A multi-lane car-following control method based on preceding vehicle state information and safety constraints

Abstract

The application discloses a multi-lane following control method based on preceding vehicle state information and safety constraints, which comprises the following steps: acquiring the speed and yaw rate of the preceding vehicle; identifying the lane through a visual sensor and determining whether the following task is same-lane following or adjacent-lane following; when the same-lane following, directly calculating the trajectory curvature of the preceding vehicle through the acquired speed and yaw rate of the preceding vehicle; when the adjacent-lane following, projecting the preceding vehicle to the lane of the following vehicle according to the lane width and calculating the trajectory curvature after the projection; converting the trajectory curvature of the preceding vehicle or the trajectory curvature after the projection of the preceding vehicle from the time dimension to the distance dimension as the target curvature of the ego vehicle; designing the lateral controller of the ego vehicle to obtain the ideal wheel rotation angle of the ego vehicle so as to track the target path; designing the longitudinal speed controller of the ego vehicle to obtain the ideal wheel torque of the ego vehicle and control the acceleration and deceleration motion of the ego vehicle, thereby ensuring the ideal inter-vehicle distance between the ego vehicle and the preceding vehicle.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of car-following technology of unmanned vehicles, and relates to a multi-lane car-following control method based on front vehicle state information and safety constraints. BACKGROUND

[0002] Car-following technology of unmanned vehicles is one of the core technologies in the research of unmanned vehicles, and plays a crucial role in improving road utilization rate and driving safety. Among them, front vehicle trajectory estimation and ego vehicle control are key components of car-following technology, and play a decisive role in completing the car-following task of unmanned vehicles in a specific driving environment. Car-following control of unmanned vehicles refers to obtaining front vehicle state information, analyzing and processing its historical trajectory, and calculating wheel angle and torque by combining control algorithm, so as to realize accurate following of the ego vehicle to the front vehicle and maintain the ideal inter-vehicle distance.

[0003] At present, car-following technology is increasingly widely used in the field of unmanned vehicle control, and the performance requirements are gradually improved. First of all, safety is the primary consideration of car-following control. Ensuring that the lateral and longitudinal errors are always within the safety range during control not only improves the stability of the system, but also significantly enhances the safety of the car-following process. Secondly, the driving environment of unmanned vehicles is complex and variable, and the car-following control method needs to adapt to various road conditions and achieve flexible car-following across lanes in different situations. Finally, the core of car-following control is to accurately track the front vehicle trajectory. Rapid and accurate acquisition of front vehicle trajectory information is an important factor to speed up the system response, which is crucial to improve the control effect.

[0004] In the technical solution of Chinese patent "CN 115683116 A front vehicle trajectory generation method and module", when constructing the front vehicle trajectory, it mainly reconstructs the front vehicle trajectory by screening each trajectory point of the front vehicle to realize car-following control. However, polynomial fitting is required during trajectory construction, and the calculation process is relatively complex. In the technical solution of Chinese patent "CN 118289014 A car-following control method, device and equipment based on MPC and storage medium", adaptive cruise control is mainly realized by using ego vehicle information and front vehicle information. However, the adjacent lane and error constraints are not considered during car-following control, which has certain conservatism.

[0005] Therefore, it is of great significance to design a car-following control method that responds quickly, can flexibly switch lanes according to actual driving conditions, and has high safety and adaptability for the development of unmanned technology. SUMMARY

[0006] To solve the above technical problems, the purpose of the present application is to provide a multi-lane car-following control method based on front vehicle state information and safety constraints.

[0007] The application provides a multi-lane car following control method based on preceding vehicle state information and safety constraints, comprising:

[0008] Step 1: obtaining the speed and yaw rate of the preceding vehicle;

[0009] Step 2: identifying the lane through a visual sensor and determining whether the car following task is same-lane car following or adjacent-lane car following;

[0010] Step 3: when the same-lane car following, directly calculating the trajectory curvature of the preceding vehicle through the speed and yaw rate of the preceding vehicle obtained in step 1; when the adjacent-lane car following, projecting the preceding vehicle to the lane of the following vehicle according to the lane width and calculating the trajectory curvature after the projection;

[0011] Step 4: converting the trajectory curvature of the preceding vehicle or the trajectory curvature after the projection of the preceding vehicle from the time dimension to the distance dimension as the target curvature of the ego vehicle;

[0012] Step 5: designing a lateral controller of the ego vehicle to obtain the ideal wheel rotation angle of the ego vehicle to track the target path;

[0013] Step 6: designing a longitudinal speed controller of the ego vehicle to obtain the ideal wheel rotation torque of the ego vehicle to control the acceleration and deceleration of the ego vehicle and ensure the ideal inter-vehicle distance between the ego vehicle and the preceding vehicle.

[0014] The multi-lane car following control method based on preceding vehicle state information and safety constraints has at least the following beneficial effects:

[0015] (1) The control method directly calculates the trajectory curvature through the speed and yaw rate of the preceding vehicle, without the need of reconstructing or fitting the trajectory, thereby reducing the calculation amount. Moreover, the curvature calculation method can project the trajectory of the preceding vehicle to the lane of the following vehicle, realizing the car following control of the ego vehicle and the preceding vehicle in the adjacent lane.

[0016] (2) The control method is flexible in obtaining the curvature of the preceding vehicle, which can directly obtain through the communication between the two vehicles or obtain through a differentiator designed by the visual sensor, thereby avoiding the privacy permission problem existing in the communication.

[0017] (3) The lateral control scheme and the longitudinal control scheme both constrain the error, preventing the vehicle from deviating from the predetermined lane in the lateral direction and ensuring the safety distance between the two vehicles and preventing the safety distance from exceeding the effective range of the sensor in the longitudinal direction. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 is a flowchart of the multi-lane car following control method based on preceding vehicle state information and safety constraints;

[0019] Figure 2is a flow chart of dimension conversion of trajectory curvature of a preceding vehicle. DETAILED DESCRIPTION

[0020] As Figure 1 shown, a multi-lane car-following control method based on preceding vehicle state information and safety constraints of the application comprises:

[0021] Step 1: obtaining the speed and yaw rate of the preceding vehicle, the step 1 specifically comprises:

[0022] Step 1.1: first obtaining the speed and yaw rate information of the preceding vehicle through vehicle-to-vehicle communication; if it cannot be obtained due to privacy permission problems, obtaining the relative distance and relative angle between the two vehicles through a vision sensor.

[0023] Step 1.2: designing a differentiator through the relative distance and relative angle between the two vehicles to obtain the speed and yaw rate information of the preceding vehicle, the step 1.2 specifically comprises:

[0024] Step 1.2.1: according to the vehicle kinematics model, calculating the speed of the preceding vehicle through the ego vehicle as follows:

[0025]

[0026] wherein d is the relative distance between the two vehicles, is the derivative of d; x P represents the distance of the preceding vehicle moving in the x-axis direction, y P represents the distance of the preceding vehicle moving in the y-axis direction; x H represents the distance of the ego vehicle moving in the x-axis direction, is the derivative of x H , y H represents the distance of the ego vehicle moving in the y-axis direction, is the derivative of y H ; ψ P is the yaw angle of the preceding vehicle.

[0027] Step 1.2.2: designing the following differentiator to calculate

[0028]

[0029] wherein, is the estimated value of , is the estimated value of d, sig n (x)=x n sgn(x), k d1 , k d2 and k d3 are adjustable parameters of the designed differentiator.

[0030] Step 1.2.3: Calculate the yaw angle of the front vehicle using the following formula:

[0031] ψ p =ψ H +ψ I

[0032] Where, ψ H ψ is the yaw angle of the vehicle. I The relative angle between the two vehicles is obtained from a vision sensor.

[0033] Step 1.2.4: The yaw rate of the front vehicle is obtained by designing the following differentiator:

[0034]

[0035] Where, γ P The yaw rate of the vehicle in front. For γ P The estimated value, It is the yaw angle ψ of the front vehicle. p The estimated value, k γ1 k γ2 k γ3 These are the adjustable parameters of the differentiator.

[0036] Step 2: Identify the lane using a visual sensor to determine whether the car-following task is to follow within the same lane or in an adjacent lane.

[0037] Step 3: When following another vehicle in the same lane, directly calculate the trajectory curvature of the preceding vehicle based on its speed and yaw rate; when following another vehicle in an adjacent lane, project the preceding vehicle onto the following lane based on the lane width, and calculate the projected trajectory curvature. Specifically, Step 3 involves:

[0038] Step 3.1: When following another vehicle in the same lane, calculate the curvature of the trajectory of the vehicle in front using the following formula:

[0039]

[0040] Step 3.2: When following another vehicle in the adjacent lane, calculate the curvature of the projected trajectory of the preceding vehicle using the following formula:

[0041]

[0042] Where B is the lane width; the curvature of the lane where the vehicle is currently located is determined by a visual sensor. If the curvature of the vehicle's lane is greater than that of the lane in front, the sign is "-", otherwise it is "+".

[0043] Step 4: The currently obtained trajectory curvature information of the leading vehicle is a variable in the time dimension. It is necessary to convert the trajectory curvature into information that changes according to the driving distance. Convert the trajectory curvature of the leading vehicle or the trajectory curvature after projection of the leading vehicle from the time dimension to the distance dimension as the target curvature of the host vehicle, as Figure 2 shown. The specific steps of Step 4 are as follows:

[0044] Step 4.1: Sample the trajectory curvature and store it in an array K in the order of sampling time. That is, the trajectory curvature at the i-th time sampling point is stored at the i-th position of the curvature array K.

[0045] Step 4.2: Calculate the driving trajectory length of the leading vehicle at the corresponding time point according to the sampling time of the trajectory curvature, and store the driving trajectory length of the leading vehicle in the array S in order. That is, store the driving trajectory length corresponding to the i-th time sampling point at the i-th position of the trajectory array S.

[0046] Step 4.3: Calculate the current driving trajectory length S H of the host vehicle and compare it with the initial distance d0 between the two vehicles. If S H < d0, the target curvature of the host vehicle is 0; if S H > d0, then compare S H - d0 with the driving trajectory lengths in the trajectory array S, traverse the data in S to find the position index i H closest to S d , and obtain the curvature κ(i d ) at the i d -th position in the curvature array K as the target curvature of the host vehicle, and complete the correspondence between the driving distance of the rear vehicle and the target curvature.

[0047] ...Step 5: Design a lateral controller for the host vehicle to obtain the ideal wheel angle of the host vehicle to make it track the target path. The specific steps of Step 5 are as follows:

[0048] Step 5.1: Calculate the distance error e d between the current position of the host vehicle and the projection point and the angle error e ψ between the yaw angle of the host vehicle and the tangent direction of the projection point according to the following formula: ... [[ID=...]]... is the speed of the projection point in the tangent direction of the ideal path.

[0051] ​​​​​​​

[0052] Step 5.2: Calculate tracking error e using preview error method p = e d + p L e ψ , p L is preview distance, in order to further ensure the safety of car following control, the ego vehicle is constrained within its target lane width, the following constraint function is introduced:

[0053]

[0054] Where t is the ego vehicle travel time, T m is the set stabilization time; ρ(0) and ρ(∞) are set constants.

[0055] Step 5.3: Reconstruct error variable as follows:

[0056]

[0057] Where k e is a constant;

[0058] Step 5.4: Finally design the ego vehicle lateral controller as:

[0059]

[0060] Where δ f is the ego vehicle front wheel steering angle, k1, k2, p a1 , p a12 , p a2 , p a22 , p d , p dd , p b and p bb are lateral controller parameters.

[0061] Step 6: Design the ego vehicle longitudinal speed controller to obtain the ideal ego vehicle wheel torque, control the ego vehicle acceleration and deceleration motion, and ensure the ideal inter-vehicle distance between the ego vehicle and the front vehicle, the step 6 is specifically:

[0062] Step 6.1: When implementing car following control, in addition to ensuring that the lateral error makes the ego vehicle track the historical path of the front vehicle, it is also necessary to ensure the distance between the ego vehicle and the front vehicle to ensure vehicle safety, the inter-vehicle distance is calculated using the time-distance method as follows:

[0063] S d = S1v P + S0

[0064] Where S d is the ideal inter-vehicle distance, S1 is the time-distance coefficient, and S0 is the minimum inter-vehicle safety distance.

[0065] Step 6.2: Actual two-vehicle distance and ideal distance error e s is expressed as

[0066] e s = S P -S H -S d +d0

[0067] wherein S P is the front vehicle driving path length, S H is the self-vehicle driving path length, and d0 is the initial position distance difference between the two vehicles.

[0068] Step 6.3: When designing the longitudinal controller, the constraints are introduced as:

[0069]

[0070] e s = S1v P

[0071]

[0072] Step 6.4: The final self-vehicle longitudinal speed controller is designed as:

[0073]

[0074]

[0075] wherein Ta is the wheel torque, e s is the upper and lower bounds of the distance error, which is determined according to the vehicle safety distance S1 and the sensor effective distance, k v1 ,k v2 ,k a ,p bv ,p bbv ,p dv ,p ddv are the longitudinal speed controller parameters.

[0076] The multi-lane car following control method based on the front vehicle state information and safety constraints of the present application directly constructs the front vehicle trajectory curvature through the speed and yaw rate information of the front vehicle, converts the car following control into trajectory tracking control based on the curvature, and simultaneously constrains the lateral error and longitudinal error of the vehicle to ensure the safety of the car following control. The method not only has simple calculation, but also can realize car following in different lanes of the front vehicle and the self-vehicle.

[0077] The above merely describes preferred embodiments of the present application and is not used to limit the idea of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for multi-lane car-following control based on preceding vehicle state information and safety constraints, characterized in that, The application relates to a method for controlling a vehicle to follow a preceding vehicle, comprising the following steps: Step 1: obtaining the speed and yaw rate of the preceding vehicle; Step 2: identifying the lane through a visual sensor and determining whether the following task is same-lane following or adjacent-lane following; Step 3: when the same-lane following, directly calculating the trajectory curvature of the preceding vehicle according to the speed and yaw rate of the preceding vehicle; when the adjacent-lane following, projecting the preceding vehicle to the lane behind the vehicle according to the lane width and calculating the trajectory curvature after the projection; Step 4: converting the trajectory curvature of the preceding vehicle or the trajectory curvature after the projection of the preceding vehicle from the time dimension to the distance dimension as the target curvature of the ego vehicle; Step 5: designing the lateral controller of the ego vehicle to obtain the ideal wheel angle of the ego vehicle to track the target path; Step 6: designing the longitudinal speed controller of the ego vehicle to obtain the ideal wheel torque of the ego vehicle to control the acceleration and deceleration of the ego vehicle and ensure the ideal inter-vehicle distance between the ego vehicle and the preceding vehicle. The step 1 is specifically as follows: 2.The multi-lane car following control method based on preceding vehicle state information and safety constraints of claim 1, wherein, Step 1.1: first obtaining the speed and yaw rate information of the preceding vehicle through inter-vehicle communication; if the information cannot be obtained due to privacy permission, obtaining the relative distance and relative angle between the two vehicles through a visual sensor; Step 1.2: designing a differentiator through the relative distance and relative angle to obtain the speed and yaw rate information of the preceding vehicle. The step 1.2 is specifically as follows: 3.The multi-lane car following control method based on preceding vehicle state information and safety constraints of claim 2, wherein, Step 1.2.1: according to the vehicle kinematics model, calculating the speed of the preceding vehicle through the ego vehicle as follows: Step 1.2.3: calculating the yaw angle of the preceding vehicle according to the following formula: where d is the relative distance between the two vehicles, is the derivative of d; x P represents the distance that the leading vehicle moves in the x-axis direction, y P represents the distance that the leading vehicle moves in the y-axis direction; x H represents the distance that the ego vehicle moves in the x-axis direction, is the derivative of x H , y H represents the distance that the ego vehicle moves in the y-axis direction, is the derivative of y H ; ψ P is the yaw angle of the leading vehicle; Step 1.2.2: Design the following differentiator calculation wherein is an estimate of is an estimate of d, sig n (x) = x n sgn(x), k d1 ,k d2 and k d3 are designable differentiator adjustable parameters; Step 1.2.4: obtaining the yaw rate of the preceding vehicle through the following designed differentiator: ψ p = ψ H + ψ I wherein ψ H is the yaw angle of the ego vehicle; ψ I is the relative angle between the two vehicles, obtained from the vision sensor; The step 3 is specifically as follows: where γ P is the yaw rate of the front vehicle, is the estimated value of γ P , is the estimated value of the yaw angle ψ p of the front vehicle, k γ1 , k γ2 , k γ3 are the adjustable parameters of the differentiator.

4. The multi-lane car following control method based on preceding vehicle state information and safety constraints of claim 1, wherein, Step 3.1: when the same-lane following, calculating the trajectory curvature of the preceding vehicle according to the following formula: Step 3.2: when the adjacent-lane following, calculating the trajectory curvature after the projection of the preceding vehicle according to the following formula: Wherein, B is the lane width; judging the curvature of the lane where the ego vehicle is located through a visual sensor, if the curvature of the lane where the ego vehicle is located is greater than the curvature of the lane where the preceding vehicle is located, the sign is taken as "-", otherwise "+". The step 4 is specifically as follows:

5. The multi-lane car following control method based on preceding vehicle state information and safety constraints of claim 1, wherein, Step 4.1: sampling the trajectory curvature and storing the trajectory curvature in the curvature array K in the order of sampling time, that is, the trajectory curvature of the i-th time sampling point is stored in the i-th position of the curvature array K; Step 4.2: calculating the length of the driving trajectory of the preceding vehicle at the corresponding time point according to the sampling time of the trajectory curvature and storing the length of the driving trajectory of the preceding vehicle in the trajectory array S in sequence, that is, the length of the driving trajectory corresponding to the i-th time sampling point is stored in the i-th position of the trajectory array S; The step 5 is specifically as follows: Step 4.3: Calculate the length S of the current driving trajectory of the host vehicle H And compare it with the initial distance d0 between the two vehicles. If S H < d0, the target curvature of the host vehicle is 0; if S H > d0, then compare S H - d0 with the driving trajectory lengths in the trajectory array S, traverse the data in S to find the position index i H closest to S d , and correspondingly obtain the curvature κ(i d ) at the i d -th position in the curvature array K as the target curvature of the host vehicle, thus completing the correspondence between the driving distance of the following vehicle and the target curvature.

6. The multi-lane car following control method based on preceding vehicle state information and safety constraints of claim 1, wherein, Step 5.3: reconstructing the error variable as follows: Step 5.1: Calculate the distance error e of the current ego vehicle position and the projection point according to the following formula d and the angle error e of the angle between the ego vehicle yaw angle and the tangent direction of the projection point ψ : where v x represents the forward speed of the ego vehicle, v y represents the lateral speed of the ego vehicle; β is the side slip angle of the ego vehicle; γ is the yaw rate of the ego vehicle, κ(s) represents the target curvature, v s is the tangential direction speed of the projection point on the ideal path; Step 5.2: Calculate tracking error e with preview error method p = e d + p L e ψ , p L is preview distance, in order to further ensure the safety of car following control, the ego vehicle is constrained within its target lane width, the following constraint function is introduced: where t is the travel time of the ego vehicle, T m is the set stabilization time; p(0) and p(∞) are set constants; Step 5.4: finally designing the lateral controller of the ego vehicle as follows: wherein k e is a constant; The step 6 is specifically as follows: where δ f is the front wheel steering angle of the ego vehicle, k1, k2, p a1 , p a12 , p a2 , p a22 , p d , p dd , p b and p bb are lateral controller parameters.

7. The multi-lane car following control method based on preceding vehicle state information and safety constraints of claim 1, wherein, Step 6.1: the distance between the ego vehicle and the preceding vehicle needs to be ensured to ensure the safety of the vehicle, the inter-vehicle distance is calculated by the time-distance method and the formula is as follows: Step 6.3: considering the safety constraint and the effective distance constraint of the sensor, introducing the constraint when designing the longitudinal controller as follows: S d = S1v P + S0 Where S d is the ideal inter-vehicle distance, S1 is the time-distance coefficient, and S0 is the minimum inter-vehicle safety distance. Step 6.2: Actual two-car distance and ideal distance error e s is represented as e s = S P - S H - S d + d0 wherein S P is the length of the driving path of the preceding vehicle, S H is the length of the driving path of the ego vehicle, and d0is the distance difference between the initial positions of the two vehicles. Step 6.4: finally designing the longitudinal speed controller of the ego vehicle as follows: e s = S1v P ​ z v2 = v x - a v where Ta is the wheel torque, e s is an upper bound and a lower bound of the distance error, respectively, determined from the vehicle safety distance S1 and the sensor effective distance, k v1 ,k v2 ,k a ,p bv ,p bbv ,p dv ,p ddv is a longitudinal speed controller parameter.

Images