Vehicle and method for controlling cut-in response

Abstract

The present invention relates to a vehicle and a method of controlling cut-in response. The present invention relates to a vehicle capable of effectively responding to cut-in of a nearby vehicle under various road conditions and a cut-in response control method thereof. The method of controlling cut-in response of the vehicle of the embodiment of the root invention includes the steps of: acquiring driving situation information; deriving an integrated lane by selectively applying a lane link, a lane side, and a point level path (PLP) based on the acquired driving situation information; determining a cut-in target based on a predicted path of each of the at least one nearby vehicle and the integrated lane; calculating a control point to be followed by driving control of the host vehicle based on a point of intersection of the predicted path of the cut-in target and the integrated lane; generating a speed profile and a driving path based on the calculated control point; and performing driving control based on parameters corresponding to the speed profile and the driving path.

Description

[0001] Cross-reference to related applications

[0002] This application claims the benefit of Korean Patent Application No. 10-2020-0186461, filed on December 29, 2020, which is incorporated herein by reference as if fully set forth herein. Technical Field

[0003] The present invention relates to a vehicle capable of effectively responding to the cutting-in of nearby vehicles under various road conditions, and a cutting-in response control method for the vehicle. Background Technology

[0004] Autonomous vehicles employing Advanced Driver Assistance Systems (ADAS) not only free drivers from simple tasks such as steering wheel and pedal operation during driving, but also prevent accidents caused by driver negligence, thus attracting increasing attention recently.

[0005] Such autonomous vehicles generate paths using the section change lines (nodes), lane center lines (lanelinks), or lane lines (lanesides) that make up the high-definition map, and perform autonomous driving control by following these paths. These paths are generally generated by reprocessing vector data collected from the high-definition map using various functional expressions, and sequentially using each point of the vector data.

[0006] However, standardized road shape assumptions state that a typical lane link exists in the center of the lane, lanes are continuous, lane widths are constant, and curvature is within a certain range. Therefore, when determining whether a nearby vehicle is in a transition state to change lanes into the vehicle's lane (its own driving lane), the following problems arise: In cases where the lane link is biased towards one lane, lanes are discontinuous, lane widths are irregular, or special sections such as U-turn zones / roundabouts exist, incorrect identification and failure to identify the vehicle frequently occur. Summary of the Invention

[0007] The purpose of this invention is to provide a vehicle capable of effectively responding to the cutting-in of nearby vehicles under various road conditions, and a cutting-in response control method for the vehicle.

[0008] Specifically, the object of the present invention is to provide a vehicle capable of improving cut-in response and ride comfort by deriving lanes that can stably respond to various road types, and a cut-in response control method for the vehicle.

[0009] The technical problems to be solved by the present invention are not limited to those mentioned above. Other technical problems not mentioned will be clearly understood by those skilled in the art from the following description.

[0010] To address the aforementioned technical problems, a method for controlling vehicle cut-in response according to an embodiment of the present invention may include the following steps: acquiring driving condition information; deriving an integrated lane based on the acquired driving condition information by selectively applying lanelinks, lanesides, and point-level paths (PLPs); determining a cut-in target based on the integrated lane and the predicted paths of at least one nearby vehicle; calculating control points to be followed by the vehicle's driving control based on the intersection of the predicted path of the cut-in target and the integrated lane; generating a speed distribution and driving path based on the calculated control points; and performing driving control based on parameters corresponding to the speed distribution and driving path.

[0011] Furthermore, the vehicle used to perform cut-in response control may include a fusion information generator, a control parameter generator, and a driving controller. The fusion information generator acquires driving information. The control parameter generator, based on the acquired driving information, derives an integrated lane by selectively applying Lanelink, Laneside, and Point-Level Path (PLP). It determines a cut-in target based on the integrated lane and the predicted paths of at least one nearby vehicle. It calculates the control points to be followed by the vehicle's driving control based on the intersection of the predicted path of the cut-in target and the integrated lane, and generates a speed distribution and driving path based on the calculated control points. The driving controller performs driving control based on parameters corresponding to the speed distribution and the point-level path.

[0012] The vehicle configured as described above, in relation to at least one embodiment of the present invention, can effectively respond to the cutting-in of nearby vehicles under various road conditions, thereby improving ride comfort.

[0013] Specifically, according to an embodiment of the invention, various road types can be stably responded to by deriving integrated lanes and calculating control points with continuity.

[0014] The effects achievable by this invention are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art through the following description. Attached Figure Description

[0015] Figure 1 An example of a vehicle configuration according to one implementation scheme is shown.

[0016] Figure 2 An example configuration of a control parameter generator according to one implementation scheme is shown.

[0017] Figure 3 An example of the control process for a cut-in response according to one implementation is shown.

[0018] Figure 4 This is a diagram illustrating situations where false detections and undetected events occur when using Lanelink-based virtual lanes.

[0019] Figure 5 An example of lane detection based on Laneside is shown.

[0020] Figure 6 An example of lane detection based on Lanelink is shown.

[0021] Figure 7 An example of lane detection based on PLP is shown.

[0022] Figure 8 An example of the process for determining an integrated lane according to one implementation scheme is shown.

[0023] Figure 9 This is a diagram illustrating the necessity of correcting and integrating lanes between frames.

[0024] Figure 10 An example of the process of correcting integrated lanes between frames is shown according to one implementation scheme.

[0025] Figure 11 This is a schematic diagram illustrating the selection of candidate vehicles based on the current location according to an implementation scheme.

[0026] Figure 12 This is a schematic diagram illustrating the parameters used to generate a predicted path according to one implementation scheme.

[0027] Figure 13A An example of curve-based prediction for generating a predicted path is shown according to one implementation scheme. Figure 13B An example of lane change timing modeled based on the bias value of nearby vehicles is shown.

[0028] Figure 14 An example of cut-in determination based on the future location of a predicted path is shown in one implementation.

[0029] Figure 15 An example of valid target verification based on a signal is shown according to one implementation scheme.

[0030] Figure 16 An example of valid target verification based on predicted path intersections is shown according to one implementation scheme.

[0031] Figure 17 An example of control point selection in a U-turn path according to one implementation is shown.

[0032] Figure 18 An example of control point selection in a straight path according to one implementation scheme is shown.

[0033] Figure 19 An example is shown of determining the rate of progress of other vehicles entering the vehicle's lane according to one implementation scheme.

[0034] Figure 20 This is a flowchart illustrating an example of the process for extracting control points according to one implementation scheme. Detailed Implementation

[0035] In the following description, embodiments of the invention will be detailed with reference to the accompanying drawings, enabling those skilled in the art to readily implement these embodiments. However, the invention can be implemented in various different embodiments and is not limited to those described herein. Furthermore, for clarity of explanation in the drawings, components unrelated to the description have been omitted, and similar reference numerals have been added to similar components throughout the specification.

[0036] Throughout this specification, when a component "includes" a certain assembly, it means that other components may be included, not excluded, unless otherwise stated. Furthermore, components indicated by the same reference numerals throughout this specification refer to the same assemblies.

[0037] Embodiments of this invention propose deriving integrated lanes by effectively utilizing lanesides, lane links, and point-level paths (PLPs), and determining cut-in targets by comparing the predicted paths and future positions of nearby vehicles. Furthermore, embodiments of this invention propose calculating control points based on the intersections of the predicted paths of the determined cut-in targets and the integrated lanes, thereby enabling stable responses to various road types.

[0038] Figure 1 An example of a vehicle configuration according to one implementation scheme is shown.

[0039] Reference Figure 1 The vehicle 100 according to the present invention includes an identification sensor 110, a high-definition map transmission module 120, a GPS 130, a communication unit 140, a fusion information generator 150, a control parameter generator 160, and a driving controller 170.

[0040] It is obvious to those skilled in the art that Figure 1Each component shown is illustrated primarily in relation to an embodiment of the invention, and may include fewer or more components in a real vehicle implementation. Each component will be described in detail below.

[0041] The identification sensor 110 includes a LiDAR 111, a camera 112, and a radar 113, and the identification sensor 110 can collect information about the road, the environment of the road, and nearby vehicles to provide identification information.

[0042] The high-definition map sending module 120 provides a high-definition map of the area surrounding the vehicle 100.

[0043] GPS 130 can receive signals from GPS satellites (not shown) and use the received signals to calculate the current position of the vehicle 100.

[0044] The communication unit 140 is a device for sending and receiving information with the interior and exterior of the vehicle 100. For example, in-vehicle communication can be performed by a transceiver that supports vehicle communication protocols (CAN, CAN-FD, LIN, Ethernet, etc.), and external communication can be performed by a general wireless communication protocol (3G / LTE / 5G, etc.) or a modem that supports the V2X protocol, but is not necessarily limited to these.

[0045] The fusion information generator 150 may include a location recognition module 151, a road information fusion module 152, and an object fusion module 153.

[0046] The location recognition module 151 can compare the recognition information received from the recognition sensor 110, the vehicle's current location information received from the GPS 130, and the high-definition map of nearby vehicles received from the high-definition map sending module 120, and generate high-definition information of the vehicle's current location and location recognition reliability information. The road information fusion module 152 can generate a high-definition map of the vehicle's environment using the high-definition information of the vehicle's current location received from the location recognition module 151 and the high-definition map of the vehicle's environment received from the high-definition map sending module 120. The object fusion module 153 can generate fused object information using the recognition information and the high-definition map of the vehicle's environment received from the road information fusion module 152. The fused object information can refer to map information including the position and speed of objects (i.e., nearby vehicles other than the vehicle itself).

[0047] The control parameter generator 160 generates control parameters for the driving controller 170 by utilizing the fused object information received from the fused information generator 150, and detailed configuration and functions will be described later.

[0048] The driving controller 170 can control the driving state of the vehicle (e.g., acceleration / deceleration, steering, braking, etc.) using the vehicle control values ​​corresponding to the control parameters received from the control parameter generator 160.

[0049] Figure 2 An example configuration of a control parameter generator according to one implementation scheme is shown.

[0050] Reference Figure 2 According to one implementation, the control parameter generator 160 may include an integrated lane calculation module 161, a box point position calculation module 162, a position-based cut-in candidate determination module 163, a path-based cut-in target determination module 164, a final effective cut-in target determination module 165, a cut-in target control point calculation module 166, a speed distribution generation module 167, a driving path generation module 168, and a control parameter output module 169.

[0051] The integrated lane calculation module 161 can selectively utilize Lanelink, Laneside, and PLP to calculate the final lane for cut-in response determination, i.e., the integrated lane, based on a high-definition map of the vehicle's environment received from the road information fusion module 152 and according to the current driving conditions.

[0052] The box point location calculation module 162 can export the integrated lane interior location of the four vertices of nearby objects (such as nearby vehicles), i.e., box point location information.

[0053] The location-based cut-in candidate determination module 163 can calculate cut-in candidate targets by using the box point location information of nearby objects to determine the longitudinal position based on the vehicle.

[0054] The path-based entry target determination module 164 can select the final entry target based on the predicted path of the entry candidate target.

[0055] The final effective entry target determination module 165 can ultimately determine the entry target by utilizing traffic information (traffic light signal information, etc.) and the intersection information between the predicted paths.

[0056] The cut-in target control point calculation module 166 can calculate the control points that the vehicle's driving control should follow when a cut-in target is selected on all types of roads, including curved roads.

[0057] The speed distribution generation module 167 can calculate a speed distribution, which is a set of target speeds that the vehicle must follow at each time point for tracking control of the control points.

[0058] The driving path generation module 168 can calculate the point-level path (PLP) that the vehicle should follow for lateral control based on bias and path holding based on in-pass.

[0059] The control parameter output module 169 can output the parameters of each control element determined in the above manner to the driving controller 170.

[0060] Figure 3 An example of the control process for a cut-in response according to one implementation is shown.

[0061] Reference Figure 3 First, driving situation information can be obtained in the fusion information generator 150 based on information acquired from at least one of the identification sensor 110, the high-definition map transmission module 120, the GPS 130, and the communication unit 140 (S301). This process (S301) corresponds to a preparation process in which nearby objects around the vehicle's lane are selected as candidates to minimize the computational load of identification sensor information about nearby objects, and detailed information about the corresponding candidates is calculated.

[0062] The integrated lane calculation module 161 of the control parameter generator 160 can calculate integrated lane information based on driving condition information (S302).

[0063] The location-based cut-in candidate determination module 163 can determine the longitudinal position of the vehicle based on the box point position information of nearby objects derived by the box point position calculation module 162, and calculate the cut-in candidate target (S303).

[0064] The path-based entry target determination module 164 can determine the entry target by calculating the predicted path of the entry candidate target (S304), comparing the predicted path of the entry candidate target with the future position of the intersection point of the integrated lane (S305), and comprehensively considering the comparison result of the future position of the vehicle (S306).

[0065] The final effective entry target determination module 165 can filter out unnecessary targets and determine the final entry target by utilizing traffic information (traffic light signal information, etc.) and intersection information between the predicted paths (i.e., mutual influence between nearby vehicles) (S307).

[0066] The cut-in target control point calculation module 166 can calculate the control points that the vehicle's driving control should follow when a cut-in target is selected on all types of roads, including curved roads (S308). This calculation of control points can play a very important role on curved roads or when only a portion of the vehicles enter the vehicle's lane.

[0067] Subsequently, the speed distribution generation module 167 can calculate the speed distribution (S309), the driving path generation module 168 can calculate the point level path (PLP) (S310), and when the control parameter output module 169 outputs the parameters of each control element to the driving controller 170, driving control can be executed (S311).

[0068] In the following text, reference will be made to Figures 4 to 10 Describe the integrated lane according to an implementation plan.

[0069] Figure 4 This is a diagram illustrating situations where false detections and undetected events occur when using LaneLink-based virtual lanes.

[0070] In urban areas, although there may be no significant differences to the naked eye, there are many situations where lane widths are not constant, such as... Figure 4 As shown on the left. For example, if the lane width at the location of vehicle 100 is 3 meters, but the lane width 50 meters ahead is 3.3 meters, the human eye will not easily perceive the change in lane width. Here, when deriving virtual lanes based on Lanelink, the following error detection problem exists: even if other vehicles 11 do not encroach on the actual lane where vehicle 100 is traveling, deceleration control is still performed because other vehicles 11 cross the virtual lane lines.

[0071] Additionally, when the lane width is as follows Figure 4 When the lane widens as shown on the right, the following situation may occur: even if another vehicle 11 encroaches on the driving lane, deceleration control is not performed because it does not cross the Lanelink-based virtual lane line, so the occupants feel that no other vehicle 11 is detected.

[0072] Figure 5 An example of lane detection based on Laneside is shown.

[0073] Because occupants perceive the actual ground conditions for a determined entry point, the entry point can be determined based on whether the vehicle will cross lane lines in a typical urban lane area. Therefore, as Figure 5 As shown, if the lane width itself is greater than the safety margin for vehicle passage, it is preferable to determine the entry target based on the Laneside, that is, it is preferable to calculate the integrated lane based on the Laneside.

[0074] Figure 6 An example of lane detection based on Lanelink is shown.

[0075] Reference Figure 6This illustrates a situation where a pocketlane with discontinuous lanes appears starting from the left side of the vehicle 100. In cases of such pocketlanes or laneside irregularities, utilizing virtual lanes with predetermined intervals calculated based on Lanelink may be easier than determining laneside-based lanes.

[0076] Figure 7 An example of lane detection based on point-level paths is shown.

[0077] In such Figure 7 In situations such as U-turns, at intersections, left / right turns, P-turns, or in wide lanes of bus stop areas, there are instances where actual lane markings (lanesides) are absent or, even if present, vehicles do not follow the lane markings as intended.

[0078] Therefore, if a point-level path (PLP) for the vehicle's tracking control has been calculated in the previous frame, a cut-in target determination can be performed within a predetermined interval where the vehicle can pass to the left and right of the corresponding point-level path. However, since the point-level path is derived at the end of the corresponding frame, information from the previous frame is used when deriving the integrated lane based on the point-level path. Therefore, in stages where the driving strategy is not fixed (the step of determining whether to change lanes), a cut-in determination can be performed based on Laneside or Lanelink, or the determination in the corresponding frame can be retained, and the point-level path-based determination of the previous frame can be finally confirmed in the next frame. In this case, a one-frame delay may occur, but considering that the determination delay due to inaccurate sensor identification is generally three frames, a one-frame delay can be ignored.

[0079] When point-level paths change continuously (e.g., path changes during lane changes), it is desirable to generate integrated lanes in the area, taking into account the direction and extent of the point-level paths.

[0080] In summary, as the lane to be used when determining the entry point, i.e., the integrated lane, lanes based on high-resolution maps can be given priority, while integrated lanes based on point-level paths (PLPs) can be given lower priority. This is because point-level paths themselves may contain errors and therefore may not reflect the actual ground conditions.

[0081] In lanes based on high-definition maps, Lanelink-based integrated lanes may differ from the actual visible lanes. Therefore, Laneside-based integrated lanes can have the highest priority, followed by Lanelink-based integrated lanes, and point-level paths can be the lowest priority.

[0082] However, if a high-definition map has not yet been built, or if it is not possible to follow a high-definition map due to construction or an accident, it is obviously possible to perform the determination based on the lanes detected by camera 112.

[0083] Figure 8 This is a flowchart illustrating the determination of basic information about the aforementioned integrated lane.

[0084] Figure 8 An example of the process for determining an integrated lane according to one implementation scheme is shown.

[0085] Reference Figure 8 The integrated lane calculation module 161 determines the lane based on the laneside in general cases except for the following: the laneside does not exist (S810 is), the laneside is discontinuous (S820 is), the shape of the laneside is not constant (S830 is), or the object is biased within the laneside (S840 is).

[0086] On the other hand, if the Laneside does not exist (yes in S810), the Laneside is discontinuous (yes in S820), the shape of the Laneside is not constant (yes in S830), or the object is biased within the Laneside (yes in S840), then the integrated lane calculation module 161 determines whether there is a difference between the point-level path (PLP) and the Lanelink (S850), and if they are the same or the difference is within a certain range (no in S850), the integrated lane can be derived based on the Lanelink (S860B). Here, the point-level path PLP can be a path calculated from the previous frame.

[0087] Furthermore, when there is a difference (greater than a certain extent) between the point-level path (PLP) and the lane link, the integrated lane calculation module 161 can derive the integrated lane based on the point-level path (PLP) (S860C).

[0088] Obviously, in addition to the above references Figure 8 In addition to the above methods, the basis for deriving integrated lanes can be predetermined based on the road type (urban area, highway, etc.) and interval (general interval, lane variable interval, special interval, etc.) shown in Table 1 below.

[0089] Table 1

[0090]

[0091] The classification criteria shown in Table 1 are exemplary, and it will be apparent to those skilled in the art that various different criteria can be set.

[0092] Meanwhile, since the point-level path (PLP) is determined in the final stage of the previous frame, the determination of the integrated lane needs to be performed after observing the driving strategy. The identification information from the identification sensor 110 is input in the form of relative distance, while the high-definition map has absolute coordinates (WGS84, UTM coordinate system, etc.). Therefore, in the case of an integrated lane that is fixed on the high-definition map, the integrated lane must be corrected (i.e., moved) according to changes in the vehicle's direction of travel and position. (Refer to...) Figure 9 Describe the necessity of such integrated lane correction.

[0093] Figure 9 This is a diagram illustrating the necessity of correcting and integrating lanes between frames.

[0094] Separately, in Figure 9 The left side shows the situation of the previous (N-1) frame. Figure 9 The right side shows the current (N) frame.

[0095] like Figure 9 As shown on the right, if the integrated lane is not corrected according to the vehicle's position changes, an error of tens of centimeters may occur because the vehicle's direction of travel and position changes are not reflected in a single frame. When it is determined that a cut-in with a resolution in centimeters is required, it cannot be determined that other vehicles 11 do not intersect with the integrated lane based on the integrated lane before correction (the uncorrected integrated lane), but the intersection point of other vehicles 11 with the integrated lane can be correctly determined through correction.

[0096] Reference Figure 10 Describe the correction method.

[0097] Figure 10 An example of the process of correcting integrated lanes between frames is shown according to one implementation scheme.

[0098] Reference Figure 10 In the above reference Figure 3 During the cut-in response control process, when determining the cut-in target (S302 to S307) in frame N, this determination can be achieved by transforming the point-level path (PLP) generated in frame N-1 from the local coordinate system to the global coordinate system using coordinate transformation. In other words, the integrated lane in the previous frame (N-1) can be fixed (stored) on the map as global coordinates, and the integrated lane fixed (stored) in the global coordinate system can be used in the next frame (N). In this way, in turning situations such as U-turns, P-turns, or lane changes, the relative position of the integrated lane can be correctly corrected through coordinate system re-transformation.

[0099] In the following text, reference will be made to Figures 11 to 16Describe in more detail the process of determining the final entry point (i.e., Figure 3 (S303 to S307).

[0100] To select a candidate vehicle for entry based on its current location (S303), the relative coordinates of each vehicle's body point need to be checked based on the integrated lane. This will be referenced... Figure 11 This will be described.

[0101] Figure 11 This is a schematic diagram illustrating the selection of candidate vehicles based on the current location according to an implementation scheme.

[0102] Reference Figure 11 The system predicts that the first other vehicle 11 will enter the driving lane of vehicle 100. However, if, during actual vehicle testing, the vehicle's deceleration control is applied based on a vehicle that lags too far behind, that vehicle could potentially reduce ride comfort. Therefore, using the rear bumper of vehicle 100 as a reference point, i.e., only other vehicles located in front of the rear bumper of vehicle 100 can be identified as targets. Subsequently, if the first other vehicle 11 intends to actually cut in, it will reach in front of the rear bumper of vehicle 100, thus eliminating any issues in actual identification.

[0103] The second other vehicle 12 can be normally identified as a candidate for entry.

[0104] In the case of the third other vehicle 13, since the other vehicle 16 occupies the target lane (i.e., the driving lane of this vehicle 100), it is physically impossible to enter within T seconds (determined reference time) even if the direction is towards the driving lane, so the third other vehicle 13 can be excluded from the candidates.

[0105] Even in the case of the fourth other vehicle 14, since it cannot be guaranteed that the distance between vehicles is sufficient for cutting in front of the fourth other vehicle 14, cutting in is physically impossible and can be excluded from the candidates. Obviously, when the distance between vehicles is subsequently guaranteed, it can be determined again whether to exclude the fourth other vehicle 14 from the candidates.

[0106] Next, we will refer to Figures 12 to 13B The calculation describes the predicted paths of nearby vehicles.

[0107] Figure 12 This is a schematic diagram illustrating the parameters used to generate a predicted path according to one implementation scheme.

[0108] Reference Figure 12 To the left, the predicted path is generated by considering the speed v of each vehicle and the bias value (lateral distance from the center of the lane to the center of the vehicle) of other vehicles 11.

[0109] In addition, such as Figure 12 As shown on the right, a predicted path is generated by considering the driving direction of other vehicles 11 (i.e., the driving direction angle θ), the driving lane link of other vehicles 11, the set of point coordinates of each point of the target lane link, and lane information.

[0110] When each of the above parameters is obtained, the path-based entry target determination module 164 can generate a predicted path for each entry target candidate determined by the location-based entry candidate determination module 163, based on non-training or based on dynamic training and high-definition map information.

[0111] When generating predicted paths for other vehicles using non-training techniques, the path-based entry target determination module 164 can consider the mutual influence between vehicles to determine the expected positions of other vehicles in each frame.

[0112] When using non-training techniques, the desired output path or lane change completion time can be calculated in tabular form for the input parameter set of other vehicles.

[0113] In this context, by utilizing the pre-stored predicted path or lane change times for each parameter set, mapping can be performed to follow a pre-planned mathematical model (n-order Bezier curve, 3rd-order polynomial, etc.) within the corresponding required time. Examples of mathematical models are shown in... Figure 13A and Figure 13B As shown in the image.

[0114] Figure 13A An example of curve-based prediction for generating a predicted path is shown according to one implementation scheme. Figure 13B An example of lane change timing modeled based on the bias value of nearby vehicles is shown.

[0115] Figure 13A This illustrates the modeling form of the predicted path based on the 5th-order Bezier curve for other vehicles. Figure 13B An example is shown where the time required for lane changes is modeled in the form of a grid diagram when the deviation value of other vehicles is 0.8 meters.

[0116] In the above mathematical modeling methods, besides referring to Figure 12 In addition to the parameters described, further considerations can be given to coordinate history sets, current velocity / acceleration, matching sensor information, high-resolution maps, and past high-resolution map matching history.

[0117] Clearly, the computation of such predicted paths can be performed by training the aforementioned parameter set using deep learning parameters and replacing it with a time series prediction problem using methods such as Convolutional Neural Networks (CNNs) or Long Short-Term Memory Networks (LSTMs). Furthermore, the driving intentions of other vehicles and the time required for lane changes can not only be determined once, but the reliability can also be improved through observations of multiple samples.

[0118] On the other hand, when the desired lane is derived based on dynamic training and high-resolution map information, the training results can be directly used as time-series location information. Only partial training is performed on the time required for lane changes, allowing mapping of the actual predicted path to follow the aforementioned mathematical model (n-order Bézier curve, third-order polynomial, etc.). When training on the dynamic information of other vehicles and high-resolution map information, neural networks used for predicting time-series data with CNNs and LSTMs are typically employed, but are not limited to these. It will be apparent to those skilled in the art that any neural network can be used as long as it can predict the predicted path or the time required for lane changes.

[0119] When determining the predicted path for each potential cutting-in target, the cutting-in determination can be performed based on the future positions of the current vehicle and other vehicles. (Refer to...) Figure 14 This will be described.

[0120] Figure 14 An example of cut-in determination based on the future location of a predicted path is shown in one implementation.

[0121] When determining the entry point, both the desired location of the target vehicles and the desired location of your own vehicle should be considered simultaneously. That is, when determining the entry point, the desired locations of nearby vehicles and your own vehicle need to be considered systematically. Figure 14 In the case where the predicted path eventually enters the driving lane of vehicle 100 after T seconds, the number of vehicles is a total of 5 vehicles, from the first other vehicle 11 to the fifth other vehicle 15.

[0122] When the vehicle 100 continues to travel in the driving lane, the actual threatening candidate for entry is the second other vehicle 12, so the second other vehicle 12 can be identified as the entry target.

[0123] The reason is that, although the second other vehicle 12 is currently behind the vehicle 100, the second other vehicle 12 is in front of the rear bumper, and position 12' after T seconds is in front of position 100' of the vehicle, so the second other vehicle 12 is taken into consideration.

[0124] Furthermore, in the case of the first other vehicle 11, although the first other vehicle 11 is currently in front of the vehicle 100, its position 11' after T seconds will be behind the vehicle 100', therefore the first other vehicle 11 is not a threat. However, even if the position 11' after T seconds is behind the vehicle 100, since the speed is variable, there may be a situation where a collision could occur within T seconds, but since it is recalculated every frame, this situation can be reclassified as a threat in the next frame.

[0125] In the case of the third other vehicle 13, the third other vehicle 13 is located at a considerable distance in front of the vehicle 100, and the fourth other vehicle 14 is located even further in front of the third other vehicle 13, so both the third other vehicle 13 and the fourth other vehicle 14 are excluded from the candidates.

[0126] For the same reason, the fifth other vehicle 15 was also excluded from the candidates.

[0127] On the other hand, regarding the second other vehicle 12 that is the target of the cut-in, since position 12' after T seconds becomes the adjacent front of position 100' of this vehicle, it can be determined that a dangerous deceleration response will be implemented. In the case of the third other vehicle 13, the fourth other vehicle 14, and the fifth other vehicle 15, a normal deceleration response can be executed instead of a dangerous deceleration response. That is, the cut-in deceleration response can be executed based on the closest future cut-in target.

[0128] Next, we will refer to Figure 15 and Figure 16 Describe the operation of the final effective entry target determination module 165.

[0129] Figure 15 An example of valid target verification based on a signal is shown according to one implementation scheme.

[0130] When traveling at intersections or other locations based on traffic light signals, vehicle 100 does not follow the designated route except under specified signal conditions. For example, such as... Figure 15 As shown, when waiting to make a U-turn, if there is a signal other than the signal that allows U-turns (depending on the intersection, such as the left turn signal, the go signal, etc.), this vehicle will not proceed to the point-level path corresponding to the U-turn.

[0131] Therefore, under all signals except the designated signal, if entry targets 11 and 12 to the integrated lane are identified and braking is applied, there is a problem that unnecessary braking may even hinder progress to the entry point. To avoid such a problem, it is necessary to additionally verify the validity of the target to be entered based on the signal. The final valid entry target determination module 165 can apply the traffic signal to the identified entry target to verify its validity based on whether the vehicle 100 has actually moved forward. However, when the final valid entry target determination module 165 itself performs filtering on the entry targets, it cannot respond to vehicles that violate the signal and can therefore treat them as "vehicles with invalid signals".

[0132] Figure 16 An example of valid target verification based on predicted path intersections is shown according to one implementation scheme.

[0133] Since the predicted path is primarily determined by the behavior of other vehicles, there may be instances where the interactions between other vehicles are not reflected. Therefore, if a collision occurs between vehicles on the predicted path, some or all of them may come to a stop and may not continue along the predicted path.

[0134] For example, such as Figure 16 As shown, when vehicles 12 and 13 make a U-turn, they will not make the U-turn if their predicted paths intersect with the predicted paths of vehicles 11 traveling straight along the centerline. That is, when other vehicles 11 are simultaneously traveling around the path of vehicle 100, and the predicted paths of other vehicles 12 and 13 are cutting into vehicle 100, other vehicles 12 and 13 can be excluded from the cutting-in target if their predicted paths are blocked by other vehicles 11.

[0135] As a result, refer to Figure 15 and Figure 16 As described above, for the entry target determined by the path-based entry target determination module 164, the final effective entry target determination module 165 performs effective verification by determining whether there is an intersection between the predicted paths of traffic signals or other nearby vehicles, and thus the final entry target can be determined.

[0136] In the following text, reference will be made to Figures 17 to 20 Describe the operation of the target control point calculation module 166.

[0137] Control points, or control target points, can serve as reference points for performing longitudinal control relative to an in-lane target. For example, in an in-lane pass, the distance to the center of the rear bumper and the speed at which the target passes within the lane can be a reference point. However, selecting control points can be challenging when choosing an in-lane target, as the current position of the in-lane target and the predicted future position where the path intersects with the integrating lane are also considered. Therefore, the selection of control points needs to take into account the continuous change from the initial position of the in-lane target to the point in time when the in-lane target intrudes into the integrating lane.

[0138] Even during the process of a vehicle cutting into the lane and passing through the area, it is preferable that the control points be continuous. This is because if the control points are discontinuous, it may cause the vehicle to brake suddenly or make a clicking sound. For this purpose, the predicted path, the intersection of other vehicle boxes with the integrated lane, and the shortest point (i.e., the orthogonal point) of the box point in the integrated lane on the vehicle's lane can be selected as control points.

[0139] Figure 17 An example of control point selection in a U-turn path according to one implementation is shown.

[0140] First refer to Figure 17 On the left side, when vehicle 100 makes a U-turn, calculate the intersection points 1710 and 1720 of the predicted path of the cutting target 11 with the integrated lane, and among the orthogonal points of intersection points 1710 and 1720, select the orthogonal point on the lane of vehicle 100 that is the shortest distance from vehicle 100 as the control point CP.

[0141] Subsequently, as the journey continues, when the target 11 approaches the vehicle's lane, it merges into the first intrusion point at the actual entry point.

[0142] For example, in Figure 17 On the right side, immediately after the cutting target 11 enters the lane of this vehicle, since the shortest point on the lane of this vehicle (i.e., control point CP) is the same as the first intrusion point 1710 of the cutting target 11 among the intersection point 1710' of the integrated lane and the box body and the orthogonal points 1730 and 1740 of the box body, according to the control point reference of the vehicle passing through the lane, it can be seen that even in the process of switching from the cutting target to the passing target in the lane, the continuity of the control point is guaranteed.

[0143] Lateral control is applied to veerging targets that will not encroach on the lane, rather than longitudinal control, thus preventing sudden braking or screeching. However, since veerging targets that will encroach on the lane are identified as cutting targets, continuity can be ensured as described above. (Refer to...) Figure 18 Let me describe this.

[0144] Figure 18 An example of control point selection in a straight path according to one implementation scheme is shown.

[0145] Reference Figure 18 In the upper left corner, when the vehicle 100 is traveling straight, calculate the intersection point 1810 of the predicted path of the cut-in target 11 and the integrated lane. Among the orthogonal points of the intersection point 1810, the orthogonal point that is closest to the vehicle 100 in the vehicle's lane can be selected as the control point CP.

[0146] In such Figure 18 In the case of the upper right, as the driving progresses, the intersection of the lane and the box body at point 1820 has the shortest distance on the vehicle's lane, and is therefore selected as the control point CP.

[0147] Next, in such Figure 18 In the case of the lower right, in the integrated lane, the orthogonal point of the rear box point 1830 of the target 11 has the shortest distance in the lane of this vehicle, and is therefore selected as the control point CP.

[0148] In addition, it can be seen that, as Figure 18 As shown in the lower left, when the cut-in target changes to a passing target within the lane, the continuity of the control point CP is guaranteed even in a straight path.

[0149] Furthermore, since a deviating target that has entered the lane of the vehicle but has not yet been determined to pass within the lane follows the same control point calculation standard as passing within the lane, the continuity of control points can also be guaranteed in this case.

[0150] On the other hand, when inaccurate identification information during control point selection leads to instability in the travel direction of other vehicles, the predicted path may change, and there is a risk of repeatedly identifying a vehicle as a cutting target and then disengaging from it. In this situation, to address the clicking sound of the vehicle, even for cutting targets, the longitudinal control response rate of the cutting target can be determined by calculating the rate of progress of entering the vehicle's lane. (Refer to...) Figure 19 This will be described.

[0151] Figure 19 An example is shown of determining the rate of progress of other vehicles entering the vehicle's lane according to one implementation scheme.

[0152] Reference Figure 19 On the left side, when other vehicles 11 approach vehicle 100 in the lateral direction and are identified as targets for cutting in, it can be done as follows: Figure 19The right side shows the calculation of the rate at which other vehicles enter the vehicle's lane, so that longitudinal control is gradually implemented based on the ratio of vehicles approaching the vehicle's lane in the lateral direction. For example, when 50% of the vehicles have entered the vehicle's lane from the initial position, the longitudinal control target speed can only reflect 50% compared to the final entry.

[0153] So far, reference Figures 17 to 19 The control point selection process described is summarized in the following flowchart.

[0154] When performing longitudinal control in response to a target approach, the main parameters are the longitudinal / lateral distance and speed of the vehicle ahead. Therefore, the distance from the vehicle's path (i.e., PLP) to the control point and the velocity component of the corresponding object at the control point on the vehicle's path (i.e., the component orthogonally projected with the velocity vector of other vehicles as the control point) correspond to the parameters used for the longitudinal speed control of the vehicle 100.

[0155] Therefore, to determine control points, it is necessary to calculate the position of the control point on the vehicle's path according to the control point determination method described above, and to calculate the path distance to the control point on the vehicle's path, as well as the speed components of other vehicles at the control point. The flowchart for this process is as follows: Figure 20 As shown.

[0156] Figure 20 This is a flowchart illustrating an example of a control point extraction process according to one implementation scheme.

[0157] Reference Figure 20 For the final entry target determined by the final effective entry target determination module 165, the entry target control point calculation module 166 can calculate the intersection point of the integrated lane with the predicted path starting from each vertex of the final entry target (S2010).

[0158] The target control point calculation module 166 can orthogonally project the intersection point onto the vehicle path (S2020). In this case, if the box point is within the integrated lane (i.e., an internal point), an orthogonal point can also be obtained for it.

[0159] Subsequently, the target control point calculation module 166 is engaged to determine the shortest point on the path of the vehicle among the orthogonal points (S2030), and calculates the cumulative distance to the corresponding point (S2040).

[0160] In addition, the target control point calculation module 166 can calculate the rate at which other vehicles enter the vehicle's lane based on the ratio of vehicles approaching the lane in the lateral direction (S2050) so as to gradually implement longitudinal control.

[0161] The target control point calculation module 166 can extract the velocity component at the orthogonal point (S2060). Here, the velocity component means the velocity of other vehicles in the lane of this vehicle, which is a scalar value obtained by orthogonally projecting the velocity vectors of other vehicles in the local coordinate system of this vehicle onto the tangent vector of the control point position in the lane of this vehicle.

[0162] The above process ultimately extracts the control points (S2070), and the speed distribution and control path can be calculated based on the extracted control points. For example, using the distance to the control point as the target distance and the speed at the control point as the target speed, the speed distribution and control path can be calculated using tracking control methods such as PID control.

[0163] By employing the cut-in response control method described so far, even at points of lane discontinuity, lane changes, and Lanelink anomalies, cut-in can be determined in the same manner as in normal lanes. In other words, a consistent integrated lane can be derived without exception each time a special road condition occurs.

[0164] Furthermore, because the control points for cutting in, passing within the lane, and deflection remain continuous, it effectively prevents clicking and sudden braking during longitudinal control.

[0165] In addition, the accuracy of target identification can be improved by considering the changing trend of the predicted path of the target candidate from the current to the future position and the integration of lanes, and the accuracy of target identification can be improved by using signal information and whether there are intersections in the predicted path.

[0166] The present invention described above can be implemented as computer-readable code on a medium for recording programs. Computer-readable media include various recording devices that store data readable by a computer system. Examples of computer-readable media include hard disk drives (HDDs), solid-state drives (SSDs), silicon disk drives (SDDs), ROMs, RAMs, CD-ROMs, magnetic tapes, floppy disks, optical data storage devices, etc.

[0167] Accordingly, the above detailed description should not be construed as restrictive in all respects, but rather as exemplary. The scope of the invention should be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the invention are included within its scope.

Claims

1. A method for controlling a vehicle's cut-in response, comprising the following steps: Obtain driving status information; Based on the acquired driving information, integrated lanes are derived by selectively applying road segment, lane edge, and point-level paths. The cut-in target is determined based on the predicted path of each of the integrated lanes and at least one nearby vehicle; The control points that the vehicle's driving control should follow are calculated based on the intersection of the predicted path of the target and the integrated lane. Speed ​​distribution and driving path are generated based on the calculated control points; Driving control is performed based on parameters corresponding to the speed distribution and driving path. The steps for exporting the integrated lane include the following: Determine whether the lane edge meets the preset conditions; If the preset conditions are met, the integrated lane is exported based on the lane edge; If the preset conditions are not met, determine whether there is a difference between the road segment and the point-level path; If discrepancies exist, the integrated lane is derived based on the point-level path; If there are no differences, then the integrated lane is derived based on the road segment.

2. The method for controlling vehicle cut-in response according to claim 1 further includes the following steps: When deriving integrated lanes based on point-level paths, the point-level path of the previous frame is transformed from the local coordinate system to the global coordinate system before determining the cut-in target.

3. The method for controlling vehicle cut-in response according to claim 1, wherein, The steps to determine the target to enter include the following: At least one cut-in candidate target is determined based on the relative coordinates of the box point of each of at least one nearby vehicle with respect to the integration lane. At least one entry target is determined based on the future location of the intersection of the predicted path of each of the at least one entry candidate target and the integrated lane, as well as the future location of the vehicle. Identify the final effective entry target among at least one entry target.

4. The method for controlling vehicle cut-in response according to claim 3, wherein, The steps to determine the final valid entry target include the following: filtering at least one entry target based on traffic light information and whether there is an intersection between the predicted paths of at least one nearby vehicle.

5. The method for controlling vehicle cut-in response according to claim 1, wherein, The steps for calculating control points include the following: For the intersection and at least one box point of the target cutting into the integrated lane, determine the orthogonal point on the vehicle's lane; Determine the shortest point on the vehicle's lane from the identified orthogonal points.

6. The method for controlling vehicle cut-in response according to claim 5, wherein, The steps for calculating control points further include the following: Determine the path distance to the shortest point; Determine the rate at which the target vehicle enters the lane of this vehicle; Extract the velocity component at the shortest point.

7. The method for controlling vehicle cut-in response according to claim 6, wherein, The steps of generating speed distribution and driving path are performed by setting the path distance to the shortest point as the target distance and the extracted speed component as the target speed, respectively.

8. A computer-readable recording medium having a program recorded thereon for performing a method for controlling a vehicle cut-in response according to any one of claims 1 to 7.

9. A vehicle for performing cut-in response control, comprising: A fusion information generator is used to acquire driving status information; The control parameter generator, based on the acquired driving information, derives an integrated lane by selectively applying road segment, lane edge, and point-level paths, determines an entry target based on the integrated lane and the predicted path of each of at least one nearby vehicle, calculates the control points to be followed by the vehicle's driving control based on the intersection of the predicted path of the entry target and the integrated lane, and generates a speed distribution and driving path based on the calculated control points. as well as The driving controller performs driving control based on parameters corresponding to the speed distribution and point-level path; The control parameter generator determines whether the lane edge meets the preset conditions, and if the preset conditions are met, the control parameter generator derives the integrated lane based on the lane edge. If the preset conditions are not met, the control parameter generator determines whether there is a difference between the road segment and the point-level path. If there is a difference, the control parameter generator derives the integrated lane based on the point-level path. If there is no difference, the control parameter generator derives the integrated lane based on the road segment.

10. The vehicle for performing cut-in response control according to claim 9, wherein, When deriving an integrated lane based on a point-level path, the control parameter generator performs a coordinate transformation to change the point-level path of the previous frame from the local coordinate system to the global coordinate system before determining the cut-in target.

11. The vehicle for performing cut-in response control according to claim 9, wherein, The control parameter generator determines at least one cut-in candidate based on the relative coordinates of the box point of each of at least one nearby vehicle with respect to the integration lane, determines at least one cut-in target based on the future position of the intersection of the predicted path of each of the at least one cut-in candidate and the integration lane and the future position of the vehicle, and determines the final valid cut-in target among the at least one cut-in target.

12. The vehicle for performing cut-in response control according to claim 11, wherein, The control parameter generator filters at least one cut-in target based on traffic light information and whether there is an intersection between the predicted paths of at least one nearby vehicle, thereby determining the final valid cut-in target.

13. The vehicle for performing cut-in response control according to claim 9, wherein, The control parameter generator determines the orthogonal point on the vehicle's lane for the intersection point and at least one box point of the target cutting into the integrated lane, and determines the shortest point on the vehicle's lane among the determined orthogonal points.

14. The vehicle for performing cut-in response control according to claim 13, wherein, The control parameter generator determines the path distance to the shortest point, determines the rate at which the target enters the vehicle's lane, and extracts the speed component at the shortest point.

15. The vehicle for performing cut-in response control according to claim 14, wherein, The control parameter generator generates speed distribution and driving path by setting the path distance to the shortest point as the target distance and the extracted speed component as the target speed, respectively.

Images