Personalized co-driving type intelligent vehicle man-machine sharing control method, system, medium and equipment

Abstract

The invention discloses a personalized co-driving type intelligent vehicle man-machine sharing control method and system, a medium and equipment, and relates to the field of intelligent auxiliary driving, and the method comprises the steps: obtaining a risk tolerance coefficient and a sensitivity zooming coefficient of a driver; acquiring a personalized driving risk field according to the risk tolerance coefficient and the sensitivity zoom coefficient of the driver; using the personalized driving risk field to calculate driving risks of different types of drivers; according to the driving risks of the different types of drivers, the driver control weights of the different types of drivers are obtained through fuzzy logic speculation; and establishing a man-machine non-cooperative game model, and realizing personalized man-machine sharing control based on the driver control weights of different types of drivers and the man-machine non-cooperative game model. When the man-machine conflict exists, the driving requirements of different types of drivers can be met while the vehicle driving safety is guaranteed, and personalized man-machine sharing control is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle control, and particularly relates to a personalized co-driving intelligent vehicle human-machine sharing control method, system, medium and device. Background Art

[0002] Autopilot can reduce human errors, reduce the incidence of traffic accidents, relieve traffic congestion, reduce commuting time and fuel consumption. However, fully autonomous driving is restricted by various factors such as vehicle safety, laws and policies, and it is difficult to be fully implemented in the short term. Therefore, it is necessary to research high-level driving assistance. Among them, human-machine co-driving, as a high-level driving assistance, the driver and the machine jointly control the vehicle during the driving process; however, there is a human-machine conflict problem during human-machine co-driving, which may reduce driving safety and the driving experience of the driver. Therefore, it is necessary to carry out personalized human-machine co-driving research. Summary of the Invention

[0003] To solve the technical problems in the background art, the present invention proposes a personalized co-driving intelligent vehicle human-machine sharing control method, system, medium and device.

[0004] A personalized co-driving intelligent vehicle human-machine sharing control method proposed by the present invention includes:

[0005] Obtain the risk tolerance coefficient and sensitivity scaling coefficient of the driver;

[0006] According to the risk tolerance coefficient and sensitivity scaling coefficient of the driver, use the personalized driving risk field to calculate the driving risks of different types of drivers;

[0007] According to the driving risks of different types of drivers, use fuzzy logic inference to obtain the driver control weights of different types of drivers;

[0008] Establish a human-machine non-cooperative game model, and realize personalized human-machine sharing control based on the driver control weights of different types of drivers and the human-machine non-cooperative game model.

[0009] Preferably, the expression of the driving risk of different types of drivers is

[0010]

[0011] where, E pr is the driving risk of different types of drivers; τ is the risk tolerance coefficient; α is the sensitivity scaling coefficient; f rel represents the relative speed influence factor; σ x and σ yrepresents the standard deviation of the Gaussian distribution; dx is the effective distance of the host vehicle from the obstacle vehicle in the vehicle driving direction; dy is the effective distance of the host vehicle from the obstacle vehicle in the direction perpendicular to the vehicle driving direction;

[0012] Among them, In the formula, λ is the influence coefficient of the relative speed on the risk; v ref is the reference speed; v rel_mag is the magnitude of the relative speed.

[0013] Preferably, according to the driving risks of different types of drivers, the driver control weights of different types of drivers are inferred using fuzzy logic, specifically including: creating a Mamdani-type fuzzy logic system; among them, the input variables of the Mamdani-type fuzzy logic system include the intensity of the personalized driving risk field and the driver preference, and the output variable of the Mamdani-type fuzzy logic system is the driver control weight; obtaining the driver preferences of different types of drivers; normalizing the driving risks of different types of drivers to obtain the intensity of the personalized driving risk field of different types of drivers; inputting the intensity of the personalized driving risk field and the driver preference of different types of drivers into the Mamdani-type fuzzy logic system to obtain the driver control weights of different types of drivers.

[0014] Preferably, a non-cooperative human-machine game model is established, and personalized human-machine shared control is realized based on the driver control weights of different types of drivers and the non-cooperative human-machine game model, specifically including:

[0015] Combining the dynamic model and kinematic model of the vehicle to obtain a simplified vehicle system model: Among them, the expression of the vehicle system model is

[0016]

[0017] In the formula, m is the vehicle mass; t is the time; XOY is the earth coordinate system; xoy is the coordinate system following the vehicle body; Vx is the longitudinal speed of the vehicle; v y is the lateral speed of the vehicle; C f is the cornering stiffness of the front tire; C r is the cornering stiffness of the rear tire; δ is the front wheel angle; ψ is the vehicle heading angle; r is the vehicle yaw rate; I z is the moment of inertia of the vehicle about the centroid axis; a, b are the distances between the vehicle centroid and the front and rear axles;

[0018] Based on the vehicle system model, a state space model is established; among them, the expression of the state space model is

[0019]

[0020] In the formula, u = βu h +(1 - β)u m ; u h is the human control input; u m is the machine control input; β is the driver control weight for different types of drivers;

[0021] The state - space model is discretized using the Euler method to obtain the discretized state - space model: Among them, the expression of the discretized state - space model is In the formula, x(k + 1) is the state of the system at the next moment, x(k) is the state of the system at the current moment, and u(k) is the input of the system at the current moment; B = A c dt, where dt is the prediction time step;

[0022] The discretized state - space model is continuously iterated to obtain the state prediction formula: Among them, the expression of the state prediction formula is X(k)=A p x(k)+B p U(k);

[0023] In the formula,

[0024] Construct the MPC prediction matrices Ψ and γ, and obtain the prediction output sequence according to the MPC prediction matrices Ψ and γ; Among them, the expression of the prediction output sequence is Z(k)=Ψx(k)+γ h U h (k)+γ m U m (k); In the formula, Z(k) represents the predicted value of the output sequence, γ h = γβ; γ m = γ(1 - β); γ and Ψ are the MPC prediction matrices;

[0025]

[0026] β is the driver control weight; u h (k) is the control input of humans at the current moment; u m (k) is the control input of the machine at the current moment; N p is the prediction step number; N u is the control step number;

[0027] Establish the human cost function J h and the machine cost function J m ; Among them,

[0028]

[0029] Among them, J h is the human cost function; J m is the machine cost function; Z(k) is the predicted value of the output sequence; Z ref,h (k) is the reference value of the human output sequence; Z(k) - Z ref,h (k) is the human error; Z ref,m (k) is the reference value of the machine output sequence; Z(k) - Z ref,m (k) is the machine error; Q h is the human error weight; Q m is the machine error weight; λ h is the human input cost weight; λ m is the machine input cost weight; u h (k) is the human control input at the current moment; u m (k) is the machine control input at the current moment; N p is the prediction step; N u is the control step;

[0030] Based on game theory, a non - cooperative game model between humans and machines is established. Both humans and machines expect to maximize their own benefits, that is, minimize their costs, namely minimizing the human cost function and minimizing the machine cost function; among them,

[0031]

[0032] Perform Nash game iteration; among them, at each MPC step, by giving the input of the other party, the optimal input of the own party is obtained to minimize the cost of the own party, and then by giving the new input of the own party, the input of the other party is updated to minimize the cost of the other party, and so on until the non - cooperative game model between humans and machines converges, obtaining the optimal control sequence of humans and the optimal control sequence of machines;

[0033] Take the first item of the optimal control sequence of humans as the optimal input of humans at the current moment, take the first item of the optimal control sequence of machines as the optimal input of machines, and obtain the optimal control sequence of humans and machines at the current moment after weighting by the driver control weight β of different types of drivers, so as to realize personalized human - machine shared control.

[0034] On the second aspect, the present invention also proposes a co - driving intelligent vehicle human - machine shared control system, including:

[0035] A driving risk acquisition module, configured to acquire a risk tolerance coefficient and a sensitivity scaling coefficient of a driver; according to the risk tolerance coefficient and the sensitivity scaling coefficient of the driver, acquire a personalized driving risk field, and use the personalized driving risk field to calculate the driving risks of different types of drivers;

[0036] A control weight acquisition module, configured to use fuzzy logic inference to obtain a driver control weight of different types of drivers according to the driving risks of different types of drivers;

[0037] A human-machine shared control module, configured to establish a human-machine non-cooperative game model, and implement personalized human-machine shared control based on the driver control weights of different types of drivers and the human-machine non-cooperative game model.

[0038] Preferably, the process of the human-machine shared control module implementing personalized human-machine shared control specifically includes:

[0039] Combining the dynamic model and the kinematic model of the vehicle to obtain a simplified vehicle system model; wherein, the vehicle system model is

[0040]

[0041] In the formula, m is the vehicle mass; t is the time; XOY is the earth coordinate system; xoy is the coordinate system following the vehicle body; Vx is the longitudinal speed of the vehicle; v y is the lateral speed of the vehicle; C f is the cornering stiffness of the front tire; C r is the cornering stiffness of the rear tire; δ is the front wheel angle; ψ is the vehicle heading angle; r is the vehicle yaw rate; I z is the moment of inertia of the vehicle about the centroid axis; a and b are the distances between the vehicle centroid and the front and rear axles;

[0042] Establishing a state space model based on the vehicle system model; wherein, the state space model is

[0043]

[0044] In the formula, u = βu h +(1 - β)u m ; u h is the human control input; u m is the machine control input; β is the driver control weight of different types of drivers;

[0045] Using the Euler method to discretize the state space model to obtain a discretized state space model: wherein, the expression of the discretized state space model is Where \(x(k + 1)\) is the state of the system at the next moment, \(x(k)\) is the state of the system at the current moment, and \(u(k)\) is the input of the system at the current moment; B = A c dt, where dt is the prediction time step;

[0046] Perform continuous iteration on the discretized state - space model to obtain the state prediction formula: Among them, the expression of the state prediction formula is \(X(k)=A\) p x(k)+B p U(k);

[0047] In the formula,

[0048] Construct the MPC prediction matrices \(\varPsi\) and \(\gamma\), and obtain the prediction output sequence according to the MPC prediction matrices \(\varPsi\) and \(\gamma\); Among them, the expression of the prediction output sequence is \(Z(k)=\varPsi x(k)+\gamma\) h U h (k)+\gamma m U m (k); In the formula, \(Z(k)\) represents the predicted value of the output sequence, \(\gamma\) h =\(\gamma\beta\); \(\gamma\) m =\(\gamma(1 - \beta)\); \(\gamma\) and \(\varPsi\) are MPC prediction matrices;

[0049] \(\beta\) is the driver control weight; u h (k) is the control input of humans at the current moment; u m (k) is the control input of the machine at the current moment; N p is the prediction step; N u is the control step;

[0050] Establish the human cost function \(J\) h and the machine cost function \(J\) m ; Among them,

[0051]

[0052] Among them, \(J\) h is the human cost function; \(J\) m is the machine cost function; \(Z(k)\) is the predicted value of the output sequence; \(Z\) ref,h (k) is the reference value of the human output sequence; \(Z(k)-Z\) ref,h (k) is the human error; \(Z\) ref,m (k) is the reference value of the machine output sequence; \(Z(k)-Z\) ref,m (k) is the machine error; Qh is the human error weight; Q m is the machine error weight; λ h is the human input cost weight; λ m is the machine input cost weight; u h (k) is the human control input at the current moment; u m (k) is the machine control input at the current moment; N p is the prediction step number; N u is the control step number;

[0053] Based on game theory, a non - cooperative human - machine game model is established. Both the human and the machine expect to maximize their own benefits, that is, minimize their costs, namely, minimize the human cost function and minimize the machine cost function; among them,

[0054]

[0055] Perform Nash game iteration on the non - cooperative human - machine game model; at each MPC step, by giving the input of the other party, find the optimal input of the own party to minimize the own cost, and then give the new input of the own party to update the input of the other party to minimize the cost of the other party, and so on until the non - cooperative human - machine game model converges, obtaining the optimal control sequence of the human and the optimal control sequence of the machine;

[0056] Take the first item of the optimal control sequence of the human as the optimal input of the human at the current moment, take the first item of the optimal control sequence of the machine as the optimal input of the machine, and obtain the optimal human - machine control sequence at the current moment after weighting by the driver control weight β of different types of drivers, so as to realize personalized human - machine shared control.

[0057] In the third aspect, the present invention also proposes a computer - readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the personalized co - driving intelligent vehicle human - machine shared control method described in any item of the first aspect.

[0058] In the fourth aspect, the present invention also proposes an electronic device, including: a processor and a memory. The memory is used to store one or more programs; when one or more programs are executed by the processor, it implements the personalized co - driving intelligent vehicle human - machine shared control method described in any item of the first aspect.

[0059] In the present invention, a personalized co-driving intelligent vehicle human-machine sharing control method, system, medium and device are proposed. By using a personalized driving risk field, the driving risks of different types of drivers are calculated; according to the driving risks of different types of drivers, the driver control weights of different types of drivers are inferred using fuzzy logic; a human-machine non-cooperative game model is established; and personalized human-machine sharing control is realized based on the driver control weights of different types of drivers and the human-machine non-cooperative game model. The present invention can, when there is a human-machine conflict, ensure the driving safety of the vehicle while meeting the driving needs of different types of drivers, and realize personalized human-machine sharing control. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic flowchart of a personalized co-driving intelligent vehicle human-machine sharing control method in an embodiment proposed by the present invention.

[0061] Figure 2 It is a schematic diagram of membership functions of personalization, driver preference and driver control weight in a Mamdani fuzzy logic system in an embodiment proposed by the present invention.

[0062] Figure 3 It is a three-dimensional schematic diagram of the driver control weight in a Mamdani fuzzy logic system in an embodiment proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0063] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the drawings and in combination with the embodiments.

[0064] Referring to Figure 1 , a personalized co-driving intelligent vehicle human-machine sharing control method proposed by the present invention includes: obtaining the risk tolerance coefficient τ and the sensitivity scaling coefficient α of the driver; obtaining a personalized driving risk field according to the risk tolerance coefficient τ and the sensitivity scaling coefficient α of the driver; calculating the driving risks of different types of drivers by using the personalized driving risk field; inferring the driver control weights of different types of drivers by using fuzzy logic according to the driving risks of different types of drivers; establishing a human-machine non-cooperative game model; and realizing personalized human-machine sharing control based on the driver control weights of different types of drivers and the human-machine non-cooperative game model.

[0065] The present invention calculates the driving risks of different types of drivers by using a personalized driving risk field; based on the driving risks of different types of drivers, the driving control weights of different types of drivers are inferred by using fuzzy logic; a non-cooperative human-machine game model is established; and personalized human-machine shared control is realized based on the driving control weights of different types of drivers and the non-cooperative human-machine game model. When there is a human-machine conflict, the present invention can ensure the driving safety of the vehicle while meeting the driving needs of different types of drivers, and realize personalized human-machine shared control.

[0066] In this embodiment, the expression of the driving risk of different types of drivers is

[0067]

[0068] where, E pr is the driving risk of different types of drivers; τ is the risk tolerance coefficient; when τ < 1, it means that the risk value is higher at the same distance; when τ > 1, the risk value is lower at the same distance; α is the sensitivity scaling coefficient; when α > 1, the overall risk value becomes larger; when α < 1, the overall risk value decreases; f rel represents the relative speed influence factor; σ x and σ y represent the standard deviations of the Gaussian distribution; dx is the effective distance of the host vehicle from the obstacle vehicle in the vehicle driving direction; dy is the effective distance of the host vehicle from the obstacle vehicle perpendicular to the vehicle driving direction.

[0069] where, in the formula, λ is the influence coefficient of the relative speed on the risk; v ref is the reference speed; v rel_mag is the magnitude of the relative speed.

[0070] In this embodiment, based on the driving risks of different types of drivers, the driving control weights of different types of drivers are inferred by using fuzzy logic, which specifically includes:

[0071] Create a Mamdani-type fuzzy logic system; where, select the intensity En of the personalized driving risk field as the first input variable of the Mamdani-type fuzzy logic system, and its value range is [0, 1], and the fuzzy subsets are sequentially divided into {L, M, B} from small to large; where, L represents low, M represents medium, and B represents high; select the driver preference Pr as another input variable of the Mamdani-type fuzzy logic system, and the value range of the variable is [0.5, 2], and the fuzzy subsets are sequentially divided into {C, B} from small to large; select the driver control weight Au as the output variable of the Mamdani-type fuzzy logic system, and its value range is [0, 0.5], and the fuzzy subsets are sequentially divided into {L, M, B} from small to large.

[0072] Obtain the driver preferences of different types of drivers; normalize the driving risks E of different types of drivers to obtain the personalized driving risk field intensity E of different types of drivers pr ; among them, n ; where, In the formula, E n The personalized driving risk field intensity of different types of drivers, with a value range of [0,1]; max(E pr ) represents the maximum value in E pr for normalization; E pr is the driving risk of different types of drivers; input the personalized driving risk field intensity En and driver preference Pr of different types of drivers into the Mamdani fuzzy logic system, and finally use the centroid method for defuzzification to output the specific driving weight coefficient, so as to obtain the control weights of different types of drivers.

[0073] The Mamdani fuzzy system in this embodiment is irreplaceable in scenarios of human-machine collaboration and requiring interpretability, while other types are more suitable for environments with high real-time performance, high noise or contradictory information.

[0074] It should be understood that the control weight of the driver is affected by the driving risk and driver preference, where the driving risk can be measured by the personalized driving risk field intensity E pr . Driver preferences can be divided into three categories: conservative, normal, and aggressive. Conservative drivers do not trust their own driving skills and have a low driver control weight; normal drivers have a moderate attitude towards machine assistance and have a medium driver control weight; aggressive drivers pursue their own control of the vehicle and hope to have a higher driver control weight, and their driver control weight is relatively high.

[0075] Among them, the Mamdani fuzzy logic rules of the Mamdani fuzzy logic system are shown in Table 1.

[0076] Table 1

[0077]

[0078] Among them, the membership functions of the personalized driving risk field intensity (Risk), driver preference (Preference), and driver control weight (Authority) in fuzzy control are as Figure 2 shown. The three-dimensional diagram of the driver control weight β of different types of drivers in fuzzy control is as Figure 3 shown.

[0079] In this embodiment, a human-machine non-cooperative game model is established; personalized human-machine shared control is realized based on the driver control weights of different types of drivers and the human-machine non-cooperative game model, specifically including:

[0080] Combining the dynamic model and kinematic model of the vehicle, a simplified vehicle system model is obtained; among them, the expression of the vehicle system model is

[0081]

[0082] In the formula, m is the vehicle mass; t is the time; XOY is the earth coordinate system; xoy is the coordinate system following the vehicle body; Vx is the longitudinal speed of the vehicle; v y is the lateral speed of the vehicle; C f is the cornering stiffness of the front tire; C r is the cornering stiffness of the rear tire; δ is the front wheel angle; ψ is the vehicle heading angle; r is the vehicle yaw rate; I z is the moment of inertia of the vehicle about the centroid axis; a and b are the distances between the vehicle centroid and the front and rear axles;

[0083] Based on the vehicle system model, a state space model is established. Taking y, ψ, v y , r as state variables, and taking y and ψ as outputs, we can get Among them, u = βu h +(1 - β)u m ; u h is the human control input; u m is the machine control input; β is the driver control weight of different types of drivers;

[0084] Using the Euler method to discretize the state space model, the discretized state space model is obtained: among them, the expression of the discretized state space model is In the formula, x(k + 1) is the state of the system at the next moment, x(k) is the state of the system at the current moment, and u(k) is the input of the system at the current moment; B = A c dt, dt is the prediction time step;

[0085] Performing continuous iteration on the discretized state space model, a state prediction formula is obtained: among them, the expression of the state prediction formula is X(k) = A p x(k)+B p U(k);

[0086] In the formula,

[0087] Construct the MPC prediction matrices Ψ and γ, and obtain the predicted output sequence according to the MPC prediction matrices Ψ and γ; where the expression of the predicted output sequence is Z(k) = Ψx(k) + γ h U h (k) + γ m U m (k); in the formula, Z(k) represents the predicted value of the output sequence, γ h = γβ; γ m = γ(1 - β); γ and Ψ are MPC prediction matrices;

[0088] β is the driver control weight; u h (k) is the control input of humans at the current moment; u m (k) is the control input of the machine at the current moment; N p is the prediction step; N u is the control step;

[0089] Establish the human cost function J h and the machine cost function J m ; where,

[0090]

[0091] where, J h is the human cost function; J m is the machine cost function; Z(k) is the predicted value of the output sequence; Z ref,h (k) is the reference value of the human output sequence; Z(k) - Z ref,h (k) is the human error; Z ref,m (k) is the reference value of the machine output sequence; Z(k) - Z ref,m (k) is the machine error; Q h is the human error weight; Q m is the machine error weight; λ h is the human input cost weight; λ m is the machine input cost weight; u h (k) is the control input of humans at the current moment; u m (k) is the control input of the machine at the current moment; N p is the prediction step; N u is the control step;

[0092] Based on game theory, a human-machine non-cooperative game model is established. At this time, the strategies of both the human and the machine are individual optimal strategies, that is, minimizing human cost and minimizing machine cost; among them,

[0093]

[0094] Perform Nash game iteration on the human-machine non-cooperative game model; at each MPC step, by giving the input of the other party, find the optimal input of the local party to minimize the local cost, and then give the new input of the local party to update the input of the other party to minimize the cost of the other party, and so on until the human-machine non-cooperative game model converges, obtaining the optimal control sequence of the human and the optimal control sequence of the machine;

[0095] Take the first item of the optimal control sequence of the human as the optimal input of the human at the current moment, take the first item of the optimal control sequence of the machine as the optimal input of the machine, and obtain the optimal human-machine control sequence at the current moment after weighting by the driver control weight β of different types of drivers, so as to realize personalized human-machine shared control.

[0096] During the driving process, both the driver and the machine hope to maximize their own revenue functions, and human-machine conflicts may occur in this process. The revenue distribution between the driver and the machine during the driving process can be described by a non-cooperative game model in game theory. When the non-cooperative game between the driver and the machine is in the Nash equilibrium state, the participants cannot improve their revenue by changing their own strategies. At this time, their strategies are individual optimal strategies, that is, minimizing human cost and minimizing machine cost, so as to alleviate human-machine conflicts.

[0097] In the second aspect, the present invention also proposes a co-driving intelligent vehicle human-machine shared control system, including:

[0098] A driving risk acquisition module, configured to acquire the risk tolerance coefficient τ and the sensitivity scaling coefficient α of the driver; according to the risk tolerance coefficient τ and the sensitivity scaling coefficient α of the driver, acquire a personalized driving risk field, and use the personalized driving risk field to calculate the driving risks of different types of drivers;

[0099] A control weight acquisition module, configured to infer the driver control weights of different types of drivers by using fuzzy logic according to the driving risks of different types of drivers;

[0100] A human-machine shared control module, configured to establish a human-machine non-cooperative game model, and realize personalized human-machine shared control based on the driver control weights of different types of drivers and the human-machine non-cooperative game model.

[0101] In this embodiment, the expression of the driving risks of different types of drivers is

[0102]

[0103] Among them, E pr is the driving risk of different types of drivers; τ is the risk tolerance coefficient; when τ < 1, it means that the risk value is higher at the same distance; when τ > 1, the risk value is lower at the same distance; α is the sensitivity scaling coefficient; when α > 1, the overall risk value becomes larger; when α < 1, the overall risk value decreases; f rel represents the relative speed impact factor; σ x and σ y represent the standard deviation of the Gaussian distribution; dx is the effective distance of the host vehicle from the obstacle vehicle in the vehicle driving direction; dy is the effective distance of the host vehicle from the obstacle vehicle in the direction perpendicular to the vehicle driving direction;

[0104] Among them, In the formula, λ is the influence coefficient of relative speed on risk; v ref is the reference speed; v rel_mag is the magnitude of the relative speed.

[0105] In this embodiment, the process of obtaining the driver control weight of different types of drivers specifically includes: creating a Mamdani-type fuzzy logic system; selecting the personalized driving risk field intensity En as the first input variable of the Mamdani-type fuzzy logic system, whose value range is [0,1], and the fuzzy subsets are sequentially divided into {L, M, B} from small to large; selecting the driver preference Pr as the other input variable of the Mamdani-type fuzzy logic system, the value range of the variable is [0.5, 2], and the fuzzy subsets are sequentially divided into {C, B} from small to large; selecting the driver control weight Au as the output variable of the Mamdani-type fuzzy logic system, whose value range is [0, 0.5], and the fuzzy subsets are sequentially divided into {L, M, B} from small to large;

[0106] Obtain the driver preferences of different types of drivers; normalize the driving risk E pr of different types of drivers to obtain the personalized driving risk field intensity E n of different types of drivers; among them, In the formula, E n represents the personalized driving risk field strength of different types of drivers, and the value range is [0,1]; max(E pr ) represents the maximum value in E pr for normalization; E pr is the driving risk of different types of drivers; input the personalized driving risk field intensity En and driver preference Pr of different types of drivers into the Mamdani-type fuzzy logic system to obtain the driver control weight of different types of drivers.

[0107] Among them, the process of the human-machine shared control module realizing personalized human-machine shared control specifically includes:

[0108] Combining the dynamic model and kinematic model of the vehicle to obtain a simplified vehicle system model; among them, the expression of the simplified vehicle system model is

[0109]

[0110] In the formula, m is the vehicle mass; t is the time; XOY is the earth coordinate system; xoy is the coordinate system following the vehicle body; Vx is the longitudinal vehicle speed; v y is the lateral vehicle speed; C f is the cornering stiffness of the front tire; C r is the cornering stiffness of the rear tire; δ is the front wheel steering angle; ψ is the vehicle heading angle; r is the vehicle yaw rate; I z is the moment of inertia of the vehicle about the centroid axis; a and b are the distances between the vehicle centroid and the front and rear axles;

[0111] Based on the vehicle system model, a state space model is established. Taking y, ψ, v y , r as state variables, and taking y and ψ as outputs, we can obtain

[0112] Among them, u = βu h +(1 - β)u m ; u h is the human control input; u m is the machine control input; β is the driver control weight of different types of drivers;

[0113] Using the Euler method to discretize the state space model to obtain the discretized state space model: among them, the expression of the discretized state space model is In the formula, x(k + 1) is the state of the system at the next moment, x(k) is the state of the system at the current moment, and u(k) is the input of the system at the current moment; B = A c dt, where dt is the prediction time step;

[0114] Performing continuous iteration on the discretized state space model to obtain a state prediction formula: among them, the expression of the state prediction formula is X(k) = A p x(k)+B p U(k);

[0115] In the formula,

[0116] Construct the MPC prediction matrices Ψ and γ, and obtain the predicted output sequence based on the MPC prediction matrices Ψ and γ; where the expression for the predicted output sequence is Z(k) = Ψx(k) + γ h U h (k) + γ m U m (k); in the formula, Z(k) represents the predicted value of the output sequence, γ h = γβ; γ m = γ(1 - β); γ and Ψ are the MPC prediction matrices;

[0117] β is the driver control weight; u h (k) is the control input of humans at the current moment; u m (k) is the control input of the machine at the current moment; N p is the prediction step; N u is the control step;

[0118] Establish the human cost function J h and the machine cost function J m ; where,

[0119]

[0120] where, J h is the human cost function; J m is the machine cost function; Z(k) is the predicted value of the output sequence; Z ref,h (k) is the reference value of the human output sequence; Z(k) - Z ref,h (k) is the human error; Z ref,m (k) is the reference value of the machine output sequence; Z(k) - Z ref,m (k) is the machine error; Q h is the human error weight; Q m is the machine error weight; λ h is the human input cost weight; λ m is the machine input cost weight; u h (k) is the control input of humans at the current moment; u m (k) is the control input of the machine at the current moment; N p is the prediction step; N u is the control step;

[0121] Based on game theory, establish a non - cooperative game model between humans and machines. Both humans and machines expect to maximize their own benefits, that is, minimize the cost, that is

[0122] Perform Nash game iteration on the human-machine non-cooperative game model; at each MPC step, obtain the optimal input of the local party by given the input of the other party to minimize the cost of the local party, and then update the input of the other party by given the new input of the local party to minimize the cost of the other party, and so on until the human-machine non-cooperative game model converges, obtaining the optimal control sequence of the human and the optimal control sequence of the machine; take the first item of the optimal control sequence of the human as the optimal input of the human at the current moment, take the first item of the optimal control sequence of the machine as the optimal input of the machine, and obtain the optimal human-machine control sequence at the current moment after weighting by the driver control weight β of different types of drivers, so as to realize personalized human-machine shared control.

[0123] In this embodiment, a human-machine non-cooperative game model is established, with minimizing the human cost function and minimizing the machine cost function as the objective functions. Through such a human-machine non-cooperative game model for human-machine shared control, it is possible to meet the driving needs of different types of drivers while ensuring the driving safety of the vehicle, and realize personalized human-machine shared control.

[0124] In a third aspect, the present invention also proposes a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the personalized co-driving intelligent vehicle human-machine shared control method described in any item of the first aspect.

[0125] In a fourth aspect, the present invention also proposes an electronic device, including: a processor and a memory, where the memory is used to store one or more programs; when the one or more programs are executed by the processor, the personalized co-driving intelligent vehicle human-machine shared control method described in any item of the first aspect is implemented.

[0126] As mentioned above, the above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and all should be covered within the protection scope of the present invention.

Claims

1. A personalized co-driving intelligent vehicle human-machine sharing control method, characterized in that, Including: Obtain the risk tolerance coefficient and sensitivity scaling coefficient of the driver; Build a personalized driving risk field according to the risk tolerance coefficient and sensitivity scaling coefficient of the driver; Use the personalized driving risk field to calculate the driving risks of different types of drivers; According to the driving risks of different types of drivers, use fuzzy logic inference to obtain the driver control weights of different types of drivers; Establish a human-machine non-cooperative game model, and realize personalized human-machine shared control based on the driver control weights of different types of drivers and the human-machine non-cooperative game model; 2. The personalized co-driving intelligent vehicle human-machine sharing control method according to claim 1, wherein The expression of the driving risk of different types of drivers is Among them, E pr is the driving risk of different types of drivers; τ is the risk tolerance coefficient; α is the sensitivity scaling coefficient; f rel represents the relative speed influence factor; σ x and σ y represent the standard deviation of the Gaussian distribution; dx is the effective distance of the host vehicle from the obstacle vehicle in the vehicle driving direction; dy is the effective distance of the host vehicle from the obstacle vehicle perpendicular to the vehicle driving direction; Among them, In the formula, λ is the influence coefficient of relative speed on risk; v ref is the reference speed; v rel_mag is the magnitude of the relative speed.

3. The personalized co-driving intelligent vehicle human-machine sharing control method according to claim 1, characterized in that, According to the driving risks of different types of drivers, use fuzzy logic inference to obtain the driver control weights of different types of drivers, specifically including: Create a Mamdani fuzzy logic system; among them, the input variables of the Mamdani fuzzy logic system include the intensity of the personalized driving risk field and the driver preference, and the output variable of the Mamdani fuzzy logic system is the driver control weight; Classify the driver risk preferences of drivers into two categories: conservative and aggressive according to the different driving behaviors of different drivers when facing the same risk; Normalize the driving risks of different types of drivers to obtain the intensity of the personalized driving risk field of different types of drivers; Input the intensity of the personalized driving risk field and the driver preference of different types of drivers into the Mamdani fuzzy logic system, and finally use the centroid method for defuzzification to output the specific driving weight coefficient to obtain the driver control weights of different types of drivers; 4. The personalized co-driving intelligent vehicle human-machine sharing control method according to claim 1, wherein Establish a human-machine non-cooperative game model, and realize personalized human-machine shared control based on the driver control weights of different types of drivers and the human-machine non-cooperative game model, specifically including: Combine the dynamic model and kinematic model of the vehicle to obtain a simplified vehicle system model: among them, the expression of the vehicle system model is where m is the vehicle mass; t is the time; XOY is the earth coordinate system; xoy is the coordinate system following the vehicle body; Vx is the longitudinal vehicle speed; v y is the lateral vehicle speed; C f is the cornering stiffness of the front wheel tires; C r is the cornering stiffness of the rear wheel tires; δ is the front wheel angle; ψ is the vehicle heading angle; r is the vehicle yaw rate; I z is the moment of inertia of the vehicle about the centroid axis; a and b are the distances between the vehicle centroid and the front and rear axles; Establish a state space model based on the vehicle system model; among them, the expression of the state space model is wherein, u = βu h +(1 - β)u m ; u h is the human control input; u m is the machine control input; β is the driver control weight of different types of drivers; Use the Euler method to discretize the state space model to obtain the discretized state space model: among them, the expression of the discretized state space model is where \(x(k + 1)\) is the state of the system at the next moment, \(x(k)\) is the state of the system at the current moment, and \(u(k)\) is the input of the system at the current moment; B = A c dt, where dt is the prediction time step; Continuously iterate the discretized state-space model to obtain the state prediction formula: where the expression of the state prediction formula is X(k) = A p x(k) + B p U(k); In the formula, Construct the MPC prediction matrices Ψ and γ, and obtain the predicted output sequence according to the MPC prediction matrices Ψ and γ; where the expression of the predicted output sequence is Z(k) = Ψx(k) + γ h U h (k) + γ m U m (k); in the formula, Z(k) represents the predicted value of the output sequence, γ h = γ β ; γ m = γ(1 - β); γ and Ψ are MPC prediction matrices; β is the driver control weight; u h (k) is the control input of humans at the current moment; u m (k) is the control input of the machine at the current moment; N p is the prediction step number; N u is the control step number; Establish the human cost function J h and the machine cost function J m ; where Among them, J h is the human cost function; J m is the machine cost function; Z(k) is the predicted value of the output sequence; Z ref,h (k) is the reference value of the human output sequence; Z(k) - Z ref,h (k) is the human error; Z ref,m (k) is the reference value of the machine output sequence; Z(k) - Z ref,m (k) is the machine error; Q h is the human error weight; Q m is the machine error weight; λ h is the human input cost weight; λ m is the machine input cost weight; u h (k) is the human control input at the current moment; u m (k) is the machine control input at the current moment; N p is the prediction step number; N u is the control step number; Based on game theory, establish a human-machine non-cooperative game model. Both parties expect to maximize their own benefits, that is, minimize the cost. That is, minimize the human cost function and minimize the machine cost function; among them, Perform Nash game iteration; among them, at each MPC step, find the optimal input of this party by giving the input of the other party to minimize the cost of this party, and then give the new input of this party to update the input of the other party to minimize the cost of the other party, and repeat this process until convergence to obtain the optimal control sequence of the human and the optimal control sequence of the machine; Take the first item of the optimal control sequence of the human as the optimal input of the human at the current moment, take the first item of the optimal control sequence of the machine as the optimal input of the machine, and obtain the optimal human-machine control sequence at the current moment through the weighted driver control weight β of different types of drivers, so as to realize personalized human-machine shared control; 5. A co-driving type intelligent vehicle human-machine sharing control system, characterized in that, Including: A driving risk acquisition module for obtaining the risk tolerance coefficient and sensitivity scaling coefficient of the driver; According to the risk tolerance coefficient and sensitivity scaling coefficient of the driver, a personalized driving risk field is obtained, and the driving risk of different types of drivers is calculated using the personalized driving risk field; A control weight acquisition module is used to obtain the driver control weights of different types of drivers by using fuzzy logic inference according to the driving risks of different types of drivers; The human-machine shared control module is used to establish a human-machine non-cooperative game model and realize personalized human-machine shared control based on the driver control weights of different types of drivers and the human-machine non-cooperative game model.

6. The personalized co-driving intelligent vehicle human-machine sharing control system according to claim 5, wherein The realization of personalized human-machine shared control of the human-machine shared control module specifically includes: Combining the vehicle's dynamic model and kinematic model, a simplified vehicle system model is obtained; the vehicle system model is Where m is the vehicle mass; t is the time; XOY is the earth coordinate system; xoy is the coordinate system following the vehicle body; Vx is the longitudinal vehicle speed; v y is the lateral vehicle speed; C f is the cornering stiffness of the front wheel tire; C r is the cornering stiffness of the rear wheel tire; δ is the front wheel steering angle; ψ is the vehicle heading angle; r is the vehicle yaw rate; I z is the moment of inertia of the vehicle about the centroid axis; a and b are the distances between the vehicle centroid and the front and rear axles; The state space model is established based on the vehicle system model; the state space model is wherein, u = βu h +(1 - β)u m ; u h is the human control input; u m is the machine control input; β is the driver control weight for different types of drivers; The state space model is discretized using the Euler method to obtain the discretized state space model: The expression of the discretized state space model is: Where x(k + 1) is the state of the system at the next moment, x(k) is the state of the system at the current moment, and u(k) is the input of the system at the current moment; B = A c dt, where dt is the prediction time step; Continuously iterate the discretized state-space model to obtain the state prediction formula: Among them, the expression of the state prediction formula is X(k) = A p x(k) + B p U(k); In the formula, Construct the MPC prediction matrices Ψ and γ, and obtain the predicted output sequence based on the MPC prediction matrices Ψ and γ; where the expression for the predicted output sequence is Z(k) = Ψx(k) + γ h U h (k) + γ m U m (k); where Z(k) represents the predicted value of the output sequence, γ h = γ β ; γ m = γ(1 - β); γ and Ψ are the MPC prediction matrices; β is the driver control weight; u h (k) is the control input of humans at the current moment; u m (k) is the control input of the machine at the current moment; N p is the prediction step number; N u is the control step number; Establish the human cost function J h and the machine cost function J m ; where Among them, J h is the human cost function; J m is the machine cost function; Z(k) is the predicted value of the output sequence; Z ref,h (k) is the reference value of the human output sequence; Z(k) - Z ref,h (k) is the human error; Z ref,m (k) is the reference value of the machine output sequence; Z(k) - Z ref,m (k) is the machine error; Q h is the human error weight; Q m is the machine error weight; λ h is the human input cost weight; λ m is the machine input cost weight; u h (k) is the control input of the human at the current moment; u m (k) is the control input of the machine at the current moment; N p is the prediction step number; N u is the control step number; Based on game theory, a human-machine non-cooperative game model is established. Both the human and the machine expect their own benefits to be the highest, that is, the cost to be the lowest, that is, to minimize the human cost function and minimize the machine cost function; among them, Perform Nash game iteration on the human-computer non-cooperative game model; In each MPC step, the optimal input of the party is obtained by giving the other party's input to minimize the cost of the party, and then the new input of the party is given to update the other party's input to minimize the cost of the other party. This is repeated until the human-machine non-cooperative game model converges, and the optimal control sequence of humans and the optimal control sequence of the machine are obtained; The first item of the optimal control sequence of humans is taken as the optimal input of humans at the current moment, and the first item of the optimal control sequence of machines is taken as the optimal input of machines. The optimal control sequence of humans and machines at the current moment is obtained by weighting the driver control weights β of different types of drivers, thereby realizing personalized human-machine shared control.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the personalized human-machine sharing control method for a co-driving smart car as described in any one of claims 1 to 4.

8. An electronic device, characterized in that, include: A processor and a memory, the memory being used to store one or more programs; When one or more programs are executed by the processor, the personalized human-machine sharing control method for a co-driving smart car as described in any one of claims 1 to 4 is implemented.

Images