Driving control device and driving control method

Abstract

The driving control device 10 includes a generation section 122 that generates a vehicle model indicating a relationship between a weight, a center-of-gravity position, a speed, a steering angle, a lateral deviation, a yaw angle deviation, and a curvature of a vehicle at a predetermined control cycle, and a calculation section 123 that calculates a steering angle that minimizes or maximizes an output value of an evaluation function including an estimated lateral deviation, an estimated yaw angle deviation, a steering angle, and a change amount of the steering angle with respect to an immediately preceding control cycle calculated on the basis of the vehicle model as an optimal steering angle. The calculation section 123 makes a coefficient of a term corresponding to the change amount smaller after an initial period elapses from a time when the vehicle is started than a weighting coefficient of the term corresponding to the change amount until the predetermined initial period elapses.

Description

Technical Field

[0001] This invention relates to a driving control device and a driving control method. Background Technology

[0002] Traditionally, driving control systems that enable vehicles to travel along a target trajectory are known. Japanese Patent No. 4297123 discloses a travel control system that determines the curvature of the target trajectory based on the vehicle's speed. Summary of the Invention

[0003] The problem to be solved by the present invention

[0004] Traditional driving control systems calculate the steering angle for a vehicle to travel along a target trajectory using a vehicle model in which the vehicle's weight and center of gravity are pre-input as fixed values. However, when the number of passengers or the vehicle's load changes, the vehicle's weight and center of gravity differ from those input as fixed values ​​in the vehicle model. As a result, the driving control system cannot calculate the steering angle for traveling along the target trajectory with high accuracy, leading to the problem that the vehicle may not be able to follow the target trajectory.

[0005] To enable a vehicle to travel along a target trajectory, it is conceivable to drive the vehicle at a steering angle that minimizes or maximizes the output value of an evaluation function, which includes estimated lateral deviation, estimated azimuth deviation, steering angle, and the change in steering angle calculated based on the vehicle model. However, when attempting to travel along the target trajectory, the steering angle changes significantly immediately after the vehicle begins to move, resulting in a deterioration in driver perception (i.e., driving comfort).

[0006] The present invention was made in view of the above circumstances, and one object of the present invention is to enable the vehicle to travel along a target trajectory and improve the driver's feeling when starting the vehicle.

[0007] Problem-solving methods

[0008] A driving control device according to a first aspect of the present invention includes: an acquisition unit for acquiring the weight of a vehicle, the center of gravity position of the vehicle, the speed of the vehicle, the steering angle of the vehicle, the lateral deviation of the vehicle, the azimuth deviation of the vehicle, and the curvature of the road surface on which the vehicle travels; a generation unit for generating a vehicle model at a predetermined control cycle, the vehicle model indicating the relationship between the weight, center of gravity position, speed, steering angle, lateral deviation, azimuth deviation, and curvature; and a calculation unit for calculating a steering angle that minimizes or maximizes the output value of an evaluation function as an optimal steering angle, the evaluation function including an estimated lateral deviation, an estimated azimuth deviation, a steering angle, and the amount of change of the steering angle relative to the immediately preceding control cycle calculated based on the vehicle model. The calculation unit makes the weighting coefficient of the term corresponding to the amount of change after an initial cycle from the time the vehicle starts smaller than the weighting coefficient of the term corresponding to the amount of change from the time the vehicle starts until the predetermined initial cycle.

[0009] The calculation unit can make the weighting factor of the term corresponding to the steering angle after the initial period from the time the vehicle starts less than the weighting factor of the term corresponding to the steering angle from the time the vehicle starts until the predetermined initial period has elapsed.

[0010] The calculation unit can change the weighting coefficients of the terms corresponding to the amount of change during a predetermined transition period starting from the initial period.

[0011] The calculation unit can determine the transition period based on the difference between the weighting coefficients before and after the change.

[0012] The calculation unit can make the weighting factor of the term corresponding to the estimated lateral deviation after the initial cycle from the time the vehicle starts greater than the weighting factor of the term corresponding to the estimated lateral deviation up to the predetermined initial cycle.

[0013] The calculation unit can make the weighting factor of the term corresponding to the estimated azimuth deviation after the initial cycle from the time the vehicle starts greater than the weighting factor of the term corresponding to the estimated azimuth deviation up to the time until the predetermined initial cycle has elapsed.

[0014] The calculation unit can determine at least one weighting coefficient of one or more terms of the evaluation function by referring to the storage unit, which stores the vehicle type associated with the weighting coefficient of one or more terms of the evaluation function.

[0015] When selecting a mode that prioritizes good feel over the ability to immediately follow the target trajectory after vehicle start-up, the calculation unit can assign a larger weighting coefficient to the term related to the steering angle input to the evaluation function up to the initial cycle, compared to selecting a mode that prioritizes the ability to immediately follow the target trajectory after vehicle start-up.

[0016] The computation unit can perform optimization calculations to minimize the output value of the evaluation function shown in the following equation.

[0017]

[0018] Satisfy δ min ≤δ[k+k t ]≤δ max

[0019] Δδ min ≤Δδ[k+k t ]≤Δδ max

[0020] Where p represents the prediction level, e z e represents the estimated lateral deviation. θ This indicates the estimated azimuth angle deviation, where δ represents the steering angle input, Δδ represents the difference between the steering angle input and the immediately preceding control cycle, and “max” and “min”, which are subscripts of each input and output variable, are the upper and lower limits of the signal, and Q1, Q2, R1, and R2 are weighting coefficients.

[0021] When the mode that prioritizes good feeling over the ability to immediately follow the target trajectory after vehicle startup is selected, the calculation unit can make the weighting coefficients R1 and R2 in the evaluation function larger up to the time of the initial cycle compared to the mode that prioritizes the ability to immediately follow the target trajectory after vehicle startup.

[0022] When prioritizing the convergence of lateral deviation and azimuth deviation to 0, compared to prioritizing the convergence of lateral deviation and azimuth deviation to 0, the calculation unit may set at least one of the weighting coefficients Q1 and Q2 to be larger, or set at least one of the weighting coefficients R1 and R2 to be smaller.

[0023] When prioritizing the reduction of the change in tire steering angle, the calculation unit can set the weighting coefficient R2 to be larger than when not prioritizing the reduction of the change in tire steering angle.

[0024] The driving control method according to a second aspect of the present invention is a computer-executed driving control method, comprising: acquiring the weight of a vehicle, the center of gravity position of the vehicle, the speed of the vehicle, the steering angle of the vehicle, the lateral deviation of the vehicle, the azimuth deviation of the vehicle, and the curvature of the road surface on which the vehicle travels; generating a vehicle model at a predetermined control cycle, the vehicle model indicating the relationship between weight, center of gravity position, speed, steering angle, lateral deviation, azimuth deviation, and curvature; and calculating a steering angle that minimizes or maximizes the output value of an evaluation function as an optimal steering angle, the evaluation function including an estimated lateral deviation, an estimated azimuth deviation, a steering angle, and a change in the steering angle relative to the immediately preceding control cycle calculated based on the vehicle model. This calculation includes making the weighting factor of the term corresponding to the change after an initial cycle from vehicle startup less than the weighting factor of the term corresponding to the change from vehicle startup until the predetermined initial cycle.

[0025] The effects of the invention

[0026] According to the present invention, a vehicle can be made to travel along a target trajectory and the driver's feeling when starting the vehicle can be improved. Attached Figure Description

[0027] Figure 1 This is a diagram used to illustrate the key points of the driving control system S.

[0028] Figure 2 This is a diagram showing the configuration of the driving control system S.

[0029] Figure 3 This is a diagram used to illustrate the vehicle model generated by the generation unit 122.

[0030] Figure 4 This is a flowchart illustrating an example of the operation of the driving control device 10.

[0031] Figure 5 This is a flowchart of the process for determining the steering angle.

[0032] Figure 6 It is a simulation block diagram. Detailed Implementation

[0033] [Overview of Driving Control Systems]

[0034] Figure 1 This is a diagram illustrating the key points of the driving control system S. The driving control system S is a system for causing a vehicle to travel along a target trajectory by controlling the vehicle's steering angle, and is, for example, a system included in a vehicle. The target trajectory is a predetermined trajectory and includes multiple driving positions that are the target of the vehicle and directions corresponding to the multiple driving positions that are the target of the vehicle.

[0035] The driving control system S calculates the optimal tire steering angle u under a rule-based control cycle using model predictive control of a vehicle model, where the vehicle model indicates the relationship between the vehicle's speed (hereinafter referred to as "vehicle speed"), lateral deviation, azimuth deviation, and the curvature of the road surface on which the vehicle travels. The driving control system S has an initiation model predictive control unit (Initiation MPC) and a follow-up model predictive control unit (Follow-up MPC). The initiation MPC is used after the vehicle starts until a predetermined time period has elapsed, and the follow-up MPC is used after the predetermined time period has elapsed. As an example, the initiation MPC and the follow-up MPC calculate the tire steering angle u1 and tire steering angle u2 in parallel.

[0036] Until the predetermined time period has elapsed, the driving control system S uses the tire steering angle u1 calculated by the initiation MPC to control the vehicle, and after the predetermined time period has elapsed, it uses the tire steering angle u2 calculated by the following MPC to control the vehicle. Compared to following the target trajectory, the initiation MPC prioritizes driver feel and determines the tire steering angle u1 so that the change per unit time does not become too large. On the other hand, the following MPC prioritizes following the target trajectory and determines the tire steering angle u2. Because the driving control system S switches between initiation MPC and following MPC before and after the predetermined time period has elapsed since vehicle startup, it can achieve both good driver feel and adherence to the target trajectory.

[0037] [Configuration of the driving control system]

[0038] Figure 2 This diagram illustrates the configuration of the driving control system S. The driving control system S includes a status recognition device 1, a driving control device 2, and a driving control device 10.

[0039] The state recognition device 1 identifies parameters indicating the vehicle's state in a regular control cycle. These parameters include, for example, the vehicle's weight, center of gravity position, speed, steering angle, lateral deviation, azimuth deviation, and road surface curvature. Lateral deviation is the difference between the vehicle's current position and its target position in a direction orthogonal to the vehicle's direction of travel. Azimuth deviation is the difference between the vehicle's direction at its current position and the target direction corresponding to that position.

[0040] The state recognition device 1 measures, for example, the weight of a person riding in the vehicle and the weight of a load loaded on the vehicle. Based on the measured weight of the person and the load, as well as the vehicle's weight, the state recognition device 1 determines the weight of the moving vehicle. Based on the identified vehicle weight and the vehicle's wheelbase, the state recognition device 1 identifies the location of the vehicle's center of gravity.

[0041] The state recognition device 1 identifies longitudinal and lateral velocities based on, for example, the vehicle speed measured by a speed sensor (not shown) installed on the vehicle. The longitudinal speed is the speed in the vehicle's forward direction, and the lateral speed is the speed in a direction orthogonal to the vehicle's forward direction. Furthermore, the state recognition device 1 acquires the vehicle's steering angle, measured, for example, by a steering angle sensor installed on the vehicle. The steering angle obtained by the state recognition device 1 is the rotation angle of the steering wheel shaft, or the difference between the vehicle's direction and the direction of the vehicle's tires.

[0042] The state recognition device 1 identifies the vehicle's position and orientation, for example, by acquiring GPS (Global Positioning System) signals. The state recognition device 1 identifies the vehicle's lateral deviation based on the identified vehicle position and the corresponding target driving position. The state recognition device 1 identifies the vehicle's azimuth deviation based on the identified vehicle orientation and the corresponding target driving direction.

[0043] For example, the state recognition device 1 identifies the curvature of the road surface corresponding to a specified position of the vehicle based on map information stored in its storage unit. The state recognition device 1 outputs the identified center of gravity, center of gravity position, longitudinal velocity, lateral velocity, steering angle, lateral deviation, azimuth deviation, and road surface curvature to the driving control device 10 in a regular control cycle.

[0044] The driving control unit 2 controls the speed and direction of the vehicle. In the next control cycle, the driving control unit 2 controls the direction of the vehicle based on the steering angle (hereinafter referred to as the "tire steering angle"). The tire steering angle is output by the driving control unit 10 in regular control cycles.

[0045] The driving control unit 10 generates a vehicle model corresponding to the vehicle state input from the state recognition device 1 in a regular control cycle. The driving control unit 10 uses the generated vehicle model in the regular control cycle to calculate the tire steering angle so that the vehicle travels in the target direction. The regular control cycle is a sampling cycle in model predictive control. The driving control unit 10 inputs the calculated tire steering angle to the driving control unit 2, thereby causing the vehicle to travel in the target direction. The configuration and operation of the driving control unit 10 will be described in detail below.

[0046] [Configuration of Driving Control Device 10]

[0047] The driving control unit 10 includes a storage unit 11 and a control unit 12. The control unit 12 includes an acquisition unit 121, a generation unit 122, a calculation unit 123, and a driving control unit 124. The driving control unit 10 generates a vehicle model based on parameters indicating the vehicle state output from the state recognition device 1, calculates the tire steering angle using an evaluation function corresponding to the generated vehicle model in a regular control cycle, and outputs the tire steering angle to the driving control unit 2.

[0048] The storage unit 11 includes a storage medium, such as ROM (Read-Only Memory), RAM (Random Access Memory), and SSD (Solid State Drive). The storage unit 11 stores programs executed by the control unit 12. For example, the control unit 12 is a CPU (Central Processing Unit). The control unit 12 operates as an acquisition unit 121, a generation unit 122, a calculation unit 123, and a driving control unit 124 by executing the programs stored in the storage unit 11.

[0049] The acquisition unit 121 acquires parameters indicating the vehicle state output from the state recognition device 1 in a rule control cycle. The acquisition unit 121 acquires the vehicle's weight, the position of the vehicle's center of gravity, the longitudinal velocity as the speed in the vehicle's travel direction, the lateral velocity as the speed in a direction orthogonal to the vehicle's travel direction, the vehicle's steering angle, the vehicle's lateral deviation, the azimuth deviation as the difference between the vehicle's direction at its current position and the target direction of the vehicle corresponding to that position, and the curvature of the road surface on which the vehicle travels.

[0050] The generation unit 122 generates a vehicle model that indicates the relationship between weight, center of gravity position, speed, steering angle, lateral deviation, azimuth deviation and curvature in a rule-controlled cycle. Figure 3 This is a diagram used to illustrate the vehicle model generated by the generation unit 122. For example, the generation unit 122 generates a model that is similar to... Figure 3 The vehicle model corresponding to the reference point shown.

[0051] The generation unit 122 can generate a vehicle model in response to the acquisition unit 121 acquiring at least one of the vehicle's weight, center of gravity position, longitudinal speed, and lateral speed in a rule-based control cycle. This vehicle model corresponds to a control cycle following the control cycle at the moment when the acquisition unit 121 acquires at least one of the weight, center of gravity position, longitudinal speed, and lateral speed. For example, the generation unit 122 generates a vehicle model that updates at least one of the vehicle's weight, center of gravity position, longitudinal speed, and lateral speed in a rule-based control cycle.

[0052] The generation unit 122 generates a vehicle factory model indicating the vehicle's motion and a route-following model indicating the vehicle's trajectory as a continuous-time vehicle model. Subsequently, the generation unit 122 derives discrete-time state equations from the generated continuous-time vehicle model. In this embodiment, as an example, an equivalent two-wheeled model is used to generate the vehicle model.

[0053] First, the vehicle factory model will be described. (Corresponding to...) Figure 3 The vehicle motion at the reference point shown can be determined by using the vehicle's longitudinal velocity v. x The lateral velocity v of the vehicle yThe following equations (1) and (2) represent the yaw angle ψ, vehicle speed v, and steering angle input δ.

[0054]

[0055]

[0056] However, assuming the vehicle's longitudinal velocity v x If the deviation angle is constant and sufficiently small, then the following equation (3) is considered to be satisfied.

[0057] v y =v x sin(β(t))≈v x β(t)···(3)

[0058] The coefficients aij (i, j = 1, 2 or 3) used in equations (1) and (2) are obtained by using the cornering coefficient K of the front wheels. f The steering resistance coefficient K of the rear wheels r The distance from the center of gravity to the front wheel l f The distance l from the center of gravity to the rear wheel r The weight m and moment of inertia I of the vehicle are calculated using the following equations (4) to (9).

[0059]

[0060]

[0061]

[0062]

[0063]

[0064]

[0065] Based on equations (1) and (2), the equation for calculating the curvature κ indicating the trajectory along which the vehicle travels is represented by the following equation (10).

[0066]

[0067] Next, the route-following model will be described. When the vehicle factory model is represented by a state-space model, the following equations (11) to (14) are obtained.

[0068]

[0069]

[0070]

[0071]

[0072] Assuming that i) the line connecting the reference point and the vehicle is orthogonal to ii) the tangent to the reference point, the path length Sr is represented by the following equation (15), which uses Figure 3 The marked distance z is shown as the distance between the vehicle and the reference point. The marked distance z is represented by the following equation (16).

[0073]

[0074]

[0075] Figure 3 The azimuth deviation θ shown is obtained by using the deviation angle β, yaw angle ψ, and attitude angle of the reference point. The following equation (17) is used to calculate.

[0076]

[0077] By substituting equations (1) and (2) into equation (12), the generation unit 122 can also use the following equation (18) to calculate the azimuth deviation θ. Furthermore, it is assumed that the following equation (19) applies to the rate of change of the vehicle's lateral deviation and the rate of change of the vehicle's azimuth deviation.

[0078]

[0079]

[0080] By redefining the state equations using equations (11) to (14) and equations (17) to (19), the following equations (20) to (25) are established. By using equations (20) to (25), the generation unit 122 can calculate the trajectory of the vehicle based on the curvature and the steering angle.

[0081]

[0082]

[0083]

[0084]

[0085]

[0086]

[0087] The generation unit 122 calculates the distance l from the center of gravity to the front wheel based on the vehicle's weight m and the position of the center of gravity of the vehicle. f and the distance l from the center of gravity to the rear wheel r Longitudinal velocity v x and lateral velocity v y The input is fed into the continuous-time vehicle model generated in the manner described above to update the continuous-time vehicle model.

[0088] By having the generation unit 122 update the vehicle model in this manner, the generation unit 122 can update the vehicle model based on at least one of the vehicle's weight, center of gravity position, and speed obtained for each rule control cycle. As a result, the generation unit 122 can generate a vehicle model corresponding to changes in vehicle weight, center of gravity position, and speed without delay.

[0089] Next, the generation unit 122 derives the discrete-time state equations. The discretized state equations are represented by the following equations (26) to (29). T represents the control period (sampling period).

[0090] x[k+1]=A d x[k]+B d u[k]···(26)

[0091] y[k]=C d x[k]···(27)

[0092] x[k]=x(kT)···(28)

[0093] u[k]=u(kT)···(29)

[0094] Since the solution to the continuous-time state equation is equation (30) below, equations (31) and (32) below can be derived by substituting equation (30) into equation (26).

[0095]

[0096]

[0097]

[0098] When zero-order hold is represented by equation (33) below, the state variable x for each control cycle can be represented by equation (34) below by substituting equation (33) into equation (32).

[0099] u(t)=u(kT)=u[k]=constant(kT≤t<(k+1)T···(33)

[0100]

[0101] Furthermore, equation (35) can be derived from equation (34).

[0102]

[0103] Next, by defining the following equations (36) and (37), the discrete-time state equation can be represented by the following equation (38).

[0104] λ=Tt···(36)

[0105] t=τ+kT···(37)

[0106]

[0107] By comparing the coefficients of equations (26) and (38), the coefficient matrix of the state equation can be represented by the following equations (39) and (40).

[0108] A d =e AT ···(39)

[0109]

[0110] As described above, the discrete-time state-space equation is derived from the state equation of the continuous-time vehicle model. However, since equation (40) is an equation that includes integration, it includes the vehicle's motion in past control cycles. As a result, the vehicle model in the current control cycle may not be generated with high accuracy. On the other hand, the generation unit 122 can derive the following equations (42) and (43) by using the following equation (41) established in the matrix exponential function of A and B.

[0111]

[0112] A d =M 11 ···(42)

[0113] B d =M 12 ···(43)

[0114] In this case, the generation unit 122 generates a discrete-time vehicle model using formulas (41) to (43). By operating the generation unit 122 in this way, the generation unit 122 can generate a discrete-time vehicle model with high accuracy. Furthermore, since the generation unit 122 can omit the calculation of vehicle motion in past control cycles, the calculation time can be shortened. As a result, the generation unit 122 can generate a high-precision vehicle model with regular control cycles.

[0115] Next, the operation of the calculation unit 123 will be described. The calculation unit 123 calculates the estimated lateral deviation and estimated azimuth deviation, excluding noise included in the lateral deviation and azimuth deviation acquired by the acquisition unit 121, and calculates the tire steering angle for optimizing the evaluation function including the estimated lateral deviation and estimated azimuth deviation. Noise is, for example, observation noise or system noise, and includes measurement errors included when the state recognition device 1 specifies the lateral deviation or azimuth deviation.

[0116] First, by inputting the steering angle and curvature acquired by the acquisition unit 121 into the state space model, the calculation unit 123 estimates the noise included in the lateral deviation and azimuth deviation acquired by the acquisition unit 121, wherein the state space model corresponds to the vehicle model corresponding to the longitudinal velocity, lateral velocity, weight, and center of gravity position acquired by the acquisition unit 121. The calculation unit 123 specifies the estimated lateral deviation and estimated azimuth deviation from which the estimation noise is excluded. The calculation unit 123 calculates the estimated lateral deviation and estimated azimuth deviation, for example, by using state equations utilizing a linear Kalman filter.

[0117] Assuming that the observation noise ν and system noise ω excluded by the computing unit 123 are white noise, the state equations using linear Kalman filtering can be represented by the following equations (44) and (45).

[0118] x[k+1]=A d x[k]+B d u[k]+w[k]···(44)

[0119] y[k]=C d x[k]+v[k]···(45)

[0120] Subsequently, based on equations (44) and (45), the calculation unit 123 can calculate the prior estimate xε of the state variable x, as shown in equation (46) below. Furthermore, when a deviation occurs between the prior estimate xε and the actual state variable due to system noise, the calculation unit 123 can use equation (47) below to correct the prior estimate xε. The variable h in equation (47) is the innovation gain.

[0121] x e [k|k-1]=A d x e [k-1]+B d u[k-1]···(46)

[0122] x e [k]=x e [k|k-1]+h[k](y[k]-Cd x e [k|k-1])···(47)

[0123] In equation (47), when the influence of observation noise decreases in response to changes in the value of variable h, the influence of system noise on the prior estimate xε increases, while when the influence of observation noise increases, the influence of system noise on the prior estimate xε decreases. On the other hand, the calculation unit 123 updates the prior variance and posterior variance to optimize variable h. The prior variance can be represented by the following equation (48), and the posterior variance can be represented by the following equation (50).

[0124] P[k|k-1]=A d P[k-1]A d T +B d v[k]A d T ···(48)

[0125]

[0126] P[k]=(Ih[k]C T )P[k|k-1]···(50)

[0127] The calculation unit 123 optimizes the weights of observation noise and system noise by adjusting the variable h, which is the innovation gain, using equations (48) to (50). For example, when the weight of observation noise increases, the denominator of equation (49) increases, thereby decreasing the innovation gain. On the other hand, when the weight of system noise increases, the numerator of equation (49) increases, thereby increasing the innovation gain.

[0128] Since the state matrices Ad and Bd are defined as time-invariant, if the observation noise and system noise are white noise, it is assumed that the state variable x of the linear time-invariant state equation with infinite time converges to a stable value. Because the calculation unit 123 identifies the observation noise and system noise by optimizing their weights, and calculates estimated lateral deviation and estimated azimuth deviation excluding the observation noise and system noise, the driving control device 10 can calculate the tire steering angle with high accuracy based on the estimated lateral deviation and estimated azimuth deviation.

[0129] The calculation unit 123 calculates the steering angle that minimizes or maximizes the output value of the evaluation function as the optimal steering angle. This evaluation function includes estimated lateral deviation, estimated azimuth deviation, steering angle, and the change in steering angle relative to the immediately preceding control cycle, calculated based on the vehicle model. Specifically, the calculation unit 123 first inputs the calculated estimated lateral deviation and estimated azimuth deviation, steering angle, and the change in steering angle into the evaluation function corresponding to the vehicle model, which corresponds to the longitudinal velocity, lateral velocity, weight, and center of gravity position acquired by the acquisition unit 121. Then, the calculation unit 123 calculates the tire steering angle that minimizes or maximizes the output value of the evaluation function.

[0130] Here, when the state variable x in the state-space equation is represented by equation (51) below, the observed output y is represented by equation (52). Here, v y , ψ, e z and e θ These are lateral velocity, yaw rate, estimated lateral deviation, and estimated azimuth deviation.

[0131]

[0132]

[0133] The computation unit 123 uses a steady-state Kalman filter to estimate the state variable x and uses the evaluation function shown in the following equation (53) to calculate the optimization problem of model predictive control. In equation (53), p represents the prediction level, δ represents the steering angle input, Δδ represents the difference between the steering angle input and the steering angle input of the immediately preceding control cycle, and the subscripts “max” and “min” as input and output variables represent the upper and lower limits of the signal. Q1, Q2, R1, and R2 are weighting factors.

[0134]

[0135] Satisfy δ min ≤δ[k+k t ]≤δ max

[0136] Δδ min ≤Δδ[k+k t ]≤Δδ max

[0137] The calculation unit 123 performs optimization calculations to minimize the output value J of the evaluation function shown in equation (53) to calculate the tire steering angle in real time, thereby achieving vehicle following of the target trajectory. By using the calculation unit 123 to calculate the tire steering angle in this way, the driving control device 10 can make the vehicle drive at a position with small error relative to the target trajectory at each moment in multiple control cycles.

[0138] The driving control unit 124 drives the vehicle based on the tire steering angle calculated by the calculation unit 123. The driving control unit 124 outputs the tire steering angle calculated by the calculation unit 123 to the driving control device 2 in a regular control cycle, so that the vehicle drives at the calculated tire steering angle.

[0139] Here, as an example, when prioritizing the convergence of lateral deviation and azimuth deviation to 0, compared to the case where lateral deviation and azimuth deviation are not prioritized to converge to 0, the calculation unit 123 sets at least one of the weighting coefficients Q1 and Q2 to be larger, and at least one of the weighting coefficients R1 and R2 to be smaller. When prioritizing the reduction of the change in tire steering angle, compared to the case where the change in tire steering angle is not prioritized, the calculation unit 123 sets R2 to be larger.

[0140] To improve the driver's immediate feeling after vehicle start-up and ensure the vehicle's ability to follow a target trajectory, the calculation unit 123 makes the weighting coefficient R1 (steering angle coefficient) of the term corresponding to the tire steering angle δ after the initial cycle from vehicle start-up smaller than the weighting coefficient R1 of the term corresponding to the tire steering angle δ up to the end of the initial cycle. Furthermore, the calculation unit 123 makes the weighting coefficient R2 (steering angle change coefficient) of the term corresponding to the change in tire steering angle Δδ after the initial cycle from vehicle start-up smaller than the weighting coefficient R2 of the term corresponding to the change in tire steering angle Δδ up to the end of the initial cycle.

[0141] The calculation unit 123 can estimate the lateral deviation e from the time the vehicle starts until after the initial cycle. z The weighting factor Q1 (lateral deviation factor) of the term is greater than the factor corresponding to the estimated lateral deviation e up to the end of the initial period. z The weighting coefficient Q1 for the term. Furthermore, the calculation unit 123 can calculate the estimated azimuth deviation e after the initial cycle from vehicle startup. θ The weighting coefficient Q2 (azimuth deviation coefficient) of the term is greater than the one corresponding to the estimated azimuth deviation e up to the end of the initial period. θ The weighting coefficient Q2 for the term.

[0142] Specifically, the calculation unit 123 makes the weighting coefficients Q1 and Q2 smaller before the initial cycle than after the initial cycle, and makes the weighting coefficients R1 and R2 larger before the initial cycle than after the initial cycle, in order to suppress excessive increase in tire steering angle or excessive increase in the amount of change in tire steering angle, thereby improving driver feel. On the other hand, after the initial cycle, the calculation unit 123 increases the weighting coefficients Q1 and Q2 and decreases the weighting coefficients R1 and R2, thereby preferentially reducing lateral deviation and azimuth deviation and improving the ability to follow the target trajectory.

[0143] Based on the mode set by the driver, the calculation unit 123 can determine the value of each weighting coefficient up to the end of the initial cycle and the value of each weighting coefficient after the end of the initial cycle. For example, when the driver selects a mode that prioritizes the driver's perception over the ability to immediately follow the target trajectory after the vehicle starts, the calculation unit 123 makes the weighting coefficients R1 and R2 up to the end of the initial cycle larger than when the driver selects a mode that prioritizes the ability to immediately follow the target trajectory after the vehicle starts.

[0144] Furthermore, if the weighting coefficient changes only at the moment the initial period ends, there is a possibility that the driver will perceive a rapid change. Therefore, the calculation unit 123 can change the weighting coefficient during a predetermined transition period after the initial period ends. For example, the predetermined transition period is determined based on the magnitude of the difference between the weighting coefficient before and after the change, and the predetermined transition period is the time required to slowly change the weighting coefficient to a degree that makes the change in the weighting coefficient less noticeable to the driver.

[0145] Furthermore, since the characteristics differ depending on the vehicle type, the optimal weighting coefficients in the evaluation function are different. Therefore, for example, the calculation unit 123 can identify the vehicle type based on vehicle identification information stored in the storage unit 11, and the calculation unit 123 can determine at least one weighting coefficient among the weighting coefficients of one or more terms of the evaluation function by referring to the storage unit 11, which stores the vehicle type associated with the weighting coefficients of one or more terms of the evaluation function. Because the calculation unit 123 is configured in this way, the control unit 12 can control the vehicle by executing a program using weighting coefficients suitable for each of the multiple vehicle types.

[0146] [Flowchart of Driving Control Device 10]

[0147] Figure 4 This is a flowchart illustrating an example of the operation of the driving control device 10. Figure 4 The flowchart shown illustrates the operation of the driving control unit 10 in calculating the tire steering angle based on parameters indicating the vehicle state obtained from the state recognition device 1.

[0148] The acquisition unit 121 acquires the vehicle's state variables (step S11), such as the vehicle's weight, center of gravity position, longitudinal velocity, lateral velocity, steering angle, lateral deviation, azimuth deviation, and curvature of the road surface. The generation unit 122 generates a vehicle model corresponding to the vehicle's state variables acquired by the acquisition unit 121 (step S12).

[0149] The calculation unit 123 calculates the estimated lateral deviation and estimated azimuth deviation, excluding observation noise and system noise included in the lateral deviation and azimuth deviation of the vehicle acquired by the acquisition unit 121, by inputting the vehicle's steering angle and the curvature of the road surface on which the vehicle travels into the state space model corresponding to the vehicle model generated by the generation unit 122 (step S13).

[0150] The calculation unit 123 inputs the curvature acquired by the acquisition unit 121 and the estimated lateral deviation and estimated azimuth deviation identified by the calculation unit 123 into the evaluation function corresponding to the vehicle model generated by the generation unit 122, and calculates the tire steering angle that minimizes the output value of the evaluation function (step S14).

[0151] If the end-of-process operation is not performed ("No" in step S15), the driving control device 10 repeats the processing from steps S11 to S14 to calculate the tire steering angle based on parameters indicating the vehicle state obtained in the next control cycle. If the end-of-process operation is performed ("Yes" in step S15), the driving control device 10 terminates the processing.

[0152] Figure 5 This is a flowchart illustrating an example of how the calculation unit 123 calculates the tire steering angle. Figure 5 The flowchart shown begins with the process of monitoring whether the vehicle is started (step S21).

[0153] If it is determined that the vehicle has started ("Yes" in step S21), the calculation unit 123 sequentially performs the calculation of the starting steering angle u1 (S22) and the calculation of the following steering angle u2 (step S23) for each control cycle. The order of these processes within the control cycle is arbitrary, and the calculation unit 123 can use multiple processors to execute these processes in parallel.

[0154] When calculating the starting steering angle u1 and the following steering angle u2, the calculation unit 123 determines whether the time elapsed since the vehicle started has exceeded a threshold (step S24). If the elapsed time is less than the threshold ("No" in step S24), the calculation unit 123 selects the starting steering angle u1 as the tire steering angle u (step S25), and then returns the process to step S22.

[0155] If the elapsed time exceeds the threshold ("Yes" in step S24), the calculation unit 123 selects the steering angle u2 as the tire steering angle u (step S26), and then determines whether the vehicle has stopped moving (step S27). If it is determined that the vehicle has not stopped moving ("No" in step S27), the calculation unit 123 returns to step S22. If it is determined that the vehicle has stopped moving ("Yes" in step S27), the calculation unit 123 determines whether the engine has stopped (step S28). If the calculation unit 123 determines that the engine has not stopped ("No" in step S28), the calculation unit 123 returns to step S21.

[0156] [Confirmed through simulation results]

[0157] The inventors confirmed the effectiveness of the driving control device 10 according to this embodiment through simulation. Figure 6 It is a simulation block diagram.

[0158] Using a pre-prepared vehicle model, a computer is used to calculate and estimate lateral deviation, azimuth deviation, and curvature. The upcoming curvature is calculated based on a map model, and optimization calculations are performed based on an evaluation function to calculate the tire steering angle. At startup, weighting coefficients Q1 and R1 are set to 23.25, Q2 to 7.5, R1 to 200, and R2 to 200. Subsequently, weighting coefficients Q1 and R2 are set to 23.25, Q2 to 7.5, R1 to 4, and R2 to 75. The initial value for the vehicle's lateral deviation is set to 1m, and the initial values ​​for the azimuth deviation and vehicle speed are set to 0. Then, verification is performed to determine if the vehicle can start smoothly from this state.

[0159] When the weighting factor is applied from the moment the vehicle starts, the lateral deviation becomes 0.1m or less due to vehicle movement, resulting in overshoot relative to the target trajectory and the vehicle deviating from the driving path. On the other hand, when the weighting factor used at vehicle start is applied from the moment the vehicle starts until 5 seconds have elapsed, the vehicle gradually approaches the target trajectory and is able to move along it. Afterwards, a switch to the next weighting factor is executed, and it is then confirmed that the vehicle is traveling along the target trajectory.

[0160] [Effect of Driving Control Device 10]

[0161] As described above, the calculation unit 123 of the driving control device 10 calculates the steering angle that minimizes or maximizes the output value of the evaluation function as the optimal tire steering angle. The evaluation function includes estimated lateral deviation, estimated azimuth deviation, steering angle, and the amount of change of steering angle relative to the immediately preceding control cycle, calculated based on the vehicle model. Then, the calculation unit 123 switches the weighting coefficients of the evaluation function based on whether the current period is within a predetermined initial cycle since vehicle startup or whether a predetermined initial cycle has passed. For example, the calculation unit 123 makes the weighting coefficient of the term corresponding to the change of steering angle after the initial cycle since vehicle startup smaller than the weighting coefficient of the term corresponding to the change of steering angle up to the end of the initial cycle. By operating the calculation unit 123 in this way, it is possible to improve both the driver's immediate feeling after vehicle startup and the follow-through ability after the initial cycle since vehicle startup.

[0162] Although embodiments of the present invention have been described above, the technical scope of the present invention is not limited to the embodiments described above, and various modifications and changes can be made without departing from the scope of the present invention. For example, all or part of the device may be functionally or physically distributed and integrated in any unit. Furthermore, new embodiments resulting from any combination of multiple embodiments are also included in the embodiments of the present invention. The effects of new embodiments caused by combinations have the effects of the original embodiments.

[0163] [Symbol Description]

[0164] 1. Status recognition device

[0165] 2. Stroke control device

[0166] 10. Driving control device

[0167] 11. Storage Department

[0168] 12. Control Department

[0169] 121. Acquisition Department

[0170] 122. Generation Department

[0171] 123. Calculation Department

[0172] 124. Driving Control Unit

Claims

1. A driving control device, including: The acquisition unit acquires the vehicle's weight, center of gravity position, speed, steering angle, lateral deviation, azimuth deviation, and curvature of the road surface on which the vehicle travels. The generation unit generates a vehicle model at a predetermined control cycle, the vehicle model indicating the relationship between the weight, the center of gravity position, the speed, the steering angle, the lateral deviation, the azimuth deviation, and the curvature; as well as The calculation unit calculates the steering angle that minimizes or maximizes the output value of the evaluation function as the optimal steering angle. The evaluation function includes estimated lateral deviation, estimated azimuth deviation, the steering angle, and the change in the steering angle relative to the immediately preceding control cycle, all calculated based on the vehicle model. The calculation unit makes the weighting coefficient of the term corresponding to the change amount after an initial period from when the vehicle is started less than the weighting coefficient of the term corresponding to the change amount from when the vehicle is started until after the predetermined initial period.

2. The driving control device according to claim 1, wherein, The calculation unit makes the weighting factor of the term corresponding to the steering angle after the initial cycle from when the vehicle is started less than the weighting factor of the term corresponding to the steering angle from when the vehicle is started until the predetermined initial cycle.

3. The driving control device according to claim 1 or 2, wherein, The calculation unit changes the weighting coefficient of the term corresponding to the amount of change during a predetermined transition period starting from the initial period.

4. The driving control device according to claim 3, wherein, The calculation unit determines the transition period based on the difference between the weighting coefficients before and after the change.

5. The driving control device according to claim 1 or 2, wherein, The calculation unit makes the weighting factor of the term corresponding to the estimated lateral deviation after the initial period from when the vehicle is started greater than the weighting factor of the term corresponding to the estimated lateral deviation up to the predetermined initial period.

6. The driving control device according to claim 1 or 2, wherein, The calculation unit makes the weighting factor of the term corresponding to the estimated azimuth deviation after the initial period from when the vehicle is started greater than the weighting factor of the term corresponding to the estimated azimuth deviation up to the predetermined initial period.

7. The driving control device according to claim 1 or 2, wherein The calculation unit determines at least one weighting coefficient of one or more terms of the evaluation function by referring to the storage unit, which stores the type of vehicle associated with the weighting coefficient of one or more terms of the evaluation function.

8. The driving control device according to claim 1 or 2, wherein, When a mode that prioritizes a good feel over the ability to immediately follow the target trajectory after the vehicle starts is selected, the calculation unit assigns a larger weighting coefficient to the term related to the steering angle input to the evaluation function up to the initial cycle, compared to the case where a mode prioritizing the ability to immediately follow the target trajectory after the vehicle starts is selected.

9. The driving control device according to claim 1 or 2, wherein, The computation unit performs optimization calculations to minimize the output value of the evaluation function shown in the following equation. Satisfy δ min ≤δ[k+k t ]≤δ max Dd min ≤Δδ[k+k t ]≤Δδ max Where p represents the prediction level, e z e represents the estimated lateral deviation. θ This indicates the estimated azimuth angle deviation, δ represents the steering angle input, Δδ represents the difference between the steering angle input and the steering angle input of the immediately preceding control cycle, "max" and "min" as subscripts of each input and output variable are the upper and lower limits of the signal, and Q1, Q2, R1 and R2 are weighting coefficients.

10. The driving control device according to claim 9, wherein, When a mode that prioritizes a good feeling over the ability to immediately follow the target trajectory after the vehicle starts is selected, the calculation unit makes the weighting coefficients R1 and R2 in the evaluation function larger up to the initial cycle compared to the case where a mode that prioritizes the ability to immediately follow the target trajectory after the vehicle starts is selected.

11. The driving control device according to claim 9, wherein When prioritizing the convergence of the lateral deviation and the azimuth deviation to 0, compared to the case where the convergence of the lateral deviation and the azimuth deviation is not prioritized to 0, the calculation unit sets at least one of the weighting coefficients Q1 and Q2 to be larger, or sets at least one of the weighting coefficients R1 and R2 to be smaller.

12. The driving control device according to claim 9, wherein When prioritizing reducing the change in the steering angle, the calculation unit sets the weighting coefficient R2 to be larger than when not prioritizing reducing the change in the steering angle.

13. A driving control method executed by a computer, including: The vehicle's weight, center of gravity position, speed, steering angle, lateral deviation, azimuth deviation, and curvature of the road surface on which the vehicle travels are obtained. A vehicle model is generated at a predetermined control cycle, the vehicle model indicating the relationship between the weight, the center of gravity position, the speed, the steering angle, the lateral deviation, the azimuth deviation, and the curvature; as well as The optimal steering angle is calculated by minimizing or maximizing the output value of an evaluation function, which includes an estimated lateral deviation, an estimated azimuth deviation, the steering angle, and the change of the steering angle relative to the immediately preceding control cycle, all calculated based on the vehicle model. The calculation includes making the weighting factor of the term corresponding to the change after an initial period from when the vehicle is started less than the weighting factor of the term corresponding to the change from when the vehicle is started until after the predetermined initial period.

Images