Lane line optimization target screening method with state hysteresis

Abstract

The application discloses a lane line optimization target screening method with state hysteresis, and belongs to the technical field of intelligent driving and vehicle longitudinal control. The method first acquires lane line parameters and vehicle motion parameters through a camera and a vehicle body sensor, calculates a lane center line, a short-time trajectory of the vehicle and a lateral error, and extracts a first derivative of the lane center line; then, a state machine containing six behavior states is used to realize accurate behavior recognition in combination with a lateral error and a first derivative threshold constraint, so that road bending and real lane changing misjudgment are avoided; when the behavior state is switched from lane changing having crossed a line to centering / offset keeping, a preview deviation limiter timer is started, and the sudden change of the preview lateral error is suppressed within a timing window; and finally, a path is screened based on an optimized lane center line fitting, and target screening is completed by adaptively scaling a channel width. The application improves the stability and accuracy of target screening under complex working conditions, reduces unnecessary acceleration and deceleration, and enhances driving comfort and system robustness.

Description

Technical Field

[0001] This invention relates to the field of intelligent driving and vehicle longitudinal control technology, and in particular to a lane line optimization target selection method with state hysteresis. Background Technology

[0002] Adaptive cruise control (ACC) is one of the most widely used longitudinal driving assistance functions in current mass-produced vehicles. After the driver sets the cruise speed and following distance, ACC controls the vehicle's acceleration or braking based on the presence and speed of the target vehicle ahead, enabling the vehicle to maintain a constant speed or automatically follow the vehicle ahead while meeting safe distance requirements.

[0003] In ACC control, the "target selection algorithm" is the core component. It is responsible for identifying target vehicles related to the vehicle's lane from multiple forward targets output by sensing devices such as cameras and millimeter-wave radar, and transmitting parameters such as relative distance and relative speed to the ACC controller. Incorrect target selection can lead to two serious consequences: first, it may mistakenly select unrelated vehicles in adjacent lanes as following targets, causing unnecessary deceleration or even braking, affecting driving comfort and safety; second, it may miss targets within the vehicle's lane that actually pose a collision risk, causing a delay in ACC response and creating safety hazards.

[0004] In existing technologies, target screening methods are mainly divided into three categories:

[0005] 1. Traditional lane-line-based target selection method: When lane lines are clear, the lane centerline is calculated directly using the left and right lane lines identified by the camera as boundaries. Targets are selected by judging whether the lateral position of the vehicle in front is within a certain range to the left and right of the centerline. This method works reasonably well when the vehicle is centered and has no intention to change lanes. However, when the vehicle deviates from the lane, changes lanes, or is on a curve, the vehicle's movement trend deviates from the static lane centerline, which can easily lead to problems such as selecting the vehicle in front of the original lane when changing lanes and misjudging vehicles in adjacent lanes when deviating from the lane.

[0006] 2. Target selection method based on vehicle motion trends: This method reduces reliance on actual lane lines and constructs the vehicle's future trajectory based on its longitudinal speed, yaw rate, and vehicle geometry. The trajectory is then expanded laterally to form a "driving channel" to determine if the target is within the channel. While this method reflects the driver's steering intention, it ignores the shape and width of actual lane lines. Significant differences in road geometry can lead to channels that are too wide or too narrow, resulting in misselection or omission.

[0007] 3. Target Selection Method Integrating Lane Lines and Vehicle Motion Status: This method predicts the vehicle's trajectory using a lateral motion model, calculates the lateral error between the current position, the pre-aimed position, and the lane centerline, and then performs offsetting and polynomial fitting on the lane centerline after determining the vehicle's behavior status. The optimized centerline is used as the selection criterion. While this method improves the rationality of the selection, it still has shortcomings: during the steering wheel return phase after a lane change, a large yaw rate leads to sensitivity to lateral deviations in the pre-aimed view, easily causing the selected path to "jump back" and resulting in misselection; on curves with large curvatures, road bends are easily misjudged as lane changes, leading to frequent switching of behavior status and oscillating selection paths; judging behavior status solely based on a single-cycle threshold lacks "delay confirmation" and "error limiting" mechanisms, failing to suppress the influence of sensor noise and transient dynamics. Therefore, it is urgent to introduce targeted optimization mechanisms based on existing comprehensive methods to improve the stability and robustness of ACC target selection under lane change, curve, and steering return conditions.

[0008] Therefore, it is urgent to introduce a targeted optimization mechanism on the basis of existing comprehensive methods to improve the stability and robustness of ACC target selection under lane change, curve and return-to-center steering conditions. Summary of the Invention

[0009] This invention aims to solve the problems of target misselection and path sway in existing ACC target selection methods during the steering wheel return phase after lane change and in curved sections. By introducing the first derivative threshold constraint of the lane centerline, state hysteresis timing and pre-aiming deviation limiting mechanism, the ACC following target selection is stabilized, thereby improving the safety and comfort of vehicle longitudinal control.

[0010] The technical solution of this invention: a lane line optimization target selection method with state hysteresis, comprising the following steps:

[0011] S1: Acquire vehicle sensor information, including left and right lane line fitting parameters, vehicle longitudinal speed, yaw rate, and vehicle geometric parameters;

[0012] S2: Calculate the lane centerline based on the fitting parameters of the left and right lane lines, and construct the short-term motion trajectory of the vehicle based on the vehicle's longitudinal speed, yaw rate and vehicle geometric parameters;

[0013] S3: At the current position of the vehicle and the preset aiming position, calculate the lateral deviation between the short-term motion trajectory of the vehicle and the center line of the lane, respectively, to obtain the initial lateral error and the aiming lateral error, and calculate the first derivative of the center line of the lane at the current position.

[0014] S4: The lane centerline heading angle cutting-out trend is obtained by comparing the first derivative with the lane change threshold. The vehicle position deviation cutting-out trend is judged by the initial lateral error and the pre-aiming lateral error. Only when the vehicle position deviation and the lane centerline heading angle simultaneously indicate a cutting-out trend to the same side is it determined that the lane change has not crossed the line. This constrains the jump of the behavior state machine and determines the current behavior state of the vehicle through the behavior state machine.

[0015] S5: When the behavior state is detected to change from left lane change crosses the line or right lane change crosses the line to center hold or offset hold, start the pre-aiming deviation limiting timer, and limit the amplitude of the pre-aiming lateral error during the effective period of the timer to obtain the limited pre-aiming lateral error.

[0016] S6: Update the lane centerline used for target selection based on the current behavior state and the pre-aiming lateral error after the amplitude limit;

[0017] S7: Based on the updated lane centerline, perform path fitting to generate a center path for target selection, and adaptively scale the lane width according to vehicle speed and distance.

[0018] S8: Based on the central path and the scaled channel width, determine whether each target ahead is located within the vehicle's driving channel, and designate the targets that meet the conditions as the following targets for adaptive cruise control.

[0019] Preferably, in step S4, the condition for the behavior state to change from centering or offset to left lane change without crossing the line is: the initial lateral error is negative, the difference between the pre-aiming lateral error and the initial lateral error is less than a negative first threshold, and the first derivative is greater than a preset positive lane change threshold.

[0020] Preferably, in step S4, the condition for the behavior state to change from centering or offset to right lane change without crossing the line is: the initial lateral error is positive, the difference between the pre-aiming lateral error and the initial lateral error is greater than a positive first threshold, and the first derivative is less than a preset negative lane change threshold.

[0021] Preferably, in step S4, the condition for the behavior state to exit from the left lane change cross-line or right lane change cross-line and return to the center hold or offset hold state includes: the absolute value of the first derivative is less than a preset second threshold.

[0022] The formula for calculating the amplitude limit of the aiming lateral error is as follows:

[0023] ;in, To anticipate lateral error, The preset amplitude limit value, This refers to the lateral error of the aiming after the amplitude is limited.

[0024] Preferably, in step S5, if the behavior state changes from offset holding to non-offset holding again during the effective period of the pre-aiming deviation limiting timer, the timer is immediately cleared and the corresponding limiting operation is terminated.

[0025] Preferably, step S6 specifically includes:

[0026] When the behavior state is offset hold, the lane centerline is laterally offset according to the pre-aiming lateral error after the amplitude limit;

[0027] When the behavior status is "left lane change without crossing the line" or "right lane change without crossing the line", the lane center line will be shifted to the center position of the corresponding target lane.

[0028] When the behavior status is centering, left lane change has crossed the line, or right lane change has crossed the line, the lane centerline is kept in the center of the current driving lane.

[0029] Preferably, in step S7, the path fitting specifically involves: using a fifth-order polynomial to fit the center path segment from the current position of the vehicle to the lane change pre-aiming point, and extending the far-end path segment beyond the effective range of the lane line using a cubic polynomial or Taylor expansion based on the lane center line.

[0030] Compared with existing technologies, the beneficial effects of this invention are:

[0031] Reduce the probability of incorrect selections during the steering wheel return-to-center phase after lane change: triggered by state transitions. The amplitude limiter timer suppresses the severe impact of the pre-aiming deviation on the center line of the lane within a short window immediately after a lane change, avoiding the misselection of vehicles in the original lane or adjacent lanes as following targets, and reducing unnecessary deceleration and braking.

[0032] Improve behavior recognition and path stability on curved road sections: Introduce lane centerline heading angles when the behavior state machine enters and exits lane-changing states. The threshold conditions accurately distinguish between the road's own curves and actual lane-changing behavior, reducing path swaying caused by behavioral state jitter in curve scenarios.

[0033] Enhanced longitudinal control comfort and driving experience of ACC: A more precise and stable target selection path makes the longitudinal acceleration and deceleration behavior of ACC smoother in complex conditions such as lane changes, curves and deviation driving, reducing frequent and abrupt speed changes, improving ride comfort and driver confidence in the system. Attached Figure Description

[0034] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0035] Unless otherwise defined, the technical or scientific terms used in this application shall have the general meaning understood by one of ordinary skill in the art to which this application pertains. Words such as “a,” “an,” “an,” “the,” “the,” and “these” used in this application do not indicate quantitative limitation and may be singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or modules (units) is not limited to the listed steps or modules (units) but may include steps or modules (units) not listed, or may include other steps or modules (units) inherent to these processes, methods, products, or devices. Words such as “connected,” “linked,” and “coupled” used in this application are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. Normally, the character " / " indicates that the objects before and after it are in an "or" relationship. The terms "first," "second," "third," etc., used in this application are merely to distinguish similar objects and do not represent a specific order of objects.

[0036] This invention provides a method for selecting lane line optimization targets with state hysteresis, such as... Figure 1 As shown, the process includes the following steps:

[0037] S1: Acquire vehicle sensor information, including left and right lane line fitting parameters, vehicle longitudinal speed, yaw rate, and vehicle geometric parameters;

[0038] S2: Calculate the lane centerline based on the fitting parameters of the left and right lane lines, and construct the short-term motion trajectory of the vehicle based on the vehicle's longitudinal speed, yaw rate and vehicle geometric parameters;

[0039] S3: At the current position of the vehicle and the preset aiming position, calculate the lateral deviation between the short-term motion trajectory of the vehicle and the center line of the lane, respectively, to obtain the initial lateral error and the aiming lateral error, and calculate the first derivative of the center line of the lane at the current position.

[0040] S4: The lane centerline heading angle cutting-out trend is obtained by comparing the first derivative with the lane change threshold. The vehicle position deviation cutting-out trend is judged by the initial lateral error and the pre-aiming lateral error. Only when the vehicle position deviation and the lane centerline heading angle simultaneously indicate a cutting-out trend to the same side is it determined that the lane change has not crossed the line. This constrains the jump of the behavior state machine and determines the current behavior state of the vehicle through the behavior state machine.

[0041] Specifically, in step S4, the condition for the behavior state to change from centering or offset to right lane change without crossing the line is: the initial lateral error is positive, the difference between the pre-aiming lateral error and the initial lateral error is greater than a positive first threshold, and the first derivative is less than a preset negative lane change threshold.

[0042] The conditions for exiting the behavior state from the left lane change crossover or the right lane change crossover and returning to the center hold or offset hold state include: the absolute value of the first derivative is less than a preset second threshold.

[0043] S5: When the behavior status is detected to change from left lane change crosses the line or right lane change crosses the line to center hold or offset hold, start the pre-aiming deviation limiting timer, and limit the amplitude of the pre-aiming lateral error during the effective period of the timer to obtain the limited pre-aiming lateral error.

[0044] S6: Update the lane centerline used for target selection based on the current behavior state and the pre-aiming lateral error after the amplitude limit;

[0045] S7: Based on the updated lane centerline, perform path fitting to generate a center path for target selection, and adaptively scale the lane width according to vehicle speed and distance.

[0046] S8: Based on the central path and the scaled channel width, determine whether each target ahead is located within the vehicle's driving channel, and designate the targets that meet the conditions as the following targets for adaptive cruise control.

[0047] The overall process of this invention includes basic calculation of vehicle motion and lane lines, behavior state recognition, lane centerline offset and path fitting, channel width calculation and target selection. The core improvements are in the behavior recognition and lane line offset parts, and the specific technical solutions are as follows:

[0048] Vehicle motion and lane line basic calculation: In this embodiment, in each control cycle, the vehicle system obtains the required information from the CAN bus or sensor interface, including the cubic polynomial fitting coefficients of the left and right lane lines and the lane line confidence level provided by the forward-looking camera; as well as parameters such as the vehicle's longitudinal speed Vx, yaw rate ω, and wheelbase L obtained from the vehicle dynamic sensors.

[0049] Specifically, the left and right lane lines are obtained from the camera recognition results, and a cubic polynomial of the lane center line is fitted to obtain the following:

[0050] ;

[0051] in, These are the polynomial coefficients.

[0052] Subsequently, parameters such as the vehicle's longitudinal velocity Vx, yaw rate ω, and wheelbase plus front overhang length L were obtained from the vehicle's sensors. Using a lateral constant acceleration model, the approximate lateral trajectory of the vehicle's front bumper center over a short period of time was derived.

[0053] ; ;

[0054] Where C1 and C2 are weighting coefficients, and K is curvature.

[0055] At the starting point x=0, the lateral position of the vehicle is defined as 0. The initial lateral error is calculated based on the intercept and slope of the lane centerline:

[0056] ;

[0057] Pre-aiming point From the maximum lane change distance A certain proportion and a fixed minimum distance are combined to determine the value, as shown in the example:

[0058] ;

[0059] exist At this point, calculate the vehicle's trajectory. Lane center line and its derivative The aiming lateral error is obtained as follows:

[0060] .

[0061] Behavioral state machine transition constraints: In this invention, when the first derivative a1_lane is greater than the positive lane-changing threshold ka1, the initial lateral error is considered. and aiming lateral error The change in a1_lane indicates that the vehicle is deflecting to the left relative to the lane centerline, showing a tendency to change lanes to the left; when a1_lane is less than a negative lane-changing threshold, it means that the lane centerline is deflecting to the right relative to the vehicle's direction of travel. and The change in the vehicle's position indicates a tendency to change lanes to the right.

[0062] This invention, by introducing the first derivative a1_lane and its threshold judgment, can more accurately distinguish between the situation where "the vehicle body has a certain heading angle relative to the center line of the lane and tends to cut out to a certain side" and the situation where "the lane itself is curved but the vehicle is still in the current lane".

[0063] Specifically, the transitions of the behavior state machine are constrained to make the transition states more consistent with the actual vehicle states. The values ​​of the autonomous vehicle behavior state EgoBehavior are defined as follows: 0 (centering), 1 (offset), 2 (left lane change without crossing the line), 3 (right lane change without crossing the line), 4 (left lane change crossed the line), and 5 (right lane change crossed the line).

[0064] Based on the existing and Based on the transition conditions, this invention introduces a threshold constraint on the first derivative of the lane line, a1_lane, in the following key transitions;

[0065] The transition from centering / offset to lane change without crossing the line will only change the behavior state from 0 or 1 to 2 or 3 if the following conditions are met simultaneously:

[0066] To enter state 2 (left lane change without crossing the line), the following conditions must be met simultaneously:

[0067] Initial lateral error Negative (vehicle is located to the left of the lane center line) A value less than the negative threshold Thr1 (indicating a significant increase in lateral deviation to the left within the pre-aiming distance) and a first derivative a1_lane greater than the threshold ka1 (indicating that the lane centerline is tilted to the left in the direction of the vehicle's movement, and there is a certain left yaw angle between the vehicle body and the lane line, which geometrically supports the intention to change lanes to the left).

[0068] To enter state 3 (right lane change without crossing the line), the following conditions must be met simultaneously:

[0069] Initial lateral error Positive (vehicle is located to the right of the lane center line) A value greater than the positive threshold Thr1 indicates a significant increase in lateral deviation to the right within the pre-aiming distance, while a first derivative a1_lane is less than the threshold ka1 (indicating that the lane centerline is tilted to the right in the direction of the vehicle's movement, and there is a certain rightward heading angle between the vehicle body and the lane line, which geometrically supports the intention to change lanes to the right).

[0070] By combining the above conditions, only when the vehicle position deviation and the lane heading angle simultaneously indicate a tendency to cut out to the same side will it be determined that the lane change did not cross the line, thus effectively distinguishing the road's own curvature from the actual lane change operation.

[0071] The transition from "Left / Right Lane Change Crossed" to "Center / Offset Hold" is allowed to return to state 1 (Offset Hold) or 0 (Center Hold) only if the following conditions are met when EgoBehavior is 4 or 5 (Left / Right Lane Change Crossed):

[0072] The absolute value of the first derivative of the lane centerline, a1_lane, is less than a smaller threshold. When the angle between the lane centerline and the vehicle's direction of travel has decreased, the road is becoming straighter, or the steering wheel has been straightened, and the initial and pre-aiming lateral errors indicate that the vehicle is now stably positioned on one side of the new lane, a transition from state 4 or 5 to state 0 or 1 is permitted. Through bidirectional constraints, the behavioral state machine becomes more stable on curved sections and during lane-change completion phases, reducing state oscillations.

[0073] State hysteresis timing and anticipation deviation limiting: This invention adds a state hysteresis timing to limit anticipation deviations that meet specific conditions. Furthermore, to suppress path bounce caused by anticipation deviations during the steering wheel return phase immediately after a lane change, this invention further introduces a state hysteresis timer and... Limiting logic:

[0074] Set a persistent variable: EgoBehavior_last (behavior state in the previous control cycle). (Pre-aiming deviation limiting timer).

[0075] At the end of each cycle, update EgoBehavior_last with the current EgoBehavior.

[0076] When the behavior status changes from 4 or 5 (lane change has crossed the line) to 0 or 1 (centering / offset holding), it is assumed that the vehicle has just completed the lane change and the steering wheel is still in the process of returning to center. At this time, the amplitude limit timer is started, for example, with a value of 40 control cycles.

[0077] In each cycle, if If it is greater than 0, then it is decremented by 1; if the behavior state changes from 1 (offset held) to another state again during the timing period, then it is immediately reset to zero and the limiting is terminated.

[0078] When the offset holding state EgoBehavior=1, the following strategy is used when updating the lane centerline offset:

[0079] like >0 indicates pre-aiming deviation Amplitude limiting, and according to Update channel centerline constant term ;

[0080] like, Then, the unlimited version will be restored. renew .

[0081] The above mechanism only takes effect within the critical short time window of "recovering from lane change after crossing the line to centering / offset". It is specifically designed to address the transient deviation during the steering wheel return phase after lane change and effectively prevents the target selection path centerline from briefly jumping back between the original lane and the new lane.

[0082] Finally, this method performs path fitting and target selection to obtain the optimized lane centerline offset. After obtaining the higher-order coefficients a1, a2, and a3, the existing polynomial path fitting method can be used:

[0083] 1. Within the interval 0 to xpo (where xpo is the target position for lane change), use the position and derivative of the current point and the target point to fit a smooth center path using a fifth-order polynomial;

[0084] 2. In the region where x > xpo, use a cubic polynomial to extend the path near the lane centerline or the vanishing point using Taylor expansion;

[0085] 3. Based on the distance and vehicle speed, the width of the channels on both sides of the path is adaptively scaled to form a PathWidth sequence, thereby obtaining a complete target filtering channel;

[0086] 4. The ACC target selection module uses the center path PathPoints, the channel width PathWidth, and the vehicle's geometric characteristics to determine whether each perceived target is within the channel range, thereby selecting the target required for longitudinal control.

[0087] This invention, based on existing vehicle motion state and lane line optimization target selection methods in urban roads, congested sections, and frequent lane-changing scenarios, can reduce the probability of misselection during the steering wheel return phase after lane changing, improve behavior recognition and path stability on curved road sections, and enhance the comfort and driving experience of ACC longitudinal control.

[0088] Furthermore, the method of the present invention operates in the ACC controller with a control cycle of 20-50Hz, and the specific implementation steps are as follows:

[0089] Step 1: In each cycle, read the latest left and right lane line fitting parameters, lane confidence, vehicle longitudinal speed, yaw rate, and vehicle geometric parameters from the camera and vehicle body sensors. Calculate the lane centerline polynomial coefficients and the short-term trajectory approximation of the vehicle to obtain the initial lateral error. Lateral error of aiming And the first derivative of the lane centerline, a1_lane;

[0090] Step 2: Using an improved behavioral state machine, based on , And the size relationship of a1_lane, determine which of the following the vehicle is currently in: centering, offset, left lane change without crossing the line, right lane change without crossing the line, left lane change crossed the line or right lane change crossed the line, and update the behavior status if necessary;

[0091] Step 3: Check if the behavior status changes from "Left / Right lane change has crossed the line" to "Center / Offset Hold". If so, start. It takes effect within a preset time period; during the time period, it applies to... Implement amplitude limiting and use the amplitude-limited value for lane centerline constant term offset; stop amplitude limiting when the timer ends or the behavior jumps out of offset holding again;

[0092] Step 4: Select an appropriate lane centerline offset strategy based on the EgoBehavior status:

[0093] In center-keeping mode, do not deviate from the lane centerline; in offset-keeping mode, adjust according to the width limit. Slightly shift off the center line; switch the center line to the target lane position when changing lanes without crossing the line; maintain the center line of the new lane when changing lanes and crossing the line.

[0094] Step 5: Based on the updated centerline, fit the path segment between the current point and the lane change preview point (or target point) using a fifth-order polynomial, extend it at a greater distance using a cubic polynomial or Taylor expansion, and adaptively scale the channel width according to the vehicle speed and distance to generate a complete target selection path and channel.

[0095] Step 6: Based on the above path and channel information, the ACC target selection module determines whether each target ahead is located within the vehicle's driving channel, selects the nearest target vehicle that meets the safety constraints as the following target, and transmits its distance and relative speed parameters to the longitudinal control module.

[0096] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0097] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0098] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A lane line optimization target screening method with lane state hysteresis, characterized by, The method comprises the following steps: S1: acquiring vehicle sensor information; S2: calculating a lane center line according to left lane line and right lane line fitting parameters, and constructing a short-time motion trajectory of the ego vehicle based on the vehicle sensor information; S3: calculating a lateral deviation between the short-time motion trajectory of the ego vehicle and the lane center line at a current position and a preset preview position respectively to obtain a starting lateral error and a preview lateral error, and calculating a first derivative of the lane center line at the current position; S4: obtaining a lane center line heading angle cut-in trend by comparing the first derivative with a lane change threshold, and obtaining a vehicle position deviation cut-in trend by the starting lateral error and the preview lateral error, and determining that the lane change is not off-track only when the vehicle position deviation and the lane center line heading angle simultaneously represent a cut-in trend to the same side, so as to constrain the jump of a behavior state machine, and determining a current ego vehicle behavior state by the behavior state machine; S5: when it is detected that the behavior state changes from left lane change off-track or right lane change off-track to center keeping or offset keeping, starting a preview deviation amplitude limiting timer, and limiting the amplitude of the preview lateral error during the validity period of the timer to obtain an amplitude-limited preview lateral error; S6: updating the lane center line for target screening according to the current behavior state and the amplitude-limited preview lateral error; S7: performing path fitting based on the updated lane center line to generate a center path for target screening, and adaptively scaling a lane width according to vehicle speed and distance; S8: judging whether each front target is located in a driving lane of the ego vehicle according to the center path and the scaled lane width, and determining a target meeting the condition as a following target of adaptive cruise control.

2. A lane line optimization goal screening method with state hysteresis according to claim 1, characterized in that, In step S4, the condition for the behavior state entering a left lane change off-track state from center keeping or offset keeping is that the starting lateral error is negative, the difference between the preview lateral error and the starting lateral error is less than a negative first threshold, and the first derivative is greater than a preset positive lane change threshold.

3. The lane line optimization target screening method with state hysteresis according to claim 2, characterized in that, In step S4, the condition for the behavior state entering a right lane change off-track state from center keeping or offset keeping is that the starting lateral error is positive, the difference between the preview lateral error and the starting lateral error is greater than a positive first threshold, and the first derivative is less than a preset negative lane change threshold.

4. The lane line optimization target screening method with state hysteresis according to claim 3, characterized in that, In step S4, the condition for the behavior state exiting from left lane change off-track or right lane change off-track and returning to center keeping or offset keeping state includes that the absolute value of the first derivative is less than a preset second threshold.

5. The lane line optimization target screening method with state hysteresis according to claim 1, characterized in that, In step S5, the amplitude of the preview lateral error is limited to obtain an amplitude-limited preview lateral error, and the calculation method is as follows: ; wherein is a preview lateral error, is a preset clipping value, is a clipped preview lateral error.

6. The lane line optimization target screening method with state hysteresis of claim 1, wherein, In step S5, if the behavior state changes from offset keeping to non-offset keeping again during the validity period of the preview deviation amplitude limiting timer, the timer and the corresponding amplitude limiting operation are immediately cleared and terminated.

7. The lane line optimization target screening method with state hysteresis according to claim 1, wherein step S6 specifically comprises: when the behavior state is offset keeping, performing lateral offset on the lane center line according to the amplitude-limited preview lateral error. when the behavior state is left-lane-change-not-crossing or right-lane-change-not-crossing, offsetting the lane centerline to the center position of the corresponding target lane; when the behavior state is center-keeping, left-lane-change-crossing or right-lane-change-crossing, keeping the lane centerline in the center of the current lane. 8.The lane line optimization target screening method with state hysteresis according to claim 1, wherein in step S7, the path fitting is specifically: using a quintic polynomial to fit the center path segment from the current position of the ego vehicle to the lane change preview point, and using a cubic polynomial or Taylor expansion based on the lane centerline to extend the distal path segment beyond the effective range of the lane line.

Images