Automatic driving vehicle dynamic drift obstacle avoidance control method based on working condition identification

Abstract

The invention discloses an automatic driving vehicle dynamic drift obstacle avoidance control method based on working condition identification, which belongs to the field of automatic driving vehicle control, and comprises the following steps: sensing multi-dimensional dynamic characteristics of a vehicle and a surrounding environment, predicting information entropy of historical multi-dimensional dynamic characteristics by using a graph neural network, and determining the dynamic drift obstacle avoidance of the automatic driving vehicle. The sampling frequency of the sensor is adaptively adjusted according to the prediction error; based on the sensor sampling data after self-adaptive adjustment, a fuzzy logic enhanced time sequence convolutional neural network is adopted to judge whether the vehicle is in a drifting critical zone or not; adaptively adjusting a state weight matrix and a control weight matrix of a model prediction controller according to whether the vehicle is in a drifting critical zone or not, and calculating to obtain a vehicle front wheel steering angle and rear wheel driving force control instruction; and the control instruction is sent to a steering actuator and a driving actuator, and vehicle obstacle avoidance and trajectory tracking are achieved. The invention aims to improve the control capability and safety of the automatic driving vehicle under complex working conditions and improve the overall dynamic performance of the vehicle.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving vehicle control, and in particular to a dynamic drift obstacle avoidance control method for an autonomous driving vehicle based on working condition identification. Background Art

[0002] With the rapid development of autonomous driving technology, the level of vehicle intelligence is constantly improving, and more and more professional driving behaviors have been successfully developed and applied to autonomous driving control systems. Drift control, as a special type of driving skill, is mostly used in racing and stunt performances and is often regarded as an extreme driving technique. However, with the further development of autonomous driving technology, drift control is gradually applied to the obstacle avoidance system of autonomous driving vehicles, providing vehicles with greater control flexibility and safety in emergency situations.

[0003] At present, the drift and obstacle avoidance technology of autonomous vehicles mainly relies on path planning and obstacle detection under normal driving conditions. However, in complex traffic environments, especially in adverse conditions such as slippery roads, icy or rainy and snowy weather, when the vehicle needs to perform emergency obstacle avoidance or trajectory tracking, the traditional rule-based drift control strategy is difficult to ensure the stability and safety of the vehicle's driving, and may even cause the vehicle to lose control. This is because traditional drift control methods are usually based on pre-set rules and cannot effectively cope with complex and changeable actual driving environments.

[0004] In addition, traditional drift control methods often ignore the impact of sensor data acquisition on control performance. Under extreme conditions such as drift, the vehicle state changes dramatically, and higher requirements are placed on the real-time and accuracy of sensor data. However, the traditional fixed sampling frequency sensor data acquisition method cannot meet the needs of drift control and may cause control delays or instability.

[0005] Therefore, it is necessary to develop a new drift obstacle avoidance control method for autonomous driving vehicles, which can adaptively adjust the sensor sampling frequency, accurately identify the vehicle's driving conditions, and adaptively adjust the control strategy according to the conditions, so as to improve the vehicle's handling ability and safety under extreme conditions such as drifting. Summary of the invention

[0006] The purpose of the present invention is to provide a dynamic drift obstacle avoidance control method for an autonomous driving vehicle based on operating condition recognition, so as to solve the problem of insufficient drift obstacle avoidance control accuracy and stability in complex environments in the prior art, and to improve the vehicle's handling performance and safety under different operating conditions.

[0007] To achieve the above object, the present invention provides a dynamic drift obstacle avoidance control method for an autonomous driving vehicle based on working condition recognition, comprising the following steps:

[0008] Step S10, sensing multi-dimensional dynamic characteristics of the vehicle and surrounding scene domains through vehicle-mounted sensors;

[0009] Step S20, predicting the current information entropy through a graph neural network according to the information entropy of the multi-dimensional dynamic features obtained in S10, and adaptively adjusting the sensor sampling frequency according to the prediction error to implement an adaptive variable frequency sampling strategy for the vehicle-mounted sensor;

[0010] Step S30, based on the adaptively adjusted sensor sampling data of step S20, a fuzzy control enhanced temporal convolutional neural network is used to perform real-time reasoning on the current vehicle state to determine whether it is in a drift critical zone or a stable driving condition;

[0011] Step S40, according to the possibility that the vehicle is in the drift critical zone determined in step S30, adaptively adjusting the state weight matrix Q and the control weight matrix R of the model predictive controller to achieve stable control of the vehicle under different working conditions;

[0012] Step S50, according to step S40, the front wheel steering angle and rear wheel driving force control instructions of the vehicle are calculated, and these instructions are sent to the steering actuator and the driving actuator respectively, so as to achieve safe obstacle avoidance and track the planned trajectory of the vehicle during driving.

[0013] Preferably, in step S10, the vehicle's multi-dimensional dynamic characteristics include vehicle speed, vehicle longitudinal acceleration, vehicle lateral acceleration, vehicle yaw angular velocity, and front wheel turning angle; the surrounding scene domain's multi-dimensional dynamic characteristics include surrounding obstacle distance, surrounding obstacle relative speed, road curvature, lane line deviation degree, and road adhesion coefficient; and the vehicle-mounted sensors include: speed sensor, acceleration sensor, gyroscope, steering angle sensor, lidar, and camera.

[0014] Preferably, step S20 includes the following steps:

[0015] Step S201, at each sampling time t, all vehicle-mounted sensor data are integrated to form a feature vector. Different sensors have different sampling frequencies. Data interpolation is performed based on timestamps to construct a normalized dynamic feature vector x(t) under a unified time base t:

[0016]

[0017] Where v(t) is the vehicle speed at time t; a x (t) is the longitudinal acceleration of the vehicle at time t; a y (t) is the lateral acceleration of the vehicle at time t; is the vehicle yaw rate at time t; δ(t) is the front wheel turning angle at time t; d i (t) is the distance of the ith surrounding obstacle at time t; vrel,i (t) is the relative speed with the ith surrounding obstacle at time t; κ(t) is the road curvature at time t; D lane (t) is the lane line deviation degree at time t; μ(t) is the road adhesion coefficient at time t;

[0018] Step S202, calculate the information entropy H(t) of the dynamic feature vector x(t) at time t:

[0019]

[0020] In the formula, p(x j (t)) is the j-th dimension component x of the dynamic feature vector x(t) at time t j (t); n is the dimension of the dynamic feature vector x(t);

[0021] Step S203: construct a graph neural network GNN model, whose input is the historical information entropy sequence H history :

[0022] H history ={H(tm),H(t-m+1),…,H(t-1)};

[0023] In the formula, m is the number of samples in the historical information entropy sequence;

[0024] Step S204: convert the historical information entropy sequence H history Input GNN model to predict the information entropy at time t

[0025]

[0026] Step S205: Calculate the actual perceived information entropy H(t) and the predicted information entropy at time t The absolute error e(t) between them is:

[0027]

[0028] Step S206: Adaptively adjust the sampling frequency f of the i-th vehicle-mounted sensor based on the prediction error e(t). i (t):

[0029]

[0030] In the formula, f i,min is the minimum sampling frequency of the ith sensor; f i,max is the maximum sampling frequency of the i-th sensor; e i,max The preset maximum error threshold for the ith sensor.

[0031] Preferably, step S30 includes the following steps:

[0032] Step S301, normalize the time series data obtained after the adaptive sensor variable frequency sampling strategy is adjusted in step S20, and the formula is as follows:

[0033]

[0034] In the formula, x i (t) is the i-th original feature data at time t, including vehicle speed and vehicle longitudinal acceleration; is the normalized value of the ith feature at time t; min(x i ) and max(x i ) represent the minimum and maximum values ​​of the i-th feature in the entire time series;

[0035] Step S302, combining the normalized feature data into a feature vector at time t:

[0036]

[0037] In the formula, n is the number of features; construct the feature vector sequence in chronological order Where N is the time window length;

[0038] Step S303, fuzzy processing is performed on the normalized feature vector sequence X:

[0039] ① Each eigenvector in the eigenvector sequence X Each component Converted into fuzzy membership, the triangle membership function is used for fuzzification:

[0040]

[0041] In the formula, For variables The membership degree of the fuzzy set A; a, b, c are the parameters that define the triangular membership function;

[0042] ②Use Mamdani fuzzy rules for fuzzy reasoning:

[0043]

[0044] In the formula, A i is the fuzzy set corresponding to the i-th feature;

[0045] ③Use the centroid method to defuzzify and obtain the fuzzy reasoning result:

[0046]

[0047] In the formula, p fuzzy is the defuzzified output value, indicating the possibility that the vehicle is in the drift critical area at time t; c j is the output center value of the jth fuzzy rule;

[0048] ④ The fuzzy reasoning result p fuzzy (t) is added to the feature vector In the above example, we get the feature enhancement vector And construct the sequence of enhanced feature vectors:

[0049]

[0050] Step S304, input the feature enhancement vector sequence X′ into the temporal convolutional network TCN model to predict the possibility that the vehicle is in the drift critical area:

[0051] p tcn (t) = TCN(X′);

[0052] In the formula, p tcn (t) is the output result of the TCN model, indicating the possibility that the vehicle is in the drift critical area at time t;

[0053] Step S305, using the weighted average method, based on the TCN model prediction results and the fuzzy reasoning results, the possibility of the vehicle being in the drift critical area is comprehensively obtained:

[0054] p final (t)=(w tcn ·p tcn (t))+(w fuzzy ·p fuzzy (t));

[0055] In the formula, p final (t) is the final probability that the vehicle is in the drift critical zone; w tcn and w fuzzy are the weights of the TCN model and fuzzy rules respectively;

[0056] Step S306: according to the fused probability p final (t), determine whether the vehicle is in the drift critical area:

[0057]

[0058] Where θ is the critical region judgment threshold.

[0059] Preferably, step S40 includes the following steps:

[0060] Step S401: construct a vehicle center of mass lateral error e based on the vehicle's three-degree-of-freedom model.r The dynamic model with the heading angle error ξ as state variables:

[0061]

[0062] In the formula, is the derivative of the lateral error; is the derivative of the heading angle error; is the derivative of the longitudinal velocity; is the derivative of the sideslip angle of the center of mass; is the derivative of the yaw rate; V is the longitudinal velocity; β is the sideslip angle of the center of mass; ξ is the heading error; r is the yaw rate; k is the road curvature; e r is the lateral error; m is the vehicle mass; F yf , F yr are the lateral forces of the front and rear wheels respectively; F xr is the rear wheel driving force; δ is the front wheel turning angle; I z is the moment of inertia of the vehicle body along the z-axis; l r ,l f are the distances from the center of mass to the front and rear axles of the vehicle, respectively;

[0063] Step S402: nonlinear dynamics model Linearization and discretization are performed; the linearization uses Taylor expansion, and the discretization uses forward Euler method:

[0064]

[0065] In the formula, is the discretized state matrix; is the discretized control matrix; the state vector is x(k)=[e r (k),ξ(k),V(k),β(k),r(k)] T ,u(k)=[δ(k),F xr (k)] T is the control vector at time k;

[0066] Step S403, constructing the optimization problem of the model predictive controller. The optimization goal is to find the optimal control input in the prediction time domain to minimize the following objective function:

[0067]

[0068] Where NP is the prediction time domain length of the model predictive controller; x sg is the desired state vector; x(k) is the state vector at time k; Δu(k) is the change in control increment; Q is the state weight matrix; R is the control weight matrix;

[0069] During the optimization process, the constraints on the model state and control quantities are as follows:

[0070]

[0071] Where Δx(k) is the change of the state variable; Δx min , Δx max are the minimum and maximum values ​​of the state variable changes respectively; Δu min , Δu max are the minimum and maximum values ​​of the control variable changes respectively; u min ,u max are the minimum and maximum values ​​of the control amount respectively.

[0072] Step S404, solving the optimization problem while satisfying the constraints, and obtaining the optimal control sequence of the vehicle in the prediction time domain:

[0073] U * =[u(k),u(k+1),...,u(k+NP-1)] T .

[0074] Preferably, the state weight matrix Q and the control weight matrix R are adaptively adjusted:

[0075] ①Determine the basic state weight matrix Q base and the control weight matrix R base for:

[0076]

[0077] In the formula, is the weight coefficient of the lateral error; q ξ is the weight coefficient of heading angle error; q V is the weight coefficient of the vehicle longitudinal velocity; q β is the weight coefficient of the vehicle's center of mass sideslip angle; q r is the weight coefficient of the vehicle's yaw rate; r δ is the weight coefficient of the front wheel turning angle; is the weight coefficient of rear wheel driving force;

[0078] ② According to the final probability p of the vehicle being in the drift critical zone obtained in step S30 final , adaptively adjust the state weight matrix Q and control weight matrix R:

[0079]

[0080] In the formula, K Q K is the proportional coefficient matrix for adjusting the weight of the Q matrix; R The proportional coefficient matrix for adjusting the weights of the R matrix.

[0081] The system of the automatic driving vehicle dynamic drift obstacle avoidance control method based on working condition recognition described in the present invention comprises:

[0082] Feature perception and adaptive sensor variable frequency sampling module: perceives the multi-dimensional dynamic features of the vehicle and surrounding scene domain through on-board sensors, predicts information entropy based on graph neural network, and adaptively adjusts the sensor sampling frequency;

[0083] Working condition recognition module: Uses fuzzy logic enhanced time series convolutional neural network to judge the working condition of vehicle drift or stable driving;

[0084] Operating condition adaptive model predictive control module: According to the vehicle operating conditions, the model predictive controller parameters are adaptively adjusted and the control quantity is calculated, and the control instructions are output to the steering and drive actuators.

[0085] Therefore, the present invention adopts the above-mentioned automatic driving vehicle dynamic drift obstacle avoidance control method based on working condition recognition, which has the following beneficial effects:

[0086] (1) This paper proposes an adaptive sensor variable frequency sampling strategy based on information entropy prediction, which uses graph neural networks to learn the relationship between historical multidimensional dynamic features and information entropy, and predict the change trend of future information entropy. According to the prediction error, the sensor sampling frequency is adaptively adjusted to increase the sampling frequency to obtain more key information features when the multidimensional dynamic features of the vehicle and the surrounding scene domain change dramatically, and reduce the sampling frequency to save computing resources when the vehicle state is stable, ultimately improving the control accuracy.

[0087] (2) The present invention adopts a working condition identification method based on a time series convolutional neural network enhanced by fuzzy logic. By combining fuzzy control and time series convolutional neural network, it is determined whether the vehicle is in the drift critical zone. The introduction of fuzzy logic enhances the model's ability to handle uncertainty and fuzzy information, and improves the robustness of working condition identification.

[0088] (3) The present invention adaptively adjusts the state weight matrix Q and control weight matrix R of the model predictive controller according to the working condition (drifting or stable driving) of the vehicle. This strategy enables the controller to be dynamically adjusted according to the real-time state of the vehicle, thereby achieving stable control of the vehicle under different working conditions.

[0089] (4) The present invention proposes a new integration method of the drift obstacle avoidance control system, including a feature perception and adaptive sensor variable frequency sampling module, a working condition identification module, and a working condition adaptive model predictive control module, which realizes closed-loop control from perception to control and improves the overall performance of the system.

[0090] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] Figure 1 is a flow chart of a control method according to an embodiment of the present invention;

[0092] Figure 2 This is a flow chart of adaptive sensor sampling based on information entropy prediction according to an embodiment of the present invention;

[0093] Figure 3 This is a flow chart of a method for identifying working conditions based on a fuzzy logic enhanced temporal convolutional neural network according to an embodiment of the present invention;

[0094] Figure 4 The figure is a flow chart of a method of adaptive model predictive controller according to an embodiment of the present invention. DETAILED DESCRIPTION

[0095] In order to make the purpose, technical scheme and advantages disclosed in the embodiments of the present invention clearer, the embodiments of the present invention are further described in detail in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the embodiments of the present invention and are not intended to limit the embodiments of the present invention. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions.

[0096] It should be noted that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or inherent to these processes, methods, products or devices.

[0097] Like reference numerals and letters denote similar items in the following drawings, and thus, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0098] In the description of the present invention, it should be noted that the terms "upper", "lower", "inside", "outside", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings, or are directions or positional relationships in which the product of the invention is usually placed when in use. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as a limitation on the present invention.

[0099] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "setting", "installation" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0100] Example

[0101] like Figure 1 As shown, the present invention provides a method for controlling dynamic drifting and obstacle avoidance of an autonomous driving vehicle based on working condition recognition, comprising the following steps:

[0102] Step S10, sensing the multi-dimensional dynamic characteristics of the vehicle and the surrounding scene domain through the vehicle-mounted sensors.

[0103] In step S10, the vehicle's multi-dimensional dynamic characteristics include vehicle speed, vehicle longitudinal acceleration, vehicle lateral acceleration, vehicle yaw angular velocity, front wheel turning angle, etc.; the surrounding scene domain multi-dimensional dynamic characteristics include surrounding obstacle distance, surrounding obstacle relative speed, road curvature, lane line deviation degree, and road adhesion coefficient, etc.; the vehicle-mounted sensors include: speed sensor, acceleration sensor, gyroscope, steering angle sensor, lidar, camera, etc.

[0104] like Figure 2 As shown, in step S20, according to the information entropy of the multi-dimensional dynamic features obtained in S10, the current information entropy is predicted through the graph neural network, and the sensor sampling frequency is adaptively adjusted according to the prediction error to realize the adaptive variable frequency sampling strategy of the vehicle-mounted sensor.

[0105] Step S20 includes the following steps:

[0106] Step S201, at each sampling time t, all vehicle-mounted sensor data are integrated to form a feature vector. Considering that different sensors may have different sampling frequencies, data interpolation is required based on timestamps, and then a normalized dynamic feature vector x(t) is constructed under a unified time base t:

[0107]

[0108] Where v(t) is the vehicle speed at time t; a x (t) is the longitudinal acceleration of the vehicle at time t; a y (t) is the lateral acceleration of the vehicle at time t; is the vehicle yaw rate at time t; δ(t) is the front wheel turning angle at time t; d i(t) is the distance of the ith surrounding obstacle at time t; v rel,i (t) is the relative speed with the ith surrounding obstacle at time t; κ(t) is the road curvature at time t; D lane (t) is the lane line deviation degree at time t; μ(t) is the road adhesion coefficient at time t;

[0109] Step S202, calculate the information entropy H(t) of the dynamic feature vector x(t) at time t:

[0110]

[0111] In the formula, p(x j (t)) is the j-th dimension component x of the dynamic feature vector x(t) at time t j (t); n is the dimension of the dynamic feature vector x(t);

[0112] Step S203: construct a graph neural network (GNN) model, whose input is the historical information entropy sequence H history :

[0113] H history ={H(tm),H(t-m+1),...,H(t-1)};

[0114] In the formula, m is the number of samples in the historical information entropy sequence;

[0115] Step S204: convert the historical information entropy sequence H history Input GNN model to predict the information entropy at time t

[0116]

[0117] Step S205: Calculate the actual perceived information entropy H(t) and the predicted information entropy at time t The absolute error e(t) between them is:

[0118]

[0119] Step S206: Adaptively adjust the sampling frequency f of the i-th vehicle-mounted sensor based on the prediction error e(t). i (t):

[0120]

[0121] In the formula, f i,min is the minimum sampling frequency of the ith sensor; f i,max is the maximum sampling frequency of the i-th sensor; e i,max The preset maximum error threshold for the ith sensor.

[0122] like Figure 3 As shown, step S30, based on the sensor sampling data after adaptive adjustment in step S20, a fuzzy control enhanced temporal convolutional neural network is used to perform real-time reasoning on the current vehicle state to determine whether it is in a drift critical area or a stable driving condition.

[0123] Step S30 includes the following steps:

[0124] Step S301, normalize the time series data obtained after the adaptive sensor variable frequency sampling strategy is adjusted in step S20, and the formula is as follows:

[0125]

[0126] In the formula, x i (t) is the i-th original feature data at time t, including vehicle speed, vehicle longitudinal acceleration, etc.; is the normalized value of the ith feature at time t; min(x i ) and max(x i ) represent the minimum and maximum values ​​of the i-th feature in the entire time series;

[0127] Step S302, combining the normalized feature data into a feature vector at time t:

[0128]

[0129] In the formula, n is the number of features. Then, construct the feature vector sequence in chronological order Where N is the time window length;

[0130] Step S303, fuzzy processing is performed on the normalized feature vector sequence X:

[0131] ① Each eigenvector in the eigenvector sequence X Each component Converted into fuzzy membership, the triangle membership function is used for fuzzification:

[0132]

[0133] In the formula, For variables The membership degree of the fuzzy set A; a, b, c are the parameters that define the triangular membership function;

[0134] ②Use Mamdani fuzzy rules for fuzzy reasoning:

[0135]

[0136] In the formula, Ai is the fuzzy set corresponding to the i-th feature;

[0137] ③Use the centroid method to defuzzify and obtain the fuzzy reasoning result:

[0138]

[0139] In the formula, p fuzzy is the defuzzified output value, indicating the possibility that the vehicle is in the drift critical area at time t; c j is the output center value of the jth fuzzy rule;

[0140] ④ The fuzzy reasoning result p fuzzy (t) is added to the feature vector In the above example, we get the feature enhancement vector And construct the sequence of enhanced feature vectors:

[0141]

[0142] Step S304, input the feature enhancement vector sequence X′ into the temporal convolutional network TCN model to predict the possibility that the vehicle is in the drift critical area:

[0143] p tcn (t) = TCN(X′);

[0144] In the formula, p tcn (t) is the output result of the TCN model, indicating the possibility that the vehicle is in the drift critical area at time t;

[0145] Step S305, using the weighted average method, based on the TCN model prediction results and the fuzzy reasoning results, the possibility of the vehicle being in the drift critical area is comprehensively obtained:

[0146] p final (t)=(w tcn ·p tcn (t))+(w fuzzy ·p fuzzy (t));

[0147] In the formula, p final (t) is the final probability that the vehicle is in the drift critical zone; w tcn and w fuzzy are the weights of the TCN model and fuzzy rules respectively;

[0148] Step S306: according to the fused probability p final (t), determine whether the vehicle is in the drift critical area:

[0149]

[0150] Where θ is the critical region judgment threshold.

[0151] like Figure 4 As shown, step S40, based on the possibility that the vehicle is in the drift critical zone determined in step S30, adaptively adjusts the state weight matrix Q and the control weight matrix R of the model predictive controller to achieve stable control of the vehicle under different operating conditions (drifting or stable driving).

[0152] Step S40 includes the following steps:

[0153] Step S401: construct a vehicle center of mass lateral error e based on the vehicle's three-degree-of-freedom model. r The dynamic model with the heading angle error ξ as state variables:

[0154]

[0155] In the formula, is the derivative of the lateral error; is the derivative of the heading angle error; is the derivative of the longitudinal velocity; is the derivative of the sideslip angle of the center of mass; is the derivative of the yaw rate; V is the longitudinal velocity; β is the sideslip angle of the center of mass; ξ is the heading error; r is the yaw rate; k is the road curvature; e r is the lateral error; m is the vehicle mass; F yf , F yr are the lateral forces of the front and rear wheels respectively; F xr is the rear wheel driving force; δ is the front wheel turning angle; I z is the moment of inertia of the vehicle body along the z-axis; l r ,l f are the distances from the center of mass to the front and rear axles of the vehicle, respectively;

[0156] Step S402: nonlinear dynamics model Linearization and discretization are performed; the linearization uses Taylor expansion, and the discretization uses forward Euler method:

[0157]

[0158] In the formula, is the discretized state matrix; is the discretized control matrix; the state vector is x(k)=[e r (k),ξ(k),V(k),β(k),r(k)] T ,u(k)=[δ(k),F xr (k)] T is the control vector at time k;

[0159] Step S403, constructing the optimization problem of the model predictive controller. The optimization goal is to find the optimal control input in the prediction time domain to minimize the following objective function:

[0160]

[0161] Where NP is the prediction time domain length of the model predictive controller; x sg is the desired state vector; x(k) is the state vector at time k; Δu(k) is the change in control increment; Q is the state weight matrix; R is the control weight matrix;

[0162] During the optimization process, the constraints of the model state and control quantities are as follows:

[0163]

[0164] Where Δx(k) is the change of the state variable; Δx min , Δx max are the minimum and maximum values ​​of the state variable changes respectively; Δu min , Δu max are the minimum and maximum values ​​of the control variable changes respectively; u min ,u max are the minimum and maximum values ​​of the control amount respectively.

[0165] Step S404, solving the optimization problem while satisfying the constraints, and obtaining the optimal control sequence of the vehicle in the prediction time domain:

[0166] U * =[u(k),u(k+1),...,u(k+NP-1)] T .

[0167] Adaptively adjust the state weight matrix Q and control weight matrix R:

[0168] ①Determine the basic state weight matrix Q base and the control weight matrix R base for:

[0169]

[0170] In the formula, is the weight coefficient of the lateral error; q ξ is the weight coefficient of heading angle error; q V is the weight coefficient of the vehicle longitudinal velocity; q β is the weight coefficient of the vehicle's center of mass sideslip angle; q r is the weight coefficient of the vehicle's yaw rate; r δ is the weight coefficient of the front wheel turning angle; is the weight coefficient of rear wheel driving force;

[0171] ② According to the final probability p of the vehicle being in the drift critical zone obtained in step S30 final , adaptively adjust the state weight matrix Q and control weight matrix R:

[0172]

[0173] In the formula, K Q K is the proportional coefficient matrix for adjusting the weight of the Q matrix; R The proportional coefficient matrix for adjusting the weights of the R matrix.

[0174] Step S50, the front wheel steering angle and rear wheel driving force control instructions of the vehicle are calculated according to step S40, and these instructions are sent to the steering actuator and the driving actuator respectively to achieve safe obstacle avoidance of the vehicle during driving and accurately track the planned trajectory.

[0175] The system of the automatic driving vehicle dynamic drift obstacle avoidance control method based on working condition recognition described in the present invention comprises:

[0176] Feature perception and adaptive sensor variable frequency sampling module: perceives the multi-dimensional dynamic features of the vehicle and surrounding scene domain through on-board sensors, predicts information entropy based on graph neural network, and adaptively adjusts the sensor sampling frequency;

[0177] Working condition recognition module: Uses fuzzy logic enhanced time series convolutional neural network to judge the working condition of vehicle drift or stable driving;

[0178] Operating condition adaptive model predictive control module: According to the vehicle operating conditions, the model predictive controller parameters are adaptively adjusted and the control quantity is calculated, and the control instructions are output to the steering and drive actuators.

[0179] Therefore, the present invention adopts the above-mentioned method for dynamic drift obstacle avoidance control of an autonomous driving vehicle based on working condition identification to solve the problem of insufficient drift obstacle avoidance control accuracy and stability in complex environments in the prior art, and to improve the vehicle's handling performance and safety under different working conditions. Multi-sensor fusion is used to perceive the multi-dimensional dynamic characteristics of the vehicle and its surrounding environment in real time, and the graph neural network is used to predict the change in feature information entropy, and then the sensor sampling frequency is adaptively adjusted. Then, the present invention uses a fuzzy logic-enhanced time-series convolutional neural network to identify the vehicle's driving conditions in real time, and according to the identification results, the parameters of the model predictive controller are adaptively adjusted, the vehicle control instructions are calculated and executed, and the vehicle's stable control under different working conditions is achieved, including steady-state drift and precise trajectory tracking.

[0180] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A dynamic drift obstacle avoidance control method for an autonomous driving vehicle based on working condition recognition, characterized in that: The following steps are involved: Step S10, sensing multi-dimensional dynamic characteristics of the vehicle and surrounding scene domains through vehicle-mounted sensors; Step S20, predicting the current information entropy through a graph neural network according to the information entropy of the multi-dimensional dynamic features obtained in S10, and adaptively adjusting the sensor sampling frequency according to the prediction error to implement an adaptive variable frequency sampling strategy for the vehicle-mounted sensor; Step S30, based on the adaptively adjusted sensor sampling data of step S20, a fuzzy control enhanced temporal convolutional neural network is used to perform real-time reasoning on the current vehicle state to determine whether it is in a drift critical zone or a stable driving condition; Step S40, according to the possibility that the vehicle is in the drift critical zone determined in step S30, adaptively adjusting the state weight matrix Q and the control weight matrix R of the model predictive controller to achieve stable control of the vehicle under different working conditions; Step S50, according to step S40, the front wheel steering angle and rear wheel driving force control instructions of the vehicle are calculated, and these instructions are sent to the steering actuator and the driving actuator respectively to achieve safe obstacle avoidance and track the planned trajectory of the vehicle during driving.

2. The method for dynamic drift obstacle avoidance control of an autonomous driving vehicle based on working condition recognition according to claim 1, characterized in that: In step S10, the vehicle multi-dimensional dynamic features include vehicle speed, vehicle longitudinal acceleration, vehicle lateral acceleration, vehicle yaw angular velocity, and front wheel turning angle; the surrounding scene domain multi-dimensional dynamic features include surrounding obstacle distance, surrounding obstacle relative speed, road curvature, lane line deviation degree, and road adhesion coefficient; Vehicle-mounted sensors include: speed sensor, acceleration sensor, gyroscope, steering angle sensor, lidar, and camera.

3. The method for dynamic drift obstacle avoidance control of an autonomous driving vehicle based on working condition recognition according to claim 2 is characterized in that: Step S20 includes the following steps: Step S201, at each sampling time t, all vehicle-mounted sensor data are integrated to form a feature vector. Different sensors have different sampling frequencies. Data interpolation is performed based on timestamps to construct a normalized dynamic feature vector x(t) under a unified time base t: Where v(t) is the vehicle speed at time t; a x (t) is the longitudinal acceleration of the vehicle at time t; a y (t) is the lateral acceleration of the vehicle at time t; is the vehicle yaw rate at time t; δ(t) is the front wheel turning angle at time t; d i (t) is the distance of the ith surrounding obstacle at time t; v rel,i (t) is the relative speed with the ith surrounding obstacle at time t; κ(t) is the road curvature at time t; D lane (t) is the lane line deviation degree at time t; μ(t) is the road adhesion coefficient at time t; Step S202, calculate the information entropy H(t) of the dynamic feature vector x(t) at time t: In the formula, p(x j (t)) is the j-th dimension component x of the dynamic feature vector x(t) at time t j (t); n is the dimension of the dynamic feature vector x(t); Step S203: construct a graph neural network GNN model, whose input is the historical information entropy sequence H history : H history ={H(t-m),H(t-m+1),…,H(t-1)}; In the formula, m is the number of samples in the historical information entropy sequence; Step S204: convert the historical information entropy sequence H history Input GNN model to predict the information entropy at time t Step S205: Calculate the actual perceived information entropy H(t) and the predicted information entropy at time t The absolute error e(t) between them is: Step S206: Adaptively adjust the sampling frequency f of the i-th vehicle-mounted sensor based on the prediction error e(t). i (t): In the formula, f i,min is the minimum sampling frequency of the ith sensor; f i,max is the maximum sampling frequency of the i-th sensor; e i,max The preset maximum error threshold for the ith sensor.

4. The method for dynamic drift obstacle avoidance control of an autonomous driving vehicle based on working condition recognition according to claim 3 is characterized in that: Step S30 includes the following steps: Step S301, normalize the time series data obtained after the adaptive sensor variable frequency sampling strategy is adjusted in step S20, and the formula is as follows: In the formula, x i (t) is the i-th original feature data at time t, including vehicle speed and vehicle longitudinal acceleration; is the normalized value of the ith feature at time t; min(x i ) and max(x i ) represent the minimum and maximum values ​​of the i-th feature in the entire time series; Step S302, combining the normalized feature data into a feature vector at time t: In the formula, n is the number of features; construct the feature vector sequence in chronological order Where N is the time window length; Step S303, fuzzy processing is performed on the normalized feature vector sequence X: ① Each eigenvector in the eigenvector sequence X Each component Converted into fuzzy membership, the triangle membership function is used for fuzzification: In the formula, For variables The membership degree of the fuzzy set A; a, b, c are the parameters that define the triangular membership function; ②Use Mamdani fuzzy rules for fuzzy reasoning: In the formula, A i is the fuzzy set corresponding to the i-th feature; ③Use the centroid method to defuzzify and obtain the fuzzy reasoning result: In the formula, p fuzzy is the defuzzified output value, indicating the possibility that the vehicle is in the drift critical area at time t; c j is the output center value of the jth fuzzy rule; ④ The fuzzy reasoning result p fuzzy (t) is added to the feature vector In the above example, we get the feature enhancement vector And construct the enhanced feature vector sequence: Step S304, input the feature enhancement vector sequence X′ into the temporal convolutional network TCN model to predict the possibility that the vehicle is in the drift critical area: p tcn (t)=TCN(X′); In the formula, p tcn (t) is the output result of the TCN model, indicating the possibility that the vehicle is in the drift critical area at time t; Step S305, using the weighted average method, based on the TCN model prediction results and the fuzzy reasoning results, the possibility of the vehicle being in the drift critical area is comprehensively obtained: p final (t)=(w tcn ·p tcn (t))+(w fuzzy ·p fuzzy (t)); In the formula, p final (t) is the final probability that the vehicle is in the drift critical zone; w tcn and w fuzzy are the weights of the TCN model and fuzzy rules respectively; Step S306: according to the fused probability p final (t), determine whether the vehicle is in the drift critical area: Where θ is the critical region judgment threshold.

5. The method for dynamic drift obstacle avoidance control of an autonomous driving vehicle based on working condition recognition according to claim 4 is characterized in that: Step S40 includes the following steps: Step S401: construct a vehicle center of mass lateral error e based on the vehicle's three-degree-of-freedom model. r The dynamic model with the heading angle error ξ as state variables: In the formula, is the derivative of the lateral error; is the derivative of the heading angle error; is the derivative of the longitudinal velocity; is the derivative of the sideslip angle of the center of mass; is the derivative of the yaw rate; V is the longitudinal velocity; β is the sideslip angle of the center of mass; ξ is the heading error; r is the yaw rate; k is the road curvature; e r is the lateral error; m is the vehicle mass; F yf , F yr are the lateral forces of the front and rear wheels respectively; F xr is the rear wheel driving force; δ is the front wheel turning angle; I z is the moment of inertia of the vehicle body along the z axis; l r ,l f are the distances from the center of mass to the front and rear axles of the vehicle, respectively; Step S402: nonlinear dynamics model Linearization and discretization are performed; the linearization uses Taylor expansion, and the discretization uses forward Euler method: In the formula, is the discretized state matrix; is the discretized control matrix; the state vector is x(k)=[e r (k),ξ(k),V(k),β(k),r(k)] T ,u(k)=[δ(k),F xr (k)] T is the control vector at time k; Step S403, constructing the optimization problem of the model predictive controller. The optimization goal is to find the optimal control input in the prediction time domain to minimize the following objective function: Where NP is the prediction time domain length of the model predictive controller; x sg is the desired state vector; x(k) is the state vector at time k; Δu(k) is the change in control increment; Q is the state weight matrix; R is the control weight matrix; During the optimization process, the constraints on the model state and control quantities are as follows: Where Δx(k) is the change of the state variable; Δx min , Δx max are the minimum and maximum values ​​of the state variable changes respectively; Δu min , Δu max are the minimum and maximum values ​​of the control variable changes respectively; u min ,u max are the minimum and maximum values ​​of the control quantity respectively; Step S404, solving the optimization problem while satisfying the constraints, and obtaining the optimal control sequence of the vehicle in the prediction time domain: U * =[u(k),u(k+1),...,u(k+NP-1)] T 。 6. The method for dynamic drift obstacle avoidance control of an autonomous driving vehicle based on working condition recognition according to claim 5 is characterized in that: Adaptively adjust the state weight matrix Q and control weight matrix R: ①Determine the basic state weight matrix Q base and the control weight matrix R base for: In the formula, q er is the weight coefficient of the lateral error; q ξ is the weight coefficient of heading angle error; q V is the weight coefficient of the vehicle longitudinal velocity; q β is the weight coefficient of the vehicle's center of mass sideslip angle; q r is the weight coefficient of the vehicle's yaw rate; r δ is the weight coefficient of the front wheel turning angle; is the weight coefficient of rear wheel driving force; ② According to the final probability p of the vehicle being in the drift critical zone obtained in step S30 final , adaptively adjust the state weight matrix Q and control weight matrix R: In the formula, K Q K is the proportional coefficient matrix for adjusting the weight of the Q matrix; R The proportional coefficient matrix for adjusting the weights of the R matrix.

Images