Front target vehicle cut-in identification method with pre-estimation function

Abstract

The invention discloses a front target vehicle cut-in identification method with a pre-estimation function, and the improved front target vehicle cut-in identification method is verified by a real vehicle to accurately identify a cut-in target vehicle in a front adjacent lane no matter whether the vehicle is in the middle of a lane, stably deviates from the left side, stably deviates from the right side or turns or changes the lane. And the switching of the vehicle can be estimated in advance, and the information of the target vehicle switched into the lane is sent to the ACC controller in advance, so that the control of the ACC controller is more agile, and the longitudinal control of the vehicle is more comfortable. After the method is adopted, the sudden brake condition of the vehicle caused by the cut-in of the front vehicle can be avoided.

Description

Technical Field

[0001] The present invention belongs to the field of vehicle longitudinal control in intelligent driving, and in particular relates to a front target vehicle cut-in recognition method with a prediction function. Background Art

[0002] The main function of longitudinal control in the intelligent driving system is the adaptive cruise control (ACC). When the driver turns on the ACC function, the auxiliary driving controller controls the vehicle's brakes and throttle and the longitudinal speed of the vehicle according to the cruising speed set by the driver and the information of the target vehicle ahead detected by the camera or millimeter-wave radar.

[0003] When the ACC function is turned on, the target screening algorithm will screen out the target of the vehicle ahead based on the target information detected by the sensor, the lane line information and the movement of the vehicle. The accuracy of target screening has a great impact on the control of ACC. Misselection of the target ahead often causes unnecessary deceleration of ACC. Selecting the target too late will cause the vehicle to decelerate significantly after the target is selected. Especially when the vehicle ahead enters the lane, when the vehicle ahead is selected often has a great impact on the control of ACC. The conventional solutions for cutting in the target ahead are mainly the following:

[0004] When the vehicle in the adjacent lane ahead cuts into the lane within a certain distance, the target is judged as cutting in, and then the nearest target is selected as the target to follow based on the longitudinal distance of the vehicles in the lane. Alternatively, based on the lateral collision time, longitudinal collision time and overlap between the vehicle and the adjacent lane ahead, when a certain threshold is reached, the adjacent lane vehicle is judged to have cut into the lane, and then the nearest target is selected in the lane.

[0005] At present, the two schemes mainly have the following problems: the front vehicle is considered to have cut into the lane only when part of it cuts into the lane, which often results in the late selection of the fast-cutting vehicle, and the ACC controller cannot respond in advance. The late selection of the close-range target often requires the ACC to generate a relatively large deceleration after selection, which affects the driver's physical sensation. The threshold of the collision time is not easy to adjust for selecting the target vehicle to cut in based on the collision time and overlap, and a large amount of test calibration is required. In addition, the parameters of the lane line are not considered, which often leads to inaccurate judgment on curves. Therefore, the present invention proposes a front target vehicle cut-in recognition method with an estimation function to solve at least one of the above-mentioned problems. Summary of the invention

[0006] In order to solve at least one of the problems mentioned in the above background technology, the present invention provides a method for identifying a front target vehicle cutting in with a prediction function.

[0007] To achieve the above object, the present invention provides the following technical solution: a method for identifying a front target vehicle cut-in with an estimation function, comprising the following steps:

[0008] Step 1: Activate the longitudinal control function of the adaptive cruise control function ACC and obtain the relevant input parameters of the vehicle, the target vehicle in front and the camera;

[0009] Step 2: Obtain the lane centerline equation based on the camera input;

[0010] Step 3: Calculate and determine the motion state of the vehicle using the vehicle information parameters in step 1. The motion state of the vehicle includes: stable left deviation, stable right deviation, centering, turning and lane changing.

[0011] Step 4: According to different working conditions, optimize the lane centerline equation, use the lane line with stable offset, without offset, and estimate the driving trajectory of the vehicle; update the lane centerline equation parameters, and update the lane line width according to different working conditions;

[0012] Step 5: Based on the information in step 4 and the forward target information in step 1, calculate whether the target cuts into the lane from the left adjacent lane or from the right lane by using the estimated time method;

[0013] Step 6: Based on the targets cutting into the lane from the left adjacent lane and the targets cutting into the lane from the right lane, select the target with the shortest longitudinal distance as the target vehicle cutting into the front adjacent lane.

[0014] The front target vehicle cut-in identification method of the present invention has been verified by a real vehicle. Regardless of whether the vehicle is driving in the center of the lane, stably offset to the left, stably offset to the right, or turning or changing lanes, it can accurately identify the target vehicle cutting into the front adjacent lane, and can predict the vehicle's cut-in in advance, and give the information of the target vehicle cutting into the lane to the adaptive cruise function ACC controller in advance, so that the control of the adaptive cruise function ACC controller is more agile and the longitudinal control of the vehicle is more comfortable. After adopting this algorithm, the occurrence of vehicle emergency braking caused by the front vehicle cutting in can be avoided.

[0015] Compared with the prior art, the invention has the following beneficial effects: the intelligence of the longitudinal control function of the adaptive cruise control function ACC is improved, and the vehicle cutting in from the adjacent lane can be identified in advance, so that the system is more in line with the driver's operation intention; the comfort of the longitudinal control function of the adaptive cruise control function ACC is improved, and the adaptive cruise control function ACC controller can respond to the cutting-in target in advance, so that the braking of the vehicle is more comfortable, and the adaptive cruise control function ACC will not suddenly brake the vehicle due to the sudden cutting-in of the target in front;

[0016] By using an estimated time method, the identification of the target entering the adjacent lane ahead is made more accurate. The accuracy and efficiency of the longitudinal control target identification of the adaptive cruise control function ACC are improved. By considering the movement trend of the vehicle, the lane line is optimized according to the vehicle, making the selection of the target entering the front more efficient, avoiding vehicle deceleration caused by misidentification, making the entire longitudinal control more comfortable, and increasing the number of times the driver turns on the adaptive cruise control function ACC and its efficiency. The safety of the adaptive cruise control function ACC system is improved, and the target cutting into the lane can be accurately identified whether on a straight road or a curve, thereby improving the safety of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0018] Figure 1 It is a flow chart of a method for identifying a front target vehicle cut-in with an estimation function;

[0019] Figure 2 Filter graph for target vehicles when the ego vehicle is turning. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0021] First, refer to Figure 1 to Figure 2 A method for identifying a front target vehicle cut-in with an estimation function according to an embodiment of the present invention is described in detail.

[0022] A method for identifying a front target vehicle cut-in with an estimation function comprises the following steps:

[0023] Step 1: Activate the longitudinal control function of the adaptive cruise control function ACC and obtain the relevant input parameters of the vehicle, the target vehicle in front and the camera;

[0024] Step 2: Obtain the lane centerline equation based on the camera input;

[0025] Step 3: Calculate and determine the motion state of the vehicle using the vehicle information parameters in step 1. The motion state of the vehicle includes: stable left deviation, stable right deviation, centering, turning and lane changing.

[0026] Step 4: According to different working conditions, optimize the lane centerline equation, use the lane line with stable offset, without offset, and estimate the driving trajectory of the vehicle; update the lane centerline equation parameters, and update the lane line width according to different working conditions;

[0027] Step 5: Based on the information in step 4 and the forward target information in step 1, calculate whether the target cuts into the lane from the left adjacent lane or from the right lane by using the estimated time method;

[0028] Step 6: Based on the targets cutting into the lane from the left adjacent lane and the targets cutting into the lane from the right lane, select the target with the shortest longitudinal distance as the target vehicle cutting into the front adjacent lane.

[0029] In order to make the screening of the adjacent lane target in front more timely and accurate when cutting into the lane, in this algorithm, we must first judge whether the vehicle is driving stably in the center area of ​​the lane or deviating from the lane for a certain distance to drive stably based on the parameters of the lane ahead detected by the sensor and the relevant motion parameters of the vehicle. When the vehicle is stably offset in the lane, the lane line needs to be optimized, the center line of the lane is also stably offset, and the lane width is reduced to the vehicle width plus 0.4m. The optimized lane line is used as the basis for the adjacent lane target to cut in. The adjacent lane target is considered to have cut into the target only after it cuts into the optimized lane line. The following algorithms are mainly used to estimate the movement trend of the vehicle:

[0030] According to the information from the camera and the left and right lane lines of the lane, the center line of the lane is obtained, which can be specifically expressed by the following equation:

[0031] y=a 0 +a 1 *x+a 2 *x 2 +a 3 *x 3 (1)

[0032] W lane is the width of the lane line.

[0033] According to the motion parameters of the vehicle, the longitudinal velocity v of the vehicle is obtained. x_ego , longitudinal acceleration a x_ego , lateral velocity v y_ego , lateral acceleration a y_ego , vehicle width W ego , the steering wheel angle of the vehicle is A ego, and then use the constant acceleration model to estimate the vehicle's next moving position based on the vehicle's current motion parameters.

[0034] Calculate the current coordinate position of the vehicle at time 0. In the initial state, the vehicle position is x 0_ego =0,y 0_ego = 0, the position of the lane line is y 0_lane =a 0 ;

[0035] Then at the starting position, the distance between the center position of the vehicle and the lane line is:

[0036] Δy 0 =y 0_ego -y 0_lane =-a 0 (2)

[0037] Based on the vehicle's speed, the preview position of the vehicle 1 second ahead, and the position of the lane line, a constant acceleration model is used to estimate the vehicle's state. In order to avoid the preview distance 1 second ahead being too close when the vehicle is at a low speed, which will lead to deviations in the calculation of the vehicle's movement trend, the minimum value of the vehicle's preview distance needs to be set to 10m.

[0038] The longitudinal position of the vehicle after 1 second is:

[0039] x ego_preview =v x_ego *t+0.5*a x_ego *t 2 (3)

[0040] Substituting t=1 into the equation:

[0041] x ego_preview =v x_ego +0.5*a x_ego

[0042] And x ego_preview The minimum value of is restricted to obtain:

[0043] x ego_preview_final =max(10,x ego_preview )

[0044] Then calculate the lateral position of the vehicle after 1 second:

[0045] y ego_preview =v y_ego *t+0.5*a y_ego *t 2 (4)

[0046] Substituting t=1 into the equation:

[0047] yego_preview =v y_ego +0.5*a y_ego (5)

[0048] At the vehicle preview point, the position of the lane line is:

[0049]

[0050] Then at the preview point, the distance between the vehicle's position and the lane line is:

[0051] Δy preview =y ego_preview -y lane_preview (7)

[0052] Use the above parameters to determine the movement trend of the vehicle:

[0053] Stable left deviation: When the vehicle is in the initial state Δy 0 <0, and at the preview point, the vehicle is within a certain range to the left of the lane line, Δy preview <-0.5, and the absolute value of the steering wheel angle A ego <30°, the vehicle is considered to be deviating to the stable left side;

[0054] Stable right deviation: When the vehicle is in the initial state Δy 0 >0, and at the preview point, the vehicle is within a certain range to the left of the lane line, Δy preview >0.5, and the absolute value of the steering wheel angle A ego <30°, the vehicle is considered to be deviating to the right in a stable manner;

[0055] Centered driving: When the vehicle has no stable left deviation, no stable right deviation, and the steering wheel angle absolute value A ego <30°, the vehicle is considered to be driving in the center;

[0056] Vehicle changes lanes or turns: When the vehicle's steering wheel angle absolute value A ego When the angle is ≥30°, the vehicle is considered to be changing lanes or turning stably.

[0057] The lane line processing strategies for different vehicle motion states are as follows:

[0058] Lane line optimization during stable left deviation: Since the vehicle is performing a stable left deviation, the lane line needs to be moved left.

[0059] When the vehicle is driving on the left side of the lane and the vehicle ahead cuts into the right lane at a relatively close distance,

[0060] Driving on the left side of the lane, there is no risk of collision with vehicles cutting in from the right at a short distance, so the target vehicle can be unselected. In the case of the vehicle moving like this, it is appropriate to shift the lane line to the left to achieve this goal. At the same time, since the vehicle is driving on the left side of the lane line, the vehicle ahead cutting in from the left lane line needs to be detected more sensitively to prevent the vehicle from colliding with the vehicle cutting in from the left. The left shift of the lane line is mainly achieved by the following method, considering the offset Δy of the vehicle at the starting time 0 and the lane deviation Δy at the preview point preview , these two values ​​can reflect the severity of the lateral deviation of the vehicle in the lane, so the lane line equation after the left deviation is:

[0061] y=a 0 +0.5*(Δy 0 +Δy preview )+a 1 *x+a 2 *x 2 +a 3 *x 3

[0062] =0.5*Δy preview -0.5*a 0 +a 1 *x+a 2 *x 2 +a 3 *x 3 (8)

[0063] The width of the lane line also needs to be optimized. The width of the vehicle is added to 0.2m on one side. This method is mainly used to prevent the lane from being too wide, which may easily lead to the misselection of the target in front. Using the width of the vehicle as the basis for the lane line width will make the target selection in front more accurate. In addition, because the entire lane line is shifted to the left, the width of the lane line is reduced, which can avoid misselection of the target vehicle in the adjacent lane on the left. The optimized lane width is: W lane =W ego +0.4;

[0064] Lane line optimization during stable right deviation: Since the vehicle is performing a stable right deviation, the lane line needs to be moved to the right;

[0065] When the vehicle is driving on the right side of the lane line and the vehicle in front cuts into the left lane line at a relatively close distance, since the ego vehicle is driving on the right side of the lane, there is no risk of collision with the vehicle cutting in from the left at a short distance, so the target vehicle can be unselected. In this moving condition of the ego vehicle, in order to achieve this goal, it is appropriate to shift the lane line to the right. At the same time, since the vehicle is driving on the right side of the lane line, the vehicle in front cutting in from the right lane line needs to be detected more sensitively to prevent the ego vehicle from colliding with the vehicle cutting in from the right. The right shift of the lane line is mainly achieved by the following method, considering the offset Δy of the ego vehicle at the starting time 0 and the lane deviation Δy at the preview point preview , these two values ​​can reflect the severity of the lateral deviation of the vehicle in the lane, so the lane line equation after the right deviation is:

[0066] y=a 0 +0.5*(Δy 0 +Δy preview )+a 1 *x+a 2 *x 2 +a 3 *x 3

[0067] =0.5*Δy preview -0.5*a 0 +a 1 *x+a 2 *x 2 +a 3 *x 3 (9)

[0068] The width of the lane line also needs to be optimized. The width of the vehicle is added to 0.2m on one side. This method is mainly used to prevent the lane from being too wide, which may easily lead to the misselection of the target in front. Using the width of the vehicle as the basis for the lane line width will make the target selection in front more accurate. In addition, because the entire lane line is moved to the right, the width of the lane line is reduced, which can avoid misselection of the target vehicle in the adjacent lane on the right. The optimized lane width is: W lane =W ego +0.4;

[0069] When the vehicle changes lanes or turns: Since the direction of the vehicle changes quickly at this time, and the process of changing lanes or turning is often short, a lateral constant acceleration model is used, and the center line of the vehicle is used to represent the center line of the lane:

[0070] y=v y_ego *t+0.5*a y_ego *t 2 (10)

[0071] Since the vehicle's trajectory changes rapidly during lane change or turning, in order to prevent misselection of the target ahead, the width of the lane line is directly based on the vehicle's width: W lane =W ego .

[0072] When the vehicle is driving in the center, the lane line is not optimized, and the lane line collected by the camera is directly used. Using formula (1), the width of the lane line is the width W of the lane line detected by the camera. lane .

[0073] Using the optimized lane line, the algorithm principle for selecting the front cut-in target is as follows: Figure 2 As shown, according to the sensor, the longitudinal velocity v of the target is obtained x_obstacle , lateral velocity v y_obstacle , vertical position x obstacle , horizontal position y obstacle , then the target's heading angle is calculated as:

[0074] A obstacle =atan2(v y_obstacle ,v x_obstacle ) (11)

[0075] The method for calculating the angle of the lane centerline at the target position is to use the longitudinal position of the target to find the point where the lane centerline is parallel to the longitudinal position of the target, and then take the position parameters of the points 1m in front and 1m behind in the longitudinal direction.

[0076] x obstacle +1 and x obstacle -1 According to the motion conditions of the vehicle, we substitute into the above formulas (8), (9), and (10) to obtain y lane_obstacle+1 and lane_obstacle-1 , then the lane centerline angle near the target is:

[0077] A lane_obstacle =atan2((y lane_obstacle+1 -y lane_obstacle-1 ),2) (12)

[0078] The angle between the target and the center line of the lane is:

[0079] A obstacle2lane =A obstacle -A lane_obstacle (13)

[0080] The lateral speed from the target to the center line of the lane is calculated as:

[0081]

[0082] After obtaining the corrected motion parameters of the front target, the front cut-in target is determined based on the relevant parameters of the front target. The specific algorithm is as follows:

[0083] Left adjacent lane target cut-in algorithm: For the target on the left, first determine that the target on the left is not selected as the following target, and the target is located on the left side of the lane, that is, y obstacle <-0.5*W lane , the target is to cut into the own lane, that is, v y_obstacle >0.01m / s 2 , the target ahead is on the left side of the lane, and the lateral speed is greater than a certain value. If it cuts into the lane, it is estimated whether the vehicle will cut into the lane in the next 1s (some compensation can be made according to the lateral speed of the cut-in). If it cuts in, the target is considered as the cut-in target. Since the lateral acceleration of the target is generally small and the error is relatively large, this algorithm uses a lateral uniform velocity model for the target ahead:

[0084] y obstacle_preview =y obstacle +1*v y_obstacle2lane (15)

[0085] If after 1s, the target's position meets the condition: y obstacle_preview >-0.5*W lane , it is considered that the left target cuts into the lane of the road and is selected as the cut-in target of the left adjacent lane.

[0086] Right adjacent lane target cut-in algorithm: For the target on the right, first determine that the target on the right is not selected as the following target, and the target is located on the right side of the lane, that is, y obstacle >0.5*W lane , the target is to cut into the own lane, that is, v y_obstacle <-0.01m / s 2 , the target ahead is on the right side of the lane, and the lateral speed is greater than a certain value. If the vehicle cuts into the lane, it is estimated whether the vehicle will cut into the lane in the next 1s (some compensation can be made according to the lateral speed of the cut-in). If it cuts in, the target is considered as the cut-in target. Since the lateral acceleration of the target is generally small and the error is relatively large, this algorithm uses a lateral uniform velocity model for the target ahead:

[0087] y obstacle_preview =y obstacle +1*v y_obstacle2lane (15)

[0088] If after 1s, the target's position meets the condition: y obstacle_preview <0.5*W lane , it is considered that the target on the right cuts into the lane of the road and is selected as the cut-in target of the right adjacent lane.

[0089] Finally, based on all the left-side and right-side target vehicles selected in the previous calculation, the target with the closest distance is selected as the following target for cutting in front, and the relevant parameters of the cutting-in target vehicle are sent to the adaptive cruise function ACC controller to perform longitudinal control of the vehicle.

[0090] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for identifying a target vehicle cutting in front of you with an estimation function, characterized in that: The steps include: Step 1: Activate the longitudinal control function of the adaptive cruise control function ACC and obtain the relevant input parameters of the vehicle, the target vehicle in front and the camera; Step 2: Obtain the lane centerline equation based on the camera input; Step 3: Calculate and determine the motion state of the vehicle using the vehicle information parameters in step 1. The motion state of the vehicle includes: stable left deviation, stable right deviation, centering, turning and lane changing. Step 4: According to different working conditions, optimize the lane centerline equation, use the lane line with stable offset, without offset, and estimate the driving trajectory of the vehicle; update the lane centerline equation parameters, and update the lane line width according to different working conditions; Step 5: Based on the information in step 4 and the forward target information in step 1, calculate whether the target cuts into the lane from the left adjacent lane or from the right lane by using the estimated time method; Step 6: Based on the targets cutting into the lane from the left adjacent lane and the targets cutting into the lane from the right lane, select the target with the shortest longitudinal distance as the target vehicle cutting into the front adjacent lane.

2. The method for identifying a front target vehicle cut-in with an estimation function according to claim 1, characterized in that: The input parameters in step 1 include: parameter information of the speed, acceleration and steering angle of the vehicle; distance information, speed information and acceleration information of the target vehicle in front.

3. The method for identifying a target vehicle cutting in front of the vehicle with an estimation function according to claim 1, characterized in that: In step 2, the camera image is collected and corrected, and then image preprocessing including grayscale processing and filtering and noise reduction is performed.

4. The method for identifying a front target vehicle cut-in with an estimation function according to claim 1, characterized in that: The step 3: based on the information in step 2 and the vehicle information parameters in step 1, calculate and determine the motion state of the vehicle: stable left offset, stable right offset, centered driving, turning or changing lanes, the calculation method includes but is not limited to the previous calculation formula, wherein the motion parameters of the vehicle can be calculated using the lateral constant acceleration model in this algorithm, or using the vehicle's dynamic formula and kinematic formula to calculate the motion parameters of the vehicle.

5. The method for identifying a front target vehicle cut-in with an estimation function according to claim 1, characterized in that: The step 4: based on the information in step 3, according to different working conditions, optimize the lane centerline equation, use the lane line with stable offset, without offset, and estimate the driving trajectory of the vehicle; update the lane centerline equation parameters, and update the lane line width according to different working conditions; regarding the lane center offset method, this algorithm can be used, or a polynomial fitting method can be used to fit the lane line using the motion parameters of the starting point and the preview point of the vehicle.

6. The method for identifying a front target vehicle cut-in with an estimation function according to claim 1, characterized in that: According to the target cutting into the lane from the left adjacent lane and the target cutting into the lane from the right lane, the target with the closest longitudinal distance is screened out as the target vehicle cutting into the adjacent lane ahead; the target vehicle cutting into the adjacent lane ahead with the closest longitudinal distance is screened out, including but not limited to the target with the closest longitudinal distance in this method.

Images