Automatic emergency braking key target selection method
By integrating lane lines, target trajectories and bicycle dynamic trajectories, the collision time and probability are calculated, and the existing AEB technology has solved the problem of degradation in the lane line perception or curved conditions, achieving higher accuracy and safety in AEB target selection.
Patent Information
- Application Number
- CN202510329525.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-20
- Publication Date
- 2025-06-27
AI Technical Summary
The existing automatic emergency braking (AEB) key target selection technology has reduced accuracy in the case of poor lane line perception or curves, which can easily lead to false detection and false braking.
By obtaining perceived target information, lane line information and vehicle motion information, using lane line, target trajectory and vehicle dynamic trajectory for fusion estimates, calculating collision time and collision probability, and then selecting key AEB targets.
It improves the estimation accuracy of the bicycle motion trajectory, reduces the error detection rate of AEB target selection, and ensures the accurate trigger of the AEB function in complex scenarios.
Smart Images
Figure CN120207320A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of intelligent driving technology, and in particular to a method for selecting key targets for automatic emergency braking. Background Art
[0002] Autonomous Emergency Braking (AEB) is an automotive active safety technology that warns the driver and takes corresponding braking measures after detecting potential collision risks to avoid or mitigate collision accidents. The selection of AEB key targets is a key prerequisite for the accurate triggering of the AEB function. Currently, the main methods for selecting AEB key targets include: 1. Judging whether there is a collision risk between the host vehicle and the target based on the movement trajectory of the host vehicle and the movement state of the target, and then selecting the AEB target; 2. Determining the AEB target based on the probability of each target being within the host vehicle's lane and the overlap rate between the target and the host vehicle.
[0003] In the existing AEB key target selection technology, the calculation of the probability of a target being within the host vehicle's lane depends on the accurate recognition of lane lines. In some scenarios where lane lines are poorly perceived (lane lines are worn, or one-sided lane lines are missing), the accuracy of AEB target selection decreases; and false detections of AEB targets are prone to occur in curves, resulting in false braking of the vehicle. Summary of the Invention
[0004] The present invention aims to solve the problems existing in the background art.
[0005] To this end, the present invention provides a method for selecting key targets for automatic emergency braking.
[0006] The technical solution adopted by the present invention to solve its technical problems is as follows:
[0007] A method for selecting key targets for automatic emergency braking, comprising the following steps:
[0008] S1. Obtain perception target information, lane line information, and vehicle movement information;
[0009] S2. Use lane lines, target trajectories, and the dynamic trajectory of the host vehicle to perform fusion estimation on the movement trajectory of the host vehicle;
[0010] S3. Calculate the time to collision and the probability of collision according to the movement trajectory of the host vehicle and each target trajectory;
[0011] S4. Calculate the time to collision and the probability of collision according to the current movement states of the host vehicle and each target;
[0012] S5. Complete the selection of AEB key targets according to the time to collision and the probability of collision calculated in S3 and S4.
[0013] Further, in the step S2, the nearest target trajectory is determined, and the historical ego-vehicle motion trajectory, the left lane line, the right lane line, the ego-vehicle dynamics trajectory, and the nearest target trajectory are translated to pass through the origin of the ego-vehicle coordinate system. For each trajectory, feature points are sampled at equal intervals with the same polar radius, and using the polynomial of each trajectory, the polar angles of multiple feature points corresponding to each trajectory are calculated. Then, according to each trajectory error model, the variance of the polar angle of each point is calculated. The polar angle of each feature point of the ego-vehicle motion trajectory is filtered and updated, and the filtered and updated multiple feature points of the ego-vehicle motion trajectory are represented in Cartesian coordinates after being converted from polar coordinates, and a cubic curve fitting is performed to obtain the current ego-vehicle motion trajectory curve polynomial.
[0014] Further, in the step S3, according to the longitudinal relative velocity and longitudinal relative distance between the ego-vehicle and the target, the predicted time to collision (TTC) and the collision point position are calculated.
[0015] Further, in the step S3, according to the lateral position variance of the ego-vehicle at the position after the predicted time to collision and the lateral position variance of the target collision point, the lateral position variance of the collision point relative to the ego-vehicle position is calculated. The lateral position y of the target satisfies a Gaussian distribution with a mean of y c1 and a standard deviation of σ1. The collision probability between the ego-vehicle and the target is calculated according to the following formula: where l and r respectively represent the left and right boundaries of the collision region, l = -0.5·object_width - safety distance, r = 0.5·object_width + safety distance, and safety distance is the safety distance from the target.
[0016] Further, in the step S4, according to the lateral and longitudinal relative velocities and relative accelerations of the current target, the position of the target after the target motion TTC is calculated based on the constant acceleration motion model. Combining with the angle rotated by the ego-vehicle at the current yaw rate within the TTC, the lateral position y of the target in the ego-vehicle coordinate system after the TTC is obtained after coordinate rotation c2 , and according to the standard deviations of the target's current position, relative velocity, and acceleration, and the standard deviations of the ego-vehicle's speed and yaw rate, based on the error propagation principle, the standard deviation σ2 of the lateral position of the target after the target motion TTC is calculated. The lateral position y of the target satisfies a Gaussian distribution with a mean of y c2 and a standard deviation of σ2. The collision probability between the ego-vehicle and the target is calculated according to the following formula: where l and r respectively represent the left and right boundaries of the collision region, l = -0.5·object_width - safety distance, r = 0.5·object_width + safety distance, and safety distance is the safety distance from the target.
[0017] Further, in the step S5, according to the collision probability of the target, the conditions for each target to be a candidate for the AEB key target are as follows: If the target was not selected as the AEB key target at the previous moment, then the P of this target c1 and P c2 need to be greater than the set probability threshold p th1 ; if the target was selected as the AEB key target at the previous moment, then the P of this target c1 and P c2 need to be greater than the set probability threshold p th2 ; the setting of p th2 is less than p th1 , to ensure that the AEB key target at the previous moment has a greater possibility of being a candidate for the current AEB key target; among all the candidates for the AEB key target, select the target with the smallest TTC as the AEB key target.
[0018] The beneficial effects of the present invention are that this application uses multiple trajectory curves including lane lines to perform fusion estimation on the motion trajectory of the host vehicle, improving the estimation accuracy of the motion trajectory of the host vehicle; and calculates the collision probability with the target based on the motion trajectory of the host vehicle and the current motion state of the host vehicle respectively. When both probabilities meet the conditions, this target can be used as a candidate for the AEB key target, reducing the false detection rate. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] The present invention will be further described below with reference to the drawings and embodiments.
[0020] Figure 1 is a schematic flowchart of the method for selecting the key target for automatic emergency braking in the present invention.
[0021] Figure 2 is a schematic diagram of determining the nearest target trajectory in the present invention.
[0022] Figure 3 is a schematic diagram of updating the motion trajectory of the host vehicle based on the left lane line in the present invention.
[0023] Figure 4 is a schematic diagram of the target position after TTC in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0024] The present invention will now be described in further detail with reference to the drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.
[0025] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, the features defined as "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, the meaning of "a plurality" is two or more.
[0026] In the description of the present invention, it should be noted that unless otherwise clearly specified and limited, the terms "installation", "connection", and "connection" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.
[0027] Referring to Figure 1 , an automatic emergency braking key target selection method includes the following steps:
[0028] S1, obtaining perception target information, lane line information, and vehicle motion information.
[0029] The perception target information specifically includes: the number of valid targets, the ID of each target, longitudinal distance, lateral distance, longitudinal relative speed, lateral relative speed, acceleration, heading angle, width, confidence level, motion state, etc.
[0030] The lane line information includes the cubic polynomial coefficients of the left lane line and the right lane line. The vehicle motion information specifically includes: the speed of the host vehicle, yaw angular velocity, and longitudinal acceleration.
[0031] S2, using the lane line, target trajectory, and host vehicle dynamic trajectory to perform a fusion estimation on the host vehicle motion trajectory.
[0032] S201. If the historical ego-vehicle motion trajectory does not exist, initialize the ego-vehicle motion trajectory. The initial trajectory is a straight line, discretely represented as a number of equally spaced feature points in polar coordinates, with the polar angle of each point being 0°, and set the initial polar angle variance of each feature point. If the historical ego-vehicle motion trajectory already exists, convert the trajectory into the expression form in the current ego-vehicle coordinate system according to the current motion speed and yaw angular velocity of the ego-vehicle.
[0033] S202. Fit the target trajectory to a cubic polynomial curve according to the current position and historical position of the perceived target, and determine the nearest target trajectory.
[0034] Perform longitudinal equally spaced sampling on the ego-vehicle motion trajectory and each target trajectory respectively. Calculate the abscissa of the sampling points according to the expression of each trajectory curve, calculate the lateral distance difference between the sampling points of the target trajectory and the ego-vehicle motion trajectory and take the average. The trajectory (target 1 trajectory) with the smallest average lateral distance difference is the nearest target trajectory. Refer to Figure 2 .
[0035] It should be noted that the number of valid historical position points participating in the curve fitting is not less than 5. Targets with less than 5 historical position points are not candidates for the nearest target trajectory.
[0036] Translate the historical ego-vehicle motion trajectory, the left lane line, the right lane line, the ego-vehicle dynamics trajectory, and the nearest target trajectory to pass through the origin of the ego-vehicle coordinate system. For each trajectory, perform feature point sampling with the same equally spaced polar radius, calculate the polar angle of multiple feature points corresponding to each trajectory using the polynomial of each trajectory, and calculate the variance of the polar angle of each point according to the lateral error model of each trajectory.
[0037] The trajectory lateral error model is k·r 4 , where k is the proportional error coefficient and r is the polar radius. The trajectory lateral position error is proportional to r 4 . Calculate the lateral position error according to the polar radius of the trajectory feature points, and then approximately convert it into the polar angle variance.
[0038] It should be noted that all trajectories are cubic polynomial curves, and their general formula is y = C0 + C1·x + C2·x 2 + C3·x 3 ; where x represents the longitudinal distance, y represents the lateral distance, C0 is the lateral offset between the trajectory and the ego-vehicle center, C1 is the tangent value of the angle between the trajectory and the driving direction at the current position of the ego-vehicle, C2 is the coefficient related to the current curvature of the trajectory, and C3 is the coefficient related to the curvature change rate of the trajectory. The ego-vehicle kinematic trajectory is calculated based on the ego-vehicle speed and yaw angular velocity.
[0039] S204. For each feature point of the historical ego-vehicle motion trajectory, using the polar angles of the feature points with the same polar radius on other trajectories as measurements, the polar angles of the feature points are filtered and updated using a one-dimensional Kalman filter algorithm.
[0040] If the left lane line, right lane line, ego-vehicle dynamics trajectory, and nearest target trajectory are all valid, each feature point of the historical ego-vehicle motion trajectory has 4 polar angle measurements and 4 filtering updates are required.
[0041] S205. Convert the multiple feature points of the ego-vehicle motion trajectory after filtering and updating from polar coordinates to rectangular coordinates, and perform a cubic curve fitting to obtain the current ego-vehicle motion trajectory curve polynomial.
[0042] Taking the left lane line as an example, the update of the ego-vehicle motion trajectory based on the left lane line is as Figure 3 shown. After translating the left lane line to the coordinate origin, sample the historical ego-vehicle motion trajectory and the translated left lane line trajectory with the same polar radius to obtain multiple pairs of feature points with the same polar radius and different polar angles. Use the feature points of the left lane line to filter and update the feature points of the historical ego-vehicle motion trajectory, and then perform a cubic curve fitting on the updated feature points to obtain the current ego-vehicle motion trajectory.
[0043] S3. Calculate the time to collision and collision probability based on the ego-vehicle motion trajectory and each target trajectory.
[0044] S301. Calculate the predicted time to collision (TTC) based on the longitudinal relative velocity and longitudinal distance of each target. For targets whose TTC meets the set conditions, continue with the following steps.
[0045] S302. Based on the speeds of the current ego-vehicle and the target, calculate the positions of the ego-vehicle and the target after moving along their respective motion trajectories for TTC, and obtain the longitudinal position x c1 and lateral position y c1 of the target in the ego-vehicle coordinate system (when the ego-vehicle moves to point o) after TTC, as Figure 4 shown.
[0046] S303. Based on point o and the polar radii of the feature points of the ego-vehicle motion trajectory, determine the interval of the feature points of the ego-vehicle motion trajectory where point o is located. Figure 4 If point o is located between feature points p1 and p2, calculate the polar angle variance of point o using linear interpolation according to the polar angle variances of p1 and p2, and then combine with the polar radius of point o to approximately obtain its lateral position variance σ o 2 ; The target trajectory is generated by fitting the historical positions of the target, and the fitting error is used as the standard deviation σ f of the lateral position of the target after moving for TTC; Then the lateral position variance σ1 2For
[0047] S304. The target lateral position y follows a Gaussian distribution with a mean of y c1 and a standard deviation of σ1. Calculate the collision probability between the host vehicle and the target according to the following formula:
[0048]
[0049] where l and r respectively represent the left and right boundaries of the collision area, l = -0.5·object_width - safety distance, r = 0.5·object_width + safety distance, and the safety distance can be configured separately for different target types. For example, the safety distance for vehicle targets can be configured as 1.5m, and the safety distance for pedestrian targets can be configured as 1.2m.
[0050] S4. Calculate the time to collision (TTC) and the collision probability according to the current motion states of the host vehicle and each target.
[0051] S401. Calculate the TTC based on the longitudinal relative velocity and longitudinal distance of each target. For targets that meet the set conditions for TTC, continue with the following steps.
[0052] S402. Based on the lateral and longitudinal relative velocities and relative accelerations of the current target, calculate the position of the target after TTC based on the constant acceleration motion model. Since the host vehicle has an angular velocity, perform a coordinate rotation according to the angle rotated by the host vehicle after moving for TTC at the current yaw angular velocity. After converting the calculated position of the target in the current host vehicle coordinate system, obtain the lateral position y of the target in the host vehicle coordinate system after TTC. c2 .
[0053] S403. The perception can provide the standard deviations of the current position, relative velocity, and acceleration of the target. The standard deviations of the host vehicle speed and yaw angular velocity are obtained through statistical analysis of test data. According to the above error parameters and based on the error propagation principle, calculate the standard deviation σ2 of the lateral position of the target after TTC.
[0054] S404. The target lateral position y follows a Gaussian distribution with a mean of y c2 and a standard deviation of σ2. Calculate the collision probability between the host vehicle and the target according to the following formula:
[0055]
[0056] where l and r respectively represent the left and right boundaries of the collision area, and their settings are the same as in step S303.
[0057] S5. Based on the collision time and collision probability calculated in S3 and S4, complete the selection of AEB key targets.
[0058] S501. According to the collision probability of the target, the conditions for each target to be a candidate for the AEB key target are as follows:
[0059] If the target was not selected as the AEB key target at the previous moment, then the P of this target c1 and P c2 need to be greater than the set probability threshold p th1 ;
[0060] If the target was selected as the AEB key target at the previous moment, then the P of this target c1 and P c2 need to be greater than the set probability threshold p th2 ;
[0061] The setting of p th2 is less than p th1 to ensure that the AEB key target at the previous moment has a greater possibility of being a candidate for the current AEB key target.
[0062] S502. Among all the candidates for the AEB key target, select the target with the minimum TTC as the AEB key target.
[0063] Inspired by the ideal embodiments of the present invention described above, through the above description, relevant staff can completely make various changes and modifications without departing from the technical idea of this invention. The technical scope of this invention is not limited to the content in the specification, and its technical scope must be determined according to the scope of the claims.
Claims
1. A method for selecting a key target for automatic emergency braking, characterized in that: The following steps are included: S1, obtains the perceived target information, lane line information and vehicle motion information; S2, using lane lines, target trajectory and vehicle dynamics trajectory, to estimate the vehicle motion trajectory; S3, calculating the collision time and collision probability according to the ego vehicle motion trajectory and each target trajectory; S4, calculating the collision time and collision probability according to the current motion state of the ego vehicle and each target; S5, completing AEB key target selection according to the collision time and collision probability calculated in S3 and S4.
2. The automatic emergency braking key target selection method according to claim 1, characterized in that: In the step S2, the nearest target trajectory is determined, the historical ego-vehicle motion trajectory, the left lane line, the right lane line, the ego-vehicle dynamics trajectory, and the nearest target trajectory are translated to pass through the origin of the ego-vehicle coordinate system, and for each trajectory, feature points are sampled at equal intervals with the same polar diameter, and the polar angles of multiple feature points corresponding to each trajectory are calculated using each trajectory polynomial, and the variance of the polar angle of each point is calculated according to each trajectory error model, and the polar angle of each feature point of the ego-vehicle motion trajectory is filtered and updated, and the multiple ego-vehicle motion trajectory feature points that have been filtered and updated are converted from polar coordinates to rectangular coordinates, and cubic curve fitting is performed to obtain the current ego-vehicle motion trajectory curve polynomial.
3. The automatic emergency braking key target selection method according to claim 1, characterized in that: In step S3, the estimated collision time is calculated based on the longitudinal relative speed and longitudinal relative distance between the vehicle and the target, and the estimated collision time and collision point position are calculated.
4. The automatic emergency braking key target selection method according to claim 3, characterized in that: In step S3, the lateral position variance of the collision point relative to the position of the vehicle is calculated based on the lateral position variance of the position of the vehicle after the estimated collision time and the lateral position variance of the target collision point. The lateral position y of the target satisfies the mean value y c1 , a Gaussian distribution with a standard deviation of σ1, and the collision probability between the vehicle and the target is calculated according to the following formula: Among them, l and r represent the left and right boundaries of the collision area respectively, l = –0.5·object_width–safety distance, r = 0.5·object_width+safety distance, and safety distance is the safety distance between the object and the target.
5. The automatic emergency braking key target selection method according to claim 4, characterized in that: In step S4, according to the current lateral and longitudinal relative speeds and relative accelerations of the target, the position of the target after TTC is calculated based on the constant acceleration motion model, and the lateral position y of the target in the ego vehicle coordinate system after TTC is obtained after the coordinate system is rotated in combination with the angle of rotation of the ego vehicle at the current yaw rate. c2 According to the standard deviation of the target's current position, relative velocity, acceleration, and the standard deviation of the vehicle's velocity and yaw rate, the standard deviation of the lateral position after the target motion TTC is calculated based on the error transmission principle. The lateral position y of the target satisfies the mean value y c2 , a Gaussian distribution with a standard deviation of σ2, and the collision probability between the vehicle and the target is calculated according to the following formula: Among them, l and r represent the left and right boundaries of the collision area respectively, l = –0.5·object_width–safety distance, r = 0.5·object_width+safetydistance, and safety distance is the safety distance between the target.
6. The automatic emergency braking key target selection method according to claim 1, characterized in that: In step S5, according to the collision probability of the target, the conditions for each target to be selected as an AEB key target candidate are as follows: if the target was not selected as an AEB key target at the previous moment, then the P of the target is c1 With P c2 Must be greater than the set probability threshold p th1 ; If the target was selected as an AEB key target at the last moment, the P of the target c1 With P c2 Must be greater than the set probability threshold p th2 ;p th2 The setting is less than p th1 , to ensure that the AEB key target at the previous moment has a greater chance of being the current AEB key target candidate; among all AEB key target candidates, the target with the smallest TTC is selected as the AEB key target.