Self-adaptive man-machine co-driving system based on mode recognition and sharing control method of self-adaptive man-machine co-driving system

Abstract

The invention discloses a self-adaptive man-machine co-driving system based on mode recognition and a sharing control method thereof. The method comprises the steps that real-time motion state parameters and driver state parameters of a vehicle are obtained; the real-time motion state parameters of the vehicle and the driver state parameters are input into a preset HMM-SVM mode recognition model for mode recognition, and the optimal working mode of the vehicle is obtained; wherein the optimal working mode of the vehicle comprises one of a minimum lateral error mode, a minimum interference mode and a minimum lateral acceleration mode; inputting the real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle into a preset man-machine co-driving model to obtain an optimal auxiliary control input instruction under the optimal working mode of the vehicle; and performing control execution on a steering motor of the vehicle according to the optimal auxiliary control input instruction. The requirements for different modes in the lane changing driving process can be met, better driving experience is provided for a driver, and lane changing driving safety is effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of automobile safety technology, and in particular to an adaptive human-machine co-driving system based on pattern recognition and a shared control method thereof. Background Art

[0002] At present, the research focus of driver assistance systems has gradually shifted from the traditional "human adaptation to machine" to how to make human-machine collaborative and shared to complete driving tasks. For driver-intelligent system shared control, how to balance driving safety and driving experience is a challenging problem, especially considering different stages of movement and different types of drivers, whose driving needs are different. Summary of the invention

[0003] In order to solve the technical problems existing in the background technology, the present invention proposes an adaptive human-machine co-driving system based on pattern recognition and a shared control method thereof.

[0004] The present invention proposes a vehicle lane-changing sharing control method based on pattern recognition, comprising:

[0005] Obtaining real-time motion state parameters of the vehicle and driver state parameters;

[0006] The real-time motion state parameters of the vehicle and the driver state parameters are input into a preset HMM-SVM pattern recognition model for pattern recognition to obtain the optimal working mode of the vehicle; wherein the optimal working mode of the vehicle includes one of a minimum lateral error mode, a minimum interference mode and a minimum lateral acceleration mode;

[0007] Among them, in the minimum lateral error mode, when the car is driving in a straight line, the minimum lateral error is used as the performance indicator of human-machine shared control:

[0008] Where J(U)1 represents the performance index, tf represents the terminal time, t0 represents the initial time, y(t) represents the actual lateral position of the vehicle at time t, and y t (t) represent the ideal lateral position of the vehicle at time t, Indicates the maximum allowable lateral error;

[0009] In the minimum interference mode, when the car needs to move laterally and the required lateral displacement is small, the minimum conflict between man and machine is used as the performance indicator:

[0010] In the formula, J(U)2 represents the performance index, δ a represents the auxiliary control input, δ d Indicates the driver's control input. Indicates the maximum value allowed for human-machine interference;

[0011] In the minimum lateral acceleration mode, when the vehicle needs to move laterally and the vehicle speed is high or the required lateral displacement is large, the minimum lateral acceleration is used as the control target:

[0012] In the formula, J(U)3 represents the control target, a y represents the lateral acceleration, Indicates the maximum value allowed for lateral acceleration;

[0013] The real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle are input into a preset human-machine co-driving model to obtain the optimal auxiliary control input instructions under the optimal working mode of the vehicle;

[0014] The vehicle's steering motor is controlled and executed according to the optimal auxiliary control input command.

[0015] Preferably, the driver state parameters include steering wheel angle / torque and brake pedal force.

[0016] Preferably, the real-time motion state parameters include real-time driving environment information and real-time driving state information;

[0017] The real-time driving status information includes: the lateral speed v of the vehicle, the longitudinal speed u of the vehicle, the yaw angular velocity ω of the vehicle r , the total mass of the vehicle m, the moment of inertia of the vehicle about the vertical axis I z , the distance a from the center of mass of the vehicle to the front axle, the distance b from the center of mass of the vehicle to the rear axle, the front wheel turning angle δ, the front wheel cornering force F yf , rear wheel cornering force F yr , front wheel driving force / braking force F xf , rear wheel driving force / braking force F xr ;

[0018] Among them, the real-time driving environment information includes: rolling resistance F f , and F f = mgf, g is the acceleration of gravity, rolling resistance coefficient f, air resistance F w , F w =C D A(3.6u) 2 / 21.15, air resistance coefficient C D , windward area A.

[0019] Preferably, before inputting the real-time motion state parameters of the vehicle and the driver state parameters into a preset HMM-SVM pattern recognition model for pattern recognition and obtaining the optimal working mode of the vehicle, the method further includes:

[0020] Construct a training model based on HMM-SVM;

[0021] Constructing a training sample set; wherein the training sample set includes the motion state of the vehicle and the corresponding optimal working mode;

[0022] The training model is trained using the training sample set, and the trained model is used as a pattern recognition model.

[0023] Preferably, the human-machine co-driving model includes a three-degree-of-freedom vehicle model, a driver model and a manipulation inverse dynamics model, and the three-degree-of-freedom vehicle model and the driver model are combined to form a driver-car model.

[0024] Preferably, the real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle are input into a preset human-machine co-driving model to obtain the optimal auxiliary control input instruction under the optimal working mode of the vehicle, specifically including:

[0025] Inputting the real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle into the driver-vehicle model to obtain the driver's manipulation input under the optimal working mode of the vehicle;

[0026] Inputting the driver's control input into the control inverse dynamics model to obtain the optimal auxiliary control input;

[0027] An optimal auxiliary manipulation input instruction is generated according to the optimal auxiliary manipulation input.

[0028] Preferably, the state variables of the three-degree-of-freedom vehicle model are expressed as X v =[v, ω, θ, x, y, δ a ] T , the state equation of the three-degree-of-freedom vehicle model is expressed as X′ v =f[X(t),U(t)],U(t)=δ a_rate is the control variable;

[0029] Where, X v represents the state variable, X′ v represents the state equation, v represents the lateral velocity, ω represents the yaw angular velocity, θ represents the heading angle, x represents the lateral displacement, y represents the longitudinal displacement, δ a represents auxiliary manipulation input, T represents matrix transpose;

[0030] The state variables of the driver-car model are represented by X d-v =[v,ω,θ,x,y,δ a ,δ d ] T , where x, y are constrained by environmental information, δ and δ d Constrained by the vehicle structure and the driver’s physiological limit, δ = δ a+δ d ; Lateral acceleration a y The constraints are: |a y |≤0.4g,|v′+uω≤0.4g;

[0031] The state variables of the inverse dynamics model are X(t) = [v(t), w(t), u(t), x(t), y(t), θ(t)] T ;

[0032] Where v(t), w(t), u(t), x(t), y(t), and θ(t) are the lateral velocity, yaw angular velocity, longitudinal velocity, lateral displacement, longitudinal displacement, and heading angle at time t, respectively; T represents matrix transpose;

[0033] The control variable Z(t) of the inverse dynamics model is the steering wheel angle δ at time t. sw (t) and front wheel driving force / braking force F xf (t);

[0034] The control task of the manipulation command generation model is to pass through the given path with minimum lateral error / minimum human-machine conflict / minimum lateral acceleration.

[0035] Preferably, controlling the steering motor of the vehicle according to the optimal auxiliary manipulation input instruction specifically includes:

[0036] Entering a driving state corresponding to the optimal working mode of the vehicle according to the optimal working mode of the vehicle;

[0037] When driving, the vehicle's steering motor is controlled and executed according to the optimal auxiliary operation input command.

[0038] The present invention also proposes a vehicle lane-changing sharing control system based on pattern recognition, comprising a perception layer, a decision layer, a manipulation layer and an execution layer connected in sequence;

[0039] The perception layer is used to obtain the real-time motion state parameters of the vehicle and the driver's state parameters;

[0040] The decision layer is used to input the real-time motion state parameters of the vehicle and the driver state parameters into a preset HMM-SVM pattern recognition model for pattern recognition to obtain the optimal working mode of the vehicle; wherein the optimal working mode of the vehicle includes one of the minimum lateral error mode, the minimum interference mode and the minimum lateral acceleration mode;

[0041] The control layer is used to input the vehicle's real-time motion state parameters, the driver's state parameters and the vehicle's optimal working mode into the preset human-machine co-driving model to obtain the optimal auxiliary control input instructions under the vehicle's optimal working mode;

[0042] The execution layer is used to control the vehicle's steering motor according to the optimal auxiliary operation input command.

[0043] Preferably, in the minimum lateral error mode, when the car is traveling in a straight line, the minimum lateral error is used as a performance indicator of human-machine shared control:

[0044] Where J(U)1 represents the performance index, tf represents the terminal time, t0 represents the initial time, y(t) represents the actual lateral position of the vehicle at time t, and y t (t) represent the ideal lateral position of the vehicle at time t, Indicates the maximum allowable lateral error;

[0045] In the minimum interference mode, when the car needs to move laterally and the required lateral displacement is small, the minimum conflict between man and machine is used as the performance indicator:

[0046] In the formula, δ a represents the auxiliary control input, δ d Indicates the driver's control input. Indicates the maximum value allowed for human-machine interference;

[0047] In the minimum lateral acceleration mode, when the vehicle needs to move laterally and the vehicle speed is high or the required lateral displacement is large, the minimum lateral acceleration is used as the control target:

[0048] In the formula, a y represents the lateral acceleration, Indicates the maximum allowed lateral acceleration.

[0049] Preferably, the human-machine co-driving model includes a three-degree-of-freedom vehicle model, a driver model and a manipulation inverse dynamics model, and the three-degree-of-freedom vehicle model and the driver model are combined to form a driver-car model.

[0050] Preferably, the control layer includes: a driver control input calculation module, an optimal auxiliary control input calculation module and an instruction generation module;

[0051] The driver manipulation input calculation module is used to input the real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle into the driver-vehicle model to obtain the driver manipulation input under the optimal working mode of the vehicle;

[0052] The optimal auxiliary control input calculation module is used to input the driver's control input into the control inverse dynamics model to obtain the optimal auxiliary control input;

[0053] The instruction generation module is used to generate an optimal auxiliary manipulation input instruction according to the optimal auxiliary manipulation input.

[0054] Preferably, the driver state parameters include steering wheel angle / torque and brake pedal force;

[0055] The real-time motion state parameters include real-time driving environment information and real-time driving state information;

[0056] The real-time driving status information includes: the lateral speed v of the vehicle, the longitudinal speed u of the vehicle, the yaw angular velocity ω of the vehicle r , the total mass of the vehicle m, the moment of inertia of the vehicle about the vertical axis I z , the distance a from the center of mass of the vehicle to the front axle, the distance b from the center of mass of the vehicle to the rear axle, the front wheel turning angle δ, the front wheel cornering force F yf , rear wheel cornering force F yr , front wheel driving force / braking force F xf and rear wheel driving force / braking force F xr ;

[0057] Among them, the real-time driving environment information includes: rolling resistance F f , and F f = mgf, g is the acceleration of gravity, rolling resistance coefficient f, air resistance F w , F w =C D A(3.6u) 2 / 21.15, air resistance coefficient C D , windward area A.

[0058] In the present invention, the proposed adaptive human-machine co-driving system based on pattern recognition and its shared control method obtain the real-time motion state parameters and driver state parameters of the vehicle, and identify the vehicle's optimal working mode when changing lanes based on the driver state parameters and the real-time motion state parameters of the vehicle, so as to obtain the optimal auxiliary control input instructions when changing lanes based on the vehicle's real-time motion state parameters, the driver state parameters and the vehicle's optimal working mode, and then control the vehicle's steering motor according to the optimal auxiliary control input instructions to meet the needs of different modes during lane changing, provide the driver with a better driving experience, effectively ensure the safety of lane changing, and take into account personalized driving needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 A schematic diagram of the structure of an adaptive human-machine co-driving system based on pattern recognition in one embodiment of the present invention.

[0060] Figure 2 A schematic diagram of training the HMM-SVM pattern recognition model in one embodiment of the present invention. DETAILED DESCRIPTION

[0061] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0062] The present invention proposes a vehicle lane-changing sharing control method based on pattern recognition, comprising:

[0063] Obtaining real-time motion state parameters of the vehicle and driver state parameters;

[0064] The real-time motion state parameters of the vehicle and the driver state parameters are input into a preset HMM-SVM pattern recognition model for pattern recognition to obtain the optimal working mode of the vehicle; wherein the optimal working mode of the vehicle includes one of a minimum lateral error mode, a minimum interference mode and a minimum lateral acceleration mode;

[0065] The real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle are input into a preset human-machine co-driving model to obtain the optimal auxiliary control input instructions under the optimal working mode of the vehicle;

[0066] The vehicle's steering motor is controlled and executed according to the optimal auxiliary control input command.

[0067] The present invention can obtain the real-time motion state parameters and driver state parameters of the vehicle, and identify the optimal working mode of the vehicle when changing lanes according to the real-time motion state parameters and driver state parameters of the vehicle, so as to obtain the optimal auxiliary operation input instruction when changing lanes according to the real-time motion state parameters, driver state parameters and the optimal working mode of the vehicle, and then control the steering motor of the vehicle according to the optimal auxiliary operation input instruction to meet the needs of different modes during lane changing, provide the driver with a better driving experience, effectively ensure the safety of lane changing, and take into account personalized driving needs.

[0068] In this embodiment, the driver state parameters include steering wheel angle / torque and brake pedal force;

[0069] The real-time motion state parameters include real-time driving environment information and real-time driving state information;

[0070] The real-time driving status information includes: the lateral speed v of the vehicle, the longitudinal speed u of the vehicle, the yaw angular velocity ω of the vehicle r , the total mass of the vehicle m, the moment of inertia of the vehicle about the vertical axis I z , the distance from the center of mass of the vehicle to the front a, the distance from the center of mass of the vehicle to the rear axle b, the front wheel turning angle δ, the front wheel cornering force F yf , rear wheel cornering force F yr, front wheel driving force / braking force F xf , rear wheel driving force / braking force F xr ;

[0071] Among them, the real-time driving environment information includes: rolling resistance F f , and F f = mgf, g is the acceleration of gravity, rolling resistance coefficient f, air resistance F w , F w =C D A(3.6u) 2 / 21.15, air resistance coefficient C D , windward area A.

[0072] Among them, F xf ≥0 is driving force, F xf <0 is braking force; F xr ≥0 is driving force, F xr <0 is the braking force.

[0073] In this embodiment, in the minimum lateral error mode, when the car is traveling in a straight line, the minimum lateral error is used as a performance indicator of human-machine shared control:

[0074] Where J(U)1 represents the performance index, tf represents the terminal time, t0 represents the initial time, y(t) represents the actual lateral position of the vehicle at time t, and y t (t) represent the ideal lateral position of the vehicle at time t, Indicates the maximum allowable lateral error;

[0075] In the minimum interference mode, when the car needs to move laterally and the required lateral displacement is small, in order to avoid human-machine conflict and to prevent lane deviation caused by the driver's inattention, the minimum conflict between man and machine is used as the performance indicator:

[0076] In the formula, δ a represents the auxiliary control input, δ d Indicates the driver's control input. Indicates the maximum value allowed for human-machine interference;

[0077] In the minimum lateral acceleration mode, when the car needs to move laterally and the speed is high or the required lateral displacement is large, in order to avoid the risk of rollover caused by excessive lateral acceleration and consider the driver's riding comfort, the minimum lateral acceleration is used as the control target:

[0078] In the formula, a y represents the lateral acceleration, Indicates the maximum allowed lateral acceleration.

[0079] In this embodiment, before inputting the real-time motion state parameters of the vehicle and the driver state parameters into a preset HMM-SVM pattern recognition model for pattern recognition and obtaining the optimal working mode of the vehicle, the following is also included:

[0080] Construct a training model based on HMM-SVM;

[0081] Constructing a training sample set; wherein the training sample set includes the motion state of the vehicle and the corresponding optimal working mode;

[0082] The training model is trained using the training sample set, and the trained model is used as a pattern recognition model to ensure the accuracy of the pattern recognition model.

[0083] Among them, the pattern recognition model based on HMM-SVM includes an HMM model and an SVM model. The HMM model is used to make a preliminary judgment on the optimal working mode of the vehicle based on the real-time motion state parameters of the vehicle and output the maximum likelihood estimation value; the SVM model performs a second identification based on the output maximum likelihood estimation value to obtain a more accurate optimal working mode of the vehicle.

[0084] Specifically, the parameter λ of the HMM model is expressed as λ = [N, M, A, B, π]. Among them, N is 3 (i.e., the three modes of minimum lateral error, minimum lateral acceleration, and minimum interference), and M is 2 (i.e., vehicle speed and lateral displacement). Therefore, during the training process, the parameters to be determined by the HMM are A, B, and π, and the parameters to be determined by the SVM are the penalty factor c and the kernel function parameter g.

[0085] like Figure 2 As shown in the figure, during the training process, the time series signals of vehicle speed and lateral displacement are first collected; then the control mode corresponding to each time series signal segment is set according to experience; finally, the values ​​of A, B and π are obtained by reverse calculation according to the time series signal and the corresponding control mode. At this point, the model parameters of the HMM have been determined. When determining the parameters of the SVM (i.e., c and g), the time series signals of vehicle speed and lateral displacement are matched with each HMM to obtain a maximum likelihood estimation value group; after normalizing the maximum likelihood estimation value group, the SVM is applied to classify and identify the required control mode. In order to obtain the best classification result, c and g are cross-validated within an interval, and then a set of c and g that can achieve the highest recognition accuracy is selected as the model parameters.

[0086] In this embodiment, the human-machine co-driving model includes a three-degree-of-freedom vehicle model, a driver model and a manipulation inverse dynamics model. The three-degree-of-freedom vehicle model and the driver model are combined to form a driver-car model.

[0087] Among them, building a human-machine co-driving model specifically includes:

[0088] Construct a three-degree-of-freedom vehicle model;

[0089] Build a driver model;

[0090] Based on the establishment of a three-degree-of-freedom vehicle model and a driver model, a human-machine co-driving model is obtained.

[0091] Among them, the three-degree-of-freedom vehicle model is

[0092] Where v is the lateral velocity of the car; u is the longitudinal velocity of the car; ω r is the yaw angular velocity of the vehicle; m is the total mass of the vehicle; I z is the moment of inertia of the vehicle around the vertical axis; a and b are the distances from the center of mass of the vehicle to the front and rear axles respectively; δ is the front wheel turning angle; F yf F is the front wheel cornering force; yr is the rear wheel cornering force; F xf is the front wheel driving force / braking force, F xf ≥0 is driving force, F xf <0 is braking force; F xr is the rear wheel driving force / braking force, F xr ≥0 is driving force, F xr <0 is braking force; F f is the rolling resistance, and F f = mgf, g is the acceleration of gravity, f is the rolling resistance coefficient; F w is the air resistance, F w =C D A(3.6u) 2 / 21.15,C D is the air resistance coefficient, and A is the windward area.

[0093] Among them, the magic tire model is as follows:

[0094]

[0095] In the formula, B, C, D, E represent the curve fitting coefficients, F yf , F yr Represent the lateral forces of the front and rear wheels respectively.

[0096] The center of mass slip angle is as follows:

[0097]

[0098] The vehicle's trajectory is as follows:

[0099]

[0100] in:

[0101] In the formula, θ represents the heading angle, and β represents the sideslip angle;

[0102] In the formula, δ a Indicates auxiliary control input.

[0103] Therefore, the state variables of the vehicle model are represented as X v =[v,ω,θ,x,y,δ a ] T .

[0104] At this time, the state equation of the vehicle model can be expressed as X′ v =f[X(t),U(t)],U(t)=δ a_rate is the control variable. When only the auxiliary control input is considered, δ = δ a .

[0105] Among them, building a driver model specifically includes:

[0106] Define f(t) as the target path, the driver's preview time is T, the preview distance is d, d = uT;

[0107] According to the preview-follow theory, it is assumed that the lateral position of the car is on the target trajectory after T time, that is:

[0108]

[0109] When the car is moving at a constant speed and the front wheel turning angle is small, assuming Depend on and get

[0110]

[0111] according to and Available

[0112] In this embodiment, the state variables of the three-degree-of-freedom vehicle model are expressed as X v =[v, ω, θ, x, y, δ a ] T , the state equation of the three-degree-of-freedom vehicle model is expressed as X′ v =f[X(t),U(t)],U(t)=δ a_rate is the control variable;

[0113] Where, X v represents the state variable, X′ vrepresents the state equation, v represents the lateral velocity, ω represents the yaw angular velocity, θ represents the heading angle, x represents the lateral displacement, y represents the longitudinal displacement, δ a represents auxiliary manipulation input, T represents matrix transpose;

[0114] The state variables of the driver-car model are represented by X d-v =[v,ω,θ,x,y,δ a ,δ d ] T , where x, y are constrained by environmental information, δ and δ d Constrained by the vehicle structure and the driver's physiological limit respectively; when considering auxiliary control input and driver control input, δ = δ a +δ d ; Lateral acceleration a y The constraints are: |a y |≤0.4g,v′+uω≤0.4g; all constraints are recorded as: Cmin≤C≤Cmax.

[0115] The state variables of the inverse dynamics model are X(t) = [v(t), w(t), u(t), x(t), y(t), θ(t)] T ;

[0116] Where v(t), w(t), u(t), x(t), y(t), and θ(t) are the lateral velocity, yaw angular velocity, longitudinal velocity, lateral displacement, longitudinal displacement, and heading angle at time t, respectively; the superscript T indicates matrix transposition;

[0117] The control variable Z(t) of the inverse dynamics model is the steering wheel angle δ at time t. sw (t) and front wheel driving force / braking force F xf (t);

[0118] The control task of the manipulation command generation model is to pass through the given path with minimum lateral error / minimum human-machine conflict / minimum lateral acceleration.

[0119] In this embodiment, the real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle are input into the preset human-machine co-driving model to obtain the optimal auxiliary control input instructions under the optimal working mode of the vehicle, specifically including:

[0120] Inputting the real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle into the driver-vehicle model to obtain the driver's manipulation input under the optimal working mode of the vehicle;

[0121] Inputting the driver's control input into the control inverse dynamics model to obtain the optimal auxiliary control input;

[0122] An optimal auxiliary manipulation input instruction is generated according to the optimal auxiliary manipulation input.

[0123] In this embodiment, the driver's control input is input into the control inverse dynamics model to obtain the optimal auxiliary control input, which specifically includes:

[0124] The driver's control input is input into the inverse dynamics model, and the optimal control solution is obtained using the pseudo-spectral method.

[0125] Specifically, the solution process using the pseudo-spectral method is:

[0126] Discretize the state variables and control variables at a set of Legendre Gauss Radau (LGR) points;

[0127] Construct global Lagrange interpolation polynomials to approximate state variables and control variables;

[0128] By replacing the dynamic differential equations with the derivatives of the state variables, the continuous optimal control problem is transformed into a nonlinear programming problem subject to a series of algebraic constraints;

[0129] The hp adaptive algorithm is used to update the nonlinear programming (NLP) problem by subdividing the time grid or increasing the order of the interpolation polynomial until the accuracy requirement is met;

[0130] Solve the nonlinear programming problem and obtain the solution.

[0131] Reference Figure 1 ,The present invention also proposes a vehicle lane-changing sharing control system based on pattern recognition, comprising a perception layer, a decision layer, a manipulation layer and an execution layer connected in sequence;

[0132] The perception layer is used to obtain the real-time motion state parameters of the vehicle and the driver's state parameters;

[0133] The decision layer is used to input the real-time motion state parameters of the vehicle and the driver state parameters into a preset HMM-SVM pattern recognition model for pattern recognition to obtain the optimal working mode of the vehicle; wherein the optimal working mode of the vehicle includes one of the minimum lateral error mode, the minimum interference mode and the minimum lateral acceleration mode;

[0134] The control layer is used to input the vehicle's real-time motion state parameters, the driver's state parameters and the vehicle's optimal working mode into the preset human-machine co-driving model to obtain the optimal auxiliary control input instructions under the vehicle's optimal working mode;

[0135] The execution layer is used to control and execute the vehicle's steering motor according to the optimal auxiliary control input command;

[0136] Among them, in the minimum lateral error mode, when the car is driving in a straight line, the minimum lateral error is used as the performance indicator of human-machine shared control:

[0137] Where J(U)1 represents the performance index, tf represents the terminal time, t0 represents the initial time, y(t) represents the actual lateral position of the vehicle at time t, and y t (t) represent the ideal lateral position of the vehicle at time t, Indicates the maximum value allowed for lateral acceleration;

[0138] In the minimum interference mode, when the car needs to move laterally and the required lateral displacement is small, in order to avoid human-machine conflict and to prevent lane deviation caused by the driver's inattention, the minimum conflict between man and machine is used as the performance indicator:

[0139] In the formula, δ a represents the auxiliary control input, δ d Indicates the driver's control input. Indicates the maximum value allowed for human-machine interference;

[0140] In the minimum lateral acceleration mode, when the car needs to move laterally and the speed is high or the required lateral displacement is large, in order to avoid the risk of rollover caused by excessive lateral acceleration and consider the driver's riding comfort, the minimum lateral acceleration is used as the control target:

[0141] In the formula, a y represents the lateral acceleration, Indicates the maximum allowed lateral acceleration.

[0142] Among them, the driver state parameters include steering wheel angle / torque and brake pedal force;

[0143] The real-time motion state parameters include real-time driving environment information and real-time driving state information;

[0144] The real-time driving status information includes: the lateral speed v of the vehicle, the longitudinal speed u of the vehicle, the yaw angular velocity ω of the vehicle r , the total mass of the vehicle m, the moment of inertia of the vehicle about the vertical axis I z , the distance a from the center of mass of the vehicle to the front axle, the distance b from the center of mass of the vehicle to the rear axle, the front wheel turning angle δ, the front wheel cornering force F yf , rear wheel cornering force F yr , front wheel driving force / braking force F xf and rear wheel driving force / braking force F xr ;

[0145] Among them, the real-time driving environment information includes: rolling resistance F f , and F f = mgf, g is the acceleration of gravity, rolling resistance coefficient f, air resistance F w , F w =C D A(3.6u) 2 / 21.15, air resistance coefficient C D , windward area A.

[0146] Therefore, the perception layer in this embodiment includes: a driver state perception module, a vehicle driving environment perception module and a vehicle driving state perception module. The driver state perception module is used to obtain driver state parameters, the vehicle driving environment perception module is used to collect vehicle driving environment signals, and the vehicle driving state perception module is used to collect vehicle driving state signals; the driver state perception module, the vehicle driving environment perception module and the vehicle driving state perception module are respectively connected to the decision layer.

[0147] Specifically, the vehicle driving state perception module includes wheel sensors, side slip sensors and lateral acceleration sensors; the vehicle driving environment perception module includes visual recognition sensors, radar sensors, ultrasonic sensors and infrared sensors; the driver state perception module includes steering wheel angle / torque sensors and brake pedal sensors.

[0148] Specifically, the decision layer presets the HMM-SVM pattern recognition model

[0149] Specifically, the parameter λ of the HMM model is expressed as λ=[N, M, A, B, π], where N is 3 (i.e., the three modes of minimum lateral error, minimum lateral acceleration, and minimum interference), and M is 2 (i.e., vehicle speed and lateral displacement).

[0150] A human-machine co-driving model is preset in the control layer. The human-machine co-driving control submodule is used to input the vehicle's real-time motion state parameters, driver state parameters and the vehicle's optimal working mode into the preset human-machine co-driving model to obtain the optimal auxiliary control input instructions under the vehicle's optimal working mode.

[0151] Specifically, the human-machine co-driving model includes a three-degree-of-freedom vehicle model, a driver model and a manipulation inverse dynamics model. The three-degree-of-freedom vehicle model and the driver model jointly form a driver-car model.

[0152] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A vehicle lane-changing sharing control method based on pattern recognition, characterized in that: include: Obtaining real-time motion state parameters of the vehicle and driver state parameters; Inputting the real-time motion state parameters and the driver state parameters into a preset HMM-SVM pattern recognition model for pattern recognition to obtain the optimal working mode of the vehicle; wherein the optimal working mode of the vehicle includes one of a minimum lateral error mode, a minimum interference mode and a minimum lateral acceleration mode; Among them, in the minimum lateral error mode, when the car is driving in a straight line, the minimum lateral error is used as the performance indicator of human-machine shared control: Where J(U)1 represents the performance index, tf represents the terminal time, t0 represents the initial time, y(t) represents the actual lateral position of the vehicle at time t, and y t (t) represent the ideal lateral position of the vehicle at time t, Indicates the maximum allowable lateral error; In the minimum interference mode, when the car needs to move laterally and the required lateral displacement is small, the minimum conflict between man and machine is used as the performance indicator: In the formula, J(U)2 represents the performance index, δ a represents the auxiliary control input, δ d Indicates the driver's control input. Indicates the maximum value allowed for human-machine interference; In the minimum lateral acceleration mode, when the vehicle needs to move laterally and the vehicle speed is high or the required lateral displacement is large, the minimum lateral acceleration is used as the control target: In the formula, J(U)3 represents the control target, a y represents the lateral acceleration, Indicates the maximum value allowed for lateral acceleration; The real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle are input into a preset human-machine co-driving model to obtain the optimal auxiliary control input instructions under the optimal working mode of the vehicle; The vehicle's steering motor is controlled and executed according to the optimal auxiliary control input command.

2. The vehicle lane-changing sharing control method based on pattern recognition according to claim 1 is characterized in that: Driver state parameters include steering wheel angle / torque and brake pedal force; The real-time motion state parameters include real-time driving environment information and real-time driving state information; The real-time driving status information includes: the lateral speed v of the vehicle, the longitudinal speed u of the vehicle, the yaw angular velocity ω of the vehicle r , the total mass of the vehicle m, the moment of inertia of the vehicle about the vertical axis I z , the distance a from the center of mass of the vehicle to the front axle, the distance b from the center of mass of the vehicle to the rear axle, the front wheel turning angle δ, the front wheel cornering force F yf , rear wheel cornering force F yr , front wheel driving force / braking force F xf and rear wheel driving force / braking force F xr ; Among them, the real-time driving environment information includes: rolling resistance F f , and F f = mgf, g is the acceleration of gravity, rolling resistance coefficient f, air resistance F w , F w =C D A(3.6u) 2 / 21.15, air resistance coefficient C D , windward area A.

3. The vehicle lane-changing sharing control method based on pattern recognition according to claim 1, characterized in that: The real-time motion state parameters of the vehicle and the driver state parameters are input into the preset HMM-SVM pattern recognition model for pattern recognition. Before obtaining the optimal working mode of the vehicle, the following steps are also included: Construct a training model based on HMM-SVM; Constructing a training sample set; wherein the training sample set includes the motion state of the vehicle and the corresponding optimal working mode; The training model is trained using the training sample set, and the trained model is used as a pattern recognition model.

4. The vehicle lane-changing sharing control method based on pattern recognition according to claim 1 is characterized in that: The human-machine co-driving model includes a three-degree-of-freedom vehicle model, a driver model and a manipulation inverse dynamics model. The three-degree-of-freedom vehicle model and the driver model jointly form a driver-car model.

5. The vehicle lane-changing sharing control method based on pattern recognition according to claim 4 is characterized in that: The real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle are input into the preset human-machine co-driving model to obtain the optimal auxiliary control input instructions under the optimal working mode of the vehicle, specifically including: Inputting the real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle into the driver-vehicle model to obtain the driver's manipulation input under the optimal working mode of the vehicle; Inputting the driver's control input into the control inverse dynamics model to obtain the optimal auxiliary control input; An optimal auxiliary manipulation input instruction is generated according to the optimal auxiliary manipulation input.

6. The vehicle lane-changing sharing control method based on pattern recognition according to claim 4 is characterized in that: The state variables of the three-degree-of-freedom vehicle model are expressed as X v =[v, ω, θ, x, y, δ a ] T , the state equation of the three-degree-of-freedom vehicle model is expressed as X′ v =f[X(t),U(t)],U(t)=δ a_rate is the control variable; Where, X v represents the state variable, X′ v represents the state equation, v represents the lateral velocity, ω represents the yaw angular velocity, θ represents the heading angle, x represents the lateral displacement, y represents the longitudinal displacement, δ a represents auxiliary manipulation input, T represents matrix transpose; The state variables of the driver-car model are represented by X d-v =[v,ω,θ,x,y,δ a ,δ d ] T , where x, y are constrained by environmental information, δ and δ d Constrained by the vehicle structure and the driver’s physiological limit, δ = δ a +δ d ; Lateral acceleration a y The constraints are: |a y |≤0.4g, |v′+uω|≤0.4g; The state variables of the inverse dynamics model are X(t) = [v(t), w(t), u(t), x(t), y(t), θ(t)] T ; Where v(t), w(t), u(t), x(t), y(t), and θ(t) are the lateral velocity, yaw angular velocity, longitudinal velocity, lateral displacement, longitudinal displacement, and heading angle at time t, respectively; T represents matrix transpose; The control variable Z(t) of the inverse dynamics model is the steering wheel angle δ at time t. sw (t) and front wheel driving force / braking force F xf (t); The control task of the manipulation command generation model is to pass through the given path with minimum lateral error / minimum human-machine conflict / minimum lateral acceleration.

7. A vehicle lane-changing sharing control system based on pattern recognition, characterized in that: It includes a perception layer, a decision layer, a manipulation layer, and an execution layer connected in sequence; The perception layer is used to obtain the real-time motion state parameters of the vehicle and the driver's state parameters; The decision layer is used to input the real-time motion state parameters of the vehicle and the driver state parameters into a preset HMM-SVM pattern recognition model for pattern recognition to obtain the optimal working mode of the vehicle; wherein the optimal working mode of the vehicle includes one of the minimum lateral error mode, the minimum interference mode and the minimum lateral acceleration mode; The control layer is used to input the vehicle's real-time motion state parameters, the driver's state parameters and the vehicle's optimal working mode into the preset human-machine co-driving model to obtain the optimal auxiliary control input instructions under the vehicle's optimal working mode; The execution layer is used to control and execute the vehicle's steering motor according to the optimal auxiliary control input command; Among them, in the minimum lateral error mode, when the car is driving in a straight line, the minimum lateral error is used as the performance indicator of human-machine shared control: Where J(U) represents the performance index, tf represents the terminal time, t0 represents the initial time, y(t) represents the actual lateral position of the vehicle at time t, and y t (t) represent the ideal lateral position of the vehicle at time t, Indicates the maximum allowable lateral error; In the minimum interference mode, when the car needs to move laterally and the required lateral displacement is small, the minimum conflict between man and machine is used as the performance indicator: In the formula, δ a represents the auxiliary control input, δ d Indicates the driver's control input. Indicates the maximum value allowed for human-machine interference; In the minimum lateral acceleration mode, when the vehicle needs to move laterally and the vehicle speed is high or the required lateral displacement is large, the minimum lateral acceleration is used as the control target: In the formula, a y represents the lateral acceleration, Indicates the maximum allowed lateral acceleration.

8. The vehicle lane-changing sharing control system based on pattern recognition according to claim 7 is characterized in that: The human-machine co-driving model includes a three-degree-of-freedom vehicle model, a driver model and a manipulation inverse dynamics model. The three-degree-of-freedom vehicle model and the driver model jointly form a driver-car model.

9. The vehicle lane-changing sharing control method based on pattern recognition according to claim 8 is characterized in that: The control layer includes: a driver control input calculation module, an optimal auxiliary control input calculation module and an instruction generation module; The driver manipulation input calculation module is used to input the real-time motion state parameters of the vehicle, the driver state parameters and the optimal working mode of the vehicle into the driver-vehicle model to obtain the driver manipulation input under the optimal working mode of the vehicle; The optimal auxiliary control input calculation module is used to input the driver's control input into the control inverse dynamics model to obtain the optimal auxiliary control input; The instruction generation module is used to generate an optimal auxiliary manipulation input instruction according to the optimal auxiliary manipulation input.

10. The vehicle lane-changing sharing control system based on pattern recognition according to claim 7, characterized in that: Driver state parameters include steering wheel angle / torque and brake pedal force; The real-time motion state parameters include real-time driving environment information and real-time driving state information; The real-time driving status information includes: the lateral speed v of the vehicle, the longitudinal speed u of the vehicle, the yaw angular velocity ω of the vehicle r , the total mass of the vehicle m, the moment of inertia of the vehicle about the vertical axis I z , the distance a from the center of mass of the vehicle to the front axle, the distance b from the center of mass of the vehicle to the rear axle, the front wheel turning angle δ, the front wheel cornering force F yf , rear wheel cornering force F yr , front wheel driving force / braking force F xf and rear wheel driving force / braking force F xr ; Among them, the real-time driving environment information includes: rolling resistance F f , and F f = mgf, g is the acceleration of gravity, rolling resistance coefficient f, air resistance F w , F w =C D A(3.6u) 2 / 21.15, air resistance coefficient C D , windward area A.

Images