Vehicle adaptive braking optimization method and device in complex scene and medium

Abstract

The invention discloses a vehicle adaptive braking optimization method and device in a complex scene and a medium, and the method comprises the steps: increasing an offset value threshold value of a pedestrian invading a vehicle driving path under the condition that the pedestrian is in a non-crossing scene, and obtaining a non-crossing scene adjustment result; under the condition that the pedestrian is in the crossing scene, filtering the transverse speed of the pedestrian to obtain a crossing scene adjustment result; and under the condition that the pedestrian is in the curve scene, reducing the curvature value of the vehicle track to a preset value to obtain a curve scene adjustment result. According to the vehicle adaptive braking optimization method and device in the complex scene and the medium, adaptive adjustment is carried out on the vehicle braking modes in the non-traversing scene, the traversing scene and the curve scene, the emergency braking adjustment result of the vehicle is jointly formed, the actual conditions and potential risks in different scenes can be fully considered, and the vehicle emergency braking adjustment efficiency is improved. The problem that collision early warning and automatic emergency braking are likely to be triggered by mistake in a complex scene of a vehicle can be solved.

Description

Technical Field

[0001] The present invention relates to autonomous driving technology, and in particular to a method, device and medium for optimizing vehicle adaptive braking in complex scenarios. Background Art

[0002] Vehicles equipped with pure driving assistance cameras, driving assistance cameras + radars or radars can detect the vehicles (motor vehicles & non-motor vehicles) or pedestrians in front in real time, determine the distance, direction and relative speed between the vehicle and the vehicle in front (pedestrians), and guide the driver to deal with it in time through voice prompts when there is a potential collision risk and the distance to the vehicle in front is less than the safe range; the equipped ADAS visual subsystem will identify pedestrians around the vehicle, detect the distance between pedestrians and vehicles, and make early warning judgments, thereby providing assistance for safe driving. The traditional pure driving assistance camera solution mainly uses cameras as sensors and detects the distance between pedestrians and vehicles through image processing algorithms. It has the advantages of low cost, easy installation, and easy integration into existing vehicle systems. It plays an important role in detecting the distance between pedestrians and vehicles.

[0003] However, during actual debugging or road testing, the pure driving assistance camera solution is not accurate enough for complex road conditions, especially pedestrians. The pure vision solution is not accurate enough to judge the longitudinal distance and speed of the target. It is easy to misselect the target in such complex scenarios, and it is also easy to mistakenly trigger the FCW false alarm or AEB function without reaching the emergency scenario, which interferes with the driver's normal driving and may even cause a rear-end collision. Summary of the invention

[0004] The present invention provides a method, device and medium for optimizing vehicle adaptive braking in complex scenarios, so as to solve the problem that vehicles are prone to false triggering of collision warning and automatic emergency braking in complex scenarios.

[0005] When the pedestrian is in a non-crossing scenario, the bias value threshold of the pedestrian intrusion into the vehicle's driving path is increased to obtain the non-crossing scenario adjustment result;

[0006] When the pedestrian is in a crossing scene, filtering the lateral speed of the pedestrian to obtain a crossing scene adjustment result;

[0007] When the pedestrian is in a curved road scene, the curvature value of the vehicle trajectory is reduced to a preset value to obtain a curved road scene adjustment result;

[0008] The emergency braking adjustment result of the vehicle is composed of the non-crossing scenario adjustment result, the crossing scenario adjustment result and the curve scenario adjustment result.

[0009] In the present invention, when pedestrians are in a non-crossing scenario, if the distance between the pedestrian and the vehicle is too close, but there is no actual risk of collision, by increasing the bias value threshold of the pedestrian intrusion into the vehicle's driving path, the vehicle's AEB system can more flexibly consider the relative position and distance between the pedestrian and the vehicle when judging whether to trigger emergency braking. In pedestrian crossing scenarios, by filtering the pedestrian's lateral speed information, the actual movement state of the pedestrian can be more accurately identified. The AEB system will only trigger emergency braking when the pedestrian is indeed at risk of crossing the vehicle's driving path; and when the pedestrian is only briefly stopped or walking slowly, the system will not be triggered by mistake. In curved scenarios, due to changes in the vehicle's driving trajectory and the forward line of sight, the traditional AEB system may misjudge non-dangerous objects on the inside of the curve or obstacles on the roadside as potential collision risks, thereby triggering emergency braking; by reducing the curvature value of the vehicle's trajectory to a preset value, the AEB system can be more flexible and accurate in judging curved scenarios.

[0010] Compared with the prior art, the present invention adaptively adjusts the vehicle braking mode in non-crossing scenarios, crossing scenario adjustments and curve scenarios, which together constitute the vehicle's emergency braking adjustment result. This comprehensive adjustment result can fully consider the actual situation and potential risks in different scenarios, making the AEB system more accurate and reliable in judging whether emergency braking needs to be triggered. Therefore, it can solve the problem of vehicles easily mistriggering collision warnings and automatic emergency braking in complex scenarios.

[0011] As a preferred solution, when the pedestrian is in a non-crossing scenario, the offset value threshold of the pedestrian intrusion into the vehicle's driving path is increased to obtain the non-crossing scenario adjustment result, specifically:

[0012] When the pedestrian is in a non-crossing scene, the bias value threshold of the vehicle driving path in the vehicle is adjusted in different dimensions to obtain a bias value threshold adjustment result;

[0013] Adding a confirmation frame with a pedestrian presence threshold time to obtain a confirmation frame adjustment result;

[0014] The non-crossing scene adjustment result is constituted by the offset value threshold adjustment result and the confirmation frame adjustment result.

[0015] This preferred solution can more accurately control the vehicle's driving path by adjusting the offset value threshold in different dimensions, so that it can make more delicate and accurate avoidance actions when encountering pedestrians. Increasing the confirmation frame of the pedestrian presence threshold time can enable the vehicle to detect the presence of pedestrians at an earlier time and make corresponding warnings and responses.

[0016] As a preferred solution, when the pedestrian is in a non-crossing scene, the bias value threshold of the vehicle driving path in the vehicle is adjusted in different dimensions to obtain the bias value threshold adjustment result, specifically:

[0017] Based on the situation that the pedestrian is in a non-crossing scenario, under one-dimensional conditions, the lateral speed of the pedestrian is set to be less than a preset value; under two-dimensional conditions, the longitudinal speed of the pedestrian is set to be less than a preset value; under three-dimensional conditions, the vehicle speed of the vehicle is set to be less than a preset value, and the bias value threshold adjustment result is obtained.

[0018] Under one-dimensional conditions, this preferred solution can accurately distinguish whether the pedestrian is stationary, walking slowly, or preparing to cross the road by setting the pedestrian's lateral speed to less than a preset value, which helps the system identify and respond in advance to avoid potential collision risks. Under two-dimensional conditions, considering the pedestrian's longitudinal speed can further refine the pedestrian's behavior pattern, such as distinguishing whether the pedestrian is accelerating, decelerating, or walking at a constant speed, so as to more accurately judge its impact on the vehicle's driving path. Under three-dimensional conditions, setting the vehicle's vehicle speed to less than a preset value can ensure that the vehicle maintains a safe speed when driving near pedestrians, reducing the risk of collision caused by excessive speed.

[0019] As a preferred solution, after obtaining the non-crossing scene adjustment result, the method further includes:

[0020] The closest angle distance and the farthest angle distance between the vehicle camera perception frame and the pedestrian are calculated according to a preset bias formula set; wherein the bias formula set includes a closest distance calculation formula and a farthest distance calculation formula;

[0021] If the closest angular distance and the farthest angular distance do not meet the preset thresholds, and the walking state of the pedestrian at the current moment and the future predicted moment does not meet the non-crossing scene adjustment result, the automatic emergency braking of the vehicle is triggered.

[0022] This preferred solution can accurately calculate the closest and farthest angle distances between the vehicle camera perception frame and pedestrians through a preset offset formula set, which helps the system to grasp the position information of pedestrians in real time and provide an accurate basis for subsequent avoidance and braking decisions. Through the preset offset formula set and threshold judgment, the system can quickly make a decision on whether to trigger automatic emergency braking in a short period of time. This efficient decision-making mechanism helps to reduce system response time and improve overall response efficiency.

[0023] As a preferred solution, the closest distance calculation formula is specifically:

[0024] closeoffset=closeoffset 1 +0.5*obj_width

[0025] Among them, closeoffset is the closest angle distance between the camera perception frame and the pedestrian after adjustment, closeoffset 1 To adjust the closest angle distance between the front camera perception frame and the pedestrian, obj_width is the pedestrian target width.

[0026] As a preferred solution, when the pedestrian is in a crossing scene, the lateral speed of the pedestrian is filtered to obtain a crossing scene adjustment result, specifically:

[0027] If the longitudinal speed of the pedestrian is greater than the lateral speed thereof, or the lateral speed of the pedestrian is less than or equal to a preset value, the vehicle braking confidence of the vehicle is set to be untrustworthy, and a confidence setting result is obtained;

[0028] Under the condition of the confidence setting result, the pedestrian recognition time of the vehicle is set to obtain the crossing scene adjustment result.

[0029] In this preferred solution, when the pedestrian's longitudinal speed is greater than the lateral speed, it may mean that the pedestrian is walking straight instead of crossing the road; at this time, if the system mistakenly identifies the pedestrian as a crossing scene and triggers emergency braking, it may cause unnecessary danger and accidents; therefore, setting the brake confidence to untrustworthy can avoid such misjudgment, thereby improving driving safety. In addition, if the pedestrian's lateral speed is less than or equal to the preset value, it may mean that the pedestrian is stationary or moving slowly. In this case, setting the brake confidence to untrustworthy can reduce emergency braking triggered by slight movement of pedestrians, avoiding unnecessary shock and discomfort to drivers and passengers.

[0030] As a preferred solution, after obtaining the curve scene adjustment result, the method further includes:

[0031] If the current vehicle speed is less than or equal to a first preset value, the curvature value of the vehicle trajectory is calculated according to the steering wheel angle of the current vehicle;

[0032] If the current vehicle speed is greater than or equal to a second preset value, the curvature value of the vehicle trajectory is calculated according to the yaw rate of the current vehicle;

[0033] If the speed of the current vehicle is greater than the first preset value and less than the second preset value, the curvature value of the current vehicle trajectory is calculated according to the steering wheel angle and the yaw rate of the current vehicle;

[0034] If the curvature value of the current vehicle trajectory is greater than the preset value, automatic emergency braking of the current vehicle is triggered.

[0035] In this preferred solution, when the vehicle is traveling at a low speed, the change in the steering wheel angle has a greater impact on the vehicle's trajectory. Therefore, calculating the trajectory curvature value through the steering wheel angle can more accurately reflect the actual driving state of the vehicle. When the vehicle is traveling at a high speed, the change in the yaw angular velocity has a more significant impact on the vehicle's trajectory. At this time, calculating the trajectory curvature value through the yaw angular velocity can more accurately predict the vehicle's driving trajectory, thereby improving driving safety. When the vehicle is traveling at a medium speed, changes in the steering wheel angle and yaw angular velocity will affect the vehicle's trajectory. Therefore, a comprehensive consideration of the two can more accurately calculate the trajectory curvature value, improving the comprehensiveness and accuracy of the calculation.

[0036] The present application also provides a vehicle adaptive braking optimization device for complex scenarios, including a non-traversing module, a traversing module, a curve module, and a comprehensive module;

[0037] The non-crossing module is used to increase the bias value threshold of the pedestrian intrusion into the vehicle's driving path when the pedestrian is in a non-crossing scene, and obtain a non-crossing scene adjustment result;

[0038] The crossing module is used to filter the lateral speed of the pedestrian when the pedestrian is in a crossing scene to obtain a crossing scene adjustment result;

[0039] The curve module is used to reduce the curvature value of the vehicle trajectory to a preset value when the pedestrian is in a curve scene, so as to obtain a curve scene adjustment result;

[0040] The comprehensive module is used to form the emergency braking adjustment result of the vehicle from the non-crossing scenario adjustment result, the crossing scenario adjustment result and the curve scenario adjustment result.

[0041] As a preferred solution, the non-traversing module includes a bias unit, a time unit and a combination unit;

[0042] The bias unit is used to adjust the bias value threshold of the vehicle driving path in the vehicle in different dimensions when the pedestrian is in a non-crossing scene, so as to obtain the bias value threshold adjustment result;

[0043] The time unit is used to increase the confirmation frame of the pedestrian presence threshold time to obtain the confirmation frame adjustment result;

[0044] The combining unit is used to form the non-crossing scene adjustment result by using the offset value threshold adjustment result and the confirmation frame adjustment result.

[0045] As a preferred solution, the biasing unit is specifically:

[0046] Based on the situation that the pedestrian is in a non-crossing scenario, under one-dimensional conditions, the lateral speed of the pedestrian is set to be less than a preset value; under two-dimensional conditions, the longitudinal speed of the pedestrian is set to be less than a preset value; under three-dimensional conditions, the vehicle speed of the vehicle is set to be less than a preset value, and the bias value threshold adjustment result is obtained.

[0047] As a preferred solution, the non-traversing module further includes a distance subunit and a trigger subunit;

[0048] The distance subunit is used to calculate the closest angle distance and the farthest angle distance between the vehicle camera perception frame and the pedestrian according to a preset bias formula set; wherein the bias formula set includes a closest distance calculation formula and a farthest distance calculation formula;

[0049] The trigger subunit is used to trigger the automatic emergency braking of the vehicle if the closest angle distance and the farthest angle distance do not meet the preset thresholds, and the walking state of the pedestrian at the current time and the future predicted time does not meet the non-crossing scene adjustment result.

[0050] As a preferred solution, the closest distance calculation formula is specifically:

[0051] closeoffset=closeoffset 1 +0.5*obj_width

[0052] Among them, closeoffset is the closest angle distance between the camera perception frame and the pedestrian after adjustment, closeoffset 1 To adjust the closest angle distance between the front camera perception frame and the pedestrian, obj_width is the pedestrian target width.

[0053] As a preferred solution, the traversal module includes a confidence unit and an identification unit;

[0054] The confidence unit is used to set the vehicle braking confidence of the vehicle to untrustworthy if the longitudinal speed of the pedestrian is greater than its lateral speed, or the lateral speed of the pedestrian is less than or equal to a preset value, to obtain a confidence setting result;

[0055] The recognition unit is used to set the pedestrian recognition time of the vehicle under the condition of the confidence setting result to obtain the crossing scene adjustment result.

[0056] As a preferred solution, the curve module further includes a low-speed unit, an overspeed unit, a medium-speed unit and a trigger unit;

[0057] Wherein, the low-speed unit is used to calculate the curvature value of the vehicle trajectory according to the steering wheel angle of the current vehicle if the current vehicle speed is less than or equal to a first preset value;

[0058] The overspeed unit is used to calculate the curvature value of the vehicle trajectory according to the yaw rate of the current vehicle if the vehicle speed of the current vehicle is greater than or equal to a second preset value;

[0059] The medium speed unit is used to calculate the curvature value of the current vehicle trajectory according to the steering wheel angle and the yaw rate of the current vehicle if the vehicle speed of the current vehicle is greater than the first preset value and less than the second preset value;

[0060] The trigger unit is used to trigger automatic emergency braking of the current vehicle if the curvature value of the current vehicle trajectory is greater than the preset value.

[0061] The present application also provides a storage medium having a computer program stored thereon. The computer program is called and executed by a computer to implement the vehicle adaptive braking optimization method for a complex scenario as described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of a vehicle adaptive braking optimization method for complex scenarios provided by an embodiment of the present application;

[0063] Figure 2 is a schematic diagram of a non-crossing scenario provided in an embodiment of the present application;

[0064] Figure 3 is a control flow chart provided in an embodiment of the present application;

[0065] Figure 4 It is a structural schematic diagram of a vehicle adaptive braking optimization device for complex scenarios provided in an embodiment of the present application. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0067] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "several" means two or more.

[0068] It should be noted that, in the description of this application, FCW refers to the forward collision warning system (Forward Collision Warning, referred to as FCW), and AEB refers to the automatic emergency braking system (Autonomous Emergency Braking, referred to as AEB).

[0069] The embodiment of the present application provides a vehicle adaptive braking optimization method for complex scenarios, which is mainly used when it is necessary to accurately judge the longitudinal distance and speed of the target object, thereby reducing the probability of falsely triggering the vehicle's FCW alarm or AEB function in complex scenarios and maintaining the driver's normal driving situation.

[0070] Embodiment 1:

[0071] See also Figure 1 The embodiment of the present application provides a vehicle adaptive braking optimization method for complex scenarios, including S1 to S4, and the specific implementation steps are as follows:

[0072] S1. When the pedestrian is in a non-crossing scenario, increase the bias value threshold of the pedestrian intrusion into the vehicle's driving path to obtain a non-crossing scenario adjustment result.

[0073] Step S1 of the embodiment of the present application includes S1.1 to S1.3, specifically:

[0074] S1.1, obtaining external information and self-driving information of the vehicle through the self-driving sensor of the vehicle to obtain comprehensive information;

[0075] Among them, the external information includes the target vehicle type, the target vehicle's lateral and longitudinal distance, the target vehicle's speed, the target vehicle's acceleration, and the lane line width C 0 , lane line longitudinal distance and lane line curvature C 1 , the ego vehicle information includes YawRate (yaw angular velocity), ego vehicle speed, steering wheel angle, ego vehicle acceleration and ego vehicle acceleration.

[0076] S1.2, based on the situation that the pedestrian is in a non-crossing scene, according to the comprehensive information, under the one-dimensional condition, the lateral speed of the pedestrian is set to be less than a preset value; under the two-dimensional condition, the longitudinal speed of the pedestrian is set to be less than a preset value; under the three-dimensional condition, the vehicle speed of the vehicle is set to be less than a preset value, and the bias value threshold adjustment result is obtained;

[0077] Add 0.2s of confirmation frame to the threshold time age=N of the pedestrian target existence, and obtain the confirmation frame adjustment result;

[0078] The non-crossing scene adjustment result is composed of the offset value threshold adjustment result and the confirmation frame adjustment result.

[0079] The specific adjustment logic is: add a new column "[-PL (lateral offset distance, - represents subtraction), 0,0,0,0,0,0,0,0,0], PL (lateral offset distance, - represents subtraction)" in the vehicle's vlat (pedestrian lateral speed) and vlgt (pedestrian longitudinal speed) two-dimensional lookup table; and, the one-dimensional condition is set to: the lateral speed is vlat, vlat is less than a certain value, such as 0.8m / s; the two-dimensional condition is set to: the longitudinal speed vlgt is 0->30m / s; the three-dimensional condition is set to: the vehicle speed is V, V is less than a certain value, such as 3.5m / s; after the modification, the current position and predicted position of the pedestrian at the time of triggering are no longer targets within the path, and AEB is no longer triggered; at the same time, a confirmation frame confirmation of 0.2s is added to the pedestrian existence threshold time age=N; wherein, the units of vlat and vlgt are m / s.

[0080] In this embodiment S1.2, under one-dimensional conditions, by setting the lateral speed of pedestrians to be less than a preset value, it is possible to accurately distinguish whether the pedestrian is stationary, walking slowly, or preparing to cross the road, which helps the system to identify and respond in advance to avoid potential collision risks. Under two-dimensional conditions, considering the longitudinal speed of pedestrians can further refine the behavior patterns of pedestrians, such as distinguishing whether pedestrians are accelerating, decelerating, or walking at a constant speed, so as to more accurately judge their impact on the vehicle's driving path. Under three-dimensional conditions, setting the vehicle speed to be less than a preset value can ensure that the vehicle maintains a safe speed when driving near pedestrians and reduce the risk of collision caused by excessive speed; therefore, by adjusting the bias value threshold in different dimensions, the vehicle's driving path can be more accurately controlled, allowing it to make more delicate and accurate avoidance actions when encountering pedestrians;

[0081] In addition, increasing the confirmation frame of the pedestrian presence threshold time can enable the vehicle to detect the presence of pedestrians at an earlier time and make corresponding warnings and responses.

[0082] S1.3, calculating the closest angle distance closeoffset and the farthest angle distance faroffset of the vehicle camera perception frame from the pedestrian according to a preset offset formula set; wherein the offset formula set includes a closest distance calculation formula and a farthest distance calculation formula, and the closest angle distance closeoffset and the farthest angle distance faroffset are combined to form a square perception frame of the vehicle;

[0083] If the closest angle distance closeoffset and the farthest angle distance faroffset do not meet the preset thresholds, and the pedestrian's walking status at the current time and the future predicted time (TTR) does not meet the non-crossing scene adjustment results, the collision warning and automatic emergency braking of the current vehicle are triggered.

[0084] For example: at the triggering moment, the target closeoffset is shortened to 0.25m and the faroffset is also shortened by half, then the pedestrian is no longer considered to be in the path of the vehicle; and then after the target vlat is judged to be less than the lateral threshold of 0.8m / s, the pedestrian needs to be in the lane at the current moment and the TTR moment to meet the inpath requirement, so the current pedestrian is no longer considered to be inpath, and FCW and AEB are no longer triggered.

[0085] The closest distance calculation formula is:

[0086] closeoffset=closeoffset 1 +0.5*obj_width

[0087] The formula for calculating the maximum distance is:

[0088] faroffset=faroffset 1 -0.5*obj_width

[0089] Among them, closeoffset is the closest angle distance between the camera perception frame of the adjusted vehicle and the pedestrian, closeoffset 1 To adjust the closest angle distance between the camera perception frame of the front vehicle and the pedestrian, obj_width is the pedestrian target width, faroffset is the farthest angle distance between the camera perception frame of the adjusted vehicle and the pedestrian, faroffset 1 To adjust the maximum angle distance between the camera perception frame of the front vehicle and the pedestrian.

[0090] To apply the embodiments of this application, please refer to Figure 2 , Figure 2 It is a schematic diagram of a non-crossing scenario provided in an embodiment of the present application, showing a target pedestrian or a target motor vehicle, and a non-crossing scenario for the current vehicle itself.

[0091] It should be noted that in the non-crossing scenario, pedestrians do not cross laterally in front of the vehicle. They may walk, stand or stand still along the road parallel to the direction of the vehicle; at this time, the relative position relationship between pedestrians and vehicles is relatively stable, and there is no direct risk of collision (unless pedestrians suddenly change direction or speed).

[0092] In this embodiment S1.3, the preset offset formula set can accurately calculate the closest angle distance and the farthest angle distance between the vehicle camera perception frame and the pedestrian, which helps the system to grasp the position information of the pedestrian in real time and provide an accurate basis for subsequent avoidance and braking decisions. Through the preset offset formula set and threshold judgment, the system can quickly make a decision on whether to trigger collision warning and automatic emergency braking in a short time. This efficient decision-making mechanism helps to reduce system response time and improve overall response efficiency.

[0093] S2. When the pedestrian is in a crossing scene, filter the pedestrian's lateral speed to obtain a crossing scene adjustment result.

[0094] Step S2 of the embodiment of the present application is specifically as follows:

[0095] In the case where the pedestrian is in a crossing scene, according to the comprehensive information, if the longitudinal speed vlgt of the pedestrian is greater than its lateral speed vlat, or the lateral speed vlat of the pedestrian is less than or equal to a preset value, the vehicle braking confidence of the vehicle is set to untrustworthy, and a confidence setting result is obtained;

[0096] Under the condition of the confidence setting result, the pedestrian recognition time of the vehicle is set, and the lateral offset compensation is increased in the same way as increasing the offset value (inpathoffset) threshold of the vehicle's driving path in the non-crossing pedestrian scenario to obtain the crossing scenario adjustment result.

[0097] The specific adjustment logic is: (target vlgt longitudinal speed > vlat lateral speed) || (vlat lateral speed <= 1.38m / s) and the vehicle target braking confidence of age < 50 will be set as untrustworthy. For pedestrians and two-wheeled vehicles identified as outside the curb, the lateral speed will be adjusted to 0. If the target age is less than 50, it will no longer be judged as an inpath dangerous target (the target position prediction at the current moment and the predicted moment no longer considers the target to be inpath).

[0098] If the driving status of the pedestrian does not meet the adjustment results of the crossing scenario, the collision warning and automatic emergency braking of the current vehicle will be triggered.

[0099] It should be noted that the situation where pedestrians are in a crossing scenario includes three possible pedestrian behaviors or states: the pedestrian is crossing the road or intends to cross the road; the pedestrian is partially invisible due to obstruction by roadside objects (such as trees, buildings, etc.); although the pedestrian mainly moves longitudinally (i.e. along the direction of the road), he also has a certain lateral speed and has not yet entered the driving path of the autonomous driving vehicle.

[0100] In this embodiment S2, when the pedestrian's longitudinal speed is greater than the lateral speed, it may mean that the pedestrian is walking straight instead of crossing the road; at this time, if the system mistakenly identifies the pedestrian as a crossing scene and triggers emergency braking, it may cause unnecessary danger and accidents; therefore, setting the brake confidence to untrustworthy can avoid such misjudgment, thereby improving driving safety. In addition, if the pedestrian's lateral speed is less than or equal to the preset value, it may mean that the pedestrian is stationary or moving slowly. In this case, setting the brake confidence to untrustworthy can reduce emergency braking triggered by slight movement of pedestrians, avoiding unnecessary shock and discomfort to drivers and passengers.

[0101] S3. When the pedestrian is in a curve scene, the curvature value of the vehicle trajectory is reduced to a preset value to obtain a curve scene adjustment result.

[0102] Step S3 of the embodiment of the present application is specifically as follows:

[0103] In the case of pedestrians on a curved road, the judgment threshold of the large curvature scene in the vehicle trajectory is increased from C to 01 Reduce to C 02 , and obtain the adjustment result of the curve scene.

[0104] If the current vehicle speed V is less than or equal to the first preset value V 1 , then the curvature value C of the vehicle trajectory is calculated according to the current vehicle steering wheel angle 0 ;

[0105] If the current vehicle speed V is greater than or equal to the second preset value V 2 , then the curvature value C of the vehicle trajectory is calculated according to the current vehicle yaw rate 0 ;

[0106] If the current vehicle speed V is greater than the first preset value V 1 and is less than the second preset value V 2 , then the curvature value C of the current vehicle trajectory is calculated based on the weighted average of the steering wheel angle and yaw angular velocity in the current vehicle 0 .

[0107] If the curvature value C of the current vehicle trajectory 0 Greater than C 02 , the collision warning and automatic emergency braking of the current vehicle will be triggered.

[0108] It should be noted that the judgment threshold of the large curvature scene is changed from C 01 Reduce to C 02, then this data change changes most of the small curvature scenes into large curvature scenes. This change reduces the driver's reaction time in large curvature scenes: the original reaction time setting is small curvature corresponding to "0 reaction time" and large curvature corresponding to "-T2 reaction time", now it is changed to look up the table according to the curvature [C 02 ,C 01 ] to determine the reaction time range [-T1, -T2]; after adjustment, the reaction time is reduced by about T 3 , which means that the system needs to calculate a smaller longitudinal deceleration to avoid a collision; therefore, the longitudinal deceleration conditions of FCW and AEB become more difficult to meet, thereby suppressing the triggering in these scenarios;

[0109] For greater than C 01 In the curvature scenario, the trigger time is not reduced. However, the system can suppress the potential risks in these scenarios by increasing the lateral offset compensation, such as increasing the offset value (inpathoffset) threshold of the pedestrian intrusion into the vehicle's driving path in the non-crossing pedestrian scenario.

[0110] The calculation formula of the curvature value is:

[0111]

[0112] Among them, C 0 is the estimated curvature of the current vehicle, K1 is the inverse of the fluctuation range of the Yawrate value when the vehicle travels along a fixed curvature route, and K 3 K is the inverse of the fluctuation amplitude of the steering wheel angle when the vehicle travels along a fixed curvature route; 2 is the filter coefficient, which is used to avoid frequent curvature jumps; Yawrate is the yaw rate of the vehicle, V is the speed of the vehicle, and V 1 and V 2 are respectively the curvature of the trajectory of the ego vehicle estimated by the steering wheel angle and the curvature of the trajectory of the ego vehicle estimated by the yaw rate of the ego vehicle; θ is the steering wheel angle, L is the wheelbase of the ego vehicle, and i is the speed ratio of the steering mechanism.

[0113] It should be noted that in the scene where pedestrians are on a curve, the vehicle is driving on the curve, and the pedestrian may be on the side of the curve or on the road. Due to the existence of the curve, the vehicle's driving trajectory will change, making the relative position relationship between the vehicle and the pedestrian complicated. In addition, there are limitations to simply relying on the curvature of the camera perception output to judge the size of the curve for target processing. When the curve recognition distance is limited and the curve curvature is prone to jump, target recognition may become unstable.

[0114] In this embodiment S3, when the vehicle is traveling at a low speed, the change of the steering wheel angle has a greater impact on the vehicle trajectory, so calculating the trajectory curvature value by the steering wheel angle can more accurately reflect the actual driving state of the vehicle. When the vehicle is traveling at a high speed, the change of the yaw angular velocity has a more significant impact on the vehicle trajectory; at this time, calculating the trajectory curvature value by the yaw angular velocity can more accurately predict the vehicle's driving trajectory, thereby improving driving safety. When the vehicle is traveling at a medium speed, the changes of the steering wheel angle and the yaw angular velocity will affect the vehicle trajectory; therefore, comprehensively considering the two can more accurately calculate the trajectory curvature value, improving the comprehensiveness and accuracy of the calculation.

[0115] S4. The emergency braking adjustment result of the vehicle is composed of the non-crossing scenario adjustment result, the crossing scenario adjustment result and the curve scenario adjustment result.

[0116] Step S4 of the embodiment of the present application is specifically as follows:

[0117] The emergency braking adjustment result of the vehicle is composed of the non-crossing scenario adjustment result, the crossing scenario adjustment result and the curve scenario adjustment result;

[0118] Once a potential collision object crossing behavior that does not meet the emergency braking adjustment results is detected, the emergency braking and collision warning of the current vehicle will be triggered immediately to minimize the possibility of collision; and the best avoidance path will be quickly planned, and the vehicle's speed and direction will be adjusted to ensure the safe passage of pedestrians and vehicles.

[0119] To apply the embodiments of this application, please refer to Figure 3 , Figure 3 : is a control flow chart provided in an embodiment of the present application, showing the general process of setting the vehicle accordingly in different scenarios in the first embodiment of the present application, specifically:

[0120] 1) For longitudinal or stationary pedestrian scenarios (pedestrians are not crossing the scene), when the lateral speed or distance is not accurately judged, the lateral distance bias of the pedestrian target is increased, that is, the misidentified pedestrian target is pushed outward from the driving path; at the same time, the lane line is used as the judgment scene, and its lateral acceleration is directly filtered to 0, so as to avoid false triggering;

[0121] 2) For scenes with short appearance time or partial occlusion (pedestrians crossing the road), the pedestrian lateral speed suppression function trigger is directly filtered out to avoid false triggering;

[0122] 3) For pedestrians on a curve, different thresholds are set to calculate the curvature of the curve to determine whether it is a curve with large curvature or a curve with small curvature, thereby reducing the system reaction time in the large curvature scenario. The smaller the longitudinal deceleration required for collision avoidance, the more difficult it is to meet the longitudinal deceleration condition, thereby avoiding false triggering.

[0123] It should be noted that the objects that trigger the current vehicle collision warning and automatic emergency braking functions involved in this application may include various potential collision objects such as pedestrians, motor vehicles and animals.

[0124] Overall, this embodiment has the following beneficial effects:

[0125] When pedestrians are in a non-crossing scenario, if the distance between the pedestrian and the vehicle is too close, but there is no actual risk of collision, by increasing the bias value threshold for pedestrians intruding into the vehicle's driving path, the vehicle's AEB system can more flexibly consider the relative position and distance between the pedestrian and the vehicle when judging whether emergency braking needs to be triggered. In pedestrian crossing scenarios, the actual movement state of pedestrians can be more accurately identified by filtering the pedestrian's lateral speed information. The AEB system will only trigger emergency braking when the pedestrian is indeed at risk of crossing the vehicle's driving path; and the system will not be triggered by mistake when the pedestrian is only briefly stopped or walking slowly. In curved scenarios, due to changes in the vehicle's driving trajectory and the line of sight ahead, traditional AEB systems may misjudge non-dangerous objects on the inside of the curve or roadside obstacles ahead as potential collision risks, thereby triggering emergency braking; by reducing the curvature value of the vehicle's trajectory to the preset value, the AEB system can be more flexible and accurate in judging curved scenarios;

[0126] In summary, this application proposes a variety of pedestrian detection and control methods for emergency braking, targeting common pedestrian AEB&FCW false triggering scenarios, which are used to reduce the problem of FCW and AEB false triggering in scenarios such as pedestrian crossing, non-crossing, and curves. It can effectively avoid the false triggering of automatic emergency braking in complex road conditions and improve driving safety. Therefore, it can solve the problem of vehicles' prone to false triggering of collision warnings and automatic emergency braking in complex scenarios.

[0127] Embodiment 2:

[0128] See also Figure 4 , the embodiment of the present application provides a vehicle adaptive braking optimization device for complex scenarios, including a non-traversing module 10, a traversing module 20, a curve module 30 and a comprehensive module 40;

[0129] Among them, the non-crossing module 10 is used to increase the bias value threshold of the pedestrian intrusion into the vehicle's driving path when the pedestrian is in a non-crossing scene, and obtain a non-crossing scene adjustment result;

[0130] A crossing module 20 is used to filter the lateral speed of the pedestrian when the pedestrian is in a crossing scene to obtain a crossing scene adjustment result;

[0131] The curve module 30 is used to reduce the curvature value of the vehicle trajectory to a preset value when the pedestrian is in a curve scene, and obtain a curve scene adjustment result;

[0132] The integration module 40 is used to form an emergency braking adjustment result of the vehicle from the non-crossing scenario adjustment result, the crossing scenario adjustment result and the curve scenario adjustment result.

[0133] In one embodiment, the non-traversing module 10 includes a data unit, a bias unit, a time unit, a combination unit, a distance sub-unit, and a trigger sub-unit;

[0134] The data unit is used to obtain the external information and the vehicle information of the vehicle through the vehicle's self-driving sensor to obtain comprehensive information;

[0135] Among them, the external information includes the target vehicle type, the target vehicle's lateral and longitudinal distance, the target vehicle's speed, the target vehicle's acceleration, and the lane line width C 0 , lane line longitudinal distance and lane line curvature C 1 , the ego vehicle information includes YawRate (yaw angular velocity), ego vehicle speed, steering wheel angle, ego vehicle acceleration and ego vehicle acceleration.

[0136] The bias unit is used to set the lateral speed of the pedestrian to be less than a preset value under one-dimensional conditions, the longitudinal speed of the pedestrian to be less than a preset value under two-dimensional conditions, and the vehicle speed of the vehicle to be less than a preset value under three-dimensional conditions, based on the situation that the pedestrian is in a non-crossing scene, and according to the comprehensive information, to obtain a bias value threshold adjustment result;

[0137] The time unit is used to increase the confirmation frame of 0.2s to the threshold time age=N of the pedestrian target existence, and obtain the confirmation frame adjustment result;

[0138] The combining unit is used to form a non-crossing scene adjustment result from the offset value threshold adjustment result and the confirmation frame adjustment result.

[0139] The specific adjustment logic is: add a new column "[-PL (lateral offset distance, - represents subtraction), 0,0,0,0,0,0,0,0,0], PL (lateral offset distance, - represents subtraction)" in the vehicle's vlat (pedestrian lateral speed) and vlgt (pedestrian longitudinal speed) two-dimensional lookup table; and, the one-dimensional condition is set to: the lateral speed is vlat, vlat is less than a certain value, such as 0.8m / s; the two-dimensional condition is set to: the longitudinal speed vlgt is 0->30m / s; the three-dimensional condition is set to: the vehicle speed is V, V is less than a certain value, such as 3.5m / s; after the modification, the current position and predicted position of the pedestrian at the time of triggering are no longer targets within the path, and AEB is no longer triggered; at the same time, a confirmation frame confirmation of 0.2s is added to the pedestrian existence threshold time age=N; wherein, the units of vlat and vlgt are m / s.

[0140] Under one-dimensional conditions, the bias unit, time unit and combination unit of this embodiment can accurately distinguish whether the pedestrian is stationary, walking slowly or preparing to cross the road by setting the pedestrian's lateral speed to be less than a preset value, which helps the system to identify and respond in advance to avoid potential collision risks. Under two-dimensional conditions, considering the pedestrian's longitudinal speed can further refine the pedestrian's behavior pattern, such as distinguishing whether the pedestrian is accelerating, decelerating or walking at a constant speed, so as to more accurately judge its impact on the vehicle's driving path. Under three-dimensional conditions, setting the vehicle's speed to be less than a preset value can ensure that the vehicle maintains a safe speed when driving near pedestrians and reduce the risk of collision caused by excessive speed; therefore, by adjusting the bias value threshold in different dimensions, the vehicle's driving path can be more accurately controlled, allowing it to make more delicate and accurate avoidance actions when encountering pedestrians;

[0141] In addition, increasing the confirmation frame of the pedestrian presence threshold time can enable the vehicle to detect the presence of pedestrians at an earlier time and make corresponding warnings and responses.

[0142] The distance subunit is used to calculate the closest angle distance closeoffset and the farthest angle distance faroffset of the vehicle camera perception frame from the pedestrian according to a preset bias formula set; wherein the bias formula set includes a closest distance calculation formula and a farthest distance calculation formula, and the closest angle distance closeoffset and the farthest angle distance faroffset are combined to form a square perception frame of the vehicle;

[0143] The trigger subunit is used to trigger the collision warning and automatic emergency braking of the current vehicle if the closest angle distance closeoffset and the farthest angle distance faroffset do not meet the preset thresholds, and the walking status of the pedestrian at the current time and the future predicted time (TTR) does not meet the non-crossing scene adjustment results.

[0144] For example: at the triggering moment, the target closeoffset is shortened to 0.25m and the faroffset is also shortened by half, then the pedestrian is no longer considered to be in the path of the vehicle; and then after the target vlat is judged to be less than the lateral threshold of 0.8m / s, the pedestrian needs to be in the lane at the current moment and the TTR moment to meet the inpath requirement, so the current pedestrian is no longer considered to be inpath, and FCW and AEB are no longer triggered.

[0145] The closest distance calculation formula is:

[0146] closeoffset=closeoffset 1 +0.5*obj_width

[0147] The formula for calculating the maximum distance is:

[0148] faroffset=faroffset 1 -0.5*obj_width

[0149] Among them, closeoffset is the closest angle distance between the camera perception frame of the vehicle and the pedestrian after adjustment, closeoffset 1 To adjust the closest angle distance between the camera perception frame of the front vehicle and the pedestrian, obj_width is the pedestrian target width, faroffset is the farthest angle distance between the camera perception frame of the adjusted vehicle and the pedestrian, faroffset 1 To adjust the maximum angle distance between the camera perception frame of the front vehicle and the pedestrian.

[0150] To apply the embodiments of this application, please refer to Figure 2 , Figure 2 It is a schematic diagram of a non-crossing scenario provided in an embodiment of the present application, showing a target pedestrian or a target motor vehicle, and a non-crossing scenario for the current vehicle itself.

[0151] It should be noted that in the non-crossing scenario, pedestrians do not cross laterally in front of the vehicle. They may walk, stand or stand still along the road parallel to the direction of the vehicle; at this time, the relative position relationship between pedestrians and vehicles is relatively stable, and there is no direct risk of collision (unless pedestrians suddenly change direction or speed).

[0152] The distance subunit and trigger subunit of this embodiment can accurately calculate the closest angle distance and the farthest angle distance between the vehicle camera perception frame and the pedestrian through the preset offset formula set, which helps the system to grasp the position information of the pedestrian in real time and provide an accurate basis for subsequent avoidance and braking decisions. Through the preset offset formula set and threshold judgment, the system can quickly make a decision on whether to trigger collision warning and automatic emergency braking in a short time. This efficient decision-making mechanism helps to reduce system response time and improve overall response efficiency.

[0153] In one embodiment, the traversal module 20 includes a confidence unit and an identification unit;

[0154] The confidence unit is used for, when a pedestrian is in a crossing scene, according to the comprehensive information, if the longitudinal speed vlgt of the pedestrian is greater than the lateral speed vlat of the pedestrian, or the lateral speed vlat of the pedestrian is less than or equal to a preset value, setting the vehicle braking confidence of the vehicle to untrustworthy, and obtaining a confidence setting result;

[0155] The recognition unit is used to set the pedestrian recognition time of the vehicle under the condition of the confidence setting result, and increase the lateral offset compensation in the same way as increasing the offset value (inpathoffset) threshold of the intrusion into the vehicle's driving path in a non-crossing pedestrian scenario, so as to obtain a crossing scenario adjustment result.

[0156] The specific adjustment logic is: (target vlgt longitudinal speed > vlat lateral speed) || (vlat lateral speed <= 1.38m / s) and the vehicle target braking confidence of age < 50 will be set as untrustworthy. For pedestrians and two-wheeled vehicles identified as outside the curb, the lateral speed will be adjusted to 0. If the target age is less than 50, it will no longer be judged as an inpath dangerous target (the target position prediction at the current moment and the predicted moment no longer considers the target to be inpath).

[0157] If the driving status of the pedestrian does not meet the adjustment results of the crossing scenario, the collision warning and automatic emergency braking of the current vehicle will be triggered.

[0158] It should be noted that the situation where pedestrians are in a crossing scenario includes three possible pedestrian behaviors or states: the pedestrian is crossing the road or intends to cross the road; the pedestrian is partially invisible due to obstruction by roadside objects (such as trees, buildings, etc.); although the pedestrian mainly moves longitudinally (i.e. along the direction of the road), he also has a certain lateral speed and has not yet entered the driving path of the autonomous driving vehicle.

[0159] In the crossing module 20 of this embodiment, when the pedestrian's longitudinal speed is greater than the lateral speed, it may mean that the pedestrian is walking straight instead of crossing the road; at this time, if the system mistakenly identifies the pedestrian as a crossing scene and triggers emergency braking, it may cause unnecessary danger and accidents; therefore, setting the braking confidence to untrustworthy can avoid such misjudgment, thereby improving driving safety. In addition, if the pedestrian's lateral speed is less than or equal to the preset value, it may mean that the pedestrian is stationary or moving slowly. In this case, setting the braking confidence to untrustworthy can reduce emergency braking triggered by slight movement of pedestrians, avoiding unnecessary shock and discomfort to drivers and passengers.

[0160] In one embodiment, the curve module 30 includes an adjustment unit, an underspeed unit, an overspeed unit, a medium speed unit, and a trigger unit;

[0161] The adjustment unit is used to adjust the judgment threshold of the large curvature scene in the vehicle trajectory from C to C according to the comprehensive information when the pedestrian is in the curved scene. 01 Reduce to C 02 , and obtain the adjustment result of the curve scene.

[0162] The low speed unit is used to determine the vehicle speed V if the current vehicle speed V is less than or equal to the first preset value V 1 , then the curvature value C of the vehicle trajectory is calculated according to the current vehicle steering wheel angle 0 ;

[0163] The overspeed unit is used to detect if the current vehicle speed V is greater than or equal to a second preset value V 2 , then the curvature value C of the vehicle trajectory is calculated according to the current vehicle yaw rate 0 ;

[0164] The medium speed unit is used to determine if the current vehicle speed V is greater than a first preset value V 1 and is less than the second preset value V 2 , then the curvature value C of the current vehicle trajectory is calculated based on the weighted average of the steering wheel angle and yaw angular velocity in the current vehicle 0 .

[0165] The trigger unit is used to trigger the curvature value C of the current vehicle trajectory. 0 Greater than C 02 , the collision warning and automatic emergency braking of the current vehicle will be triggered.

[0166] It should be noted that the judgment threshold of the large curvature scene is changed from C 01 Reduce to C 02, then this data change changes most of the small curvature scenes into large curvature scenes. This change reduces the driver's reaction time in large curvature scenes: the original reaction time setting is small curvature corresponding to "0 reaction time" and large curvature corresponding to "-T2 reaction time", now it is changed to look up the table according to the curvature [C 02 ,C 01 ] to determine the reaction time range [-T1, -T2]; after adjustment, the reaction time is reduced by about T 3 , which means that the system needs to calculate a smaller longitudinal deceleration to avoid a collision; therefore, the longitudinal deceleration conditions of FCW and AEB become more difficult to meet, thereby suppressing the triggering in these scenarios;

[0167] For greater than C 01 In the curvature scenario, the trigger time is not reduced. However, the system can suppress the potential risks in these scenarios by increasing the lateral offset compensation, such as increasing the offset value (inpathoffset) threshold of the pedestrian intrusion into the vehicle's driving path in the non-crossing pedestrian scenario.

[0168] The calculation formula of the curvature value is:

[0169]

[0170] Among them, C 0 is the estimated curvature of the current vehicle, K1 is the inverse of the fluctuation range of the Yawrate value when the vehicle travels along a fixed curvature route, and K 3 K is the inverse of the fluctuation amplitude of the steering wheel angle when the vehicle travels along a fixed curvature route; 2 is the filter coefficient, which is used to avoid frequent curvature jumps; Yawrate is the yaw rate of the vehicle, V is the speed of the vehicle, and V 1 and V 2 are respectively the curvature of the trajectory of the ego vehicle estimated by the steering wheel angle and the curvature of the trajectory of the ego vehicle estimated by the yaw rate of the ego vehicle; θ is the steering wheel angle, L is the wheelbase of the ego vehicle, and i is the speed ratio of the steering mechanism.

[0171] It should be noted that in the scene where pedestrians are on a curve, the vehicle is driving on the curve, and the pedestrian may be on the side of the curve or on the road. Due to the existence of the curve, the vehicle's driving trajectory will change, making the relative position relationship between the vehicle and the pedestrian complicated. In addition, there are limitations to simply relying on the curvature of the camera perception output to judge the size of the curve for target processing. When the curve recognition distance is limited and the curve curvature is prone to jump, target recognition may become unstable.

[0172] In the curve module 30 of this embodiment, when the vehicle is traveling at a low speed, the change of the steering wheel angle has a greater impact on the vehicle trajectory, so calculating the trajectory curvature value by the steering wheel angle can more accurately reflect the actual driving state of the vehicle. When the vehicle is traveling at a high speed, the change of the yaw angular velocity has a more significant impact on the vehicle trajectory; at this time, calculating the trajectory curvature value by the yaw angular velocity can more accurately predict the vehicle's driving trajectory, thereby improving driving safety. When the vehicle is traveling at a medium speed, the changes of the steering wheel angle and the yaw angular velocity will affect the vehicle trajectory; therefore, comprehensively considering the two can more accurately calculate the trajectory curvature value, improving the comprehensiveness and accuracy of the calculation.

[0173] In one embodiment, the integration module 40 is specifically:

[0174] The emergency braking adjustment result of the vehicle is composed of the non-crossing scenario adjustment result, the crossing scenario adjustment result and the curve scenario adjustment result;

[0175] Once a potential collision object crossing behavior that does not meet the emergency braking adjustment results is detected, the emergency braking and collision warning of the current vehicle will be triggered immediately to minimize the possibility of collision; and the best avoidance path will be quickly planned, and the vehicle's speed and direction will be adjusted to ensure the safe passage of pedestrians and vehicles.

[0176] To apply the embodiments of this application, please refer to Figure 3 , Figure 3 : is a control flow chart provided in an embodiment of the present application, showing the general process of setting the vehicle accordingly in different scenarios in the second embodiment of the present application, specifically:

[0177] 1) For longitudinal or stationary pedestrian scenarios (pedestrians are not crossing the scene), when the lateral speed or distance is not accurately judged, the lateral distance bias of the pedestrian target is increased, that is, the misidentified pedestrian target is pushed outward from the driving path; at the same time, the lane line is used as the judgment scene, and its lateral acceleration is directly filtered to 0, so as to avoid false triggering;

[0178] 2) For scenes with short appearance time or partial occlusion (pedestrians crossing the road), the pedestrian lateral speed suppression function trigger is directly filtered out to avoid false triggering;

[0179] 3) For pedestrians on a curve, different thresholds are set to calculate the curvature of the curve to determine whether it is a curve with large curvature or a curve with small curvature, thereby reducing the system reaction time in the large curvature scenario. The smaller the longitudinal deceleration required for collision avoidance, the more difficult it is to meet the longitudinal deceleration condition, thereby avoiding false triggering.

[0180] It should be noted that the objects that trigger the current vehicle collision warning and automatic emergency braking functions involved in this application may include various potential collision objects such as pedestrians, motor vehicles and animals.

[0181] Overall, this embodiment has the following beneficial effects:

[0182] When pedestrians are in a non-crossing scenario, if the distance between the pedestrian and the vehicle is too close, but there is no actual risk of collision, by increasing the bias value threshold for pedestrians intruding into the vehicle's driving path, the vehicle's AEB system can more flexibly consider the relative position and distance between the pedestrian and the vehicle when judging whether emergency braking needs to be triggered. In pedestrian crossing scenarios, the actual movement state of pedestrians can be more accurately identified by filtering the pedestrian's lateral speed information. The AEB system will only trigger emergency braking when the pedestrian is indeed at risk of crossing the vehicle's driving path; and the system will not be triggered by mistake when the pedestrian is only briefly stopped or walking slowly. In curved scenarios, due to changes in the vehicle's driving trajectory and the line of sight ahead, traditional AEB systems may misjudge non-dangerous objects on the inside of the curve or roadside obstacles ahead as potential collision risks, thereby triggering emergency braking; by reducing the curvature value of the vehicle's trajectory to the preset value, the AEB system can be more flexible and accurate in judging curved scenarios;

[0183] In summary, this application proposes a variety of pedestrian detection and control methods for emergency braking, targeting common pedestrian AEB&FCW false triggering scenarios, which are used to reduce the problem of FCW and AEB false triggering in scenarios such as pedestrian crossing, non-crossing, and curves. It can effectively avoid the false triggering of automatic emergency braking in complex road conditions and improve driving safety. Therefore, it can solve the problem of vehicles' prone to false triggering of collision warnings and automatic emergency braking in complex scenarios.

[0184] Embodiment three:

[0185] An embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the vehicle adaptive braking optimization method for a complex scenario;

[0186] Wherein, the vehicle adaptive braking optimization method for complex scenarios, if implemented in the form of a software functional unit and used as an independent product, can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0187] The above are preferred embodiments of the present invention. It should be noted that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A vehicle adaptive braking optimization method for complex scenarios, characterized in that: include: When the pedestrian is in a non-crossing scenario, the bias value threshold of the pedestrian intrusion into the vehicle's driving path is increased to obtain the non-crossing scenario adjustment result; When the pedestrian is in a crossing scene, filtering the lateral speed of the pedestrian to obtain a crossing scene adjustment result; When the pedestrian is in a curved road scene, the curvature value of the vehicle trajectory is reduced to a preset value to obtain a curved road scene adjustment result; The emergency braking adjustment result of the vehicle is composed of the non-crossing scenario adjustment result, the crossing scenario adjustment result and the curve scenario adjustment result.

2. The vehicle adaptive braking optimization method for complex scenarios as claimed in claim 1, characterized in that: When the pedestrian is in a non-crossing scenario, the bias value threshold of the pedestrian intrusion into the vehicle's driving path is increased to obtain the non-crossing scenario adjustment result, specifically: When the pedestrian is in a non-crossing scene, the bias value threshold of the vehicle driving path in the vehicle is adjusted in different dimensions to obtain a bias value threshold adjustment result; Adding a confirmation frame with a pedestrian presence threshold time to obtain a confirmation frame adjustment result; The non-crossing scene adjustment result is constituted by the offset value threshold adjustment result and the confirmation frame adjustment result.

3. The vehicle adaptive braking optimization method for complex scenarios as claimed in claim 2, characterized in that: In the case where the pedestrian is not crossing the scene, the bias value threshold of the vehicle driving path in the vehicle is adjusted in different dimensions to obtain the bias value threshold adjustment result, which is specifically: Based on the situation that the pedestrian is in a non-crossing scenario, under one-dimensional conditions, the lateral speed of the pedestrian is set to be less than a preset value; under two-dimensional conditions, the longitudinal speed of the pedestrian is set to be less than a preset value; under three-dimensional conditions, the vehicle speed of the vehicle is set to be less than a preset value, and the bias value threshold adjustment result is obtained.

4. The method for optimizing vehicle adaptive braking in complex scenarios as claimed in claim 1, characterized in that: After obtaining the non-crossing scene adjustment result, the method further includes: The closest angle distance and the farthest angle distance between the vehicle camera perception frame and the pedestrian are calculated according to a preset bias formula set; wherein the bias formula set includes a closest distance calculation formula and a farthest distance calculation formula; If the closest angular distance and the farthest angular distance do not meet the preset thresholds, and the walking state of the pedestrian at the current moment and the future predicted moment does not meet the non-crossing scene adjustment result, the automatic emergency braking of the vehicle is triggered.

5. The method for optimizing vehicle adaptive braking in complex scenarios as claimed in claim 4, characterized in that: The closest distance calculation formula is specifically: closeoffset=closeoffset1+0.5*obj_width Among them, closeoffset is the closest angle distance between the adjusted rear camera perception frame and the pedestrian, closeoffset1 is the closest angle distance between the adjusted front camera perception frame and the pedestrian, and obj_width is the pedestrian target width.

6. The method for optimizing vehicle adaptive braking in complex scenarios as claimed in claim 1, characterized in that: When the pedestrian is in a crossing scene, the lateral speed of the pedestrian is filtered to obtain a crossing scene adjustment result, specifically: If the longitudinal speed of the pedestrian is greater than the lateral speed thereof, or the lateral speed of the pedestrian is less than or equal to a preset value, the vehicle braking confidence of the vehicle is set to be untrustworthy, and a confidence setting result is obtained; Under the condition of the confidence setting result, the pedestrian recognition time of the vehicle is set to obtain the crossing scene adjustment result.

7. The vehicle adaptive braking optimization method for complex scenarios as claimed in claim 1, characterized in that: After obtaining the curve scene adjustment result, the method further includes: If the current vehicle speed is less than or equal to a first preset value, the curvature value of the vehicle trajectory is calculated according to the steering wheel angle of the current vehicle; If the current vehicle speed is greater than or equal to a second preset value, the curvature value of the vehicle trajectory is calculated according to the yaw rate of the current vehicle; If the speed of the current vehicle is greater than the first preset value and less than the second preset value, calculating the curvature value of the current vehicle trajectory according to the steering wheel angle and the yaw rate of the current vehicle; If the curvature value of the current vehicle trajectory is greater than the preset value, automatic emergency braking of the current vehicle is triggered.

8. A vehicle adaptive braking optimization device for complex scenarios, characterized in that: Including non-crossing module, crossing module, curve module and comprehensive module; The non-crossing module is used to increase the bias value threshold of the pedestrian intrusion into the vehicle's driving path when the pedestrian is in a non-crossing scene, and obtain a non-crossing scene adjustment result; The crossing module is used to filter the lateral speed of the pedestrian when the pedestrian is in a crossing scene to obtain a crossing scene adjustment result; The curve module is used to reduce the curvature value of the vehicle trajectory to a preset value when the pedestrian is in a curve scene, so as to obtain a curve scene adjustment result; The comprehensive module is used to form the emergency braking adjustment result of the vehicle from the non-crossing scenario adjustment result, the crossing scenario adjustment result and the curve scenario adjustment result.

9. The vehicle adaptive braking optimization device for complex scenarios as claimed in claim 8, characterized in that: The non-traversing module includes a bias unit, a time unit and a combination unit; The bias unit is used to adjust the bias value threshold of the vehicle driving path in the vehicle in different dimensions when the pedestrian is in a non-crossing scene, so as to obtain the bias value threshold adjustment result; The time unit is used to increase the confirmation frame of the pedestrian presence threshold time to obtain the confirmation frame adjustment result; The combining unit is used to form the non-crossing scene adjustment result by using the offset value threshold adjustment result and the confirmation frame adjustment result.

10. The vehicle adaptive braking optimization device for complex scenarios as claimed in claim 9, characterized in that: The bias unit is specifically: Based on the situation that the pedestrian is in a non-crossing scenario, under one-dimensional conditions, the lateral speed of the pedestrian is set to be less than a preset value; under two-dimensional conditions, the longitudinal speed of the pedestrian is set to be less than a preset value; under three-dimensional conditions, the vehicle speed of the vehicle is set to be less than a preset value, and the bias value threshold adjustment result is obtained.

11. The vehicle adaptive braking optimization device for complex scenarios as claimed in claim 8, characterized in that: The non-traversing module further includes a distance subunit and a trigger subunit; The distance subunit is used to calculate the closest angle distance and the farthest angle distance between the vehicle camera perception frame and the pedestrian according to a preset bias formula set; wherein the bias formula set includes a closest distance calculation formula and a farthest distance calculation formula; The trigger subunit is used to trigger the automatic emergency braking of the vehicle if the closest angle distance and the farthest angle distance do not meet the preset thresholds, and the walking state of the pedestrian at the current moment and the future predicted moment does not meet the non-crossing scene adjustment result.

12. The vehicle adaptive braking optimization device for complex scenarios as claimed in claim 11, characterized in that: The closest distance calculation formula is specifically: closeoffset=closeoffset1+0.5*obj_width Among them, closeoffset is the closest angle distance between the adjusted rear camera perception frame and the pedestrian, closeoffset1 is the closest angle distance between the adjusted front camera perception frame and the pedestrian, and obj_width is the pedestrian target width.

13. The vehicle adaptive braking optimization device for complex scenarios as claimed in claim 8, characterized in that: The traversal module includes a confidence unit and a recognition unit; The confidence unit is used to set the vehicle braking confidence of the vehicle to untrustworthy if the longitudinal speed of the pedestrian is greater than its lateral speed, or the lateral speed of the pedestrian is less than or equal to a preset value, to obtain a confidence setting result; The recognition unit is used to set the pedestrian recognition time of the vehicle under the condition of the confidence setting result to obtain the crossing scene adjustment result.

14. The vehicle adaptive braking optimization device for complex scenarios as claimed in claim 8, characterized in that: The curve module also includes a low-speed unit, an overspeed unit, a medium-speed unit and a trigger unit; Wherein, the low-speed unit is used to calculate the curvature value of the vehicle trajectory according to the steering wheel angle of the current vehicle if the current vehicle speed is less than or equal to a first preset value; The overspeed unit is used to calculate the curvature value of the vehicle trajectory according to the yaw rate of the current vehicle if the vehicle speed of the current vehicle is greater than or equal to a second preset value; The medium speed unit is used to calculate the curvature value of the current vehicle trajectory according to the steering wheel angle and the yaw rate of the current vehicle if the vehicle speed of the current vehicle is greater than the first preset value and less than the second preset value; The trigger unit is used to trigger automatic emergency braking of the current vehicle if the curvature value of the current vehicle trajectory is greater than the preset value.

15. A storage medium, characterized in that: The storage medium stores a computer program, which is called and executed by a computer to implement a vehicle adaptive braking optimization method for complex scenarios as described in any one of claims 1 to 7 above.

Images