A method, system and terminal device for recognizing a target on a curved road based on road information

Abstract

The application discloses a curve target identification method and system based on road information and a terminal device, and the method comprises the following steps: according to the obtained current vehicle speed information, the current curve radius is obtained by performing radius prediction through an initial curve radius; the current curve radius is fused with the current vehicle speed information to obtain a longitudinal distance threshold; and based on a preset judgment condition and the longitudinal distance threshold, the curve target identification is performed through the obtained vehicle driving information, and a front following target of the vehicle is identified. The application can accurately identify the following target of the side lane by predicting the curve radius where the current vehicle is located, and improve the driving safety and the driving experience.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of intelligent driving, and particularly relates to a curve target recognition method and system based on road information and a terminal device. BACKGROUND

[0002] An adaptive cruise system (ACC) is a driving assistance system that adaptively adjusts the vehicle speed of a host vehicle in multiple scenarios according to the vehicle speed and following distance set by a driver. The system constructs a road environment according to sensing modules, algorithms, data training, and other information, and issues appropriate acceleration and deceleration under the control of a regulation algorithm to maintain the vehicle speed of the host vehicle at the user-set speed in the case where no target is recognized in front of the host vehicle. When a target is recognized, i.e., a vehicle is traveling in front of the host vehicle, the system can control the host vehicle to follow the front vehicle and maintain a certain safety distance according to the set time-distance level.

[0003] However, in some special scenarios, such as a curve scenario, due to the limitations of the sensing performance of a monocular camera, there is a certain probability of causing the mis-screening of a target in a side lane, which leads to the erroneous issuance of a deceleration instruction by the intelligent driving control end, resulting in unnecessary deceleration of the vehicle and affecting the comfort and experience of the driver and passengers. SUMMARY

[0004] The application provides a curve target recognition method and system based on road information and a terminal device, which accurately recognize the following target in a side lane by predicting the radius of the curve in which the host vehicle is located, thereby improving driving safety and driving experience.

[0005] The first aspect of the application provides a curve target recognition method based on road information, which comprises the following steps:

[0006] According to the obtained current vehicle speed information, the radius of the current curve is predicted by using the initial curve radius, so as to obtain the current curve radius;

[0007] The current curve radius and the current vehicle speed information are fused to obtain a longitudinal distance threshold value;

[0008] Based on the preset judgment condition and the longitudinal distance threshold value, the curve target recognition is performed by using the obtained vehicle driving information, so as to recognize the following target in front of the vehicle.

[0009] The above scheme first predicts the radius of the current curve according to the obtained vehicle speed information, to obtain a relatively accurate current curve radius; then fuses the current curve radius with the vehicle speed information to obtain a threshold value of the longitudinal distance between the vehicle and the front vehicle relative to the current curve radius, to provide data support for subsequent determination of a vehicle following target that is close to the vehicle; and then according to a set judgment condition, identifies a vehicle following target that is closest to the vehicle, is safe in lateral distance, and has no collision risk, by using the obtained vehicle driving information, so that the vehicle does not need to urgently adjust the vehicle speed because of mistakenly identifying a vehicle in another lane as a vehicle following target, thereby reducing the passenger comfort. Moreover, because the most suitable vehicle following target is accurately identified in the curve scene, the driving safety is also improved.

[0010] In a possible implementation method of the first aspect, according to the obtained current vehicle speed information, the current curve radius is obtained by radius prediction through an initial curve radius, specifically as follows:

[0011] According to a preset vehicle speed-radius increment table, the current vehicle speed information is subjected to interpolation processing to obtain a corresponding curve radius increment;

[0012] The curve reference value is output by conditional judgment on the initial curve radius, the curve radius output in the previous period, and the curve radius increment;

[0013] The curve reference value and the curve radius output in the previous first threshold period are subjected to sliding average calculation to obtain the current curve radius in the current period.

[0014] The above scheme first finds the curve radius increment corresponding to different vehicle speeds in a standard case through table lookup interpolation of the vehicle speed-radius increment table, which can prevent the curve radius fluctuation collected by the sensor in a short time from being too large, and reduce the calculation error in the subsequent calculation. Then the initial curve radius collected in the current period and the curve radius predicted in the previous period are subjected to conditional judgment to output a curve reference value that can most reduce the calculation error, and then the curve reference value and the curve radius predicted in the previous period are subjected to sliding average calculation to obtain the most accurate current curve radius, thereby reducing the risk that other vehicles in other lanes are easily identified as vehicle following targets.

[0015] In a possible implementation method of the first aspect, the curve reference value is output by conditional judgment on the initial curve radius, the curve radius output in the previous period, and the curve radius increment, specifically as follows:

[0016] If the initial curve radius is less than or equal to the sum of the curve radius output in the previous period and the curve radius increment, and the initial curve radius is greater than or equal to the difference between the curve radius output in the previous period and the curve radius increment, the curve radius output in the previous period is taken as the curve reference value;

[0017] Otherwise, the initial curve radius is taken as the curve reference value.

[0018] The above scheme uses the curve radius increment to determine whether the curve radius data collected by the sensor in a short time fluctuates too much. If the fluctuation is too large, the sensing data of the sensor is not reliable, because the lane width cannot change greatly in a short time, so the curve radius output in the previous period is taken as the curve reference value, which is more stable. If the fluctuation is not large, the initial curve radius collected at the current time can be taken as the curve reference value.

[0019] In a possible implementation method of the first aspect, the current curve radius and the current vehicle speed information are fused to obtain a longitudinal distance threshold, specifically:

[0020] According to a preset vehicle speed-scaling factor table, the current vehicle speed information is subjected to interpolation processing to obtain a distance scaling factor;

[0021] According to a preset radius-longitudinal distance table, the current curve radius is subjected to interpolation processing to obtain a longitudinal distance standard value;

[0022] The distance scaling factor and the longitudinal distance standard value are calculated to obtain the longitudinal distance threshold.

[0023] In a possible implementation method of the first aspect, based on a preset judgment condition and the longitudinal distance threshold, a curve target is identified by using the obtained vehicle driving information to identify a front following target of the vehicle, specifically:

[0024] The vehicle driving information at the current time is collected, and a self-driving trajectory of the vehicle and a target driving trajectory of the front vehicle are simulated according to the vehicle driving information and a preset trajectory simulation formula;

[0025] Based on the judgment condition, a plurality of judgment standards are set; wherein the judgment standards include a confidence degree standard, a lateral judgment standard and a no-crossing risk standard;

[0026] It is judged whether the vehicle driving information, the self-driving trajectory and the target driving trajectory satisfy the judgment standards, and when all the judgment standards are satisfied, the target longitudinal distance is compared with the longitudinal distance threshold;

[0027] If the target longitudinal distance is greater than or equal to the longitudinal distance threshold, it is determined that the front vehicle is not the front following target;

[0028] If the target longitudinal distance is less than the longitudinal distance threshold, it is determined that the front vehicle is the front following target, and the vehicle speed of the vehicle is adjusted according to the target longitudinal distance and the longitudinal distance threshold.

[0029] The above scheme first collects the lateral distance and longitudinal distance between the vehicle and the target vehicle to determine whether there is a collision risk between the vehicle and the target vehicle and whether the target vehicle is on the side lane according to a preset judgment condition, so as to avoid identifying vehicles in other lanes; finally, when the judgment condition is met, the target longitudinal distance is compared with the longitudinal distance threshold value, and if the target longitudinal distance exceeds the longitudinal distance threshold value, it means that the vehicle is far away from the target vehicle, and at this time, it is not suitable to take the target vehicle as a following target, because the emergency speed-up in order to shorten the distance between the vehicle and the target vehicle will reduce the driving comfort.

[0030] In a possible implementation method of the first aspect, determining whether the vehicle driving information, the self-driving trajectory and the target driving trajectory satisfy the judgment standard specifically comprises:

[0031] obtaining a target longitudinal distance between the vehicle and the target vehicle from the vehicle driving information;

[0032] determining whether the confidence degree standard is satisfied according to the target longitudinal distance, the speed signal perception time and the vehicle longitudinal relative speed;

[0033] determining whether the lateral judgment standard is satisfied according to the predicted lateral collision time;

[0034] determining whether the no-crossing risk standard is satisfied according to the self-driving trajectory and the target driving trajectory.

[0035] The above scheme first determines the distance and relative speed between the vehicle and the front vehicle according to the target longitudinal distance and the like, which will not affect the driving comfort; and then determines whether the vehicles are prone to collision due to too close distance by counting the predicted lateral collision time and the driving trajectories of the two vehicles, thereby improving the driving safety.

[0036] In a possible implementation method of the first aspect, determining whether the no-crossing risk standard is satisfied according to the self-driving trajectory and the target driving trajectory specifically comprises:

[0037] performing interpolation processing on the current curve radius to obtain a lateral distance reference value according to a preset radius-lateral distance table;

[0038] generating a lateral distance range according to the lateral distance reference value, and setting the self-driving trajectory equal to the target driving trajectory for solving;

[0039] if the self-driving trajectory is equal to the target driving trajectory in the lateral distance range, the no-crossing risk standard is satisfied;

[0040] otherwise, the no-crossing risk standard is not satisfied.

[0041] In a possible implementation manner of the first aspect, the self-driving trajectory and the target driving trajectory are specifically:

[0042] The self-driving trajectory is specifically expressed as:

[0043] Y = C0 + C1*X + C2*X 2 + C3*X 3 ;

[0044] In the formula, Y is the self-driving trajectory, C0, C1, C2 and C3 are coefficients of the lane center line, and X is a longitudinal coordinate of the vehicle.

[0045] The target driving trajectory is specifically expressed as:

[0046] W = A0 + A1*X + A2*X 2 + A3*X 3 ;

[0047] In the formula, W is the target driving trajectory, A0, A1, A2 and A3 are coefficients of the target driving trajectory, and X is a longitudinal coordinate of the target vehicle.

[0048] The second aspect of the present application provides a curve target identification system based on road information, the system comprising: a radius prediction module, a data fusion module and a following target identification module;

[0049] The radius prediction module is configured to perform radius prediction based on the obtained current speed information and an initial curve radius, to obtain a current curve radius.

[0050] The data fusion module is configured to perform information fusion on the current curve radius and the current speed information, to obtain a longitudinal distance threshold.

[0051] The following target identification module is configured to perform curve target identification based on a preset judgment condition and the longitudinal distance threshold, and vehicle driving information, to identify a following target in front of the vehicle.

[0052] The third aspect of the present application provides a terminal device, the device comprising: a terminal device comprising a processor and a memory, the memory storing a computer program, and the processor implementing the steps of any one of the curve target identification methods based on road information in the embodiments of the present application when executing the computer program. BRIEF DESCRIPTION OF DRAWINGS

[0053] In order to more clearly illustrate the technical solutions of the present application, the drawings needed to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and all other drawings obtained by those of ordinary skill in the art without creative effort based on these drawings also belong to the protection scope of the present application.

[0054] Figure 1 is a specific flowchart of a road information-based curve target identification method provided by an embodiment of the present application;

[0055] Figure 2 is a curve target misidentification diagram of a road information-based curve target identification method provided by an embodiment of the present application;

[0056] Figure 3 is a curve target identification strategy of a road information-based curve target identification method provided by an embodiment of the present application;

[0057] Figure 4 is a curve target correct identification diagram of a road information-based curve target identification method provided by an embodiment of the present application;

[0058] Figure 5 is a structural diagram of a road information-based curve target identification system provided by an embodiment of the present application;

[0059] Figure 6 is a structural diagram of a terminal device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0060] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort belong to the protection scope of the present application.

[0061] It should be understood that the step numbers used herein are only for the convenience of description, and are not limited as the execution sequence of the steps.

[0062] First Embodiment

[0063] Vehicles equipped with driver assistance cameras can achieve ACC adaptive cruise control, enabling stable speed and following of the vehicle in front, adapting to the speed of the vehicle ahead, and achieving stop-and-go functionality across the entire speed range and all road types. However, when the vehicle is traveling straight and the target vehicles ahead are all in curved environments, or when the vehicle is in a curve and the target vehicles are on a straight road, the vehicle cannot directly confirm whether the target vehicle is the correct one based solely on visual judgment using the driver assistance cameras. When the camera mistakenly selects a target, causing ACC to trigger unnecessarily and decelerate, it will affect the driver's normal driving and reduce the driving experience.

[0064] like Figure 1 As shown, Figure 1 This application provides a schematic flowchart of a curve target recognition method based on road information according to a certain embodiment. The curve target recognition method based on road information in this embodiment includes steps S1 to S3, which are described in detail below:

[0065] Step S1: Based on the obtained current vehicle speed information, predict the radius using the initial curve radius to obtain the current curve radius.

[0066] Currently, vehicles equipped with ACC systems can accurately identify the vehicle ahead using auxiliary cameras. However, when the vehicle is on a straight road and the target vehicles ahead are all in a curved environment, or when the vehicle is on a curve and the target vehicles are on a straight road, the vehicle's selection of the vehicle ahead mainly relies on the camera to identify lane lines, predict the vehicle's yaw angle and the target vehicle's trajectory, and then judge the target based on the time distance to determine whether to follow the target.

[0067] If a purely visual solution is used, that is, only cameras are used, the camera may mistakenly select a target when there are blind spots on curves or when the light is too dim to recognize lane lines. This can lead to the ACC triggering the deceleration unnecessarily, affecting the driver's normal driving. Figure 2 This illustrates a scenario of misidentification of a target vehicle on a curve. In the image, black represents the driver's vehicle, and red represents the misidentified target vehicle. Within the virtual lane lines, the misidentified target vehicle is the closest vehicle to the driver. In this situation, the misidentified target vehicle is inside the curve and in the adjacent lane. When the driver's ACC system triggers unnecessarily to decelerate in order to follow the misidentified target vehicle, the driver is prone to accidentally entering the adjacent lane, leading to driving safety issues. Therefore, a more accurate target identification method is needed to accurately select the appropriate target vehicle to follow in curve driving scenarios.

[0068] In this embodiment, the curve-related signals collected by the sensor are processed to limit the target recognition distance within a certain radius, thereby improving the false recognition phenomenon in target screening and finding suitable following targets more accurately.

[0069] Firstly, the speed signal of the ego vehicle is collected by the sensor to obtain the current speed information of the vehicle, and a look-up table interpolation is performed according to a preset speed-radius increment table to obtain the curve radius increment under different speeds. In the embodiment of the application, the look-up table interpolation is a linear interpolation method, and linear interpolation is performed in each speed interval.

[0070] For example, Table 1 below is a speed-radius increment table used in the embodiment of the application. In the table, v is the vehicle speed, and dy is the curve radius increment. When the current speed information of the vehicle is 26 m / s, the corresponding curve radius increment can be determined as 60 m.

[0071] Table 1: Speed-radius increment table

[0072]

[0073] Specifically, the curve radius increment is mainly used to prevent the target vehicle from being misrecognized due to the large fluctuation of the curve radius collected by the sensor at the front end of the vehicle head in a short time. Because the change of the lane radius will not be too large in a short time, and environmental factors such as light intensity and whether the lane line is clear can cause the sensor to collect large error data, the curve radius increment is introduced in the embodiment of the application to adjust the length of the curve radius, so that the obtained curve radius value is stable, and the increase of the error such as value jump or distortion caused by poor sensor recognition accuracy is prevented.

[0074] After obtaining the corresponding curve radius increment, a conditional judgment is performed according to the initial curve radius collected by the sensor, the curve radius output in the previous period and the curve radius increment, and a curve reference value is output, specifically as follows:

[0075] If the initial curve radius >= the curve radius output in the previous period - the curve radius increment, and the initial curve radius <= the curve radius output in the previous period + the curve radius increment, the curve radius output in the previous period is taken as the curve reference value;

[0076] Otherwise, the initial curve radius is taken as the curve reference value.

[0077] The above process is a conditional judgment of the initial curve radius collected in the current period and the curve radius predicted in the previous period, and a relatively stable curve reference value that can best reduce the calculation error is output, so as to avoid using data with large fluctuation for subsequent calculation and increase the data error.

[0078] Then, the obtained curve reference value is filtered, the curve radius and the curve reference value output in the previous n periods are calculated by sliding average, and the average value is taken as the current curve radius in the current period. The specific calculation formula is as follows:

[0079] Rd_out = (Rd_fill + Rd_out(0) + Rd_out(1) +... + Rd_out(n)) / n;

[0080] In the formula, Rd_out is the current curve radius, Rd_fill is the curve reference value, and n is the number of previous periods.

[0081] Optionally, in some embodiments, n is 9, that is, the output curve radius and curve reference value in the previous nine periods are taken for sliding average calculation.

[0082] In step S2, the current curve radius and the current vehicle speed information are fused to obtain a longitudinal distance threshold.

[0083] In the embodiments of the present application, the distance scaling factor is determined according to the current vehicle speed information by table lookup interpolation through a preset vehicle speed-scaling factor table. The distance scaling factor is used to proportionally scale the longitudinal distance threshold according to the vehicle speed to ensure driving safety.

[0084] Table 2 below is the vehicle speed-scaling factor table used in the embodiments of the present application:

[0085] Table 2: Vehicle speed-scaling factor table

[0086]

[0087] In the table, v is the vehicle speed, is the distance scaling factor; when the vehicle speed is 16.7 m / s, the corresponding distance scaling factor is 0.88, that is, the longitudinal distance threshold is scaled by a proportion of 0.88.

[0088] The obtained current curve radius is then linearly interpolated through a preset radius-longitudinal distance table to obtain a longitudinal distance standard value. The radius-longitudinal distance table used in the embodiments of the present application is shown in Table 3 below:

[0089] Table 3: Radius-longitudinal distance table

[0090]

[0091] In the table, Rd_out is the current curve radius, and d_long is the basic recognition distance.

[0092] Finally, the basic recognition distance is multiplied by the distance scaling factor to obtain the corresponding longitudinal distance threshold. The longitudinal distance threshold is a condition threshold for determining a following target in the longitudinal distance of a front target when the curve radius is less than 1800 m. If the front target vehicle exceeds the longitudinal distance threshold, it means that the distance from the ego vehicle is too far, and the front target vehicle is not determined as a following target.

[0093] In step S3, a target recognition is performed on the curve based on the preset judgment condition and the longitudinal distance threshold value by using the obtained vehicle driving information, and a front following target of the vehicle is recognized.

[0094] In the embodiments of the present application, three judgment standards are set according to the preset judgment condition, which are confidence standard, lateral judgment standard and no-crossing risk standard, for judging whether the longitudinal distance between the vehicle and the front target vehicle is too close or too far, whether the lateral collision between the vehicle and the front target vehicle is possible, and whether there is a crossing risk between the two vehicles.

[0095] The vehicle driving information at the current time is collected, including the target longitudinal distance, the target lateral distance and the longitudinal relative speed between the vehicle and the front target vehicle.

[0096] It is judged whether the driving state of the vehicle meets the confidence standard: when the time for the sensor of the ego vehicle to continuously perceive the front target vehicle is more than 0.15s, the distance and the relative speed between the ego vehicle and the front target vehicle are judged. When the target longitudinal distance is less than 120m or the longitudinal relative speed is less than -5m / s, it is considered that the driving state of the vehicle meets the confidence standard.

[0097] It is judged whether the driving state of the vehicle meets the lateral judgment standard: in the curve scene, if the predicted lateral collision time between the ego vehicle and the front target vehicle is greater than 0.5s, it is considered that the lateral risk with the target vehicle is small, and it is considered that the lateral judgment standard is met. The lateral collision time is predicted based on the target lateral distance.

[0098] It is judged whether the driving state of the vehicle meets the no-crossing risk standard: according to the vehicle driving information and the preset trajectory simulation formula, the self-driving trajectory of the vehicle and the target driving trajectory of the front vehicle are simulated. Then, the lateral distance reference value is obtained by table lookup interpolation based on the current curve radius through the preset radius-lateral distance table, as shown in Table 4 below:

[0099] Table 4 Radius-lateral distance table

[0100]

[0101] In the table, X_rel is the lateral distance reference value.

[0102] According to the lateral distance reference value, a corresponding lateral distance range is generated, and then the self-driving trajectory is equal to the target driving trajectory to solve. If the self-driving trajectory is equal to the target driving trajectory within the lateral distance range, the no-crossing risk standard is met, indicating that there is no trajectory crossing risk between the ego vehicle and the front target vehicle.

[0103] Exemplarily, the lateral distance range of the embodiment of the present application is 0~X_rel.

[0104] wherein the self-driving trajectory is specifically expressed as:

[0105] Y = C0 + C1*X + C2*X 2 + C3*X 3 ;

[0106] In the formula, Y is the self-driving trajectory, C0, C1, C2, and C3 are coefficients of the lane center line, which are generally provided by the road information fusion module of the vehicle; and X is the longitudinal coordinate of the vehicle.

[0107] The target driving trajectory is specifically expressed as:

[0108] W = A0 + A1*X + A2*X 2 + A3*X 3 ;

[0109] In the formula, W is the target driving trajectory, A0, A1, A2, and A3 are coefficients of the target driving trajectory, which are generally provided by the target trajectory processing module of the vehicle; and X is the longitudinal coordinate of the target vehicle.

[0110] When the three standards are all met, it indicates that the longitudinal distance and the lateral distance between the front target vehicle and the vehicle are safe and reasonable, and there is no intersection risk between the driving trajectories of the two vehicles, and there is no collision risk or the vehicle is not a side lane vehicle, so the recognition and judgment mechanism can be started, and whether the front target vehicle can be used as a following target is determined by the target longitudinal distance and the longitudinal distance threshold.

[0111] Specifically, if the target longitudinal distance is greater than or equal to the longitudinal distance threshold, it is determined that the front vehicle is not a front following target; if the target longitudinal distance is less than the longitudinal distance threshold, it is determined that the front vehicle is a front following target, and the driving speed of the vehicle is adjusted according to the target longitudinal distance and the longitudinal distance threshold.

[0112] Figure 3 The target recognition strategy of the embodiment of the present application is provided, and the whole strategy is divided into three modules: signal filtering, information fusion, and target data. The signal filtering is to filter the speed information and the curve radius collected by the sensor of the vehicle, screen out reasonable data for data fusion, and finally output the recognized following target combined with the simulated driving trajectory. Compared with the prior art, not only the visual scheme provided by the camera is relied on, but also the predicted curve radius range is strictly controlled through the collected data, so as to avoid that the recognized following target is actually in other lanes.

[0113] The vehicle will not identify the vehicle in the side lane as a following target under the curve target identification strategy provided by the embodiments of the present application. Figure 4 The vehicle will not identify the vehicle in the side lane as a following target under the curve target identification strategy provided by the embodiments of the present application.

[0114] The embodiments of the present application have the following beneficial effects:

[0115] The embodiments of the present application first predict the curve radius of the current vehicle according to the acquired vehicle speed information, obtaining a relatively accurate current curve radius; then fuse the current curve radius with the vehicle speed information, obtaining a threshold value of the longitudinal distance of the vehicle and the front vehicle relative to the current curve radius, providing data support for subsequent determination of the following target closest to the vehicle; then according to the set judgment condition, the following target closest to the vehicle, safe in lateral distance and free of collision risk is identified through the acquired vehicle driving information, so that the vehicle does not need to adjust the speed urgently in the driving process because of mistakenly identifying the vehicle in other lanes as a following target, resulting in reduced passenger comfort. Moreover, because the most suitable following target is accurately identified in the curve scene, driving safety is also improved.

[0116] Second embodiment

[0117] Further, in order to execute the road information based curve target identification system corresponding to the above-mentioned method embodiment, to realize the corresponding functions and technical effects, Figure 5 A structural diagram of a road information based curve target identification system is provided. For ease of illustration, only the part related to the present embodiment is shown. The road information based curve target identification system provided by the embodiments of the present application comprises:

[0118] The radius prediction module 201 is configured to perform radius prediction through the initial curve radius according to the acquired current vehicle speed information, obtaining the current curve radius.

[0119] In the embodiments of the present application, the current vehicle speed information is subjected to interpolation processing according to a preset vehicle speed-radius increment table, obtaining a corresponding curve radius increment; the initial curve radius, the curve radius output in the previous period and the curve radius increment are subjected to conditional judgment, outputting a curve reference value; the curve reference value and the curve radius output in the first threshold value periods are subjected to sliding average calculation, obtaining the current curve radius in the current period.

[0120] The data fusion module 202 is configured to fuse the current curve radius with the current vehicle speed information, obtaining a longitudinal distance threshold value.

[0121] In the embodiment of the present application, the distance scaling factor is determined by looking up a preset speed-scaling factor table according to the current vehicle speed information. The distance scaling factor is used to scale the longitudinal distance threshold according to the vehicle speed, to ensure driving safety.

[0122] The following Table 5 is the speed-scaling factor table used in the embodiment of the present application:

[0123] Table 5: Speed-scaling factor table

[0124]

[0125] In the table, v is the vehicle speed, is the distance scaling factor; when the vehicle speed is 16.7 m / s, the corresponding distance scaling factor is 0.88, i.e., the longitudinal distance threshold is scaled by a factor of 0.88.

[0126] The obtained current curve radius is then linearly interpolated by a preset radius-longitudinal distance table to obtain a longitudinal distance standard value. The radius-longitudinal distance table used in the embodiment of the present application is shown in the following Table 6:

[0127] Table 6: Radius-longitudinal distance table

[0128]

[0129] In the table, Rd_out is the current curve radius, and d_long is the basic recognition distance.

[0130] Finally, the basic recognition distance is multiplied by the distance scaling factor to obtain the corresponding longitudinal distance threshold. The longitudinal distance threshold is a condition threshold for determining a following target in the longitudinal distance of the front target when the curve radius is less than 1800 m. If the front target vehicle exceeds the longitudinal distance threshold, it means that the distance from the ego vehicle is too far, and the front target vehicle is not determined as a following target.

[0131] The following target recognition module 203 is configured to perform curve target recognition based on the preset judgment condition and the longitudinal distance threshold, by using the obtained vehicle driving information, to identify a front following target of the vehicle.

[0132] In the embodiment of the present application, the vehicle driving information at the current time is collected, and a self-driving trajectory of the vehicle and a target driving trajectory of the front vehicle are simulated according to the vehicle driving information and a preset trajectory simulation formula.

[0133] Based on the judgment condition, a plurality of judgment standards are set; the judgment standards include a confidence level standard, a lateral judgment standard, and a no-crossing risk standard.

[0134] determining whether the vehicle driving information, the self-driving trajectory and the target driving trajectory satisfy the determination criteria, and comparing the target longitudinal distance with the longitudinal distance threshold when all the determination criteria are satisfied;

[0135] if the target longitudinal distance is greater than or equal to the longitudinal distance threshold, determining that the front vehicle is not a front following target;

[0136] if the target longitudinal distance is less than the longitudinal distance threshold, determining that the front vehicle is a front following target, and adjusting the vehicle speed of the vehicle according to the target longitudinal distance and the longitudinal distance threshold.

[0137] In some embodiments, the radius prediction module 201 further comprises:

[0138] The vehicle currently configured with the ACC system can accurately identify the front following target through the auxiliary camera, but when the ego vehicle is on a straight road and the front target vehicle is in a curved road environment, or the vehicle is in a curved road and the target vehicle is in a straight road, the selection of the front vehicle by the ego vehicle mainly relies on the camera to identify the lane line, the driving trajectory prediction of the vehicle yaw angle and the driving trajectory prediction of the front target vehicle, and then the target is determined according to the time distance to determine whether the target is taken as a following target.

[0139] If only the visual solution is used, that is, only the camera is used, when there is a blind area in the curve or the light is too dark and the camera cannot identify the lane line, etc., the camera is easy to misselect the target, thereby causing the ACC to trigger the false deceleration and affecting the normal driving of the driver.

[0140] In the embodiments of the present application, the target recognition distance within a certain radius range is limited by processing the curve related signals collected by the sensor, so as to improve the misidentification phenomenon of target screening and more accurately find the appropriate following target.

[0141] First, the speed signal of the ego vehicle is collected by the sensor to obtain the current speed information of the vehicle and perform table lookup interpolation according to the preset speed-radius increment table to obtain the curve radius increment under different vehicle speeds. In the embodiments of the present application, the table lookup interpolation is a linear interpolation method, and linear interpolation is performed in each vehicle speed interval.

[0142] For example, Table 7 below is the speed-radius increment table used in the embodiments of the present application. In Table 7, v is the vehicle speed, dy is the curve radius increment, and when the current speed information of the vehicle is 26 m / s, the corresponding curve radius increment can be determined as 60 m.

[0143] Table 7: Speed-radius increment table

[0144]

[0145] Specifically, the bend radius increment is mainly used to prevent the target vehicle from misrecognizing due to the too large fluctuation of the bend radius collected by the sensor at the front end of the vehicle head in a short time. Because the change of the lane radius will not be too large in a short time, and the environmental factors such as the light intensity and whether the lane line is clear will cause the sensor to collect data with large errors, the bend radius increment is introduced in the embodiment of the application to adjust the length of the bend radius, so that the obtained bend radius value is stable, and the error increase condition such as value jump or distortion caused by poor sensor recognition accuracy is prevented.

[0146] After the corresponding bend radius increment is obtained, conditional judgment is performed according to the initial bend radius currently collected by the sensor, the bend radius output in the previous period and the bend radius increment, and a bend reference value is output, specifically:

[0147] If the initial bend radius >= the bend radius output in the previous period - the bend radius increment, and the initial bend radius <= the bend radius output in the previous period + the bend radius increment, the bend radius output in the previous period is taken as the bend reference value;

[0148] Otherwise, the initial bend radius is taken as the bend reference value.

[0149] The above process is to perform conditional judgment on the initial bend radius currently collected and the bend radius predicted in the previous period, output a relatively stable bend reference value which can most reduce the calculation error, and avoid using data with too large fluctuation for subsequent calculation to increase the data error.

[0150] Then, the obtained bend reference value is filtered, the bend radii and the bend reference values output in the previous n periods are taken for sliding average calculation, and the obtained average value is taken as the current bend radius in the current period, and the specific calculation formula is:

[0151] Rd_out = (Rd_fill + Rd_out(0) + Rd_out(1) +... + Rd_out(n)) / n;

[0152] In the formula, Rd_out is the current bend radius, Rd_fill is the bend reference value, and n is the previous n periods.

[0153] In some embodiments, the following steps are further included.

[0154] Three judgment criteria are set according to the preset judgment condition, which are confidence criterion, lateral judgment criterion and no intersection risk criterion, for judging whether the longitudinal distance between the vehicle and the front target vehicle is too close or too far, whether the vehicle and the front target vehicle are likely to have a lateral collision, and whether there is an intersection risk between the two vehicles.

[0155] The vehicle driving information at the current time is collected, including target longitudinal distance, target lateral distance, longitudinal relative speed, etc. between the vehicle and the front target vehicle.

[0156] It is judged whether the driving state of the vehicle meets the confidence criterion: when the time for the ego vehicle to continuously perceive the front target vehicle is more than 0.15s, the distance and relative speed between the ego vehicle and the front target vehicle are judged. When the target longitudinal distance is less than 120m or the longitudinal relative speed is less than -5m / s, it is considered that the driving state of the vehicle meets the confidence criterion.

[0157] It is judged whether the driving state of the vehicle meets the lateral judgment criterion: in the scenario of a curve, if the predicted lateral collision time between the ego vehicle and the front target vehicle is greater than 0.5s, it is considered that the lateral risk with the target vehicle is small, and it is considered that the lateral judgment criterion is met. The lateral collision time is predicted based on the target lateral distance.

[0158] It is judged whether the driving state of the vehicle meets the non-crossing risk criterion: according to the vehicle driving information and a preset trajectory simulation formula, the ego driving trajectory of the vehicle and the target driving trajectory of the front vehicle are simulated. Then, based on the current curve radius, a radius-lateral distance table is looked up and interpolated to obtain a lateral distance reference value, as shown in Table 8 below:

[0159] Table 8 Radius-lateral distance table

[0160]

[0161] In the table, X_rel is the lateral distance reference value.

[0162] According to the lateral distance reference value, a corresponding lateral distance range is generated, and the ego driving trajectory is equal to the target driving trajectory to solve. If the ego driving trajectory is equal to the target driving trajectory within the lateral distance range, the non-crossing risk criterion is met, indicating that there is no trajectory crossing risk between the ego vehicle and the front target vehicle.

[0163] For example, the lateral distance range of the embodiment of the application is 0~X_rel.

[0164] Wherein, the specific expression of the ego driving trajectory is:

[0165] Y = C0 + C1*X + C2*X 2 + C3*X 3 ;

[0166] In the formula, Y is the self-driving trajectory, C0, C1, C2, and C3 are coefficients of the lane center line, which are generally provided by a road information fusion module of the vehicle; and X is a longitudinal coordinate of the vehicle.

[0167] The target driving trajectory is specifically expressed as:

[0168] W = A0 + A1*X + A2*X 2 + A3*X 3 ;

[0169] In the formula, W is the target driving trajectory, A0, A1, A2, and A3 are coefficients of the target driving trajectory, which are generally provided by a target trajectory processing module of the vehicle; and X is a longitudinal coordinate of the target vehicle.

[0170] When the three criteria are all met, it indicates that the longitudinal distance and the lateral distance between the front target vehicle and the vehicle are safe and reasonable, and there is no risk of crossing the driving trajectory and no risk of collision between the two vehicles, and the vehicle is not a side lane vehicle, at this time, the recognition and judgment mechanism can be started, and whether the front target vehicle can be used as a following target is determined by the target longitudinal distance and the longitudinal distance threshold.

[0171] Specifically, if the target longitudinal distance is greater than or equal to the longitudinal distance threshold, it is determined that the front vehicle is not a front following target; and if the target longitudinal distance is less than the longitudinal distance threshold, it is determined that the front vehicle is a front following target, and the driving speed of the vehicle is adjusted according to the target longitudinal distance and the longitudinal distance threshold.

[0172] The embodiment of the present application has the following beneficial effects:

[0173] The embodiment of the present application first predicts the radius of the current curve according to the acquired vehicle speed information, to obtain a more accurate current curve radius; then fuses the current curve radius with the vehicle speed information to obtain a threshold for the longitudinal distance between the vehicle and the front vehicle under the current curve radius, to provide data support for subsequent determination of a following target closest to the vehicle; and then according to the set judgment condition, a following target closest to the vehicle, safe in lateral distance, and free of collision risk is recognized through the acquired vehicle driving information, so that the vehicle does not need to be urgently adjusted in speed during driving because of mistakenly identifying a vehicle in another lane as a following target, thereby reducing the passenger comfort. Moreover, because the most suitable following target is accurately recognized in the curve scene, the driving safety is also improved.

[0174] Further, Figure 6 A structural diagram of a terminal device provided by an embodiment of the present application is shown in FIG. 1. As shown in the figure, the terminal device 3 of the embodiment includes at least one processor 30 (in the embodiment, two processors 30 are provided) and a memory 31. Figure 6 The processor 30 is configured to perform the following steps:Figure 6 The terminal device 3 can be a desktop computer, a cloud server, a notebook computer, and the like, which can include but is not limited to the processor 30, the memory 31.

[0175] The terminal device 3 can be a desktop computer, a cloud server, a notebook computer, and the like, which can include but is not limited to the processor 30, the memory 31. Figure 6 The terminal device 3 is only an example and does not constitute a limitation on the terminal device 3, and can include more or fewer components than shown.

[0176] The above-described specific embodiments further illustrate the purposes, technical solutions, and beneficial effects of the present application. It should be understood that the above-described specific embodiments are only for the purpose of the present application and do not limit the protection scope of the present application. It is particularly pointed out that any modification, equivalent replacement, improvement, etc. made by those skilled in the art within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A curve target recognition method based on road information, characterized by, The application relates to a method for identifying a front-following target of a vehicle. According to the acquired current vehicle speed information, a current curve radius is obtained by radius prediction based on an initial curve radius. The current curve radius and the current vehicle speed information are fused to obtain a longitudinal distance threshold value. Based on a preset judgment condition and the longitudinal distance threshold value, a front-following target of the vehicle is identified by acquired vehicle driving information, specifically: the vehicle driving information at the current time is collected, and a self-driving trajectory of the vehicle and a target driving trajectory of the front vehicle are simulated according to the vehicle driving information and a preset trajectory simulation formula; based on the judgment condition, a plurality of judgment standards are set; the judgment standards include a confidence degree standard, a lateral judgment standard and a non-crossing risk standard; whether the vehicle driving information, the self-driving trajectory and the target driving trajectory meet the judgment standards is judged, when all the judgment standards are met, a target longitudinal distance is compared with the longitudinal distance threshold value; if the target longitudinal distance is greater than or equal to the longitudinal distance threshold value, it is determined that the front vehicle is not the front-following target; if the target longitudinal distance is less than the longitudinal distance threshold value, it is determined that the front vehicle is the front-following target, and the vehicle speed of the vehicle is adjusted according to the target longitudinal distance and the longitudinal distance threshold value.

2. The road information-based curve target recognition method according to claim 1, characterized by, According to the acquired current vehicle speed information, a current curve radius is obtained by radius prediction based on an initial curve radius. The current curve radius and the current vehicle speed information are fused to obtain a longitudinal distance threshold value. According to a preset vehicle speed-radius increment table, the current vehicle speed information is subjected to interpolation processing to obtain a corresponding curve radius increment; The initial curve radius, the curve radius output in the previous period and the curve radius increment are subjected to conditional judgment to output a curve reference value; 3. The curve object recognition method based on road information according to claim 2, characterized by, The curve reference value and the curve radius output in the first threshold value periods are subjected to sliding average calculation to obtain the current curve radius in the current period. The initial curve radius, the curve radius output in the previous period and the curve radius increment are subjected to conditional judgment to output a curve reference value. If the initial curve radius is less than or equal to the sum of the curve radius output in the previous period and the curve radius increment, and the initial curve radius is greater than or equal to the difference between the curve radius output in the previous period and the curve radius increment, the curve radius output in the previous period is taken as the curve reference value; 4. The road information-based curve target recognition method according to claim 1, characterized by, Otherwise, the initial curve radius is taken as the curve reference value. The current curve radius and the current vehicle speed information are fused to obtain a longitudinal distance threshold value. According to a preset vehicle speed-radius increment table, the current vehicle speed information is subjected to interpolation processing to obtain a corresponding curve radius increment; According to a preset radius-longitudinal distance table, the current curve radius is subjected to interpolation processing to obtain a longitudinal distance standard value; 5. The road information-based curve target recognition method according to claim 1, characterized by, The distance scaling factor and the longitudinal distance standard value are calculated to obtain the longitudinal distance threshold value. The vehicle driving information, the self-driving trajectory and the target driving trajectory are subjected to conditional judgment to output a curve reference value. The target longitudinal distance between the vehicle and the target vehicle is obtained from the vehicle driving information; determining whether the confidence criterion is satisfied according to the target longitudinal distance, the speed signal perception time and the vehicle longitudinal relative speed; determining whether the lateral determination criterion is satisfied according to the predicted lateral collision time; determining whether the no-crossing risk criterion is satisfied according to the self-driving trajectory and the target driving trajectory.

6. The curve object recognition method based on road information according to claim 5, characterized by, The determining whether the no-crossing risk criterion is satisfied according to the self-driving trajectory and the target driving trajectory specifically comprises: interpolating the current curve radius according to a preset radius-lateral distance table to obtain a lateral distance reference value; generating a lateral distance range according to the lateral distance reference value, and setting the self-driving trajectory equal to the target driving trajectory for solving; if the self-driving trajectory is equal to the target driving trajectory within the lateral distance range, the no-crossing risk criterion is satisfied; otherwise, the no-crossing risk criterion is not satisfied.

7. The road information-based curve target recognition method according to claim 1, characterized by, The self-driving trajectory and the target driving trajectory specifically comprise: The self-driving trajectory has a specific expression as follows: Y = C0 + C1*X + C2*X 2 + C3*X 3 ; wherein Y is the self-driving trajectory, C0, C1, C2 and C3 are coefficients of a lane center line, and X is a longitudinal coordinate of the vehicle. The target driving trajectory has a specific expression as follows: W = A0 + A1*X + A2*X 2 + A3*X 3 ; wherein W is the target driving trajectory, A0, A1, A2 and A3 are coefficients of the target driving trajectory, and X is a longitudinal coordinate of a target vehicle.

8. A curve target recognition system based on road information, characterized by, It comprises: a radius prediction module, a data fusion module and a following target identification module; The radius prediction module is configured to perform radius prediction on the initial curve radius according to the obtained current speed information to obtain the current curve radius. The data fusion module is configured to perform information fusion on the current curve radius and the current speed information to obtain a longitudinal distance threshold. The following target identification module is configured to perform curve target identification on the obtained vehicle driving information based on a preset determination condition and the longitudinal distance threshold, and identify a following target in front of the vehicle, specifically: collecting the vehicle driving information at the current time, and simulating a self-driving trajectory of the vehicle and a target driving trajectory of a front vehicle according to the vehicle driving information and a preset trajectory simulation formula; setting a plurality of determination criteria based on the determination condition; the determination criteria include a confidence criterion, a lateral determination criterion and a no-crossing risk criterion; determining whether the vehicle driving information, the self-driving trajectory and the target driving trajectory satisfy the determination criteria; when all the determination criteria are satisfied, comparing a target longitudinal distance with the longitudinal distance threshold; if the target longitudinal distance is greater than or equal to the longitudinal distance threshold, it is determined that the front vehicle is not the following target; if the target longitudinal distance is less than the longitudinal distance threshold, it is determined that the front vehicle is the following target, and the speed of the vehicle is adjusted according to the target longitudinal distance and the longitudinal distance threshold.

9. A terminal device, comprising: It comprises a processor and a memory, the memory stores a computer program, and the processor executes the computer program to realize the steps of the road information-based curve target identification method in any one of claims 1 to 7.

Images