmanned vehicle; if the trajectory planning is successful, the duration of the stable trajectory is determined. If the stable time parameter is greater than several times the preset time t set , the MPC model is used to track the trajectory of the unmanned vehicle.

7. A lateral and longitudinal coordinated control system for an unmanned vehicle on a curve with a large curvature, characterized in that: include: A trajectory tracking module, which establishes a kinematic lateral and longitudinal coupled MPC model including a centripetal acceleration constraint, and performs trajectory tracking of the unmanned vehicle according to the MPC model; The obstacle avoidance module establishes an FNN model based on the laser radar point cloud data and performs real-time obstacle avoidance for the unmanned vehicle according to the FNN model; The fusion module integrates the MPC model with the FNN model to perform lateral and longitudinal coordinated control of the unmanned vehicle on a curve with large curvature to achieve safe driving; The establishing of the FNN model and performing obstacle avoidance according to the FNN model include: Collect data sets including obstacle location information, vehicle speed and corner feedback information as training sets and validation sets for the FNN model; Using laser radar as the environment to perceive the road environment and determine the fuzzy domain of the target perceived obstacle information; Combining the logical rationality and transparency of fuzzy logic control and the self-adaptation of artificial neural network parameters, a fuzzy neural network model is set up with the position information of the obstacle relative to the laser radar as input and the target speed and angle as output; The obstacle position, speed and turning angle data collected in advance are input into the fuzzy neural network model, and the actual output value of each group of data network and the expected output value in the data are obtained through self-supervised learning. The two are compared to calculate the error, and the coefficients and parameters of the fuzzy neural network model are corrected based on the error. The real-time obstacle avoidance of the unmanned vehicle is achieved through the trained fuzzy neural network model.

8. A computer-readable storage medium storing one or more programs, characterized in that: The one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any one of the methods of claims 1 to 6.

9. A computing device, characterized in that include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any one of the methods described in claims 1 to 6.

Images