Safety-first game incremental man-machine transverse sharing driving control method

Abstract

The invention discloses a safety-first game incremental man-machine transverse shared driving control method, and belongs to the technical field of automobile automatic driving and man-machine co-driving, and the method comprises the steps: obtaining a driving intention safety factor of a driver; acquiring an expected track of the driver and the intelligent system; calculating a man-machine intention consistency coefficient according to the safety factor and the discrete point position information of the man-machine expected trajectory; according to the safety factor and the man-machine intention consistency coefficient, a driving control right distribution strategy with safety priority is designed, driving control right coefficients Wh and Wm are calculated, and Wh and Wm are weight coefficients of a driver and a machine control torque item in the man-machine shared driving controller respectively; establishing a human-vehicle-road closed-loop system preview error dynamical model of tactile interaction; and combining a game theory and an incremental model prediction control theory to construct a man-machine transverse sharing driving control method based on a safety-first driving control right distribution rule.

Description

Technical Field

[0001] This invention belongs to the fields of autonomous driving technology and human-machine co-driving technology, specifically relating to a safety-first, game-theoretic incremental human-machine lateral shared driving control method. Background Technology

[0002] Autonomous driving technology has shown great potential in improving driving safety and reducing driver workload. However, limited by current technological development and public acceptance, fully autonomous driving technology is unlikely to become widespread in the short term. Level 3 autonomous driving has become an important path for the gradual development of autonomous driving technology. In the human-machine co-driving technology framework, both the autonomous driving system and the natural driver can drive the vehicle independently. They control the vehicle and complete driving tasks through information interaction, with time-sharing and authority-sharing. Human-machine shared driving control can effectively avoid the risks caused by the driver leaving the control loop and being unable to take over the vehicle in time, achieving complementary advantages between driver capabilities and intelligent systems. However, because the driver and the intelligent system are both in the control loop, they are coupled and mutually restrictive, which may even lead to control conflicts when human and machine intentions are inconsistent. Therefore, shared control methods with the ability to resolve driving rights conflicts have become a research focus in the field of human-machine co-driving.

[0003] Chinese invention patent (CN113650609A) uses fuzzy logic rules to obtain human-machine driving weight coefficients based on the lateral distance and error variation of the vehicle's center of gravity from the lane centerline. It then establishes a human-machine driving rights sharing model by linearly weighting human-machine control commands to determine the final vehicle steering wheel angle control input. However, this patent only simulates the driver's driving intentions under conditions of distraction or fatigue, and does not fully consider human-machine intentions. Figure 1 Regarding consistency, Chinese invention patent (CN118859951A) discloses a human-machine shared control method based on multi-objective model predictive control. This method adjusts the shared control weight factor based on the vehicle's arrival time at the lane boundary and the collision time, dynamically adjusting the driver's steering command to follow the target and the obstacle avoidance trajectory to follow the target. However, this method struggles to reflect the dynamic control interaction process between the driver and the intelligent system. Chinese invention patent (CN120245993A) considers the driving risks of different types of drivers, using fuzzy logic to infer the driver's control weight, and then establishes a non-cooperative game-theoretic shared control model. In this method, the driver and the intelligent system only consider their own tracking trajectories, which can easily lead to a stalemate of continuous human-machine confrontation when their intentions are inconsistent. Therefore, how to fully consider the intentions of the human and machine... Figure 1 To ensure consistency, designing dynamic allocation rules for human-machine driving rights and establishing a shared control method capable of resolving driving rights conflicts are key issues that urgently need to be addressed. Summary of the Invention

[0004] To address the problems of the aforementioned solutions, this invention provides a safety-first, game-theoretic, incremental, human-machine lateral shared driving control method, which includes: Obtain driver's driving intention safety factors and quantify the safety of driving intentions; Obtain the desired trajectory of the driver and the intelligent system; Based on the safety factor and the discrete point location information of the human-machine expected trajectory, the human-machine intention is calculated. Figure 1 Consistency coefficient, which quantitatively represents the degree of difference between human and machine driving intentions; Considering safety factors and human-machine interaction Figure 1 Consistency coefficient, design a safety-first driving control allocation strategy, and realize the driving control authority coefficient W. h and W m Dynamic adjustment; Establish a dynamic model of the aiming error of the human-vehicle-road closed-loop system of tactile interaction; Combining game theory and incremental model predictive control theory, and based on a safety-first driving control allocation rule, a human-machine lateral shared driving control method is constructed to achieve safe resolution of human-machine driving rights conflicts and smooth control of human-machine shared driving.

[0005] Human-machine interaction Figure 1 Consistency α is used to quantify the intention conflict between the driver and the intelligent system, based on human-machine intention... Figure 1 The consistency index dynamically adjusts the human-machine driving control authority coefficient W. h and W m On the other hand, human-machine interaction Figure 1 Alignment α is also used to adjust the machine's driving intentions. The machine modifies its own driving intentions based on the driver's safety intentions, so as to reduce the conflict between human and machine driving intentions from the source.

[0006] Computational human-machine interface Figure 1 The expression for the consistency coefficient is: Where γ is the driver's driving intention safety factor, γ=0 indicates that the driver has a high-risk driving intention, in which case α=1 means that the machine only uses its own driving intention as the tracking target and does not consider the driver's intention, ensuring safe vehicle operation; TIC represents the human-machine target trajectory deviation; when the driver's intention is safe, i.e. γ=1, the human-machine intention... Figure 1 Consistency α is inversely proportional to the target trajectory deviation (TIC) of the driver and the machine.

[0007] Among them, the human-machine target trajectory deviation (TIC) is defined as N. p Average deviation of expected human-machine trajectory within the step prediction domain: Where, N pThis represents the prediction step size of the incremental shared controller. N is defined p The average deviation of the human-machine expected trajectory within the predicted step; c TIC Defined as the trajectory deviation threshold, when the average deviation of the expected human-machine trajectory exceeds this threshold, the trajectory deviation reaches its maximum, i.e., TIC=1. Trajectory deviation threshold c TIC Personalized settings can be customized to suit the driver's driving preferences based on driving experience.

[0008] Where d is defined hm,i Let be the Euclidean distance between the expected human-machine trajectories at the i-th time step in the Frenet coordinate system. Specifically, it can be calculated based on the positions of the corresponding trajectory points of the expected human-machine trajectories in the Frenet coordinate system. In the formula, the lane centerline is taken as the reference line in the Frenet coordinate system, s i Let be the vertical position of the Frenet coordinate system at the i-th time step. x h,i , y h,i )and( x m,i , y m,i ) represent the discrete points of the target trajectory of the driver and the machine at the i-th time step, respectively. x h,i and y h,i The first in the geodetic coordinate system i The location of discrete points on the human driver's target trajectory at each time step; x m,i and y m,i The first in the geodetic coordinate system i The location of discrete points on the target trajectory of the intelligent system at each time step.

[0009] Based on human-machine intention Figure 1 Based on consistency and safety factors, a safety-first driving control allocation strategy is designed, namely the driving control authority coefficient. W h and W m The adjustment rules. Among them, W h and W m These are the weighting coefficients for the driver and machine control torque terms in the human-machine shared driving controller, respectively, which are adjusted... W h and W mThe relative relationship can realize human-machine control torque T h and T m The adjustment is specifically manifested in larger W m Limit the control torque of the machine.

[0010] Driving control authority coefficient W h and W m The main regulatory principle is: when the driver intends to cause danger, the principle of "safety first" is followed, and driving control is allocated to the machine, with the machine only tracking its desired safe path. Therefore... γ When =0, α The value should always be 1. W m Should be smaller W h This is to facilitate the allocation of driving control to the machine. In situations where the driver's intention is safety-oriented, and human and machine intentions conflict, the principle of "human-centeredness" is followed, allocating driving control to the driver to ensure the driver's driving intentions are met; human-machine intentions... Figure 1 At this time, the machine and the driver share driving rights to reduce the driver's workload; with the interaction between human and machine... Figure 1 As consistency decreases, control should gradually be distributed towards the driver. Therefore... γ =1 and α When =0, W m Should be greater than W h This ensures that driving control is assigned to the driver; γ =1 and α Design when =1 W m = W h Ensure that humans and machines share driving rights; γ =1 and α When the value decreases from 1 to 0, W m It should be gradually increased. W h It should be reduced gradually.

[0011] Driving control authority coefficient W h and W m The specific expression for the adjustment rule is: In the formula, W h0 andW m0 The basic control weighting coefficient; E h and E m This is a correction factor. Basic control weight coefficient. W h0 and W m0 Decision made α Human-machine driving control authority allocation when =0, introducing a correction factor. E h and E m To adjust the rate of change of the exponential function. Through extensive simulation experiments, the basic control weights and correction factors were set to... , , and .

[0012] Considering the torque coupling characteristics of human-machine control, a dynamic model of the aiming error of the human-vehicle-road closed-loop system based on tactile interaction is derived, with the expression as follows: Wherein, state x and output y are respectively and , and These represent the lateral position error and the heading angle error, respectively. v y and ω These represent the lateral velocity and yaw rate at the vehicle's center of gravity, respectively. θ Steering wheel angle; τ h and τ m The control torque applied to the steering wheel by the driver and the intelligent system respectively κ r The curvature of the reference path is considered as an environmental disturbance term; the coefficient matrix is ​​as follows: in, c f and c r These represent the equivalent lateral stiffness of the front and rear tires, respectively. l f and l r These represent the distances from the vehicle's center of gravity to the front and rear axles, respectively. v x This indicates the longitudinal velocity at the vehicle's center of gravity. Lp Pre-aiming distance; K v This is the total gain when the return torque is transmitted from the tires to the steering wheel; i sw Steering ratio; J eq It is the equivalent inertia of the driver-steering wheel interaction system; b eq This is the equivalent damping of the system; Discretizing the above-mentioned aiming error dynamics model yields: ; In the formula, , , , , Ts is the sampling time; Incremental MPC is used to update the future state of the vehicle in the prediction time domain, and an augmented state space equation is constructed based on a discrete aiming error dynamics model: In the formula, the augmented state vector ξ ( k ) represents ξ ( k )=[ x ( k ) τ h ( k ) τ m ( k )] T ;Δ τ h and Δ τ m These are the torque control increments for the driver and the machine, respectively; η ( k ) is the first k The augmented output vector of the step; the coefficient matrices are as follows: By iteratively updating the prediction equations, the shared lateral control system in the future is obtained. N p In-step error prediction output: In the formula, Y(k) is N p The step-by-step prediction of the controlled output in the time domain is expressed as: and They are respectively N c The control inputs of the driver and the machine in the step control time domain are represented as follows: in, N c To control the step size. for N p The reference path curvature in the time domain is predicted stepwise, expressed as: The coefficient matrices in the motion error prediction equation are represented as follows: Based on the motion error prediction equation, we can obtain N p The prediction steps cover the lateral position error and heading angle error of the vehicle relative to the reference line in the Frenet coordinate system within the time domain. Since the inherent conflict between human and machine intentions can cause conflict in human-machine control torque, it is also necessary to output the torques of both the driver and the machine to resolve this conflict. N c The absolute torque control quantity of the driver and the machine within the step time domain.

[0013] The designed output coefficient matrix is ​​as follows: and At that time, the predicted output of the absolute torque control quantity T for the driver and the machine is calculated. h and T m , is represented as: The coefficients are expressed as follows: The optimization objective function for human-machine lateral shared driving control is established as follows: In the formula, the output matrix Y h and Y m These represent the lateral position and heading angle errors of the vehicle relative to the driver and machine target tracking paths within the prediction domain, respectively. α Indicating human-machine interaction Figure 1 Consistency, a value between 0 and 1; error weight matrix Q h and Qm Constraining the vehicle's error relative to the target trajectory of the driver and machine affects trajectory tracking accuracy; controlling the incremental weight matrix R h and R m Constraining the incremental changes in control for both the driver and the machine helps improve control smoothness; torque weighting matrix W h and W m The control torque that constrains the driver and the machine determines the degree of effort exerted by the driver and the machine during driving tasks. T max , T min and Δ T max Δ T min These are the upper and lower limit constraints for human-machine torque and torque increment within the control time domain, respectively.

[0014] Based on the concept of "optimal response," the Nash equilibrium of incremental human-machine lateral shared control based on game theory can be expressed as: in, These are the optimal control strategies for the driver and the machine under their respective optimal strategies, and the strategy combination. This is the Nash equilibrium solution. At the Nash equilibrium point, in the machine... Under the optimal strategy, the driver's choice of other strategies will not be worse than the choice of strategy. This produces better results. The same applies to machines.

[0015] To solve the Nash equilibrium in real time, the binary agent coupled optimization control problem is formulated as a standard quadratic form, and then solved using a quadratic programming algorithm. The first control step in the solved control sequence is the Nash equilibrium of the human-machine game. To avoid the binary agent optimization control problem being unsolvable, a relaxation factor is introduced into the optimization objective. ε h and ε m The standard quadratic optimization problem is formulated as follows: in, and These are the weighting coefficients of the relaxation factor term; Cons h and Cons m This is a constant term and has no effect on the solution; it can be ignored. The other variables are expressed as follows: .

[0016] Compared with the prior art, the beneficial effects of the present invention are: The proposed human-machine shared control method based on safety-priority driving rights allocation calculates the human-machine intention based on safety factors and discrete point location information of the human-machine desired trajectory. Figure 1 Consistency coefficient; comprehensively considering safety factor coefficient and human-machine intention. Figure 1 The consistency coefficient is used to calculate the driving control rights allocation coefficient between the driver and the intelligent system. Combining game theory and incremental model predictive control theory, a human-machine lateral shared driving control method based on a safety-first driving control rights allocation strategy is established to achieve safe resolution of human-machine driving rights conflicts and smooth control of human-machine shared driving. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a game-theoretic incremental human-machine lateral shared driving control method based on a safety-first driving rights allocation strategy; Figure 2 Human-machine interface when the driver's driving intention is dangerous Figure 1 Schematic diagram of consistency quantification characterization; Figure 3 Human-machine interface when the driver intends to drive safely Figure 1 Schematic diagram of consistency quantification characterization; Figure 4 This is a schematic diagram of the adjustment rules for the human-machine pilot control coefficient. Detailed Implementation

[0019] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] Reference Figure 1 This invention proposes a safety-first, game-theoretic, incremental, human-machine lateral shared driving control method, comprising: acquiring the driver's driving intention safety factor. γ and the driver's and intelligent system's desired trajectory; based on safety factors and discrete point location information of the human-machine desired trajectory ( xh,i ( s i ), y h,i ( s i ))and( x m,i ( s i ), y m,i ( s i )), Calculate human-machine intention Figure 1 Coherence coefficient α Considering safety factors and human-machine interaction Figure 1 Consistency coefficient, design a safety-first driving control allocation strategy, and calculate the driving control authority coefficient. W h and W m Establish a dynamic model of the anti-aiming error of the human-vehicle-road closed-loop system with tactile interaction; combine game theory and incremental model predictive control theory to construct a human-machine lateral shared driving control method based on a safety-first driving control allocation strategy, so as to achieve safe resolution of human-machine driving rights conflicts and smooth control of human-machine shared driving.

[0021] In this invention, the driver's driving intention can be assessed by an upper-level risk assessment system, and the safety of the driver's intention provided by the upper-level risk assessment system is defined as a safety factor. γ ,in, γ =0 indicates that the driver's driving intentions are unsafe and the risk of collision is high; γ =1 indicates that the driver has a safe driving intention, and driving according to the driver's intention carries a low risk. Human-Machine Intention Figure 1 Desire α Used to quantify the intention conflict between the driver and the intelligent system, based on human-machine intention... Figure 1 Dynamic adjustment of human-machine driving control authority coefficient based on consistency index W h and W m On the other hand, human-machine interaction Figure 1 Desire α This is used to adjust the machine's driving intentions, enabling the machine to modify its driving intentions according to the driver's safety intentions, thereby reducing the degree of conflict between human and machine driving intentions at the source. It comprehensively considers the discrete point position information of safety factors and the desired human-machine trajectory. x h,i ( s i ), y h,i ( s i))and( x m,i ( s i ), y m,i ( s i )), Calculate human-machine intention Figure 1 Coherence coefficient α .

[0022] First, define d hm,i For the Frenet coordinate system i The Euclidean distance between the expected human-machine trajectories at each time step can be calculated based on the positions of the corresponding trajectory points in the Frenet coordinate system: In the formula, the lane centerline is taken as the reference line in the Frenet coordinate system. s i For the Frenet coordinate system in the th... i The longitudinal position of each time step, ( x h,i , y h,i )and( x m,i , y m,i ) are respectively the first i The discrete points of the target trajectory of the driver and the machine at each time step. x h,i and y h,i The first in the geodetic coordinate system i The location of discrete points on the human driver's target trajectory at each time step; x m,i and y m,i The first in the geodetic coordinate system i The location of discrete points on the target trajectory of the intelligent system at each time step.

[0023] Defined as N p The average deviation of the expected human-machine trajectory within the step prediction domain is the deviation of the human-machine target trajectory. TIC : In the formula, N p This represents the prediction step size of the incremental shared controller. Defined N p The average deviation of the human-machine expected trajectory within the predicted step; cTIC Defined as the trajectory deviation threshold, when the average deviation of the human-machine expected trajectory exceeds this threshold, the degree of human-machine trajectory deviation reaches its maximum, i.e. TIC =1. Trajectory deviation threshold c TIC Personalized settings can be customized to suit the driver's driving preferences based on driving experience.

[0024] Furthermore, based on the deviation between the human and machine's expected trajectory and the safety factor, the human-machine intention is calculated. Figure 1 α-homogeneity: In the formula, γ is the safety factor. γ=0 indicates that the driver has a high-risk driving intention. In this case, α=1 means that the machine only uses its own driving intention as the tracking target and does not consider the driver's intention, ensuring safe vehicle operation. When the driver's intention is safe, i.e., γ=1, the human-machine interaction is considered safe. Figure 1 Consistency α is inversely proportional to the deviation of the driver and the machine from the target trajectory.

[0025] In this invention, based on safety factors and human-machine interaction... Figure 1 Consistency coefficient, obtained by using human-machine driving authority adjustment rules to obtain human-machine driving control authority coefficient. W h and W m In incremental shared controllers, W h and W m These are the weighting coefficients for the driver and machine control torque terms, respectively, adjusted by... W h and W m The relative relationship can realize human-machine control torque T h and T m The adjustment is specifically manifested in larger W m Limit the machine's control torque. Based on driving intention. Figure 1 Consistency and safety factors W h and W m The adjustment rules are as follows: In the formula, W h0 and W m0 The basic control weighting coefficient; E h and E mThis is a correction factor. Basic control weight coefficient. W h0 and W m0 Decision made α Human-machine driving control authority allocation when =0, introducing a correction factor. E h and E m To adjust the rate of change of the exponential function. Through extensive simulation experiments, the basic control weights and correction factors were set to... , , and .

[0026] When the driver intends to cause danger, following the principle of "safety first," driving control is delegated to the machine, and the machine only tracks its desired safe path. Therefore... γ When =0, α The value should always be 1. W m Should be smaller W h This is to facilitate the allocation of driving control to the machine. In situations where the driver's intention is safety-oriented, and human and machine intentions conflict, the principle of "human-centeredness" is followed, allocating driving control to the driver to ensure the driver's driving intentions are met; human-machine intentions... Figure 1 At this time, the machine and the driver share driving rights to reduce the driver's workload; with the interaction between human and machine... Figure 1 As consistency decreases, control should gradually be distributed towards the driver. Therefore... γ =1 and α When =0, W m Should be greater than W h This ensures that driving control is assigned to the driver; γ =1 and α Design when =1 W m = W h Ensure that humans and machines share driving rights; γ =1 and α When the value decreases from 1 to 0, W m It should be gradually increased. W h It should be reduced gradually.

[0027] Considering the torque coupling characteristics of human-machine control, a dynamic model of the aiming error of the human-vehicle-road closed-loop system based on tactile interaction is derived, with the expression being: Among them, state x With output y They are respectively and , and These represent the lateral position error and the heading angle error, respectively. v y and ω These represent the lateral velocity and yaw rate at the vehicle's center of gravity, respectively. θ Steering wheel angle; τ h and τ m The control torque applied to the steering wheel by the driver and the intelligent system respectively κ r The curvature of the reference path is considered as an environmental disturbance term; the coefficient matrix is ​​as follows: in, c f and c r These represent the equivalent lateral stiffness of the front and rear tires, respectively. l f and l r These represent the distances from the vehicle's center of gravity to the front and rear axles, respectively. v x This indicates the longitudinal velocity at the vehicle's center of gravity. L p Pre-aiming distance; K v This is the total gain when the return torque is transmitted from the tires to the steering wheel; i sw Steering ratio; J eq It is the equivalent inertia of the driver-steering wheel interaction system; b eq This is the equivalent damping of the system.

[0028] Discretizing the above-mentioned aiming error dynamics model yields: In the formula, , , , , Ts is the sampling time.

[0029] Incremental MPC is used to update the future state of the vehicle in the prediction time domain, and an augmented state space equation is constructed based on a discrete aiming error dynamics model: In the formula, the augmented state vector ξ ( k ) represents ξ ( k )=[ x ( k ) τ h ( k ) τ m ( k )] T ;Δ τ h and Δ τ m These are the torque control increments for the driver and the machine, respectively; η ( k ) is the first k The augmented output vector of the step; the coefficient matrices are as follows: By iteratively updating the prediction equations, the shared lateral control system in the future is obtained. N p In-step error prediction output: In the formula, Y(k) is N p The step-by-step prediction of the controlled output in the time domain is expressed as: and They are respectively N c The control inputs of the driver and the machine in the step control time domain are represented as follows: in, N c To control the step size. for N p The reference path curvature in the time domain is predicted stepwise, expressed as: The coefficient matrices in the motion error prediction equation are represented as follows: Based on the motion error prediction equation, we can obtain N pThe prediction steps cover the lateral position error and heading angle error of the vehicle relative to the reference line in the Frenet coordinate system within the time domain. Since the inherent conflict between human and machine intentions can cause conflict in human-machine control torque, it is also necessary to output the torques of both the driver and the machine to resolve this conflict. N c The absolute torque control quantity of the driver and the machine within the step time domain.

[0030] The designed output coefficient matrix is ​​as follows: and At the same time, predict the absolute torque control quantity output by the driver and the machine. T h and T m , is represented as: The coefficients are expressed as follows: The optimization objective function for human-machine lateral shared driving control is established as follows: In the formula, the output matrix Y h and Y m These represent the lateral position and heading angle errors of the vehicle relative to the driver and machine target tracking paths within the prediction domain, respectively. α Indicating human-machine interaction Figure 1 Consistency, a value between 0 and 1; error weight matrix Q h and Q m Constraining the vehicle's error relative to the target trajectory of the driver and machine affects trajectory tracking accuracy; controlling the incremental weight matrix R h and R m Constraining the incremental changes in control for both the driver and the machine helps improve control smoothness; torque weighting matrix W h and W m The control torque that constrains the driver and the machine determines the degree of effort exerted by the driver and the machine during driving tasks. T max , T min and Δ T max Δ T min These are the upper and lower limit constraints for human-machine torque and torque increment within the control time domain, respectively.

[0031] Based on the concept of "optimal response," the Nash equilibrium of incremental human-machine lateral shared control based on game theory can be expressed as: in, These are the optimal control strategies for the driver and the machine under their respective optimal strategies, and the strategy combination. This is the Nash equilibrium solution. At the Nash equilibrium point, in the machine... Under the optimal strategy, the driver's choice of other strategies will not be worse than the choice of strategy. This produces better results. The same applies to machines.

[0032] To solve the Nash equilibrium in real time, the binary agent coupled optimization control problem is formulated as a standard quadratic form, and then solved using a quadratic programming algorithm. The first control step in the solved control sequence is the Nash equilibrium of the human-machine game. To avoid the binary agent optimization control problem being unsolvable, a relaxation factor is introduced into the optimization objective. ε h and ε m The standard quadratic optimization problem is formulated as follows: in, and These are the weighting coefficients of the relaxation factor term; Cons h and Cons m This is a constant term and has no effect on the solution; it can be ignored. The other variables are expressed as follows: The above formulas are all numerical calculations after removing dimensions. The formulas are obtained by software simulation based on a large amount of data and are closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulation based on a large amount of data.

[0033] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A safety-first, game-theoretic, incremental, human-machine lateral shared driving control method, characterized in that the method... include: Obtain driver's driving intention safety factor; Obtain the desired trajectory of the driver and the intelligent system; Calculate the consistency coefficient of human-machine intent based on the safety factor and the discrete point location information of the human-machine expected trajectory; Based on the safety factor and the consistency coefficient of human-machine intent, a safety-first driving control allocation strategy is designed, and the driving control authority coefficient W is calculated. h and W m , W h and W m These are the weighting coefficients for the driver and machine control torque terms in the human-machine shared driving controller, respectively. Establish a dynamic model of the aiming error of the human-vehicle-road closed-loop system of tactile interaction; Combining game theory and incremental model predictive control theory, a human-machine lateral shared driving control method based on a safety-first driving control allocation rule is constructed.

2. The safety-first, game-theoretic, incremental human-machine lateral shared driving control method according to claim 1, characterized in that, The expression for calculating the human-machine intent consistency coefficient is: Wherein, γ is the driver's driving intention safety factor, γ=0 indicates that the driver has a high-risk driving intention, at which point α=1 indicates that the machine only uses its own driving intention as the tracking target without considering the driver's intention, ensuring the safe driving of the vehicle; TIC represents the human-machine target trajectory deviation; when the driver's intention is safe, i.e. γ=1, the human-machine intention consistency α is inversely proportional to the driver and machine target trajectory deviation TIC.

3. The safety-first, game-theoretic, incremental human-machine lateral shared driving control method according to claim 2, characterized in that, Human-machine target trajectory deviation (TIC) is defined as N. p Average deviation of expected human-machine trajectory within the step prediction domain: Where, N p This represents the prediction step size of the incremental shared controller. N is defined p The average deviation of the human-machine expected trajectory within the predicted step; c TIC Defined as a trajectory deviation threshold, when the average deviation of the human-machine expected trajectory exceeds this threshold, TIC=1; defined as d hm,i Let be the Euclidean distance between the expected trajectories of the human and machine at the i-th time step in the Frenet coordinate system.

4. The safety-first, game-theoretic, incremental human-machine lateral shared driving control method according to claim 1, characterized in that, Driving control authority coefficient W h and W m The specific expression for the adjustment rule is: ； In the formula, W h0 and W m0 The basic control weighting coefficient; E h and E m Correction factor; basic control weight coefficient W h0 and W m0 Decision made α Human-machine driving control authority allocation when =0, introducing a correction factor. E h and E m To adjust the rate of change of the exponential function; Set the basic control weights and correction factors to... , , and .

5. The safety-first, game-theoretic, incremental, human-machine lateral shared driving control method according to claim 1, characterized in that, The dynamic model of the aiming error in the human-vehicle-road closed-loop system of haptic interaction is expressed as follows: Wherein, state x and output y are respectively and , and These represent the lateral position error and the heading angle error, respectively. v y and ω These represent the lateral velocity and yaw rate at the vehicle's center of gravity, respectively. θ Steering wheel angle; τ h and τ m The control torque applied to the steering wheel by the driver and the intelligent system respectively κ r The curvature of the reference path is considered as an environmental disturbance term; the coefficient matrix is ​​as follows: in, c f and c r These represent the equivalent lateral stiffness of the front and rear tires, respectively. l f and l r These represent the distances from the vehicle's center of gravity to the front and rear axles, respectively. v x This indicates the longitudinal velocity at the vehicle's center of gravity. L p Pre-aiming distance; K v This is the total gain when the return torque is transmitted from the tires to the steering wheel; i sw Steering ratio; J eq It is the equivalent inertia of the driver-steering wheel interaction system; b eq This is the equivalent damping of the system; Discretizing the above-mentioned aiming error dynamics model yields: ； In the formula, , , , , Ts is the sampling time; Incremental MPC is used to update the future state of the vehicle in the prediction time domain, and an augmented state space equation is constructed based on a discrete aiming error dynamics model: In the formula, the augmented state vector ξ ( k ) represents ξ ( k )=[ x ( k ) τ h ( k ) τ m ( k )] T ;Δ τ h and Δ τ m These are the torque control increments for the driver and the machine, respectively; η ( k ) is the first k The augmented output vector of the step; the coefficient matrices are as follows: By iteratively updating the prediction equations, the shared lateral control system in the future is obtained. N p In-step error prediction output: In the formula, Y(k) is N p The step-by-step prediction of the controlled output in the time domain is expressed as: and They are respectively N c The control inputs of the driver and the machine in the step control time domain are represented as follows: in, N c To control the step size; for N p The reference path curvature in the time domain is predicted stepwise, expressed as: The coefficient matrices in the motion error prediction equation are represented as follows: Based on the motion error prediction equation, obtain N p The lateral position error and heading angle error of the vehicle relative to the reference line in the Frenet coordinate system are predicted in the time domain. The designed output coefficient matrix is ​​as follows: and At that time, the predicted output of the absolute torque control quantity T for the driver and the machine is calculated. h and T m , is represented as: The coefficients are expressed as follows: The optimization objective function for human-machine lateral shared driving control is established as follows: In the formula, the output matrix Y h and Y m These represent the lateral position and heading angle errors of the vehicle relative to the driver and machine target tracking paths within the prediction domain, respectively. α Represents the consistency of human-machine intent, expressed as a value between 0 and 1; Error weight matrix Q h and Q m Constrain the vehicle's error relative to the target trajectory of the driver and machine; control the incremental weight matrix. R h and R m Constraining the incremental changes in control for both the driver and the machine; torque weighting matrix W h and W m Constraining the control torque of the driver and the machine; T max , T min and Δ T max Δ T min These are the upper and lower limit constraints for human-machine torque and torque increment within the control time domain, respectively.

6. The safety-first, game-theoretic incremental human-machine lateral shared driving control method according to claim 1, characterized in that, The Nash equilibrium of incremental human-machine lateral shared control based on game theory is expressed as: in, These are the optimal control strategies for the driver and the machine under their respective optimal strategies, and the strategy combination. This is a Nash equilibrium solution.

7. The safety-first, game-theoretic, incremental human-machine lateral shared driving control method according to claim 6, characterized in that, Real-time solution of Nash equilibrium, including: The binary agent coupled optimization control problem is formulated as a standard quadratic form, and a relaxation factor is introduced. ε h and ε m The standard quadratic optimization problem is formulated as follows: in, and These are the weighting coefficients of the relaxation factor term; Cons h and Cons m For constant terms; 。

Images