An adaptive unmanned serial control method

Abstract

The application relates to a self-adaptive unmanned serial control method, which comprises the following steps: inputting a reference heading angle; calculating a heading angle error; calculating a reference front wheel rotation angle by a self-adaptive controller 1; calculating a desired front wheel rotation angle by a position error calculation; calculating a front wheel rotation angle error; calculating a front wheel rotation angle control amount by a self-adaptive controller 2; inputting the front wheel rotation angle control amount into a vehicle to obtain an updated front wheel rotation angle and a heading angle. Compared with the prior art, the application is aimed at the under-actuated and strong nonlinear system of the unmanned vehicle, can realize model-free adaptive control, can stably control the front wheel rotation angle by the serial control method, can reduce tracking error, can stably synchronize the convergence of lateral error and heading error, can improve the tracking precision of the unmanned system, and can ensure the safety of operation.

Description

Technical Field

[0001] This invention belongs to the field of unmanned driving control technology, and relates to motion control methods for unmanned vehicles, particularly an adaptive unmanned driving cascade control method. Background Technology

[0002] The development of autonomous vehicle technology has been rapid in recent years, primarily due to its enormous technological potential to alleviate increasing traffic congestion, improve road safety, and conserve energy. As the core of autonomous vehicle technology, motion control is crucial, encompassing trajectory planning and control execution. These two technologies complement each other and together constitute the key technologies for autonomous vehicles. Among these, addressing the underactuation, nonlinearity, and inaccurate modeling issues inherent in vehicle control remains a current scientific challenge for motion control. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the existing technology and provide an adaptive cascade control method for unmanned driving.

[0004] The objective of this invention can be achieved through the following technical solutions:

[0005] An adaptive cascade control method for autonomous driving includes the following steps:

[0006] 1) Obtain the reference heading angle θ r ;

[0007] 2) Calculate the heading angle error Δθ;

[0008] 3) Input the heading angle error Δθ into the adaptive controller 1 to calculate the reference front wheel steering angle δ. r ;

[0009] 4) Calculate the desired front wheel steering angle δ based on the position error. e ;

[0010] 5) Calculate the front wheel steering angle error Δδ;

[0011] 6) Input the front wheel steering angle error Δδ into the adaptive controller 2 to calculate the front wheel steering angle control quantity δ. c ;

[0012] 7) Adjust the front wheel steering angle control value δ c Input the vehicle and obtain the updated front wheel steering angle δ and heading angle θ.

[0013] In step 2), the heading angle error Δθ is calculated as follows: Δθ = θ r -θ, where θ is the vehicle's actual heading angle.

[0014] In step 3), the discrete vehicle model of the adaptive controller 1 is Where k is the k-th time after discretization, v is the vehicle speed, T is the sampling period, and L is the vehicle wheelbase.

[0015] In step 3), the relationship between the heading angle and the front wheel steering angle is obtained from the discrete vehicle model of the adaptive controller 1 as θ(k+1)=f θ (θ(k),δ(k)),fθ(.) represents the nonlinear mapping relationship between the front wheel steering angle and the heading angle.

[0016] In step 3), the full-format dynamic linearized data model of the relationship between the heading angle and the front wheel steering angle is Δθ(k+1)=φ θ (k)·Δθ(k)+φ δ (k)·Δδ(k), where φ θ (k) and φδ(k) are unknown bounded pseudo gradients.

[0017] In step 3), the control law of the adaptive controller 1 can be obtained from the full-format dynamic linearized data model. Wherein, step size factor ρ θ ,ρ δ ∈(0,1], weighting factor λ θ >0.

[0018] In step 4), when the position error is e, the desired front wheel steering angle δ is... e The calculation method is as follows Among them, L w This is the aiming distance.

[0019] In step 5), the front wheel steering angle error Δδ is calculated as follows: Δδ = δ r +δ e -δ, where δ is the actual front wheel steering angle of the vehicle.

[0020] In step 6), the vehicle steer-by-wire actuator model referenced by the adaptive controller 2 is... Among them, J eq B is the equivalent rotational inertia of the steering system. eq For the equivalent damping of the steering system, C eq For the equivalent stiffness of the steering system, K eq The equivalent transmission ratio of the steering system is given by τ, where τ is the output torque of the motor, and M is the torque of the motor. z This is the tire return torque.

[0021] In step 6), the relationship between the front wheel angle and the output torque of the steering motor, obtained from the vehicle steer-by-wire actuator model referenced by the adaptive controller 2, is δ(k+1)=f τ(δ(k),δ(k-1),τ(k)). Among them, f τ (.) The nonlinear mapping relationship between the front wheel steering angle and the output torque of the steering motor.

[0022] In step 6), the front wheel steering angle control value δ is... c Equivalent to the output torque of the steering motor, the front wheel angle and the front wheel angle control quantity δ obtained by the adaptive controller 2 from the vehicle steer-by-wire actuator model are... c The relation is updated to δ(k+1)=f δ (δ(k),δ(k-1),δ c (k)). Where, f δ (.) Nonlinear mapping relationship between front wheel steering angle and front wheel steering angle control quantity.

[0023] In step 6), the front wheel steering angle and the front wheel steering angle control amount δ c A full-format dynamic linearized relational data model is Where, φ c (k)=[φ1(k),φ2(k),φ3(k)] T For an unknown bounded pseudo gradient, ΔH c (k)=[Δδ(k),Δδ(k-1),Δδ c (k)] T .

[0024] In step 6), the control law of the adaptive controller 2 can be obtained from the full-format dynamic linearized data model.

[0025]

[0026] Where the step size factors ρ1, ρ2, ρ3 ∈ (0, 1], and the weight factor λ c >0.

[0027] Compared with the prior art, the present invention has the following advantages:

[0028] 1. The adaptive control law 1 in this invention can realize model-free adaptive control of the heading angle, so that the control strategy no longer depends on the precise vehicle model parameters. The algorithm is adapted to the uncertainty of the vehicle model parameters and can be quickly deployed to unmanned vehicles to achieve real-time and rapid elimination of heading errors.

[0029] 2. The adaptive control law 2 in this invention can realize model-free adaptive control of the front wheel steering angle, so that the control strategy no longer depends on the precise actuator model parameters. The algorithm is adapted to the uncertainty of the vehicle's steer-by-wire actuator model parameters and can be quickly deployed to autonomous vehicles to achieve real-time and rapid elimination of lateral errors.

[0030] 3. The cascade control method of the present invention stabilizes the front wheel steering angle, effectively reducing tracking error, improving steering tracking accuracy, and ensuring tracking efficiency;

[0031] 4. The cascade control method of the present invention can achieve model-free adaptive control for underactuated and strongly nonlinear systems such as unmanned vehicles, and can stably bring lateral and heading errors to converge synchronously, thereby improving the tracking accuracy of the unmanned driving system and ensuring safe operation. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the adaptive unmanned cascade control method of the present invention;

[0033] Figure 2 This is a schematic diagram of trajectory tracking according to the present invention;

[0034] Figure 3 This is a schematic diagram illustrating trajectory tracking for both the pure tracking method and the method of this invention;

[0035] Figure 4 A schematic diagram of the turning angle for trajectory tracking using the pure tracking method and the method of the present invention;

[0036] Figure 5 This is a schematic diagram illustrating the errors in trajectory tracking between the pure tracking method and the method of the present invention. Detailed Implementation

[0037] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0038] Example:

[0039] like Figure 1 As shown, Figure 1 This is a schematic diagram of an adaptive cascade control method for autonomous driving according to the present invention. The specific steps of this strategy include:

[0040] Step 1: Obtain the reference heading angle θ according to the trajectory to be tracked. r .

[0041] Step 2: Calculate the heading angle error Δθ. The calculation method is Δθ = θ r -θ, where θ is the vehicle's actual heading angle.

[0042] Step 3: Place the vehicle model Discretize to obtain the reference vehicle model for adaptive controller 1. Where k is the k-th time after discretization, v is the vehicle speed, T is the sampling period, and L is the vehicle wheelbase.

[0043] Step 4: From the vehicle model of the discrete adaptive controller 1 obtained in Step 3, the relationship between the heading angle and the front wheel steering angle is θ(k+1)=fθ (θ(k),δ(k)),f θ (.) Nonlinear mapping relationship between front wheel steering angle and heading angle.

[0044] Step 5: From Step 4, the full-format dynamic linearized data model of the relationship between the heading angle and the front wheel steering angle is obtained as Δθ(k+1)=φ θ (k)·Δθ(k)+φ δ (k)·Δδ(k). Where, φ θ (k) and φ δ (k) is an unknown bounded pseudo gradient.

[0045] Step 6: Design the criterion function J for the control input of adaptive controller 1. θ As shown in formula (1),

[0046] J θ =(θ r (k+1)-θ(k+1)) 2 +λ θ ·(δ r (k)-δ r (k-1)) 2 (1)

[0047] Wherein, the weighting factor λ θ >0.

[0048] Step 7: Substitute the dynamically linearized data model from Step 5 into Formula (1) to obtain the control law of the adaptive controller 1, as shown in Formula (2).

[0049]

[0050] Wherein, step size factor ρθ,ρ δ ∈(0,1).

[0051] Step 8: Due to φ θ (k) and φ δ (k) represents the pseudo-gradient, which is time-varying, unknown, but bounded, and the estimation criterion function J. θ,δ As shown in formula (3),

[0052]

[0053] Wherein, the weighting factor μ θ >0, For φ θ The estimated value of (k-1), For φ δ The estimated value of (k-1).

[0054] Step 9: The pseudo gradient estimation algorithm is shown in formula (4).

[0055]

[0056] Wherein, step size factor η θ ∈(0,2],k θ (k) is the dynamic coefficient.

[0057] Step 10: Dynamic coefficient k θ (k) The solution method is shown in formula (5).

[0058]

[0059] Step 11: Introduce a reset mechanism, that is, when the conditions of equation (6) or (7) are met, the reset estimate is the initial value of equation (8), i.e.

[0060]

[0061] (Δθ(k-1)) 2 +(Δδ(k-1)) 2 ≤ε2 (7)

[0062]

[0063] Where ε1, ε2 > 0 are local minima. for initial value, for The initial value of .

[0064] Step 12: Input the heading angle error Δθ into the control law of adaptive controller 1, and calculate the reference front wheel steering angle δ. r .

[0065] Step 13: As Figure 2 As shown, the desired front wheel steering angle δ is calculated based on a position error of e. e The calculation method is as follows Among them, L w R is the aiming distance, and O is the turning radius. t x is the center of the circle with the turning radius. t For the tracking points of the trajectory that needs to be tracked, O v This is the center point of the vehicle's rear axle.

[0066] Step 14: Calculate the front wheel steering angle error Δδ. The calculation method is Δδ = δ r +δ e -δ, where δ is the actual front wheel steering angle of the vehicle.

[0067] Step 15: From the vehicle steer-by-wire actuator model The relationship between the front wheel steering angle of the adaptive controller 2 and the output torque of the steering motor is obtained as δ(k+1)=f τ (δ(k),δ(k-1),τ(k)). Among them, J eq B is the equivalent rotational inertia of the steering system. eq For the equivalent damping of the steering system, C eq For the equivalent stiffness of the steering system, K eq The equivalent transmission ratio of the steering system is given by τ, where τ is the output torque of the motor, and M is the torque of the motor. z f is the tire return torque. τ (.) The nonlinear mapping relationship between the front wheel steering angle and the output torque of the steering motor. Where, f τ (.) The nonlinear mapping relationship between the front wheel steering angle and the output torque of the steering motor.

[0068] Step 16: Adjust the front wheel steering angle control value δ c Equivalent to the output torque of the steering motor, the front wheel angle and the front wheel angle control quantity δ obtained by the adaptive controller 2 from the vehicle steer-by-wire actuator model are... c The relation is updated to δ(k+1)=f δ (δ(k),δ(k-1),δ c (k)). Where, f δ (.) Nonlinear mapping relationship between front wheel steering angle and front wheel steering angle control quantity.

[0069] Step 17: From Step 16, the front wheel steering angle and the front wheel steering angle control value δ can be obtained. c A full-format dynamic linearized relational data model is Where, φ c (k)=[φ1(k),φ2(k),φ3(k)] T For an unknown bounded pseudo gradient, ΔH c (k)=[Δδ(k),Δδ(k-1),Δδ c (k)] T .

[0070] Step 18: Design the criterion function J for the control input of the adaptive controller 2. δ As shown in formula (9),

[0071] J δ =(δ r (k+1)+δ e (k+1)-δ(k+1)) 2 +λ c ·(δ c (k)-δ c (k-1)) 2 (9)

[0072] Wherein, the weighting factor λ c >0.

[0073] Step 19: Substitute the formatted dynamic linearized data model from Step 17 into Equation (9) to obtain the control law of the adaptive controller 2, as shown in Equation (10):

[0074]

[0075] Where the step size factors ρ1, ρ2, ρ3 ∈ (0, 1).

[0076] Step 20: Due to φ c The pseudo-gradient represented by (k) is time-varying, unknown, but bounded. We design an estimation criterion function J. δ,δ As shown in formula (11),

[0077]

[0078] Step 21: The pseudo gradient estimation algorithm is shown in equation (12).

[0079]

[0080] Wherein, step size factor η c ∈(0,2],k c (k) is the dynamic coefficient.

[0081] Step 22: Dynamic coefficient k θ (k) The solution method is shown in formula (5).

[0082]

[0083] Step 23: Introduce a reset mechanism, that is, when the conditions of equation (14) or equation (15) are met, the reset estimate is the initial value of equation (16), i.e.

[0084]

[0085] ||ΔH c (k-1)||≤ε4 (15)

[0086]

[0087] Among them, ε3, ε4 > 0 are local minima. for The initial value of .

[0088] Step 24: Adjust the front wheel steering angle control value δ c Input the vehicle and obtain the updated front wheel steering angle δ and heading angle θ.

[0089] One embodiment of the present invention is as follows: Figure 3The image depicts a scenario of autonomous driving on a circular road. The thin dashed lines represent the reference road, and the driver tracks the reference road counter-clockwise. The thick solid lines represent the control effect of the pure tracking method, and the thick dashed lines represent the control effect of the method described in this invention. Figure 4 As shown, when using the method of this invention for tracking, the response of the front wheel steering angle is more stable and the overshoot is smaller than that of the pure tracking method. Figure 5 As shown, the peak tracking error is within 0.18m when using the method of this invention, and close to 0.48m when using the pure tracking method. The results indicate that the cascade control method proposed in this invention stabilizes the front wheel steering angle, reduces tracking error, and enables the lateral and heading errors to converge synchronously.

[0090] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An adaptive cascade control method for unmanned driving, characterized in that, Includes the following steps: 1) Obtain the reference heading angle θ r ; 2) Calculate the heading angle error Δθ; 3) Input the heading angle error Δθ into the adaptive controller 1 to calculate the reference front wheel steering angle δ. r ; 4) Calculate the desired front wheel steering angle δ based on the position error. e ; 5) Calculate the front wheel steering angle error Δδ; 6) Input the front wheel steering angle error Δδ into the adaptive controller 2 to calculate the front wheel steering angle control quantity δ. c ; 7) Adjust the front wheel steering angle control value δ c Input the vehicle information to obtain the updated front wheel steering angle δ and heading angle θ; In step 2), the heading angle error Δθ is calculated as Δθ=θr-θ, where θ is the actual heading angle of the vehicle; In step 3), the discrete vehicle model of the adaptive controller 1 is: ； Where k is the kth time after discretization, v is the vehicle speed, T is the sampling period, and L is the vehicle wheelbase; The relationship between the heading angle and the front wheel steering angle, obtained from the discrete vehicle model of adaptive controller 1, is θ(k+1)=f θ (θ(k),δ(k)),f θ (.) represents the nonlinear mapping relationship between the front wheel steering angle and the heading angle; The full-format dynamic linearized data model of the relationship between heading angle and front wheel steering angle is Δθ(k+1)=φθ(k)·Δθ(k)+φδ(k)·Δδ(k), where, θ (k) and δ (k) represents an unknown bounded pseudo-gradient; from the full-format dynamic linearized data model, the control law of adaptive controller 1 can be obtained as follows: ； Wherein, step size factor ρ θ ,ρ δ ∈(0,1], weighting factor λ θ >0.

2. The adaptive unmanned cascade control method according to claim 1, characterized in that, In step 4), when the position error is e, the desired front wheel steering angle δ is... e The calculation method is as follows: ; Among them, L w Pre-aiming distance; In step 5), the front wheel steering angle error Δδ is calculated as follows: Δδ = δ r +δ e -δ, where δ is the actual front wheel steering angle of the vehicle.

3. The adaptive unmanned cascade control method according to claim 1, characterized in that, In step 6), the vehicle steer-by-wire actuator model referenced by the adaptive controller 2 is: ； Among them, J eq B is the equivalent rotational inertia of the steering system. eq For the equivalent damping of the steering system, C eq For the equivalent stiffness of the steering system, K eq The equivalent transmission ratio of the steering system is given by τ, where τ is the output torque of the motor, and M is the torque of the motor. z This is the tire return torque; The relationship between the front wheel steering angle and the output torque of the steering motor, obtained from the vehicle steer-by-wire actuator model referenced by adaptive controller 2, is δ(k+1)=f τ (δ(k),δ(k-1),τ(k)), where, f τ (.) The nonlinear mapping relationship between the front wheel steering angle and the output torque of the steering motor.

4. The adaptive unmanned cascade control method according to claim 3, characterized in that, front wheel steering angle control value δ c Equivalent to the output torque of the steering motor, the front wheel angle and the front wheel angle control quantity δ obtained by the adaptive controller 2 from the vehicle steer-by-wire actuator model are... c The relation is updated to: δ(k+1)=f δ (δ(k),δ(k-1),δc(k)), where, f δ (.) The nonlinear mapping relationship between the front wheel steering angle and the front wheel steering angle control quantity; Front wheel steering angle and front wheel steering angle control value δ c The full-format dynamic linearized data model of relational expressions is as follows: ； in, c (k) = [1(k), 2(k), 3(k)] T For an unknown bounded pseudo gradient, ΔH c (k)=[Δδ(k),Δδ(k-1),Δδ c (k)] T .

5. The adaptive unmanned cascade control method according to claim 4, characterized in that, Based on the full-format dynamic linearized data model, the control law of adaptive controller 2 can be obtained as follows: ； Wherein, step size factors ρ1,ρ 2, ρ3∈(0,1], weight factor λc>0.

Images