Event triggering preview decision-making method for path following of autonomous vehicle

Abstract

The invention relates to an event triggering preview decision-making method for path following of an automatic driving automobile. The method comprises the following steps: S1, obtaining a to-be-followed expected speed scheme of the automatic driving automobile and selecting a preview target; s2, updating a to-be-followed pose list according to the preview path point of the preview target; s3, road points in the to-be-followed pose list are inserted into a followed road point list and deleted, meanwhile, the distance difference between the first mileage and the end mileage of the followed road point list is calculated, and insertion is stopped when the distance exceeds the shortest walking distance; s4, performing preview based on the current following waypoint list, and if the preset precision is not met, triggering a re-preview plan and inserting a preview result into a turning radius-mileage list; s5, repeating the steps S3-S4 until the to-be-followed pose list is empty, and outputting a turning radius-mileage list; and S6, based on the current driving information and the turning radius-mileage list, calculating an expected lateral acceleration and carrying out following control. According to the invention, the following dynamic precision of the automatic driving automobile can be ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving vehicles, and in particular to an event-triggered preview decision method for path following of an autonomous driving vehicle. Background Art

[0002] With the development of science and technology, autonomous vehicles have become one of the fastest-growing and most concerned technology growth points in the current automotive industry. Among them, path following is one of the important functions required by autonomous vehicles. The current working process of the autonomous driving system is mostly to first decide an expected trajectory according to the traffic environment, and then control the car to follow the trajectory. Therefore, following the predetermined ideal trajectory or path is an important part of vehicle motion control. The current driving assistance system or autonomous driving system generally has the phenomenon of high control frequency and fast change in motion control, which is inconsistent with human driving habits, resulting in a decrease in the comfort and trust of drivers and passengers, and "staying away" from autonomous driving and "buying but not using" it, which objectively inhibits the development and promotion of autonomous driving technology. Therefore, in order to solve the problem that the existing autonomous driving car frequently adjusts the control action during the process of following the predetermined trajectory or path, resulting in poor vehicle motion stability and poor driving comfort, the present invention proposes an event-triggered preview decision method for path following of an autonomous driving car. Summary of the invention

[0003] The purpose of the present invention is to provide an event-triggered preview decision method for path following of an autonomous driving vehicle. By establishing an event-triggered planning method, an event trigger mechanism is established based on following deviation and execution mileage. Driving operation adjustments are made only when the following error exceeds the allowable deviation range. The expected lateral acceleration is determined based on the preview result, and the vehicle steering system is controlled through a standard control strategy to ensure the dynamic accuracy of following.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] An event-triggered preview decision method for path following of an autonomous vehicle, comprising:

[0006] S1. Obtaining an expected speed plan of the autonomous driving vehicle to be followed, and selecting a preview target according to the expected speed plan to be followed;

[0007] S2. Update the list of postures to be followed according to the preview path points of the preview target;

[0008] S3, inserting the waypoints in the to-be-followed posture list into the following waypoint list and deleting them, and calculating the difference between the first and last mileages in the following waypoint list, and stopping the insertion when the difference between the first and last mileages exceeds the preset shortest walking distance;

[0009] S4, performing preview based on the current follow-up waypoint list. If the preview result does not meet the preset accuracy, triggering re-preview planning and inserting the preview result into the turning radius-mileage list, and returning to S3; if the preview result meets the preset accuracy, directly returning to S3;

[0010] S5, repeat S3-S4 until the list of postures to be followed is empty, and output the turning radius-mileage list;

[0011] S6. Based on the current driving information and turning radius-mileage list of the autonomous driving vehicle, the expected lateral acceleration is calculated and sent to the autonomous driving vehicle for following control.

[0012] Optionally, in S1, obtaining the expected speed scheme to be followed of the autonomous driving vehicle, and selecting a preview target according to the expected speed scheme to be followed includes:

[0013] Obtaining following path information and a current vehicle state of the autonomous vehicle;

[0014] The current vehicle state is converted into the same coordinate system as the following path to obtain the mileage range of the preview target.

[0015] Optionally, in S4, performing preview includes:

[0016] A double-arc multi-segment trajectory calculation model is constructed, and the double-arc multi-segment trajectory calculation model is solved based on the trajectory shape free variable θ to obtain the double-arc multi-segment trajectory, wherein the trajectory shape free variable θ is determined by an objective function.

[0017] Optionally, the double arc multi-segment trajectory calculation model is:

[0018]

[0019] J = A + R1 (N1 - V);

[0020] C A =A+R1N1;

[0021] C B =J+R2V;

[0022] Among them, A is the starting point, θ A is the starting direction, B is the end point, θ B is the end point direction, α is θ A The angle pointing to AB, β is the angle from AB to θ B The angle of the first arc, θ is the angle of the first arc, N1 is the normal vector of point A, V is the normal vector of point J, R1 is the radius of the first arc, R2 is the radius of the second arc, C A , CB are the centers of the two arcs.

[0023] Optionally, the objective function is:

[0024]

[0025] L1≥L min ;

[0026] L2≥L min ;

[0027] Among them, w d is the weight of the lateral deviation, d is the preview deviation, w R is the curvature weight, κ min is the minimum curvature, κ max is the maximum curvature, L1 is the length of the first arc, L2 is the length of the second arc, L min is the minimum distance.

[0028] Optionally, in S4, triggering the re-preview planning is determined by a preset trigger mechanism, and the trigger mechanism is constructed according to the following deviation and the execution mileage, and the trigger mechanism is:

[0029]

[0030] Among them, s e is the current mileage of the main vehicle, d e is the path following deviation, d max is the maximum allowable path following deviation, s max is the maximum execution mileage, and e is the trigger result.

[0031] Optionally, in S4, after triggering re-preview planning and inserting the preview result into the turning radius-mileage list, other waypoints except the last two waypoints in the current follow-up waypoint list are deleted, a new follow-up waypoint list is obtained, and the process returns to S3.

[0032] Optionally, in S6, calculating the expected lateral acceleration includes:

[0033]

[0034] Among them, a y is the lateral acceleration, V is the current vehicle speed, R n is the turning radius.

[0035] The beneficial effects of the present invention are:

[0036] The present invention can trigger the driving preview function based on the event of whether the predefined following error exceeds a predetermined range, minimize the number of operation adjustments while ensuring the following accuracy, and improve the vehicle's motion stability and driving comfort. Previewing with double arcs as motion primitives ensures that the target position and direction are achieved at the same time. A multi-segment preview method based on motion primitives is established, so that the preview error can be directly used as the preview segmentation basis during the preview process, positively ensuring the following accuracy when the following target does not match the motion primitive. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 This is a flow chart of an event-triggered preview decision method for path following of an autonomous driving vehicle according to an embodiment of the present invention;

[0039] Figure 2 Selecting a coordinate system for the preview target of the embodiment of the present invention;

[0040] Figure 3 This is a flow chart of the event-triggered planning method according to an embodiment of the present invention;

[0041] Figure 4 This is an example of a double arc on the same side of an embodiment of the present invention;

[0042] Figure 5 This is an example of a double arc on opposite sides of an embodiment of the present invention;

[0043] Figure 6 A schematic diagram of double arc symbol definition according to an embodiment of the present invention;

[0044] Figure 7 The present invention is a multi-stage following decision flow chart based on motion primitives according to an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0047] This embodiment provides an event-triggered preview decision method for path following of an autonomous driving vehicle, such as Figure 1 As shown, including:

[0048] S1. Obtaining an expected speed plan of the autonomous driving vehicle to be followed, and selecting a preview target according to the expected speed plan to be followed;

[0049] S2. Update the list of postures to be followed according to the preview path points of the preview target;

[0050] S3, inserting the waypoints in the to-be-followed posture list into the following waypoint list and deleting them, and calculating the difference between the first and last mileages in the following waypoint list, and stopping the insertion when the difference between the first and last mileages exceeds the preset shortest walking distance;

[0051] S4, performing preview based on the current follow-up waypoint list. If the preview result does not meet the preset accuracy, triggering re-preview planning and inserting the preview result into the turning radius-mileage list, and returning to S3; if the preview result meets the preset accuracy, directly returning to S3;

[0052] S5, repeat S3-S4 until the list of postures to be followed is empty, and output the turning radius-mileage list;

[0053] S6. Based on the current driving information and turning radius-mileage list of the autonomous driving vehicle, the expected lateral acceleration is calculated and sent to the autonomous driving vehicle for following control.

[0054] Specifically, this embodiment can trigger the driving preview function based on the event of whether the predefined following error exceeds a predetermined range, minimize the number of operation adjustments while ensuring the following accuracy, and improve the vehicle's motion stability and driving comfort. Previewing with double arcs as motion primitives ensures that the target position and direction are achieved at the same time. A multi-segment preview method based on motion primitives is established, so that the preview error can be directly used as the basis for preview segmentation during the preview process, positively ensuring the following accuracy when the following target does not match the motion primitive.

[0055] Furthermore, in S1, obtaining the expected speed scheme of the autonomous driving vehicle to be followed, and selecting a preview target according to the expected speed scheme to be followed includes:

[0056] Obtaining following path information and a current vehicle state of the autonomous vehicle;

[0057] The current vehicle state is converted into the same coordinate system as the following path to obtain the mileage range of the preview target.

[0058] Specifically, the preview target selection in this embodiment includes the following contents:

[0059] The paths and speeds that the driver wants to follow are varied. In order to uniformly express the target path and target speed, this embodiment describes the desired path and desired speed in the form of a discrete point sequence. i ,Y i ) represents the position coordinates of the following path, θ i Indicates the direction of the path to follow. For path following targets, you can use (X i , Y i ,θ i ) is represented in the form of a sequence of points.

[0060] When a person is driving, the expected target will move forward along the expected driving path. Therefore, assuming that the preview target is always selected along the direction of the follow-up path, a preview target selection method is proposed. Figure 2 As shown in the figure, the Frenet coordinate system is established with the following path as the s-axis. The earth coordinates of the main vehicle at the current moment are (X e , Y e ), the vehicle speed is V, the preview distance is T P , VT P is the preview distance. First, the main vehicle's geodetic coordinates are converted into frenet coordinates (s e , d e ). The preview target is [s e ,s Tp ] The following target between mileages, where s Tp =s e +VT P That is, the mileage distance in the Frenet coordinate system is used as the preview distance. Obviously, when the mileage difference remains unchanged and the curvature of the expected following path is greater, the straight-line distance from the driver to the preview point is shorter, which is consistent with the characteristics that when the driver is driving a vehicle, the greater the road curvature, the smaller the field of view, and the shorter the forward distance.

[0061] Further, in S4, performing the preview includes:

[0062] A double-arc multi-segment trajectory calculation model is constructed, and the double-arc multi-segment trajectory calculation model is solved based on the trajectory shape free variable θ to obtain the double-arc multi-segment trajectory, wherein the trajectory shape free variable θ is determined by an objective function.

[0063] A double arc is the simplest G1 continuous curve that can connect the first and last poses. In order to ensure that the position and direction are achieved at the same time, this embodiment uses a double arc as the motion primitive for lateral motion. A , C B C is the center of the two arcs, and J is the common tangent point of the two arcs. A , C B Can be on the same side, such as Figure 4 As shown, it can also be on the opposite side as Figure 5 shown.

[0064] Define the starting position as A and the direction as θ A , end point is B, direction is θ B α is θ A The angle pointing to AB, β is the angle from AB to θ B The angle of the first arc, θ is the angle of the first arc, N1 is the normal vector of point A, V is the normal vector of point J, R1 is the radius of the first arc, and R2 is the radius of the second arc. Figure 6 shown.

[0065] The calculation model of double arc multi-segment trajectory is:

[0066]

[0067] J = A + R1 (N1 - V);

[0068] C A =A+R1N1;

[0069] C B =J+R2V;

[0070] Among them, A is the starting point, θ A is the starting direction, B is the end point, θ B is the end point direction, α is θ A The angle pointing to AB, β is the angle from AB to θ B , θ is the angle of the first arc, N1 is the normal vector of point A, V is the normal vector of point J, R1 is the radius of the first arc, and R2 is the radius of the second arc.

[0071] When α and β have the same sign, they form a C-shaped double arc, such as Figure 4 , otherwise it is constructed as an S-shaped double arc, such as Figure 5 As shown. In the above formula, θ is a free variable. For the same starting and ending positions, different θ will produce different double arc curves. During planning, θ is determined by optimizing the objective function expressed by the following formula.

[0072] The objective function is:

[0073]

[0074] L1≥L min ;

[0075] L2≥L min ;

[0076] Among them, w d is the weight of the lateral deviation, d is the preview deviation, w R is the curvature weight, κ min is the minimum curvature, κ max is the maximum curvature, L1 is the length of the first arc, L2 is the length of the second arc, L min is the minimum distance.

[0077] Human drivers do not make decisions in real time when driving, but intermittently and with tolerance. Based on this, an event-triggered planning method is proposed. First, the event trigger is calculated. When the event is triggered, the re-normalization is performed, otherwise the previous planning result is continued.

[0078] Further, in S4, the triggering of the re-preview planning is determined by a preset trigger mechanism, and the trigger mechanism is constructed according to the following deviation and the execution mileage. Specifically, this embodiment establishes the following Figure 3 The event-triggered planning method shown mainly considers the path following deviation d e Design an event trigger mechanism, the trigger mechanism is:

[0079]

[0080] Among them, s e is the current mileage of the main vehicle, d e is the path following deviation, d max is the maximum allowable path following deviation, s max is the maximum execution mileage. The trigger mechanism is to re-plan when and only when the path following deviation exceeds the design allowable deviation or exceeds the maximum execution mileage of the previous plan.

[0081] Furthermore, in S4, after triggering the re-preview planning and inserting the preview result into the turning radius-mileage list, other waypoints except the last two waypoints in the current follow-up waypoint list are deleted, a new follow-up waypoint list is obtained, and the process returns to S3.

[0082] Specifically, Figure 7 As shown, in order to solve the problem of excessive following deviation when the following target does not match the motion primitive, this embodiment adopts a multi-segment following decision method based on motion primitives. When a preview target is given, the preview target is previewed through the motion primitives, and the segmentation points are determined according to the preview error. In this process, the following error can be controlled within a reasonable error range.

[0083] For path following, first, place the waypoints (X i ,Y i ,θ i ) is inserted into the list of follow-up waypoints. Since the car needs to travel a long enough distance to perform maneuvering operations, it is determined whether the difference between the first and last mileage of the follow-up waypoint exceeds the shortest distance required for maneuvering operations. If the distance is insufficient, continue to insert waypoints until the mileage difference exceeds the shortest distance required for maneuvering operations. When the distance is sufficient, preview is performed through the objective function. If the preview tracking accuracy is less than the allowable error, continue to insert more waypoints. Otherwise, the preview result of the previous step ([(R i1 ,s i1 ), (R i2 ,s i2 )] where R i1 , R i2 are the radii of the first and second arcs in the i-th double arc, s i1 ,s i2 are the starting mileages of the first and second arcs in the i-th double arc, that is, the starting position mileage and the mileage of the common tangent point. For the convenience of expression, we will no longer distinguish which double arc it belongs to, and only mark it in the execution order.) Insert the turning radius-mileage list, and delete the remaining waypoints in the follow-up waypoint list except the last two waypoints. Repeat the above steps until the list of postures to be followed is empty. Turning radius-mileage list [(R1, s1),,…, (R n ,s n )] is the decision result.

[0084] Furthermore, in S6, according to the current mileage s of the host vehicle, e Determine the turning radius R. When s n ≤s e ≤s n+1 The turning radius is R n , the expected lateral acceleration is determined according to the turning radius and the current vehicle speed V, and the expected lateral acceleration is calculated as:

[0085]

[0086] Among them, a y is the lateral acceleration, V is the current vehicle speed, R n is the turning radius. y The target of the vehicle's direction control is transmitted to the automatic driving system, and the vehicle's steering system (such as the power steering motor) is followed by standard automatic control methods (such as PID), so that the vehicle can achieve the motion goal of following a predetermined trajectory or path.

[0087] This embodiment enables the autonomous driving car to smoothly follow the expected path by establishing an event-triggered preview method; establishes a driving preview motion primitive model in a double arc mode to ensure that the position and direction of the car are achieved simultaneously; and establishes a motion primitive-based multi-segment preview method for following errors to improve the scope of application and following accuracy of the method.

[0088] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. An event-triggered preview decision method for path following of an autonomous vehicle, characterized in that: include: S1. Obtaining an expected speed plan of the autonomous driving vehicle to be followed, and selecting a preview target according to the expected speed plan to be followed; S2. Update the list of postures to be followed according to the preview path points of the preview target; S3, inserting the waypoints in the to-be-followed posture list into the following waypoint list and deleting them, and calculating the difference between the first and last mileages in the following waypoint list, and stopping the insertion when the difference between the first and last mileages exceeds the preset shortest walking distance; S4, performing preview based on the current follow-up waypoint list. If the preview result does not meet the preset accuracy, triggering re-preview planning and inserting the preview result into the turning radius-mileage list, and returning to S3; if the preview result meets the preset accuracy, directly returning to S3; S5, repeat S3-S4 until the list of postures to be followed is empty, and output the turning radius-mileage list; S6. Based on the current driving information and turning radius-mileage list of the autonomous driving vehicle, the expected lateral acceleration is calculated and sent to the autonomous driving vehicle for following control.

2. The event-triggered preview decision method for path following of an autonomous driving vehicle according to claim 1, characterized in that: In S1, obtaining the expected speed scheme of the autonomous driving vehicle to be followed, and selecting a preview target according to the expected speed scheme to be followed includes: Obtaining following path information and a current vehicle state of the autonomous vehicle; The current vehicle state is converted into the same coordinate system as the following path to obtain the mileage range of the preview target.

3. The event-triggered preview decision method for path following of an autonomous driving vehicle according to claim 1, characterized in that: In S4, performing preview includes: A double-arc multi-segment trajectory calculation model is constructed, and the double-arc multi-segment trajectory calculation model is solved based on the trajectory shape free variable θ to obtain the double-arc multi-segment trajectory, wherein the trajectory shape free variable θ is determined by an objective function.

4. The event-triggered preview decision method for path following of an autonomous driving vehicle according to claim 3, characterized in that: The double arc multi-segment trajectory calculation model is: J = A + R1 (N1 - V); C A =A+R1N1; C B =J+R2V; Among them, A is the starting point, θ A is the starting direction, B is the end point, θ B is the end point direction, α is θ A The angle pointing to AB, β is the angle from AB to θ B , θ is the angle of the first arc, N1 is the normal vector of point A, V is the normal vector of point J, R1 is the radius of the first arc, R2 is the radius of the second arc, C A , C B are the centers of the two arcs.

5. The event-triggered preview decision method for path following of an autonomous driving vehicle according to claim 4, characterized in that: The objective function is: L1≥L min ; L2≥L min ; Among them, w d is the weight of the lateral deviation, d is the preview deviation, w R is the curvature weight, κ min is the minimum curvature, κ max is the maximum curvature, L1 is the length of the first arc, L2 is the length of the second arc, L min is the minimum distance.

6. The event-triggered preview decision method for path following of an autonomous driving vehicle according to claim 1, characterized in that: In S4, triggering the re-preview planning is determined by a preset trigger mechanism, and the trigger mechanism is constructed according to the following deviation and the execution mileage, and the trigger mechanism is: Among them, s e is the current mileage of the main vehicle, d e is the path following deviation, d max is the maximum allowable path following deviation, s max is the maximum execution mileage, and e is the trigger result.

7. The event-triggered preview decision method for path following of an autonomous driving vehicle according to claim 1, characterized in that: In the S4, after triggering the re-preview planning and inserting the preview result into the turning radius-mileage list, the other waypoints except the last two waypoints in the current follow-up waypoint list are deleted, a new follow-up waypoint list is obtained, and the process returns to the S3.

8. The event-triggered preview decision method for path following of an autonomous driving vehicle according to claim 1, characterized in that: In S6, calculating the expected lateral acceleration includes: Among them, a y is the lateral acceleration, V is the current vehicle speed, R n is the turning radius.

Images