A method and device for controlling an autonomous vehicle at an intersection

Abstract

The application provides an automatic driving vehicle control method and device at a crossroad, and relates to the technical field of automatic driving. The method comprises the following steps: acquiring a collection vehicle trajectory and front and rear lane line data to be connected at a target crossroad; performing curve fitting on the front and rear lane line data under different lane changing types to obtain initial fitting lane lines under different lane changing types; determining the offset degree between the front and rear lane lines and the initial fitting lane lines under different lane changing types, and determining a lane line matching mode at the target crossroad according to the offset degree; using the collection vehicle trajectory to optimize the connecting line under the lane line matching mode of the target initial fitting lane line to obtain an optimal connecting line, and controlling an automatic driving vehicle passing through the target crossroad according to the optimal connecting line. Through the automatic driving vehicle control method and device at the crossroad, the problem of low control accuracy of the automatic driving vehicle at the crossroad is solved.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and more specifically, to an autonomous vehicle control method and device at an intersection. Background Technology

[0002] When autonomous vehicles operate in real-world or simulated environments, their decision-making systems need to consider various environmental factors, including surrounding vehicles and pedestrians. This requires the assistance of autonomous driving perception modules for object detection and lane line detection. However, lane lines are typically absent in the middle area of ​​intersections, making lane line detection impossible. Therefore, lane connectors are needed for vehicle-assisted control. Current technologies usually determine connector constraints based on the curves of the preceding and following lanes at the intersection. For example, lane points on the lane curves are used as constraints, and then the lane connectors are determined based on these constraints.

[0003] However, when the terrain at an intersection is complex, simply determining the connecting lines directly using lane lines can easily result in connecting lines that do not match the actual situation and are not accurate enough, leading to low control accuracy of autonomous vehicles at intersections. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a method and device for controlling autonomous vehicles at intersections, so as to solve the problem of low control accuracy of autonomous vehicles at intersections.

[0005] In a first aspect, embodiments of this application provide a method for controlling an autonomous vehicle at an intersection, including:

[0006] Acquire the trajectory of the data collection vehicle and the front and rear lane lines to be connected at the target intersection. The trajectory of the data collection vehicle is the actual trajectory of the data collection vehicle when it passes through the target intersection.

[0007] Curve fitting is performed on the front and rear lane line data under different lane change types to obtain the initial fitted lane lines under different lane change types.

[0008] Determine the degree of offset between the front and rear lane lines and the initially fitted lane lines under different lane change types, and determine the lane line matching method at the target intersection based on the degree of offset.

[0009] The optimal connecting line is obtained by optimizing the initially fitted lane line of the target using the trajectory of the data acquisition vehicle and then controlling the autonomous vehicle passing through the target intersection based on the optimal connecting line.

[0010] Optionally, the lane matching method includes a constant lane matching method and a lane changing matching method. The optimal connecting line is obtained by optimizing the connecting line of the target initially fitted lane line using the collected vehicle trajectory under the lane matching method. This includes: if the lane matching method is a constant lane matching method, the curve at the connecting line position in the collected vehicle trajectory is used as the sample curve; if the lane matching method is a lane changing matching method, the curve at the connecting line position in the collected vehicle trajectory, excluding the set area corresponding to the lane changing point, is used as the sample curve; the connecting line of the target initially fitted lane line is optimized using the sample curve to obtain the optimal connecting line.

[0011] Optionally, the initial fitted lane line of the target lane is optimized using the sample curve to obtain the optimal connecting line. This includes: selecting multiple trajectory sampling points on the sample curve and obtaining the coordinates and yaw angle of each trajectory sampling point; for each trajectory sampling point, determining the point spacing between the trajectory sampling point and the corresponding connecting point, and determining the position of the connecting point corresponding to the trajectory sampling point based on the point spacing and the yaw angle of the trajectory sampling point; performing cubic spline interpolation on every two adjacent connecting points to obtain candidate connecting lines corresponding to the initial fitted lane line of the target lane; and using the candidate connecting lines to determine the optimal connecting line under different lane matching methods.

[0012] Optionally, candidate connecting lines are used to determine the optimal connecting line under different lane matching methods, including: if the lane matching method is a lane-change matching method, the candidate connecting line is directly used as the optimal connecting line; if the lane matching method is a lane-change matching method, the lane-change points are updated using the candidate connecting lines, and the convergence condition is determined based on the lane-change points; if the convergence condition is not met, the curves in the candidate connecting lines other than the set area corresponding to the updated lane-change points are used as new sample curves; the process returns to the step of selecting multiple trajectory sampling points on the sample curve and obtaining the coordinates and yaw angle of each trajectory sampling point until the candidate connecting line meets the convergence condition, and the candidate connecting line that meets the convergence condition is used as the optimal connecting line.

[0013] Optionally, the location of the connection point corresponding to the trajectory sampling point is determined based on the point spacing and the yaw angle of the trajectory sampling point, including: determining the normal angle based on the yaw angle; and taking the point on the normal angle that is a distance of point spacing from the trajectory sampling point as the connection point.

[0014] Optionally, determining the normal angle based on the yaw angle includes: if the lane matching method is a non-lane matching method, or the lane matching method is a lane changing matching method and the trajectory sampling point is before the lane changing point, the angle after deflecting the yaw angle of the trajectory sampling point by 90° in the direction of the target initially fitted lane line is taken as the normal angle; if the lane matching method is a lane changing matching method and the trajectory sampling point is after the lane changing point, the angle after deflecting the yaw angle of the trajectory sampling point by 90° in the opposite direction of the direction of the target initially fitted lane line is taken as the normal angle.

[0015] Optionally, determining the distance between the trajectory sampling point and the corresponding connection point includes: determining the front distance and the rear distance, where the front distance is the distance between the lane line before the vehicle enters the target intersection and the trajectory of the data collection vehicle, and the rear distance is the distance between the lane line after the vehicle leaves the target intersection and the trajectory of the data collection vehicle; calculating the distance difference between the rear distance and the front distance, and determining the distance between the trajectory sampling point based on the distance difference and the proximity relationship between the trajectory sampling point and the first and last trajectory sampling points.

[0016] Optionally, curve fitting is performed on the front and rear lane line data under different lane change types to obtain the initial fitted lane lines under different lane change types, including: for each lane change type, determining the rear lane line matching each front lane line at the target intersection under that lane change type; for each set of matching front and rear lane lines under that lane change type, performing cubic equation fitting on the front and rear lane line data corresponding to that set of front and rear lane lines to obtain the initial fitted lane lines corresponding to that set of front and rear lane lines under that lane change type.

[0017] Optionally, determining the degree of offset between the front and rear lane lines and the initially fitted lane lines under different lane change types includes: for each set of matched front and rear lane lines under each lane change type, determining the deviation distance between the lane line sampling points on the matched front and rear lane lines and the fitting points on the corresponding initially fitted lane lines; and determining the degree of offset based on the deviation distance and the degree of preference of vehicles at the intersection for different lane change types.

[0018] Optionally, the lane matching method at the target intersection can be determined based on the degree of offset, including selecting the lane change type with the smallest degree of offset as the lane matching method.

[0019] Secondly, embodiments of this application also provide an autonomous vehicle control device for intersections, the device comprising:

[0020] The data acquisition module is used to acquire the trajectory of the data collection vehicle and the front and rear lane lines to be connected at the target intersection. The trajectory of the data collection vehicle is the actual trajectory of the data collection vehicle when it passes the target intersection.

[0021] The curve fitting module is used to perform curve fitting on the front and rear lane line data under different lane change types, and obtain the initial fitted lane lines under different lane change types.

[0022] The matching method determination module is used to determine the degree of offset between the front and rear lane lines and the initially fitted lane lines under different lane change types, and to determine the lane line matching method at the target intersection based on the degree of offset.

[0023] The vehicle control module is used to optimize the initial fitted lane line of the target using the collected vehicle trajectory, and obtain the optimal connecting line by lane matching. Based on the optimal connecting line, the module controls the autonomous vehicle passing through the target intersection.

[0024] The embodiments of this application bring the following beneficial effects:

[0025] This application provides an autonomous vehicle control method and apparatus at an intersection, which can determine the lane matching method based on the degree of deviation between the front and rear lane lines and the fitted lane lines, thereby improving the accuracy of lane matching. Furthermore, it utilizes the collected vehicle trajectory to optimize the fitted lane lines using connection lines corresponding to the lane matching method. The collected vehicle trajectory is obtained by the vehicle passing through the intersection, reflecting the actual terrain conditions at the intersection. The obtained optimal connection line also conforms to the actual terrain, improving the accuracy of the connection line. Compared with existing autonomous vehicle control methods at intersections, this method solves the problem of low control accuracy for autonomous vehicles at intersections.

[0026] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 A flowchart of an autonomous vehicle control method at an intersection provided in an embodiment of this application is shown;

[0029] Figure 2 A schematic diagram showing the positions of the front and rear lane lines at an intersection provided in an embodiment of this application is shown;

[0030] Figure 3This illustration shows a schematic diagram of the matching relationship between front and rear lane lines under different lane change types provided in the embodiments of this application;

[0031] Figure 4 A schematic diagram of sample curves under the lane change matching method provided in the embodiments of this application is shown;

[0032] Figure 5 This diagram illustrates the positional relationship between trajectory sampling points and connection points provided in an embodiment of this application.

[0033] Figure 6 A schematic diagram of the structure of the autonomous vehicle control device at an intersection provided in an embodiment of this application is shown. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0035] It is worth noting that prior to this application, when autonomous vehicles operate in real-world or simulated environments, their decision-making systems need to consider various environmental factors, including surrounding vehicles and pedestrians. This requires the use of autonomous driving perception modules for object detection and lane line detection. However, in the middle area of ​​intersections, there are usually no lane lines to mark lanes, making lane line detection impossible. Therefore, lane connectors are needed for vehicle-assisted control. In existing technologies, connector constraints are typically determined based on the curves of the lanes before and after the intersection. For example, lane points on the lane curves are used as constraints, and then the lane connectors are determined based on these constraints. The determined lane connectors are usually in the form of quadratic curves. However, when the terrain at the intersection is complex, simply using lane lines to determine the connectors can easily result in connectors that do not match the actual situation and are not accurate enough, leading to low control accuracy of autonomous vehicles at intersections.

[0036] Based on this, this application provides a method for controlling autonomous vehicles at intersections to improve the accuracy of autonomous vehicle control at intersections.

[0037] Please see Figure 1 , Figure 1 This is a flowchart illustrating an automated vehicle control method at an intersection, as provided in an embodiment of this application. Figure 1 As shown in the embodiments of this application, the method for controlling an autonomous vehicle at an intersection includes:

[0038] Step S101: Obtain the trajectory of the data collection vehicle and the front and rear lane lines to be connected at the target intersection.

[0039] In this step, to obtain the precise connecting line at the target intersection, the actual terrain conditions at the intersection need to be taken into account. This can be achieved by controlling a data collection vehicle to pass through the target intersection and acquiring its trajectory, which can then be used to calculate the precise connecting line. The data collection vehicle trajectory is the actual path the vehicle takes as it passes through the target intersection.

[0040] In this embodiment, while acquiring the trajectory of the data acquisition vehicle, it is also necessary to acquire the front and rear lane line data to be connected at the target intersection. The front and rear lane line data include front lane line data and rear lane line data. The front lane line data is the lane line data of the front lane at the target intersection, and the rear lane line data is the lane line data of the rear lane at the target intersection. Specifically, the front lane line data includes the coordinates of multiple front lane points on the front lane line and the front lane line itself; the rear lane line data includes the coordinates of multiple rear lane points on the rear lane line and the rear lane line itself; and the data acquisition vehicle trajectory includes the coordinates of multiple trajectory points and the yaw angle of each trajectory point.

[0041] The following reference Figure 2 Let's introduce the front lane markings and the rear lane markings.

[0042] Figure 2 This application provides a schematic diagram illustrating the positions of the front and rear lane lines at an intersection, as shown in the embodiment. Figure 2 As shown, the lane lines before a vehicle enters the intersection are called the front lane lines, and the lane lines after a vehicle exits the intersection are called the rear lane lines. There are three front lane lines and three rear lane lines. It should be noted that... Figure 2 This is just one example of the relative position between front and rear lane lines to illustrate the definition of front and rear lane lines. The relative position between front and rear lane lines is not limited to a straight line or a diagonal line, but may be other situations, which will not be elaborated here.

[0043] Step S102: Perform curve fitting on the front and rear lane line data under different lane change types to obtain the initial fitted lane lines under different lane change types.

[0044] In this step, different lane change types include no lane change, left lane change, and right lane change. Among them, left lane change and right lane change are also referred to as lane change types.

[0045] Since there are multiple front and rear lane lines, and the matching relationship between front and rear lane lines differs under different lane change types, it is necessary to determine the matching rear lane line for each front lane line before generating the connecting lines at the intersection. Then, the matching relationship between the front and rear lane lines is used to generate the initial fitted lane lines.

[0046] In this embodiment, the front lane lines at the intersection are denoted as A(1),...,A(m), and the data collection vehicle is located between lane lines A(p) and A(p+1). The rear lane lines at the intersection are denoted as B(1),...,B(n), and the data collection vehicle is located between lane lines B(q) and B(q+1). The lane numbering is determined based on the relative position of the lane lines in the coordinate system of the data collection vehicle, and the leftmost lane line of the data collection vehicle is numbered 1.

[0047] Here, the matching relationship between the front and rear lane lines can be determined based on the correspondence between the front lane line where the vehicle is located before entering the target intersection and the rear lane line where the vehicle is located after exiting the target intersection. After determining the matching relationship between the front and rear lane lines under different lane change types, each set of matched front and rear lane lines can be fitted to obtain the initial fitted lane lines corresponding to each set of matched front and rear lane lines. Taking the above example, if A(2) matches B(2) and A(3) matches B(3), then the initial fitted lane lines C(2) corresponding to A(2) and B(2) and C(3) corresponding to A(3) and B(3) can be generated respectively. Among them, the initial fitted lane lines include the fitted connecting lines at the target intersection.

[0048] In one feasible implementation, step S102 includes: step a1 and step a2.

[0049] Step a1: For each lane change type, determine the matching rear lane line for each front lane line at the target intersection under that lane change type.

[0050] The following reference Figure 3 This section will introduce the matching relationship between the lane lines in front and behind under different lane change types.

[0051] Figure 3 This application provides a schematic diagram illustrating the matching relationship between lane lines under different lane change types, as shown in the embodiments of this application. Figure 3 As shown, when both the front and rear lane lines are 3 lane lines, taking the lane lines on both sides of the data collection vehicle as an example: if it is a lane change, then A(2) matches B(2) and A(3) matches B(3); if it is a left lane change (changing lanes to the left), then A(2) matches B(1) and A(3) matches B(2); if it is a right lane change (changing lanes to the right), then A(1) matches B(2) and A(2) matches B(3).

[0052] The lane matching relationship under the condition of not changing lanes is as follows:

[0053] The A(p-min(p,q)+1),...,A(p),...,A(p+min(mp,nq)) in the front lane line are matched with the B(q-min(p,q)+1),...,B(q),...,B(q+min(mp,nq)) in the rear lane line, respectively.

[0054] The lane markings under left lane change conditions are as follows:

[0055] The A(p-min(p,q+1)+1),...,A(p),...,A(p+min(mp,nq-1)) in the front lane line are matched with the B(q-min(p,q+1)+2),...,B(q+1),...,B(q+1+min(mp,nq-1)) in the rear lane line, respectively.

[0056] The lane line matching relationship under right lane change is:

[0057] The A(p-min(p,q-1)+1),...,A(p),...,A(p+min(mp,n-q+1)) in the front lane line are matched with the B(q-min(p,q-1)),...,B(q-1),...,B(q-1+min(mp,n-q+1)) in the rear lane line, respectively.

[0058] As can be seen, under the lane-change type, the rear lane line matching the front lane line A(p) is B(q); under the left lane change type, the rear lane line matching the front lane line A(p) is B(q+1); under the right lane change type, the rear lane line matching the front lane line A(p) is B(q-1).

[0059] Step a2: For each set of matching front and rear lane lines under this lane change type, perform cubic equation fitting on the front and rear lane line data corresponding to the set of front and rear lane lines to obtain the initial fitted lane lines corresponding to the set of front and rear lane lines under this lane change type.

[0060] For lane-change types, if the front lane line A(p) matches the rear lane line B(q), then five lane line sampling points are selected from multiple lane line points on A(p) and B(q). For example, five lane line sampling points are selected from multiple front lane points on the front lane line, and five lane line sampling points are selected from multiple rear lane points on the rear lane line, resulting in a total of 10 lane line sampling points. A cubic equation is then fitted to these 10 lane line sampling points to obtain the initial fitted lane lines corresponding to A(p) and B(q) under lane-change types. This process is repeated to determine the initial fitted lane lines corresponding to all matched front and rear lane lines under lane-change types.

[0061] Similarly, for left lane change and right lane change types, the initial fitted lane lines corresponding to all matching front and rear lane lines under this type are determined. Assuming that there are 4 sets of matching front and rear lane lines under each lane change type, a total of 4×3=12 initial fitted lane lines can be obtained.

[0062] Step S103: Determine the degree of offset between the front and rear lane lines and the initially fitted lane lines under different lane change types, and determine the lane line matching method at the target intersection based on the degree of offset.

[0063] In this step, the lane line matching method corresponds to the lane change type. Different lane change types correspond to different lane line matching methods, including lane change matching method, left lane change matching method, and right lane change matching method.

[0064] The degree of offset is used to characterize the overall distance between all matched preceding and following lane lines and the corresponding initial fitted lane line under a certain lane change type.

[0065] In this embodiment, to obtain a more accurate connecting line, it is necessary to determine the matching relationship between the front and rear lane lines. However, the matching method between the front and rear lane lines differs under different lane change types. Therefore, it is necessary to first determine the optimal lane change type at the target intersection. Two principles must be followed when determining the lane line matching method: First, the smaller the offset corresponding to the lane change type, the more accurate the lane line matching method corresponding to that lane change type; that is, the smaller the fitting error, the better the curve fitting effect. Second, vehicles prefer not to change lanes when passing through an intersection. In this way, the offset between the front and rear lane lines and the corresponding initially fitted lane lines can be calculated, and the lane line matching method at the target intersection can be determined based on the offset and lane change preferences.

[0066] In one feasible implementation, step S103 determines the degree of offset between the front and rear lane lines and the initially fitted lane lines under different lane change types, including steps b1 and b2.

[0067] Step b1: For each set of matched front and rear lane lines under each lane change type, determine the deviation distance between the lane line sampling points on the matched front and rear lane lines and the fitting points on the corresponding initial fitted lane lines.

[0068] For each set of matched lane lines under each lane change type, the square of the absolute value of the coordinate difference between each lane line sampling point in the matched front and rear lane lines and the fitted point on the corresponding initial fitted lane line is determined. This calculation result is called the deviation value. The sum of the deviation values ​​of all lane line sampling points in the matched front and rear lane lines is taken as the deviation distance. The sum of the deviation distances of all matched front and rear lane lines is called the deviation degree. The formula for calculating the deviation degree is:

[0069]

[0070] Where k represents the number of lane line sampling points on a single lane line. For example, if there are 5 lane line sampling points on the front lane line, then k = 5. This represents the coordinates of the i-th lane line sampling point on the matched preceding and following lane lines p′ and q′. L represents the coordinates of the i-th fitted point on the initial fitted lane line corresponding to the preceding and following lane lines p′ and q′; err Indicates the degree of deviation. The calculated result is the deviation distance.

[0071] Step b2: Determine the degree of deviation based on the deviation distance and the degree of preference of vehicles at the intersection for different lane change types.

[0072] Here, hyperparameters are set to characterize the preference of vehicles at intersections for different lane-change types. The hyperparameter is 0 when no lane is changed, and a fixed value greater than 0 is used for left and right lane changes. The degree of deviation can be calculated using the following formula:

[0073]

[0074] Where α represents the hyperparameter; L change Indicates the degree of preference; L total Indicates the degree of offset.

[0075] After determining the degree of deviation, the lane matching method corresponding to the lane change type with the smallest degree of deviation is selected as the optimal lane matching method at the target intersection.

[0076] Specifically, a larger offset indicates that the initial fitted lane lines under this lane change type are generally farther apart from the corresponding preceding and following lane lines; a smaller offset indicates that the initial fitted lane lines under this lane change type are generally more similar to the corresponding preceding and following lane lines. Therefore, the lane line matching method corresponding to the lane change type with the smallest offset is taken as the optimal lane line matching method at the target intersection, i.e., L. total The lane change type with the lowest value corresponds to the optimal lane matching method.

[0077] Step S104: Optimize the connecting line of the target initially fitted lane line using the collected vehicle trajectory under the lane line matching method to obtain the optimal connecting line, and control the autonomous vehicle passing through the target intersection according to the optimal connecting line.

[0078] In this step, after determining the lane matching method, multiple front lane lines are matched with multiple rear lane lines according to the determined lane matching method, and the initial fitted lane lines corresponding to the matched front and rear lane lines are called candidate initial fitted lane lines.

[0079] Each initially fitted lane line includes an initial connecting line, which is the line connecting the lanes in front and behind in the initially fitted lane line. Since the initially fitted lane line is a curve directly obtained from multiple fitting points, it does not take into account the actual terrain conditions at the target intersection and the impact of lane change points on the connecting lines. Therefore, it is necessary to optimize the initial connecting lines in the initially fitted lane line.

[0080] In the embodiments of this application, different lane matching methods correspond to different connection line optimization methods. If it is a lane-change matching method, the initial connection line can be directly optimized using the trajectory of the collected vehicle to obtain the optimal connection line. If it is a lane-change matching method, the initial connection line needs to be optimized by updating the trajectory of the collected vehicle after the lane change point to obtain the optimal connection line.

[0081] Since there are multiple candidate initial fitting lane lines, one is selected as the target initial fitting lane line. The initial connecting lines within this target initial fitting lane line are then optimized, and the other candidate initial fitting lane lines are optimized using the same method. Here, the candidate initial fitting lane line corresponding to one of the lane lines on either side of the lane where the data acquisition vehicle is located is selected as the target initial fitting lane line.

[0082] After determining the optimal connecting lines corresponding to all matching front and rear lane lines, these optimal connecting lines are used as the connecting lines between the front and rear lane lines at the target intersection in the high-definition map. When the autonomous vehicle is running in a real or simulated environment, its control system can control the autonomous vehicle to drive through the target intersection according to the optimal connecting lines.

[0083] In one feasible implementation, the lane matching method includes a lane-unchanging matching method and a lane-changing matching method, and step S104 includes: step c1, step c2 and step c3.

[0084] Step c1: If the lane matching method is the unchanging lane matching method, then the curve at the connecting line position in the collected vehicle trajectory is used as the sample curve.

[0085] Specifically, if the lane matching method is to remain unchanged, the curve between the lane lines at the target intersection in the trajectory of the collected vehicle is selected as the sample curve.

[0086] Step c2: If the lane line matching method is the lane change matching method, the curves at the connecting line positions in the collected vehicle trajectory, excluding the set area corresponding to the lane change point, are used as sample curves.

[0087] Here, the lane change point is the intersection between the data collection vehicle's trajectory and the candidate connecting line. The defined area is the region within a preset length before and after the lane change point, centered on it. Lane change matching methods include left lane change matching and right lane change matching.

[0088] Specifically, if the lane matching method is left lane change matching or right lane change matching, the initial lane change point is first determined. The intersection of the collected vehicle trajectory and the initial connecting line in the target initially fitted lane line is taken as the initial lane change point. The area within 5 meters before and after the initial lane change point in the collected vehicle trajectory is designated as the set area. The curve at the connecting line position in the collected vehicle trajectory, excluding the set area, is taken as the sample curve, resulting in two sample curve segments.

[0089] After updating the lane change point, the area within 5 meters before and after the new lane change point in the candidate connecting line is taken as the set area, and the curves in the candidate connecting line other than the set area are taken as sample curves.

[0090] The following reference Figure 4 Let's introduce the sample curves under the lane change matching method.

[0091] Figure 4 A schematic diagram of the sample curves under the lane change matching method provided in the embodiments of this application is shown, as follows: Figure 4 As shown, taking the left lane change matching method as an example, the curve at the connecting line position (target intersection) in the collected vehicle trajectory is divided into two parts: sample curve and non-sample curve. The sample curve includes a first sample curve and a second sample curve. The first sample curve is the sample curve outside the set area before the lane change point, i.e., the curve between the front lane line and the lane change point in the figure, represented by a hollow circle. The second sample curve is the sample curve outside the set area after the lane change point, i.e., the curve between the rear lane line and the lane change point in the figure, represented by a hollow circle. Correspondingly, the target initial fitted lane line is also divided into two parts, corresponding to the sample curve and the non-sample curve respectively. The part corresponding to the sample curve is called the interpolation curve, and the part corresponding to the non-sample curve is called the non-interpolation curve.

[0092] It should be noted that in the lane-unchanging matching method, since the entire target initial fitted lane line needs to be interpolated, there is no need to distinguish between interpolated curves and non-interpolated curves.

[0093] Step c3: Use the sample curve to optimize the initial fitted lane line of the target lane line to obtain the optimal connecting line.

[0094] Based on the sample curve, the connecting line to be optimized in the initial fitted lane line can be determined. When the lane matching method is lane change matching, this connecting line to be optimized is... Figure 4 The curve corresponding to the solid black circle in the middle. Among them, step c3 includes: step c31, step c32, step c33, and step c34.

[0095] Step c31: Select multiple trajectory sampling points on the sample curve and obtain the coordinates and yaw angle of each trajectory sampling point.

[0096] A trajectory sampling point is selected every 0.5 meters or 1 meter on the sample curve. The trajectory sampling points are denoted as P(0),...,P(M). Then, the coordinates and yaw angle of each trajectory sampling point are obtained.

[0097] Step c32: For each trajectory sampling point, determine the point spacing between the trajectory sampling point and the corresponding connection point, and determine the position of the connection point corresponding to the trajectory sampling point based on the point spacing and the yaw angle of the trajectory sampling point.

[0098] The following reference Figure 5 This section will introduce the positional relationship between trajectory sampling points and connection points.

[0099] Figure 5 This illustration shows a schematic diagram of the positional relationship between trajectory sampling points and connection points provided in an embodiment of this application, as shown below. Figure 5 As shown, the data collection vehicle travels from left to right along its trajectory. The curve between the left and right lane lines is the trajectory of the data collection vehicle, represented by a dashed line. There are multiple trajectory sampling points on the trajectory, designated P(0), P(1), P(2), P(3), P(4), and P(5). Each trajectory sampling point has multiple connection points on the connecting line corresponding to the left lane line, designated Q(0), Q(1), Q(2), Q(3), Q(4), and Q(5). The distance between the trajectory sampling points and the connection points is represented by L(M). Therefore, the distance between P(0) and Q(0) is L(0), the distance between P(1) and Q(1) is L(1), and so on, with the distance between P(5) and Q(5) being L(5). The unit of length in the planar coordinate system is meters. Q(0) is the last point on the left lane before the intersection, and also the first point on the connecting line corresponding to the left lane; Q(5) is the first point on the left lane after the intersection, and also the last point on the connecting line corresponding to the left lane.

[0100] It should be noted that the connection point is a point calculated based on the trajectory sampling points. The connection point is different from the fitting point. The fitting point is a point on the initial fitted lane line, but the connection point is not necessarily located on the initial fitted lane line.

[0101] Specifically, the distance between the front and rear points is determined. The distance between the front points is the distance between the lane line before the vehicle enters the target intersection and the trajectory of the data collection vehicle, which is the distance L(0) between the first trajectory sampling point P(0) and the first connection point Q(0). The distance between the rear points is the distance between the lane line after the vehicle leaves the target intersection and the trajectory of the data collection vehicle, which is the distance L(5) between the last sampling point P(5) and the last connection point Q(5).

[0102] Then, calculate the distance difference between the distance between the last point and the distance between the first point, i.e., L(5)-L(0), and determine the distance between the sampling points of the trajectory based on the distance difference and the proximity between the sampling points of the trajectory and the sampling points of the first and last trajectories.

[0103] For example, i / M represents the distance between the trajectory sampling point and the first and last trajectory sampling points. For the first trajectory sampling point P(0), i = 0. Thus, the distance between the first trajectory sampling point and the first and last trajectory sampling points is 0 / 5 = 0, the distance between the second trajectory sampling point and the first and last trajectory sampling points is 1 / 5 = 0.2, and so on, until the distance between the last trajectory sampling point and the first and last trajectory sampling points is 1. In this way, the point spacing can be expressed as: L(i) = L(0) + i / M × (L(M) - L(0)). For example, the point spacing corresponding to the second trajectory sampling point is L(1) = L(0) + 1 / 5 × (L(5) - L(0)).

[0104] The normal angle is determined based on the yaw angle, and the point on the normal angle that is the distance between the sampling point and the sampling point of the trajectory is taken as the connection point of the sampling point of the trajectory.

[0105] Specifically, if the lane matching method is a non-lane matching method, or if the lane matching method is a lane change matching method and the trajectory sampling point is before the lane change point, the yaw angle of the trajectory sampling point is deflected by 90° towards the direction of the target initial fitted lane line as the normal angle. The direction of the target initial fitted lane line is the direction of the fitting point corresponding to the trajectory sampling point on the target initial fitted lane line. The fitting point at the corresponding position can be determined based on the order of the fitting points' positions on the target initial fitted lane line.

[0106] Taking trajectory sampling point P(1) as an example, if the lane matching method is no lane change and the target initial trajectory lane line is the left lane line of the collection vehicle, then yaw(1)+π / 2 is taken as the normal angle of trajectory sampling point P(1), where yaw(1) represents the yaw angle of trajectory sampling point P(1). If the lane matching method is left lane change matching method and the trajectory sampling point P(1) is before the lane change point, then yaw(1)+π / 2 is also taken as the normal angle of trajectory sampling point P(1). If the lane matching method is right lane change matching method and the trajectory sampling point P(1) is before the lane change point, then yaw(1)-π / 2 is also taken as the normal angle of trajectory sampling point P(1). Then, the point Q(1) at the angle yaw(1)+π / 2 that is a distance L(1) from the trajectory sampling point P(1) is taken as the connection point.

[0107] If the lane matching method is lane change matching method and the trajectory sampling point is after the lane change point, the yaw angle of the trajectory sampling point is deflected by 90° in the opposite direction of the direction of the target initially fitted lane line as the normal angle.

[0108] Taking trajectory sampling point P(1) as an example, if the lane matching method is left lane change matching method and the trajectory sampling point P(1) is after the lane change point, then yaw(1)-π / 2 is taken as the normal angle of trajectory sampling point P(1). If the lane matching method is right lane change matching method and the trajectory sampling point P(1) is after the lane change point, then yaw(1)+π / 2 is taken as the normal angle of trajectory sampling point P(1), where yaw(1) represents the yaw angle of trajectory sampling point P(1). Then, the point Q(1) that is at the yaw(1)-π / 2 angle and is a distance L(1) from the trajectory sampling point P(1) is taken as the connection point.

[0109] Step c33: Perform cubic spline interpolation for each pair of adjacent connection points to obtain candidate connection lines corresponding to the target initial fitted lane line.

[0110] Cubic spline interpolation is performed on Q(0) and Q(1), Q(1) and Q(2), Q(2) and Q(3), Q(3) and Q(4), and Q(4) and Q(5) respectively to obtain multiple interpolation curves. If the lane matching method is the non-lane matching method, all interpolation curves are connected together to obtain the optimized candidate connecting line corresponding to the target initial fitted lane line; if the lane matching method is the lane changing matching method, all interpolation curves are connected together with the non-interpolation curves in the target initial fitted lane line to obtain the optimized candidate connecting line corresponding to the target initial fitted lane line.

[0111] It should be noted that after the candidate connection is determined for the first time, the candidate connection line is used as a new sample curve. At this time, the candidate connection line is divided into two parts: interpolation curve and non-interpolation curve. Then, all the new interpolation curves obtained by cubic spline interpolation are connected together with the non-interpolation curves in the candidate connection line to obtain the optimized candidate connection line.

[0112] Step c34: Using candidate connecting lines, determine the optimal connecting line under different lane line matching methods.

[0113] If the lane matching method is the same as the lane matching method, then the candidate connecting line is directly used as the optimal connecting line.

[0114] If the lane line matching method is lane change matching method, then the lane change point is updated using the candidate connecting line. Based on the lane change point, it is determined whether the candidate connecting line meets the convergence condition. If the convergence condition is not met, the curves in the candidate connecting line other than the set area corresponding to the updated lane change point are taken as new sample curves. Then, the execution of step c31 is returned until the candidate connecting line meets the convergence condition, and the candidate connecting line that meets the convergence condition is taken as the optimal connecting line.

[0115] Here, the intersection of the vehicle trajectory and the candidate connecting line is taken as the new lane change point. It is determined whether the distance between the new lane change point and the previous lane change point is less than the distance threshold. If it is less than the distance threshold, the candidate connecting line is determined to meet the convergence condition. If it is greater than or equal to the distance threshold, the candidate connecting line is determined not to meet the convergence condition.

[0116] Alternatively, based on the original lane change point and multiple points on the previous round of candidate connecting lines, the new lane change point and multiple points on the new candidate connecting lines, determine the distance between the two candidate connecting lines, and determine whether the distance between the lines is less than the distance between the lines threshold. If it is less than the distance between the lines threshold, the candidate connecting line is determined to meet the convergence condition. If it is greater than or equal to the distance between the lines threshold, the candidate connecting line is determined to not meet the convergence condition.

[0117] Compared with existing methods for controlling autonomous vehicles at intersections, this application can determine the lane matching method based on the degree of deviation between the lane line and the fitted lane line, thereby improving the accuracy of lane matching. Furthermore, it uses the collected vehicle trajectory to optimize the fitted lane line with the corresponding connecting line according to the lane matching method. The collected vehicle trajectory is obtained by the vehicle passing through the intersection, reflecting the actual terrain at the intersection. The obtained optimal connecting line can also conform to the actual terrain, improving the accuracy of the connecting line and solving the problem of low control accuracy of autonomous vehicles at intersections.

[0118] Based on the same inventive concept, this application also provides an autonomous vehicle control device for intersections, which corresponds to the autonomous vehicle control method for intersections. Since the principle of the device in this application is similar to the autonomous vehicle control method for intersections described above, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0119] Please see Figure 6 , Figure 6 This is a schematic diagram of the structure of an automated driving vehicle control device at an intersection, provided as an embodiment of this application. Figure 6 As shown, the automated vehicle control device 200 at the intersection includes:

[0120] The data acquisition module 201 is used to acquire the trajectory of the data collection vehicle and the front and rear lane lines to be connected at the target intersection. The trajectory of the data collection vehicle is the actual trajectory of the data collection vehicle when it passes the target intersection.

[0121] The curve fitting module 202 is used to perform curve fitting on the front and rear lane line data under different lane change types to obtain the initial fitted lane lines under different lane change types.

[0122] The matching method determination module 203 is used to determine the degree of offset between the front and rear lane lines and the initially fitted lane lines under different lane change types, and to determine the lane line matching method at the target intersection based on the degree of offset.

[0123] The vehicle control module 204 is used to optimize the connecting line of the target initial fitted lane line by using the collected vehicle trajectory in the lane line matching mode to obtain the optimal connecting line, and to control the autonomous vehicle passing through the target intersection according to the optimal connecting line.

[0124] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0125] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0126] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0127] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0128] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0129] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for controlling an automated vehicle at an intersection, characterized in that, include: Acquire the trajectory of the data collection vehicle and the front and rear lane lines to be connected at the target intersection. The trajectory of the data collection vehicle is the actual trajectory of the data collection vehicle when it passes the target intersection. Curve fitting is performed on the front and rear lane line data under different lane change types to obtain the initial fitted lane lines under different lane change types. Determine the degree of offset between the front and rear lane lines and the initially fitted lane lines under different lane change types, and determine the lane line matching method at the target intersection based on the degree of offset. The initial fitted lane line of the target is optimized by using the trajectory of the collected vehicle to obtain the optimal connecting line under the lane line matching method. The autonomous driving vehicle passing through the target intersection is controlled according to the optimal connecting line. The step of optimizing the initial fitted lane line using the collected vehicle trajectory to obtain the optimal connecting line under the lane line matching method includes: According to the lane line matching method, the curves at different positions in the trajectory of the collected vehicle are used as sample curves; Multiple trajectory sampling points are selected on the sample curve. For each selected trajectory sampling point on the sample curve, the position of the corresponding connection point is determined based on the point spacing between the trajectory sampling point and the connection point at the corresponding position and the yaw angle of the trajectory sampling point. For each pair of adjacent connection points, cubic spline interpolation is performed to obtain the candidate connection lines corresponding to the target initially fitted lane lines. If the lane line matching method is the lane change matching method, then the lane change point is updated using the candidate connecting line, and it is determined whether the candidate connecting line meets the convergence condition based on the lane change point. If the convergence condition is not met, the curves in the candidate connecting lines other than the set area corresponding to the updated lane change point are taken as new sample curves, and the process of selecting multiple trajectory sampling points on the sample curves is returned until the candidate connecting lines meet the convergence condition. The candidate connecting lines that meet the convergence condition are then taken as the optimal connecting lines.

2. The method according to claim 1, characterized in that, The lane matching method includes a lane-unchanging matching method and a lane-changing matching method. The step of using curves at different positions in the collected vehicle trajectory as sample curves according to the lane matching method includes: If the lane matching method is the unchanging lane matching method, then the curve at the connecting line position in the collected vehicle trajectory is used as the sample curve; If the lane matching method is the lane change matching method, the curves at the connecting line positions in the collected vehicle trajectory, excluding the set area corresponding to the lane change point, are used as sample curves.

3. The method according to claim 1, characterized in that, The method further includes: If the lane matching method is a non-lane matching method, then the candidate connecting line is directly used as the optimal connecting line.

4. The method according to claim 1, characterized in that, The step of determining the position of the connection point corresponding to the trajectory sampling point based on the point spacing between the trajectory sampling point and the corresponding connection point, and the yaw angle of the trajectory sampling point, includes: Determine the normal angle based on the yaw angle; The point at the normal angle that is at a distance of the specified point spacing from the trajectory sampling point is taken as the connection point.

5. The method according to claim 4, characterized in that, Determining the normal angle based on the yaw angle includes: If the lane matching method is a lane-change matching method, or if the lane matching method is a lane-change matching method and the trajectory sampling point is before the lane-change point, then the yaw angle of the trajectory sampling point after being deflected 90° in the direction of the target initial fitted lane line is taken as the normal angle. If the lane matching method is the lane change matching method and the trajectory sampling point is after the lane change point, the yaw angle of the trajectory sampling point is deflected by 90° in the opposite direction to the direction of the target initial fitted lane line as the normal angle.

6. The method according to claim 1, characterized in that, The distance between the sampling points of the trajectory and the connection points at their corresponding locations is determined using the following method: Determine the front point spacing and the rear point spacing. The front point spacing is the distance between the lane line before the vehicle enters the target intersection and the trajectory of the data collection vehicle. The rear point spacing is the distance between the lane line after the vehicle leaves the target intersection and the trajectory of the data collection vehicle. Calculate the distance difference between the rear point distance and the front point distance, and determine the point distance of the trajectory sampling point based on the distance difference and the proximity relationship between the trajectory sampling point and the first and last trajectory sampling points.

7. The method according to claim 1, characterized in that, The step of performing curve fitting on the front and rear lane line data under different lane change types to obtain the initial fitted lane lines under different lane change types includes: For each lane change type, determine the matching rear lane line for each front lane line at the target intersection under that lane change type; For each set of matching front and rear lane lines under this lane change type, a cubic equation is fitted to the front and rear lane line data corresponding to this set of front and rear lane lines to obtain the initial fitted lane lines corresponding to this set of front and rear lane lines under this lane change type.

8. The method according to claim 1, characterized in that, Determining the degree of offset between the preceding and following lane lines and the initially fitted lane lines under different lane change types includes: For each set of matched front and rear lane lines under each lane change type, determine the deviation distance between the lane line sampling point on the matched front and rear lane lines and the fitting point on the corresponding initial fitted lane line. The degree of deviation is determined based on the deviation distance and the degree of preference of vehicles at the intersection for different lane-changing types.

9. An automated driving vehicle control device for an intersection, characterized in that, include: The data acquisition module is used to acquire the trajectory of the data collection vehicle and the front and rear lane lines to be connected at the target intersection. The trajectory of the data collection vehicle is the actual trajectory of the data collection vehicle when it passes the target intersection. The curve fitting module is used to perform curve fitting on the front and rear lane line data under different lane change types to obtain the initial fitted lane lines under different lane change types. The matching method determination module is used to determine the degree of offset between the front and rear lane lines and the initially fitted lane lines under different lane change types, and to determine the lane line matching method at the target intersection based on the degree of offset. The vehicle control module is used to optimize the initial fitted lane line of the target using the collected vehicle trajectory, and to obtain the optimal connecting line by lane line matching. Based on the optimal connecting line, the module controls the autonomous vehicle passing through the target intersection. The vehicle control module is specifically used for: According to the lane line matching method, the curves at different positions in the trajectory of the collected vehicle are used as sample curves; Multiple trajectory sampling points are selected on the sample curve. For each selected trajectory sampling point on the sample curve, the position of the corresponding connection point is determined based on the point spacing between the trajectory sampling point and the connection point at the corresponding position and the yaw angle of the trajectory sampling point. For each pair of adjacent connection points, cubic spline interpolation is performed to obtain the candidate connection lines corresponding to the target initially fitted lane lines. If the lane line matching method is the lane change matching method, then the lane change point is updated using the candidate connecting line, and it is determined whether the candidate connecting line meets the convergence condition based on the lane change point. If the convergence condition is not met, the curves in the candidate connecting lines other than the set area corresponding to the updated lane change point are taken as new sample curves, and the process of selecting multiple trajectory sampling points on the sample curves is returned until the candidate connecting lines meet the convergence condition. The candidate connecting lines that meet the convergence condition are then taken as the optimal connecting lines.

Images