A lateral and longitudinal integrated predictive control method for autonomous driving

Abstract

The present invention discloses a lateral and longitudinal integrated predictive control method for automatic driving, comprising: obtaining current vehicle state parameters through sensors; predicting the future state of the vehicle according to the vehicle state parameters, and obtaining the longitudinal predicted vehicle speed, lateral predicted vehicle speed, predicted yaw angular velocity and predicted center of mass sideslip angle at each moment in the vehicle prediction time domain; calculating the vehicle prediction stability margin; predicting the state of obstacles around the vehicle and determining a collision-free risk space; planning the vehicle trajectory in the collision-free risk space and obtaining the vehicle's target lateral displacement, target lateral speed, target lateral acceleration, target longitudinal displacement, target longitudinal speed and longitudinal acceleration; if the vehicle is in an understeering state in the future, increasing the lateral acceleration during the trajectory planning process; if the vehicle is in an oversteering state in the future, reducing the lateral acceleration during the trajectory planning process; and controlling the steering wheel angle and distributing the torque of the four-wheel motor according to the vehicle trajectory planning result.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving, and in particular relates to a lateral and longitudinal integrated predictive control method for autonomous driving. Background Art

[0002] With the rapid development of economy and technology over the past decades, the level of automobile intelligence has been greatly improved. At present, L2-level intelligent driving systems have been widely used in various models. For example, the AEB (Autonomous Emergency Braking) system uses millimeter-wave radar to detect the relative distance and relative speed of obstacles in front of the vehicle, and judges the degree of danger of vehicle collision, and takes partial or full braking of the vehicle to avoid collision or reduce the severity of collision; the NOA (Navigate on Autopilot) system is further upgraded on the basis of the traditional ACC (Adaptive Cruise Control) system, which can realize functions such as autonomous lane change, overtaking, and lane insertion. However, when facing dangerous scenes or extreme working conditions, the current intelligent driving systems require the driver to take over, and their intelligence level still needs to be further improved. Summary of the invention

[0003] The purpose of the present invention is to provide an integrated lateral and longitudinal predictive control method for autonomous driving, which constructs a vehicle driving safety space by comprehensively predicting the vehicle's stability state and the risk of collision with obstacles, and performs trajectory planning and vehicle motion control within the safety space considering the vehicle's stability state constraints, thereby improving the driving safety and ride comfort of smart vehicles.

[0004] The technical solution provided by the present invention is:

[0005] A lateral and longitudinal integrated predictive control method for autonomous driving, comprising:

[0006] The current vehicle status parameters are obtained through sensors: longitudinal displacement X, lateral displacement Y, longitudinal vehicle speed v x , lateral speed v y , longitudinal acceleration a x , lateral acceleration a y , heading angle θ, steering wheel angle δ sw ; Predicting the future state of the vehicle according to the current vehicle state parameters to obtain the longitudinal predicted vehicle speed, lateral predicted vehicle speed, predicted yaw angular velocity and predicted sideslip angle of the center of mass at each moment in the vehicle prediction time domain;

[0007] Calculate the vehicle predicted stability margin κ stable (k);

[0008] κ stable (k) = max(κueb (k),κ unt (k),κ untchange (k));

[0009] Among them, κ ueb (k) represents the understeering factor, κ unt (k) represents the oversteering factor, κ untchange (k) represents the corrected oversteering factor;

[0010]

[0011] In the formula, s fl 、s fr 、s rl and rr are the slip rates of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; is the steering wheel angle influence coefficient, δ sw is the steering wheel angle, δ swcri is the critical turning angle;

[0012] Predict the state of obstacles around the vehicle and determine the space without collision risk;

[0013] Performing vehicle trajectory planning in the collision-free risk space to obtain a target lateral displacement, a target lateral velocity, a target lateral acceleration, a target longitudinal displacement, a target longitudinal velocity, and a longitudinal acceleration of the vehicle;

[0014] Among them, when and When the target lateral acceleration is corrected to

[0015] when or mean(κ untchange (k)) and When the target lateral acceleration is corrected to

[0016] In the formula, is the lateral acceleration increment coefficient, ω des represents the target yaw rate; ω represents the actual yaw rate, a ytar represents the target lateral acceleration; κ uebmax is the maximum value of understeering factor;

[0017] The steering wheel angle is controlled and the torque of the four-wheel motor is distributed according to the vehicle trajectory planning results.

[0018] Preferably, the longitudinal predicted vehicle speed at each moment in the vehicle prediction time domain is Lateral predicted vehicle speed Predicted yaw rate and predicted slip angle The calculation method is:

[0019]

[0020]

[0021]

[0022]

[0023] in, represents the longitudinal predicted vehicle speed obtained from historical data, represents the lateral predicted vehicle speed obtained from historical data, represents the predicted yaw rate obtained from historical data, It represents the predicted sideslip angle of the center of mass predicted by historical data; represents the longitudinal predicted vehicle speed predicted by the linear vehicle dynamics model, represents the lateral predicted vehicle speed predicted by the linear vehicle dynamics model, represents the predicted yaw rate predicted by the linear vehicle dynamics model, represents the predicted sideslip angle of the center of mass predicted by the linear vehicle dynamics model; represents the longitudinal predicted vehicle speed predicted by the nonlinear vehicle dynamics model, represents the lateral predicted vehicle speed predicted by the nonlinear vehicle dynamics model, represents the predicted yaw rate predicted by the nonlinear vehicle dynamics model, represents the predicted sideslip angle of the center of mass predicted by the nonlinear vehicle dynamics model; and are the coefficients of each item respectively; a yc It is expressed as the corrected lateral acceleration; g is the acceleration due to gravity.

[0024] Preferably, the understeering factor calculation formula is:

[0025]

[0026] In the formula, is the proportional influence coefficient of the understeer factor; is the vehicle speed influence coefficient of understeering factor; is the road adhesion influence coefficient; a ω is the yaw rate error coefficient of the understeering factor; a β is the center of mass sideslip angle error coefficient of understeering factor; β desis the target center of mass sideslip angle; Predict the sideslip angle of the center of mass at time k; Predict the yaw rate at time k; is the predicted longitudinal vehicle speed at time k; μ0 is the corrected road adhesion coefficient.

[0027] Preferably, the calculation formula of the oversteering factor is:

[0028]

[0029] In the formula, is the vehicle speed influence coefficient of the oversteering factor; is the road adhesion influence coefficient of the oversteering factor; b ω is the yaw rate error coefficient of the oversteering factor; b β is the center of mass sideslip angle error coefficient of the oversteering factor.

[0030] Preferably, the method for determining the collision risk-free space is:

[0031] Traverse all single obstacle risk values ​​Risk i ;

[0032] If Risk i <Risk max , there is no need to constrain the collision risk space;

[0033] If Risk i ≥Risk max , then the collision risk space of a single obstacle i is predicted;

[0034] Among them, Risk max is the collision risk threshold;

[0035] Traverse the collision risk spaces of all obstacles with collision risks, subtract all collision risk spaces from the initial collision risk space, and obtain the remaining collision-free risk space;

[0036] The preliminary collision risk space is the space within the effective lane boundary line or the effective driving area within the sensor perception range.

[0037] Preferably, the risk value of a single obstacle Risk i Calculated by the following formula;

[0038]

[0039] in, is the vertical risk impact coefficient; is the lateral risk impact coefficient; d iis the distance from the vehicle to the ith obstacle; l wai is the longitudinal warning distance from the vehicle to obstacle i; l coi is the longitudinal collision danger distance from the vehicle to obstacle i; s wai is the lateral warning distance from the vehicle to obstacle i; s coi is the side collision danger distance from the vehicle to obstacle i; Δv xobsi is the longitudinal relative speed between the vehicle and obstacle i, Δv yobsi is the lateral relative speed between the vehicle and obstacle i.

[0040] Preferably, if there is no solution for vehicle trajectory planning in the collision risk-free space, the vehicle is controlled to perform emergency braking.

[0041] Preferably, if or The torque of the four-wheel motors is evenly distributed;

[0042]

[0043] Among them, T bfl 、T bfr 、T brl 、T brr They are the motor torques of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; a is the distance from the center of mass to the front axle; b is the distance from the center of mass to the rear axle; m is the mass of the vehicle.

[0044] Preferably, if and

[0045] When the vehicle turns left in the future;

[0046]

[0047]

[0048]

[0049]

[0050] In the formula, k ueωl is the left understeer correction coefficient, k ueβl is the left-turn understeer β correction coefficient; when the vehicle turns right in the future,

[0051]

[0052]

[0053]

[0054]

[0055] In the formula, k ueωr is the right understeer correction coefficient, k ueβl is the right understeer β correction coefficient. Preferably, or mean(κ untchange (k)), and When the vehicle turns left in the future:

[0056]

[0057]

[0058]

[0059]

[0060] In the formula, k unωl is the left oversteering correction coefficient, k unβl is the oversteering β correction coefficient;

[0061] When the vehicle turns right in the future,

[0062]

[0063]

[0064]

[0065]

[0066] In the formula, k unωr is the correction coefficient for right oversteering, k unβl It is the β correction coefficient of right oversteer.

[0067] The beneficial effects of the present invention are:

[0068] The integrated lateral and longitudinal predictive control method for autonomous driving provided by the present invention constructs a safe driving space for the vehicle by comprehensively predicting the vehicle's stable state and the risk of collision with obstacles, and performs trajectory planning and vehicle motion control within the safe space by considering the vehicle's stable state constraints. It can predict the vehicle's motion state and the risk of environmental collision in advance, thereby performing further pre-control, thereby improving the driving safety and riding comfort of the smart car and effectively enhancing the intelligence level of the smart car. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 This is a framework diagram of the lateral and longitudinal integrated predictive planning control system for autonomous driving described in the present invention.

[0070] Figure 2 This is a flow chart of the lateral and longitudinal integrated predictive planning control method for autonomous driving described in the present invention.

[0071] Figure 3 This is the collision risk space constraint diagram described in the present invention. DETAILED DESCRIPTION

[0072] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0073] In order to improve the autonomy and actual performance of the intelligent driving system in critical danger scenarios and extreme working conditions, the present invention provides a lateral and longitudinal integrated predictive control method for autonomous driving for intelligent distributed (four-wheel independent drive) electric vehicles, which is implemented based on the following lateral and longitudinal integrated predictive control system for autonomous driving.

[0074] like Figure 1 As shown, the lateral and longitudinal integrated predictive planning and control system for autonomous driving is composed of a data acquisition module, a central processing module, and an execution module. The data acquisition module is composed of a laser radar sensor, a camera sensor, a GNNS (Global Navigation Satellite System) sensor, and a steering wheel angle sensor. The laser radar is used to sense obstacle information in the vehicle's surroundings; the camera sensor is used to sense lane line information and obstacle information in the vehicle's surroundings; the GNNS sensor is a global navigation satellite system, which is used to locate the vehicle's position information and detect the vehicle's longitudinal speed v x , lateral speed v y , longitudinal acceleration a x , lateral acceleration a y , yaw angular velocity ω, heading angle θ; the steering wheel angle sensor is used to detect the steering wheel angle δ f ; The central processing module is composed of vehicle state prediction, driving risk prediction, and driving behavior decision; the vehicle state prediction is used to predict the state information of the vehicle in the future; the collision risk prediction is used to predict the risk of collision with surrounding obstacles in the future; the driving behavior decision makes driving behavior decisions based on the results of driving risk prediction and issues control instructions to the execution module; the execution module includes a steering wheel, a left front wheel motor, a right front wheel motor, a left rear wheel motor, and a right rear wheel motor; the steering wheel is used to control the lateral movement direction of the vehicle; the left front wheel motor, the right front wheel motor, the left rear wheel motor, and the right rear wheel motor are used to control the four-wheel torque.

[0075] like Figure 2As shown, the specific method of the lateral and longitudinal integrated predictive planning control for autonomous driving is as follows.

[0076] 1. Vehicle Status Prediction

[0077] The GNSS sensor and steering wheel angle sensor of the data acquisition module can respectively collect the longitudinal displacement X, lateral displacement Y, and longitudinal speed v of the vehicle. x , lateral speed v y , longitudinal acceleration a x , lateral acceleration a y , heading angle θ, steering wheel angle δ sw ;

[0078] In order to improve the future T sp (Vehicle prediction time domain) The accuracy of vehicle state prediction within a time period. Vehicle state prediction consists of three parts: historical data prediction, linear vehicle dynamics model prediction, and nonlinear vehicle dynamics model prediction.

[0079] Historical data forecast part:

[0080] Record the current time t before T nf Data within a time period, if the current time is less than T nf (Time threshold in the initial stage) No prediction is made;

[0081] The recorded data is:

[0082]

[0083]

[0084]

[0085] In the formula, Δt nc is the sampling period; v x (tT nf |t) as an example, it means the Tth time before the current time nf The vehicle speed at the time;

[0086] The historical center of mass slip angle can be calculated based on historical data.

[0087]

[0088] Prediction model formula:

[0089]

[0090]

[0091] ·······

[0092]

[0093] In the formula, s(k+1|t) represents the predicted value of the next cycle at the current time, s(k+T sp / Δt nc |t) represents the future k+T at the current moment sp / Δt nc The predicted value at the cycle moment; take k = t; s(kj|t) represents the value of the j cycles before the current moment, which is the historical data; in the subsequent prediction, the predicted value of the previous cycle is used as the historical data;

[0094] v x 、v y , ω, β use the above prediction model to predict the state and get the prediction array based on historical data (Longitudinal predicted vehicle speed), (lateral predicted vehicle speed), (predicted yaw rate), (predicted center of mass sideslip angle);

[0095] Linear vehicle dynamics model prediction part:

[0096] The discrete linear vehicle dynamics prediction model is as follows:

[0097]

[0098]

[0099] ······

[0100]

[0101] In the formula, δ f is the front wheel turning angle, i sw is the transmission ratio from the steering wheel to the front wheels, T is the sampling period of the prediction model, k f is the front axle cornering stiffness, k r is the rear axle cornering stiffness, m is the vehicle mass, a is the distance from the center of mass to the front axle, and b is the distance from the center of mass to the rear axle;

[0102] The array is predicted by the linear vehicle dynamics model (Longitudinal predicted vehicle speed), (lateral predicted vehicle speed), (predicted yaw rate), (predicted center of mass sideslip angle);

[0103] Nonlinear vehicle dynamics model prediction part:

[0104] The discrete nonlinear vehicle dynamics model is as follows:

[0105]

[0106]

[0107] ······

[0108]

[0109] In the formula, F cf is the lateral force of the front wheel, F cr is the rear wheel lateral force, both calculated by the tire magic formula;

[0110] The array is predicted by the nonlinear vehicle dynamics model (Longitudinal predicted vehicle speed), (lateral predicted vehicle speed), (predicted yaw rate), (predicted center of mass sideslip angle);

[0111] The three parts of the prediction results are fitted according to the following formula:

[0112]

[0113]

[0114]

[0115]

[0116] In the formula, a yc is the corrected lateral acceleration, a yc The following conditions must be met:

[0117]

[0118] g is the acceleration due to gravity;

[0119] Other fitting coefficients satisfy the following conditions:

[0120]

[0121]

[0122]

[0123]

[0124] 2. Vehicle Predicted Stability Margin Calculation

[0125] In order to better predict the vehicle's stable state and provide assistance for subsequent behavioral decisions, it is necessary to calculate the vehicle's current stable state;

[0126] First, calculate the target yaw rate and sideslip angle:

[0127]

[0128] right and Integrate to get β des and ω des ;

[0129] The understeer factor calculation formula is as follows:

[0130]

[0131] In the formula, is the proportional influence coefficient of the understeering factor, ranging from 1 to 5; is the vehicle speed influence coefficient, ranging from 0 to 1; is the road adhesion influence coefficient, ranging from 0 to 1; a ω is the yaw rate error coefficient, ranging from 0 to 1; a β is the mass center sideslip angle error coefficient, ranging from 0 to 1;

[0132] In the formula, μ0 satisfies the following restrictions:

[0133]

[0134] Among them, μ is the road adhesion coefficient, μ0 is the corrected road adhesion coefficient;

[0135] The oversteering factor calculation formula is as follows:

[0136] The oversteering factor is as follows:

[0137]

[0138] is the vehicle speed influence coefficient, ranging from 0 to 1; is the road adhesion influence coefficient, ranging from 0 to 1; b ω is the yaw rate error coefficient, ranging from 0 to 1; b β is the mass center sideslip angle error coefficient, ranging from 0 to 1;

[0139] If the vehicle speed v x Greater than low adhesion critical speed v xcri0 , steering wheel angle δ sw Greater than the critical rotation angle δ swcri, μ0 is less than the critical value μ of low adhesion coefficient cri0 , the slip rate of the four wheels s fl (left front wheel), s fr (right front wheel), s rl (left rear wheel), s rr (right rear wheel), any one of which is greater than the critical slip rate s cri , indicating that the vehicle is in a low-adhesion dangerous steering condition, and the oversteering factor needs to be further corrected. The corrected oversteering factor is as follows:

[0140]

[0141] In the formula, is the steering wheel angle influence coefficient, ranging from 0 to 1;

[0142] The vehicle stable state is κ stable (k),κ stable (k) = max(κ ueb (k),κ unt (k),κ untchange (k)).

[0143] 3. Prediction of obstacle status in effective road space

[0144] According to the obstacle information, lane line information, and road boundary line information collected by the laser radar and camera, the obstacle information within the sensing range and the effective road space is extracted according to the currently commonly used algorithms;

[0145] Extracted information includes longitudinal relative vehicle speed Δv xobs i. Lateral relative vehicle speed Δv yobsi , longitudinal relative acceleration Δa xobsi , lateral relative acceleration Δa yobsi , longitudinal relative displacement ΔX obsi , lateral relative displacement ΔY obsi , i ranges from 0 to N obsmax , N obsmax is the maximum number of perceptible obstacles;

[0146] In the forecast period T sp The vehicle kinematics model is used to predict obstacle information.

[0147]

[0148]

[0149] ·······

[0150]

[0151] 4. Collision risk detection

[0152] In the prediction time domain T sp Internally predict the risk of collision between surrounding obstacles and the vehicle;

[0153] The risk status of a single obstacle is calculated according to the following formula:

[0154]

[0155] In the formula, is the vertical risk impact coefficient, ranging from 0 to 1; is the lateral risk impact coefficient, ranging from 0 to 1; d i is the distance from the vehicle to the ith obstacle; l wai is the longitudinal warning distance from the vehicle to the i-th obstacle vehicle; l coi is the longitudinal collision danger distance from the vehicle to the i-th obstacle vehicle; s wai is the lateral warning distance from the vehicle to the i-th obstacle vehicle; s coi is the side collision danger distance from the vehicle to the i-th obstacle vehicle; l wai , l coi ,s wai ,s coi The specific expression is as follows:

[0156]

[0157]

[0158]

[0159]

[0160] In the formula, τ i is the critical collision time between the vehicle and a single obstacle, l wa0 is the critical braking distance, ranging from 3 to 5m; c sw is the lateral warning adjustment coefficient, with a value range of 0 to 1; is the side collision adjustment coefficient, ranging from 0 to 1; w d is the lane width, which is determined by the perceived lane information;

[0161] 5. Driving Behavior Decision

[0162] A: Collision risk space constraints

[0163] The risk status of a single obstacle calculated based on the four (collision risk detection) i, perform a risk value check on its traversal, if Risk i <Risk max , indicating that the risk value of the obstacle does not exceed the risk threshold, and there is no need to constrain the collision risk space. max is the collision risk threshold, ranging from 5 to 10. Otherwise, it indicates that the risk value of the obstacle exceeds the risk threshold, and there is a risk of collision. It is necessary to constrain the collision risk space. The collision risk space constraint process is as follows:

[0164] Assuming that there are obstacles with collision risks within the sensor sensing range and the effective lane boundary or effective driving area as the preliminary collision risk space, predict its collision risk space, and traverse all spaces with obstacles with collision risks, subtract all collision risk spaces from the preliminary collision risk space to obtain the remaining collision-free risk space;

[0165] by Figure 3 Taking the scenario shown as an example, obstacle 1 is vehicle numbered obstacle 1, obstacle 2 is vehicle numbered obstacle 2, and the longitudinal and lateral lengths of the collision risk space of a single obstacle are:

[0166] l ki = l wai +l cari +d carw ;

[0167] s ki =s wai +w cari +w carw ;

[0168] In the formula, l cari is the obstacle vehicle length, w cari is the width of the obstacle vehicle, obtained from the perception part, d carw The vertical reserved distance ranges from 0.2 to 0.5 m. carw It is the horizontal reserved distance, and the value range is 0.1~0.2m;

[0169] Taking obstacle 1 as an example, the longitudinal length and lateral length of its collision risk control are respectively,

[0170] l k1 = l wa1 +l car1 +d carw ;

[0171] s k1 =s wa1 +w car1 +w carw ;

[0172] This space indicates that under the current vehicle state, if the vehicle enters this space, there is a risk of collision with obstacle 1, and if it is not in this space, there is no risk of collision with obstacle 1;

[0173] After checking obstacle 1, check the collision risk spaces of other obstacles in turn, and combine all obstacle risk spaces to obtain the total collision risk space;

[0174] If there is a certain angle between the obstacle vehicle and the vehicle, the angle is greater than 2°, and the relative speed increment Δv yi >0.02m / s, then Figure 3 As shown in the figure, there is a certain angle between the obstacle vehicle 1 and the vehicle, which is more than 2°. The distance between the collision risk space boundary near the vehicle side and the vehicle boundary is The other side is where s wa1 Indicates the lateral warning distance from the vehicle to the first obstacle vehicle; otherwise, the two sides are evenly distributed.

[0175] B: Trajectory planning considering vehicle stability

[0176] Calculate the average stable state of the vehicle in the prediction time domain,

[0177]

[0178] like and Indicates that the vehicle will be in an understeering state in the future, κ uebmax The value range is 0.25~0.5. The lateral acceleration can be appropriately increased during trajectory planning. The target lateral displacement, target lateral velocity, and target lateral acceleration are [Y tar v ytar a ytar +Δa ytar ], the target longitudinal displacement, target longitudinal velocity, and target longitudinal acceleration are [X tar v xtar a xtar ];

[0179] Δa ytar is the target lateral acceleration increment,

[0180]

[0181] is the lateral acceleration increment coefficient, ranging from 0 to 2;

[0182] The horizontal and vertical directions are planned using fifth-order polynomials respectively:

[0183]

[0184] The lateral initial constraint is the current vehicle lateral state [Y v y a y ], the terminal state is the target quantity [Y tar v ytar a ytar +Δa ytar ], solve the coefficients a0, a1, a2, a3, a4, a5 according to the initial constraints and terminal goals,

[0185]

[0186] Where t is the time for path planning, and its value range is [0,T pend ],T pend To plan terminal time;

[0187] The same is true for the longitudinal direction. The longitudinal starting constraint is the current vehicle longitudinal state [X v x a x ], the terminal state is the target quantity [X tar v xtar a xtar ], solve the coefficients b0, b1, b2, b3, b4, b5 according to the initial constraints and terminal goals;

[0188] like or mean(κ untchange (k)) and Indicates that the vehicle will be in an oversteering state in the future, κ uebmax The value range is 0.25~0.5. The lateral acceleration can be appropriately reduced during trajectory planning. The target lateral displacement, lateral velocity, and lateral acceleration are [Y tar v ytar a ytar -Δa ytar ], the target longitudinal displacement, longitudinal velocity, and longitudinal acceleration are [X tar v xtar a xtar ];

[0189] Δa ytar is the target lateral acceleration increment,

[0190]

[0191] is the lateral acceleration increment coefficient, ranging from 0 to 2;

[0192] The horizontal and vertical directions are planned using fifth-order polynomials respectively:

[0193]

[0194] On the contrary, if or It indicates that the vehicle will be in a stable state in the future, and the target state of trajectory planning does not need to be adjusted. The target lateral displacement, lateral velocity, and lateral acceleration are [Y tar v ytar a ytar ], the target longitudinal displacement, longitudinal velocity, and longitudinal acceleration are [X tar v xtar a xtar ], and use quintic polynomials for horizontal and vertical planning respectively:

[0195]

[0196] If there is no solution within the collision risk constraint space, it means that within the feasible space, the collision cannot be avoided through active steering or partial braking, and the vehicle is at a greater risk of collision. At this time, emergency braking is required to avoid the collision or minimize the severity of the collision.

[0197] 6. Execution Control

[0198] In step 5 (driving behavior decision), the target lateral motion trajectory and longitudinal motion trajectory are calculated, and the target lateral trajectory is tracked by the lateral PID controller to calculate the target steering wheel angle, and the steering wheel control is used to control and follow; the longitudinal PID controller is used to track the target longitudinal vehicle speed and calculate the target total torque M. tarall , dynamically adjust and allocate the four-wheel motors, and the steps of dynamic adjustment and allocation are as follows:

[0199] If the vehicle's future motion state is in a steady state ( or ), the torque of the four-wheel motors is distributed evenly.

[0200]

[0201]

[0202]

[0203]

[0204] T bfl , T bfr , T brl , T brr They are the motor torques of the left front wheel, right front wheel, left rear wheel, and right rear wheel respectively;

[0205] If the vehicle's future motion state is understeering ( and ), and the vehicle turns left in the future, the right wheel increases torque and the left wheel decreases torque,

[0206]

[0207]

[0208]

[0209]

[0210] k ueωl is the left understeer correction coefficient, k ueβl is the left understeer β correction coefficient;

[0211] If the vehicle's future motion state is understeering ( and ), and the vehicle turns right in the future, the left wheel increases torque and the right wheel decreases torque,

[0212]

[0213]

[0214]

[0215]

[0216] k ueωr is the right understeer correction coefficient, k ueβl is the right understeer β correction coefficient;

[0217] If the vehicle's future motion state is in oversteering ( or mean(κ untchange (k)) and ), and the vehicle turns left in the future, the right wheel reduces torque and the left wheel increases torque,

[0218]

[0219]

[0220]

[0221]

[0222] k unωl is the left oversteering correction coefficient, k unβl is the left oversteering β correction coefficient;

[0223] If the vehicle's future motion state is in oversteering ( or mean(κ untchange (k)) and ), and the vehicle turns right in the future, the left wheel reduces torque and the right wheel increases torque,

[0224]

[0225]

[0226]

[0227]

[0228] k unωr is the correction coefficient for right oversteering, k unβl It is the β correction coefficient of right oversteer.

[0229] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A lateral and longitudinal integrated predictive control method for autonomous driving, characterized in that: include: The current vehicle status parameters are obtained through sensors: longitudinal displacement X, lateral displacement Y, longitudinal vehicle speed v x , lateral speed v y , longitudinal acceleration a x , lateral acceleration a y , heading angle θ, steering wheel angle δ sw ; Predicting the future state of the vehicle according to the current vehicle state parameters to obtain the longitudinal predicted vehicle speed, lateral predicted vehicle speed, predicted yaw angular velocity and predicted sideslip angle of the center of mass at each moment in the vehicle prediction time domain; Calculate the vehicle predicted stability margin κ stable (k); k stable (k)=max(κ ueb (k),k unt (k),k untchange (k)); Among them, κ ueb (k) represents the understeering factor, κ unt (k) represents the oversteering factor, κ untchange (k) represents the corrected oversteering factor; In the formula, s fl 、s fr 、s rl and rr are the slip rates of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; is the steering wheel angle influence coefficient, δ sw is the steering wheel angle, δ swcri is the critical turning angle; Predict the state of obstacles around the vehicle and determine the space without collision risk; Performing vehicle trajectory planning in the collision-free risk space to obtain a target lateral displacement, a target lateral velocity, a target lateral acceleration, a target longitudinal displacement, a target longitudinal velocity, and a longitudinal acceleration of the vehicle; Among them, when and When the target lateral acceleration is corrected to a ytar1 ′; when or mean(κ untchange (k)) and When the target lateral acceleration is corrected to a ytar2 ′; In the formula, is the lateral acceleration increment coefficient, ω des represents the target yaw rate; ω represents the actual yaw rate, a ytar represents the target lateral acceleration; κ uebmax is the maximum value of understeering factor; The steering wheel angle is controlled and the torque of the four-wheel motor is distributed according to the vehicle trajectory planning results.

2. The lateral and longitudinal integrated predictive control method for autonomous driving according to claim 1, characterized in that: Longitudinal predicted vehicle speed at each moment in the vehicle prediction time domain Lateral predicted vehicle speed Predicted yaw rate and predicted slip angle The calculation method is: in, represents the longitudinal predicted vehicle speed obtained from historical data, represents the lateral predicted vehicle speed obtained from historical data, represents the predicted yaw rate obtained from historical data, It represents the predicted sideslip angle of the center of mass predicted by historical data; represents the longitudinal predicted vehicle speed predicted by the linear vehicle dynamics model, represents the lateral predicted vehicle speed predicted by the linear vehicle dynamics model, represents the predicted yaw rate predicted by the linear vehicle dynamics model, represents the predicted sideslip angle of the center of mass predicted by the linear vehicle dynamics model; represents the longitudinal predicted vehicle speed predicted by the nonlinear vehicle dynamics model, represents the lateral predicted vehicle speed predicted by the nonlinear vehicle dynamics model, represents the predicted yaw rate predicted by the nonlinear vehicle dynamics model, represents the predicted sideslip angle of the center of mass obtained by the nonlinear vehicle dynamics model; ρ vxdata , vxdyn , vydata , vydyn , ωdata , ωdyn , βdata and ρ βdyn are the coefficients of each item respectively; a yc It is expressed as the corrected lateral acceleration; g is the acceleration due to gravity.

3. The method for lateral and longitudinal integrated predictive control for autonomous driving according to claim 2, characterized in that: The understeer factor calculation formula is: In the formula, is the proportional influence coefficient of the understeer factor; is the vehicle speed influence coefficient of understeering factor; is the road adhesion influence coefficient; a ω is the yaw rate error coefficient of the understeering factor; a β is the center of mass sideslip angle error coefficient of the understeering factor; β des is the target center of mass sideslip angle; Predict the sideslip angle of the center of mass at time k; Predict the yaw rate at time k; is the predicted longitudinal vehicle speed at time k; μ0 is the corrected road adhesion coefficient.

4. The lateral and longitudinal integrated predictive control method for autonomous driving according to claim 3, characterized in that: The calculation formula of oversteering factor is: Where b vx is the vehicle speed influence coefficient of the oversteering factor; b μ0 is the road adhesion influence coefficient of the oversteering factor; b ω is the yaw rate error coefficient of the oversteering factor; b β is the center of mass sideslip angle error coefficient of the oversteering factor.

5. The lateral and longitudinal integrated predictive control method for autonomous driving according to claim 3 or 4, characterized in that: The method to determine the collision risk-free space is: Traverse all single obstacle risk values ​​Risk i ; If Risk i <Risk max , there is no need to constrain the collision risk space; If Risk i ≥Risk max , then the collision risk space of a single obstacle i is predicted; Among them, Risk max is the collision risk threshold; Traverse the collision risk spaces of all obstacles with collision risks, subtract all collision risk spaces from the initial collision risk space, and obtain the remaining collision-free risk space; The preliminary collision risk space is the space within the effective lane boundary line or the effective driving area within the sensor perception range.

6. The method for lateral and longitudinal integrated predictive control for autonomous driving according to claim 5, characterized in that: Risk value of a single obstacle i Calculated by the following formula; in, is the vertical risk impact coefficient; is the lateral risk impact coefficient; d i is the distance from the vehicle to the ith obstacle; l wai is the longitudinal warning distance from the vehicle to obstacle i; l coi is the longitudinal collision danger distance from the vehicle to obstacle i; s wai is the lateral warning distance from the vehicle to obstacle i; s coi is the side collision danger distance from the vehicle to obstacle i; Δv xobsi is the longitudinal relative speed between the vehicle and obstacle i, Δv yobsi is the lateral relative speed between the vehicle and obstacle i.

7. The lateral and longitudinal integrated predictive control method for autonomous driving according to claim 6, characterized in that: If there is no solution for vehicle trajectory planning in the collision-free risk space, the vehicle is controlled to perform emergency braking.

8. The lateral and longitudinal integrated predictive control method for autonomous driving according to claim 7, characterized in that: if or The torque of the four-wheel motors is evenly distributed; Among them, T bfl 、T bfr 、T brl 、T brr They are the motor torques of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; a is the distance from the center of mass to the front axle; b is the distance from the center of mass to the rear axle; m is the mass of the vehicle.

9. The lateral and longitudinal integrated predictive control method for autonomous driving according to claim 8, characterized in that: if and When the vehicle turns left in the future; In the formula, k ueωl is the left understeer correction coefficient, k ueβl is the left understeer β correction coefficient; When the vehicle turns right in the future, In the formula, k ueωr is the right understeer correction coefficient, k ueβl It is the right understeer β correction coefficient.

10. The lateral and longitudinal integrated predictive control method for autonomous driving according to claim 9, characterized in that: or mean(κ untchange (k)), and When the vehicle turns left in the future: In the formula, k unωl is the left oversteering correction coefficient, k unβl is the oversteering β correction coefficient; When the vehicle turns right in the future, In the formula, k unωr is the correction coefficient for right oversteering, k unβl It is the β correction coefficient of right oversteer.

Images