Automatic driving active obstacle avoidance method based on whisker algorithm

Abstract

The invention discloses an automatic driving active obstacle avoidance method based on a whisker algorithm, belongs to the field of vehicle dynamics control, and discloses an automatic driving real-time obstacle avoidance method based on the whisker algorithm. Firstly, sensors such as a camera and a millimeter-wave radar sense obstacle information; secondly, the active obstacle avoidance system generates a parabolic whisker based on real-time sensing information and current kinematics information of the vehicle; the whisker length is related to the vehicle speed so as to ensure the vehicle driving safety; and then, the tentacles are screened by adopting comfort constraint and safety constraint, and a vehicle driving area is obtained. And finally, based on the vehicle two-degree-of-freedom model, model prediction control is adopted to track the whisker, and the optimal front wheel steering angle input is obtained after rolling optimization.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of automatic driving, and particularly relates to an automatic driving active obstacle avoidance method based on a whisker algorithm. BACKGROUND

[0002] An automatic driving obstacle avoidance system can effectively prevent or reduce collision accidents, and the key is to accurately identify and actively avoid the front obstacles. In recent years, with the continuous development of active obstacle avoidance technology, the research object gradually shifts from static obstacles to dynamic obstacles. The difficulty of real-time obstacle avoidance of dynamic obstacles lies in that when the front vehicle suddenly changes the trajectory, the traditional path planning algorithm is difficult to re-plan a new driving route to avoid the moving obstacle in a short time. While the whisker algorithm can quickly select a new whisker from the feasible region to avoid obstacles when the front vehicle suddenly changes the trajectory, thereby ensuring the real-time, safety and comfort of obstacle avoidance. SUMMARY

[0003] In view of the existing problems, the application provides an automatic driving real-time obstacle avoidance method based on a whisker algorithm. First, sensors such as cameras and millimeter wave radars perceive obstacle information. Second, the active obstacle avoidance system generates a parabolic whisker based on real-time perception information and current kinematic information of the vehicle. The length of the whisker is related to the vehicle speed to ensure the safety of vehicle driving. Then, the whisker is screened by using comfort constraints and safety constraints to obtain the drivable area of the vehicle. Finally, the model predictive control is used based on the two-degree-of-freedom model of the vehicle to track the whisker, and the optimal front wheel steering angle input is obtained after rolling optimization. BRIEF DESCRIPTION OF DRAWINGS

[0004] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of these drawings.

[0005] Figure 1 The whole idea is shown in the figure; Figure 2 The feasible region analysis diagram is shown. DETAILED DESCRIPTION

[0006] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0007] The following is in conjunction with the appendix Figure 1 —2 provides a further detailed description of this application, namely, an autonomous driving active obstacle avoidance method based on the tentacle algorithm. In the tentacle algorithm, each tentacle is an arc segment originating from the vehicle's center of mass, which can be described by a parabolic equation. The focal points of all tentacles are on the same straight line, and the endpoint position is related to the vehicle speed and the curvature of the tentacle. The specific expression is as follows: in, and These represent the horizontal and vertical coordinates of the tentacles in the geodetic coordinate system. This represents the curvature of the parabola, and the size of the parabola's opening (i.e., the tip of the tendril). (Value) with The value decreases as the curvature increases. By designing different parabolic curvatures, 21 different tentacles can be generated. The specific method is as follows: in, Number the tentacles. This is the curvature correction coefficient for the tentacles, used to adjust the spacing between adjacent tentacles. This is the reference curvature, specifically the curvature of the 11th tentacle, which can be obtained from the vehicle's steady-state yaw rate gain at the previous moment. This refers to the front wheel steering angle of the vehicle at the previous moment. and These represent the wheelbase and the vehicle speed, respectively. Indicates the vehicle stability factor. For the sake of the vehicle's quality, and These represent the distances from the vehicle's center of gravity to the front and rear axles, respectively. and These represent the lateral stiffness of the front and rear wheels, respectively. Indicates the corrected curvature. Indicates the maximum curvature of the tentacles. It is a correction factor. The length of the tentacles and The relationship with the speed of this vehicle is as follows: The maximum speed of this vehicle is set to 100 km / h. This maximum speed is divided into ten equal parts, and these parts are numbered as follows: As a speed range. For example, when At that time, the vehicle speed was 1 km / h ~ 10 km / h.

[0008] 2. Discretize each of the 21 tentacles into several points, and use a traversal method to find the point on each tentacle that is closest to the target vehicle. At the same time, set the point closest to the target vehicle as... Taking vehicle width into account, let... Filter out based on the safe distance threshold. Tentacles greater than or equal to this threshold constitute the feasible region of the tentacle algorithm, and the calculation method is shown in the following formula.

[0009] Taking comfort into account, the optimal tentacle is selected from the selected tentacles based on the smallest curvature.

[0010] 3. Model Predictive Control (MPC) is an advanced process control method. A significant feature of MPC is that, compared to LQR control, MPC can consider various constraints on the spatial state variables, while LQR, PID, and other controls can only consider various constraints on the input and output variables.

[0011] First, a two-degree-of-freedom (DOF) vehicle model is established. The two-DOF vehicle model is the most fundamental model for analyzing vehicle handling stability. It is frequently used to determine the vehicle's lateral stability based on the driver's steering intentions and is a key model in vehicle steering stability control programs. This paper studies a steering control system based on the whisker algorithm for lane changing and obstacle avoidance. Therefore, using a dynamic two-DOF model with state variables such as lateral error and heading error is highly effective.

[0012] Let the state vector be... for: , among them , , , These represent the lateral error, the derivative of the lateral error, the heading error, and the derivative of the heading error, respectively. The state-space form of the two-degree-of-freedom model established based on this state vector is as follows: in , , , , This represents the moment of inertia of the vehicle about the z-axis. This indicates the longitudinal speed of the vehicle.

[0013] Discretizing the above two-degree-of-freedom model yields the following formula: in , , , , , It is the identity matrix. Based on the above formula, a new state vector is established as follows: Thus, the augmented matrix is ​​obtained. and as follows: in , , , The new state-space equations are as follows: Based on the above formula Step-state prediction can be obtained as follows: To make the entire system clearer, the output of future times is represented in matrix form: in , , , , , .

[0014] By observing the above equation, we can see that both the state and output quantities in the prediction time domain can be determined by the current state of the system. and control increment in the control time domain This calculation forms the basis of the "prediction" function in model predictive control algorithms.

[0015] Let the reference output vector be Then the cost function can be written as: The last four terms of the equation are independent of the solution value and can be ignored in the optimization solution. Therefore, the cost function above can be written as: set up , Then the above formula can be written as: Add comfort constraints to the above cost function to ensure that the front wheel angle of the vehicle is not too large and the steering wheel is not turned too fast during the driving process. That is, the lateral acceleration of the vehicle cannot exceed the maximum value and the longitudinal impact of the vehicle cannot exceed the maximum value.

[0016] Depend on have to: Depend on have to: Ultimately, it transforms into a quadratic programming problem: The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. An autonomous driving active obstacle avoidance method based on the tentacle algorithm, wherein in the tentacle algorithm, each tentacle is an arc segment originating from the vehicle's center of mass, which can be described by a parabolic equation. The foci of all tentacles lie on the same straight line, and the endpoint position is related to the vehicle speed and the curvature of the tentacle. The specific expression is as follows: in, and These represent the horizontal and vertical coordinates of the tentacles in the geodetic coordinate system. This represents the curvature of the parabola, and the size of the parabola's opening (i.e., the tip of the tendril). (Value) with The value decreases as the curvature increases. By designing different parabolic curvatures, 21 different tentacles can be generated. The specific method is as follows: in, Number the tentacles. This is the curvature correction coefficient for the tentacles, used to adjust the spacing between adjacent tentacles. This is the reference curvature, specifically the curvature of the 11th tentacle, which can be obtained from the vehicle's steady-state yaw rate gain at the previous moment. This refers to the front wheel steering angle of the vehicle at the previous moment. and These represent the wheelbase and the vehicle speed, respectively. Indicates the vehicle stability factor. For the sake of the vehicle's quality, and These represent the distances from the vehicle's center of gravity to the front and rear axles, respectively. and These represent the lateral stiffness of the front and rear wheels, respectively. Indicates the corrected curvature. Indicates the maximum curvature of the tentacles. It is a correction factor. The length of the antennae and The relationship with the speed of this vehicle is as follows: The maximum speed of this vehicle is set to 100 km / h. This maximum speed is divided into ten equal parts, and these parts are numbered as follows: As a speed range. For example, when At that time, the vehicle speed was 1 km / h ~ 10 km / h.

2. An active obstacle avoidance method for autonomous driving based on the tentacle algorithm, characterized in that: The 21 tentacles are discretized into several points, and a traversal method is used to find the point on each tentacle that is closest to the target vehicle. At the same time, set the point closest to the target vehicle as... Taking into account the vehicle width, let Filter out based on the safe distance threshold. Tentacles greater than or equal to this threshold constitute the feasible region of the tentacle algorithm, and the calculation method is shown in the following formula. Taking comfort into account, the optimal tentacle is selected from the selected tentacles based on the smallest curvature.

3. An active obstacle avoidance method for autonomous driving based on the whisker algorithm, characterized in that: Model predictive control (MPC) is an advanced process control method. A significant feature of MPC is that, compared with LQR control, MPC can consider various constraints on the spatial state variables, while LQR, PID and other controls can only consider various constraints on the input and output variables. First, a two-degree-of-freedom (DOF) vehicle model is established. The two-DOF vehicle model is the most fundamental model for analyzing vehicle handling stability. It is frequently used to determine the vehicle's lateral stability based on the driver's steering intentions and is a key model in vehicle steering stability control procedures. This paper studies a steering control system based on the whisker algorithm for lane changing and obstacle avoidance. Therefore, a dynamic two-degree-of-freedom model using state variables such as lateral error and heading error is very effective; Let the state vector be... for: , among them , , , These represent the lateral error, the derivative of the lateral error, the heading error, and the derivative of the heading error, respectively. The state-space form of the two-degree-of-freedom model established based on this state vector is as follows: in , , , , This represents the moment of inertia of the vehicle about the z-axis. This indicates the longitudinal speed of the vehicle; Discretizing the above two-degree-of-freedom model yields the following formula: in , , , , , It is the identity matrix. Based on the above formula, a new state vector is established as follows: Thus, the augmented matrix is ​​obtained. and as follows: in , , , The new state-space equations are as follows: Based on the above formula Step-state prediction can be obtained as follows: To make the entire system clearer, the output of future times is represented in matrix form: in , , , , , . By observing the above equation, we can see that both the state and output quantities in the prediction time domain can be determined by the current state of the system. and control increment in the control time domain This calculation forms the basis of the "prediction" function in model predictive control algorithms; Let the reference output vector be Then the cost function can be written as: The last four terms of the equation are independent of the solution value and can be ignored in the optimization solution. Therefore, the cost function above can be written as: set up , Then the above formula can be written as: Add comfort constraints to the above cost function to ensure that the front wheel angle of the vehicle is not too large and the steering wheel is not turned too fast during the driving process. That is, the lateral acceleration of the vehicle cannot exceed the maximum value and the longitudinal impact of the vehicle cannot exceed the maximum value. Depend on have to: Depend on have to: Ultimately, it transforms into a quadratic programming problem: 。

Images