An automatic driving test scene risk grading method and device

Abstract

The application provides an automatic driving test scene risk grading method and device, the method comprises the following steps: acquiring a road state in a scene, and a driving state of a target vehicle under a target automatic driving level and a preset acceleration strategy set; determining a first drivable trajectory of the target vehicle according to the road state, the driving state and the preset acceleration strategy set; sampling the first drivable trajectory to obtain trajectory sampling information; grading the trajectory sampling information of each first drivable trajectory to obtain a grading result of the first drivable trajectory; and determining a risk level of the target vehicle in the scene under the target automatic driving level according to the grading result. The scheme proposes a risk grading rule for a test vehicle for an automatic driving test scene, and realizes the evaluation of the risk response capability of the test vehicle.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving, in particular to an automatic driving test scene risk grading method and device. BACKGROUND

[0002] With the continuous progress of intelligent networked automobile technology, the level of automatic driving gradually transitions from assisted driving level to full automatic driving level, and the tasks undertaken by automatic driving gradually increase, so a strict and perfect test scene becomes a necessary link before the automatic driving vehicle goes on the road. Testing according to the risk level of the scene can effectively ensure the comprehensiveness of the test results and the integrity of the design and operation domain coverage range, and provides an effective basis for the classification of automatic driving vehicles.

[0003] In the prior art, the theoretical system method for grading the risk of the automatic driving vehicle operation scene only relies on the analysis of external scene information. Among them, Chinese patent document CN110727706A discloses a risk driving scene rapid extraction and grading method for intelligent networked automobile testing, which mainly grades the driving scene risk level according to the subjective evaluation method, interprets the driving scene risk through expert experience, and classifies the scene risk into three levels of high, medium and low by subjective and objective evaluation. Chinese patent document CN110020797A discloses an evaluation method for automatic driving test scene based on perception defects, which classifies the test scene according to the perception degree of different types of vehicle-mounted sensors, and grades the risk level of each type of test scene according to the perception degree. However, these existing scene risk grading methods mostly focus on the information outside the test vehicle in the test scene, and lack objective evaluation of the response risk ability of the test subject vehicle in the scene. When the same type of scene faces different automatic driving level requirements or different types of vehicles, it is unreasonable to evaluate the risk level only by the traffic factors and natural factors in the scene. SUMMARY

[0004] The present application provides an automatic driving test scene risk grading method and device, which proposes a risk grading rule for the test vehicle for the automatic driving test scene, and realizes the evaluation of the response risk ability of the test vehicle.

[0005] In a first aspect, the present application provides an automatic driving test scene risk grading method, comprising:

[0006] obtaining the road state in the scene and the driving state and the preset acceleration strategy set of the target vehicle under the target automatic driving level;

[0007] determining a first drivable trajectory of the target vehicle according to the road state, the driving state and the preset acceleration strategy set;

[0008] sampling the first drivable trajectory to obtain trajectory sampling information;

[0009] grading the trajectory sampling information of each first drivable trajectory to obtain a grading result of the first drivable trajectory;

[0010] determining a risk level of the target vehicle in the target automatic driving level in the scene according to the grading result.

[0011] Optionally, the road state includes road adhesion, road structure, road width, lane number and road curvature; the driving state includes the road position of the vehicle, the driving speed of the vehicle, the driving speed direction of the vehicle and the vehicle body orientation; and the preset acceleration strategy set is a set of accelerations that can be reached by the target vehicle.

[0012] The determining of the first drivable trajectory of the target vehicle according to the road state, the driving state and the preset acceleration strategy set includes:

[0013] determining a trajectory endpoint coordinate of the target vehicle according to the road state, the driving state, the preset acceleration strategy set and the planning time;

[0014] obtaining a current position of the target vehicle and taking the current position as a coordinate starting point;

[0015] performing curve fitting based on the coordinate starting point and the trajectory endpoint coordinate to obtain a displacement curve and a speed curve of the first drivable trajectory.

[0016] Optionally, the trajectory endpoint coordinate is determined by the following formula:

[0017] ο goal,n×m = {(s goal,n ,l goal,m )α n ∈ [α min ,α max ], m∈ [1, k]}

[0018] wherein n represents the nth acceleration, α min and α max represent the minimum acceleration and the maximum acceleration that can be reached by the preset acceleration strategy set respectively; m represents the lane number, m∈ [1, k]; o goal,n×m represents the trajectory endpoint coordinate when the target vehicle adopts the acceleration α n in the mth lane; s goal,n represents the longitudinal coordinate of the trajectory endpoint coordinate o goal,n×m , and l goal,n represents the lateral coordinate of the trajectory endpoint coordinate o goal,n×ma lateral coordinate of the vehicle.

[0019] Optionally, the displacement curve is determined by the following formula:

[0020] f p (t) = a0 + a1t + a2t 2 +a3t 3 +a4t 4 +a5t 4

[0021] The speed curve is determined by the following formula:

[0022] f v (t) = a1 + 2a2t + 3a3t 2 +4a4t 3 +4a5t 3

[0023] Wherein, f p (t) is used to represent the displacement curve; f v (t) is used to represent the speed curve; a0, a1, a2, a3, a4, a5 are all trajectory fitting parameters.

[0024] Optionally, the first drivable trajectory is sampled to obtain trajectory sampling information, including:

[0025] For each of the first drivable trajectory, the following is performed:

[0026] Determine the sampling time of the first drivable trajectory according to a preset time interval;

[0027] Obtain the trajectory information of the first drivable trajectory at each of the sampling time; wherein the trajectory information includes coordinates, longitudinal speed, lateral speed and trajectory curvature;

[0028] Arrange the trajectory information at the sampling time in time sequence as the trajectory sampling information.

[0029] Optionally, the trajectory sampling information of each of the first drivable trajectory is graded to obtain the grading result of the first drivable trajectory, including:

[0030] S1: For each of the sampling time in the trajectory sampling information, it is judged whether the curvature at the sampling time is greater than a preset curvature, and whether the slip angle at the sampling time is greater than a preset slip angle; if the determination results of the sampling time are all no, the trajectory point at the sampling time is a feasible trajectory point; otherwise, it is determined that the trajectory point at the sampling time is an infeasible trajectory point;

[0031] S2: determining whether there is an infeasible trajectory point in each piece of trajectory sampling information; if yes, performing step S3; otherwise, performing step S4;

[0032] S3: determining the first drivable trajectory to which the trajectory sampling information belongs as an infeasible trajectory, and determining the sampling time corresponding to the first infeasible trajectory point in the trajectory sampling information as a trajectory termination time;

[0033] S4: performing collision detection on each sampling time in the trajectory sampling information to determine whether there is a sampling time at which a collision is likely to occur; if yes, performing step S5; otherwise, performing step S6;

[0034] S5: determining the first drivable trajectory to which the trajectory sampling information belongs as a possible collision trajectory, and determining the first sampling time at which a collision is likely to occur in the trajectory sampling information as a trajectory termination time in chronological order;

[0035] S6: determining the first drivable trajectory to which the trajectory sampling information belongs as a collision-free trajectory, and taking the trajectory termination time of the first drivable trajectory as a trajectory termination time; wherein the classification result comprises the trajectory termination time.

[0036] Optionally, the collision detection on each sampling time in the trajectory sampling information to determine whether there is a sampling time at which a collision is likely to occur comprises:

[0037] Obtaining the motion state of other traffic participants in the scene and a set of preset acceleration strategies;

[0038] According to the road state, the motion state of other traffic participants and the set of preset acceleration strategies, determining the second drivable trajectory of the other traffic participants;

[0039] For each of the other traffic participants, sequentially connecting the trajectory points on each of the second drivable trajectories at the same sampling time according to the sampling time of the first drivable trajectory to obtain an occupied area;

[0040] For each sampling time of the first drivable trajectory, determining whether the trajectory point on the first drivable trajectory at the sampling time is located in the occupied area of the other traffic participants at the sampling time; if yes, determining that the first drivable trajectory is likely to collide at the sampling time.

[0041] Optionally, the determination of the risk level of the target vehicle in the scene under the target automatic driving level according to the classification result comprises:

[0042] According to the grading result, a trajectory termination time of the first drivable trajectory, and a quantity of the first drivable trajectory, a risk score of a target vehicle in the scene under a target automatic driving level is calculated.

[0043] According to the risk score, the risk level is determined.

[0044] The risk score is determined by the following formula:

[0045]

[0046] Wherein, R is used to represent the risk score; M is used to represent the quantity of the first drivable trajectory; t z is used to represent the trajectory termination time of the first drivable trajectory; t i is used to represent the trajectory termination time of the first drivable trajectory; t

[0047] In a second aspect, the present application provides an automatic driving test scene risk grading device, comprising:

[0048] A perception module is configured to acquire a road state in a scene, a driving state of a target vehicle under a target automatic driving level, and a preset acceleration strategy set.

[0049] A trajectory determination module is configured to determine a first drivable trajectory of the target vehicle according to the road state, the driving state, and the preset acceleration strategy set.

[0050] A sampling module is configured to sample the first drivable trajectory to obtain trajectory sampling information.

[0051] A risk grading module is configured to grade the trajectory sampling information of each of the first drivable trajectories to obtain a grading result of the first drivable trajectory, and determine a risk level of the target vehicle in the scene under the target automatic driving level according to the grading result.

[0052] In a third aspect, an embodiment of the present application further provides a computing device, comprising a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the method of any first aspect of the present application.

[0053] In a fourth aspect, an embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the computer program, when executed in a computer, causes the computer to execute the method of any first aspect of the present application.

[0054] In a fifth aspect, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of the method according to any of the first aspects of the present specification.

[0055] The embodiment of the present application provides an automatic driving test scene risk grading method and device, which focuses on the demand of the test vehicle automatic driving level in the automatic driving test scene, and scientifically guides the test scene grading. The method takes the standard level automatic driving vehicle as the benchmark, generates a drivable trajectory set (i.e., a first drivable trajectory) by acquiring the road state in the scene, the driving state of the target vehicle under the target automatic driving level and the preset acceleration strategy set, then calculates the potential risk of each trajectory by the sampling method, evaluates the overall risk degree of the current scene under the current automatic driving level, and divides the test scene risk level, so as to supplement the evaluation of the risk response ability of the test vehicle in the test scene risk grading, realize more comprehensive risk evaluation of the test scene, and further ensure the effectiveness, accuracy and comprehensiveness of the intelligent networked vehicle test result. BRIEF DESCRIPTION OF DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0057] Figure 1 is a flow chart of an automatic driving test scene risk grading method provided by an embodiment of the present application;

[0058] Figure 2 is a hardware architecture diagram of a computing device provided by an embodiment of the present application;

[0059] Figure 3 is a structural schematic diagram of an automatic driving test scene risk grading device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0060] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0061] The specific implementation mode of the concept of the present application will be described below.

[0062] Referring to Figure 1 The embodiment of the present application provides an automatic driving test scene risk grading method, which comprises the following steps:

[0063] In step 100, the road state in the scene, the driving state of the target vehicle under the target automatic driving level and the preset acceleration strategy set are obtained.

[0064] In step 102, the first drivable trajectory of the target vehicle is determined according to the road state, the driving state and the preset acceleration strategy set.

[0065] In step 104, the first drivable trajectory is sampled to obtain trajectory sampling information.

[0066] In step 106, the trajectory sampling information of each first drivable trajectory is graded to obtain the grading result of the first drivable trajectory.

[0067] In step 108, the risk level of the target vehicle in the scene under the target automatic driving level is determined according to the grading result.

[0068] In the embodiment of the present application, the method takes the standard level automatic driving vehicle as the benchmark, obtains the road state in the scene, the driving state of the target vehicle under the target automatic driving level and the preset acceleration strategy set, generates a first drivable trajectory set, then calculates the potential risk of each trajectory by sampling method, evaluates the overall risk degree of the current scene under the current automatic driving level, and divides the test scene risk level, which supplements the evaluation of the risk response ability of the test vehicle in the test scene risk grading, realizes more comprehensive risk evaluation of the test scene, and further guarantees the effectiveness, accuracy and comprehensiveness of the intelligent networked vehicle test result.

[0069] The execution mode of each step is described below. Figure 1

[0070] Firstly, in step 100, the road state includes road adhesion, road structure, road width, lane number and road curvature; the driving state includes the road position of the vehicle, the driving speed of the vehicle, the driving speed direction of the vehicle and the vehicle body orientation; and the preset acceleration strategy set is a set of accelerations that can be reached by the target vehicle.

[0071] ​It should be noted that the driving level of the automatic driving technology refers to the SAE J3016 standard, which is divided into six levels, L0-L5; the target automatic driving level is any one of the six levels, which is determined by the requirement of measuring the automatic driving level of the test target. The test scene of the target vehicle under different target automatic driving levels is different, and at least one scene corresponding to the automatic driving level is selected based on the automatic driving level required by the test, the automatic driving system corresponding to the level is used, and the above steps 100-108 are used for risk classification for each scene; the preset acceleration strategy set represents the ability of the vehicle power system to reach the drivable trajectory under the current driving speed and road environment, which is obtained according to the inherent properties of the vehicle.

[0072] Specifically, the road state in the scene includes but is not limited to traffic data collection by roadside sensors to extract relevant road states; wherein the road state is affected by external factors such as weather environment, complex road traffic and traffic accidents, etc.

[0073] According to the road state, the driving state and the preset acceleration strategy set, the first drivable trajectory of the target vehicle is determined, including:

[0074] According to the road state, the driving state, the preset acceleration strategy set and the planning time, the trajectory endpoint coordinates of the target vehicle are determined;

[0075] The current position of the target vehicle is obtained and the current position is taken as the coordinate starting point;

[0076] Based on the coordinate starting point and the trajectory endpoint coordinates, curve fitting is performed to obtain the displacement curve and the speed curve of the first drivable trajectory.

[0077] It should be noted that the planning time is the predicted driving time of the target vehicle during the test. The first drivable trajectory of different types of target vehicles is different.

[0078] In the present application, the automatic driving decision system adopted according to the target automatic driving level can generate discrete trajectory endpoint coordinates, and then by taking the current position of the target vehicle in the current state as the coordinate starting point, combining the trajectory endpoint coordinate set, the set of drivable trajectories of the target vehicle in the current state can be obtained by using curve fitting.

[0079] In a preferred embodiment, the trajectory endpoint coordinates are determined by the following formula:

[0080] ο goal,n×m = {(s goal,n ,l goal,m )α n ∈ [α min ,α max ], m∈[1,k]}

[0081] Where n is used to characterize the nth acceleration, α min α max These are used to characterize the minimum and maximum acceleration achievable by the preset set of acceleration strategies, respectively; m is used to characterize the number of lanes, m∈[1,k]; o goal,n×m Used to characterize the target vehicle's acceleration α n The coordinates of the trajectory endpoint in lane m; s goal,n Used to characterize the coordinates of the trajectory endpoint o goal,n×m The vertical coordinate, l goal,n Used to characterize the coordinates of the trajectory endpoint o goal,n×m The horizontal coordinate.

[0082] Specifically, based on the preset acceleration strategy set, i.e., the discrete acceleration set a n ∈[a min ,a max Using an acceleration model, sampling is performed at preset time intervals within the planned time to calculate the longitudinal coordinate s of each trajectory endpoint. goal,n Simultaneously, based on the acquired road conditions, the centerline coordinates of each lane are used as the lateral coordinates of each trajectory endpoint. goal,n Then, the vertical coordinates and horizontal coordinates of each trajectory endpoint are combined to obtain the trajectory endpoint coordinate set.

[0083] In a preferred embodiment, the displacement curve is determined by the following formula:

[0084] f p (t)=a0+a1t+a2t 2 +a3t 3 +a4t 4 +a5t 4

[0085] The velocity curve is determined by the following formula:

[0086] f v (t)=a1+2a2t+3a3t 2 +4a4t 3 +4a5t 3

[0087] Among them, f p (t) is used to characterize the displacement curve; f v (t) is used to characterize the velocity curve; a0, a1, a2, a3, a4, and a5 are all trajectory fitting parameters, and t is any moment within the planning time.

[0088] It should be noted that the trajectory fitting parameters are based on the target vehicle and are preset.

[0089] In the present application, the mathematical function method is adopted to generate the drivable trajectory of the target vehicle, which reduces the difficulty of trajectory generation while ensuring bounded acceleration and bounded curvature derivative. Thus, when sampling according to the preset time interval, the coordinates, longitudinal speed, transverse speed and trajectory curvature of the trajectory points at each sampling time on the first drivable trajectory can be directly obtained.

[0090] In step 104, the first drivable trajectory is sampled to obtain trajectory sampling information, including:

[0091] For each first drivable trajectory, the following is performed:

[0092] The sampling time of the first drivable trajectory is determined according to the preset time interval;

[0093] The trajectory information of the first drivable trajectory at each sampling time is obtained; wherein the trajectory information includes coordinates, longitudinal speed, transverse speed and trajectory curvature;

[0094] The trajectory information at the sampling time arranged in time sequence is taken as the trajectory sampling information.

[0095] Specifically, for example, the planning time of each first drivable trajectory is 5s, and the preset time interval is 0.1s, then the first drivable trajectory has 50 sampling times, and the sampling times in the time range of 0-5s are 0.1s, 0.2s, 0.3s……4.8s, 4.9s, 5s, then the trajectory information of the first drivable trajectory at 0.1s is obtained, the trajectory information 01 at 0.2s is obtained, the trajectory information 03 at 0.3s is obtained, the trajectory information 48 at 4.8s is obtained, the trajectory information 49 at 4.9s is obtained, and the trajectory information 50 at 5s is obtained, then the trajectory information is arranged in time sequence to obtain the trajectory sampling information: trajectory information 01-trajectory information 02-trajectory information 03……trajectory information 48-trajectory information 49-trajectory information 50.

[0096] It should be noted that the preset time interval of sampling is determined according to the test requirements and risk accuracy, and the equal interval sampling method is adopted.

[0097] In step 106, the trajectory sampling information of each first drivable trajectory is classified to obtain the classification result of the first drivable trajectory, including:

[0098] S1: for each sampling time in the trajectory sampling information, performing: judging whether the curvature at the sampling time is greater than a preset curvature, and judging whether the slip angle at the sampling time is greater than a preset slip angle; if the judgment results of the sampling time are all not yes, the trajectory point at the sampling time is a feasible trajectory point; otherwise, it is determined that the trajectory point at the sampling time is an infeasible trajectory point;

[0099] S2: judging whether there is an infeasible trajectory point in each piece of trajectory sampling information; if yes, performing step S3; otherwise, performing step S4;

[0100] S3: determining the first drivable trajectory to which the trajectory sampling information belongs as an infeasible trajectory, and determining the sampling time corresponding to the first infeasible trajectory point appearing in the trajectory sampling information as a trajectory termination time;

[0101] S4: performing collision detection on each sampling time in the trajectory sampling information, judging whether there is a sampling time at which a collision may occur; if yes, performing step S5; otherwise, performing step S6;

[0102] S5: determining the first drivable trajectory to which the trajectory sampling information belongs as a possible collision trajectory, and determining the first sampling time at which a collision may occur in the trajectory sampling information as a trajectory termination time in time sequence;

[0103] S6: determining the first drivable trajectory to which the trajectory sampling information belongs as a collision-free trajectory, and taking the trajectory termination time of the first drivable trajectory as a trajectory termination time; wherein, the classification result includes the trajectory termination time.

[0104] Specifically, the calculation formula of the slip angle β is:

[0105]

[0106] wherein v x is the longitudinal speed of the vehicle, and v y is the lateral speed of the vehicle.

[0107] Specifically, in step S1, the trajectory point at the sampling time is an infeasible trajectory point, including the following three cases: 1) the judgment result is that the curvature at the sampling time is greater than the preset curvature, and the slip angle is not greater than the preset slip angle; 2) the judgment result is that the curvature at the sampling time is greater than the preset curvature, and the slip angle is greater than the preset slip angle; 3) the judgment result is that the curvature at the sampling time is not greater than the preset curvature, and the slip angle is greater than the preset slip angle.

[0108] As described in the foregoing example, in the first drivable trajectory A, the trajectory points at 0.2s, 0.3s and 4.8s obtained by step S1 are all unfeasible trajectory points, so the first drivable trajectory A is an unfeasible trajectory, and the trajectory termination time is 0.2s (i.e., the second sampling time).

[0109] In the present application, the trajectory sampling information is first subjected to dynamic checking, including checking the slip angle and curvature of the first drivable trajectory to determine whether the trajectory is feasible, and if the slip angle and curvature are both qualified after dynamic checking, collision checking is performed; the collision checking uses other traffic participants in the scene except the target vehicle to determine the possibility of collision, and if there is no collision, the feasible trajectory is a collision-free trajectory with the lowest risk level and the highest safety, so the trajectory termination time is the trajectory termination time. Once the dynamic checking or the collision checking fails, the drivable trajectory is at risk and the trajectory termination time needs to be determined in time. Thus, based on the dynamic checking and the collision checking, the accuracy, reliability and scientificity of the automatic driving test scene risk classification method are further improved.

[0110] Specifically, in step S6, if the first drivable trajectory passes the dynamic checking and the collision checking and no unfeasible trajectory or possible collision occurs at any sampling time, the planning time is used as the trajectory termination time, and the trajectory is classified as a collision-free trajectory.

[0111] In some preferred embodiments, in step 106, the collision checking is performed on each sampling time in the trajectory sampling information to determine whether there is a sampling time at which a possible collision occurs, including:

[0112] obtaining the motion state of each other traffic participant in the scene and a set of preset acceleration strategies;

[0113] determining a second drivable trajectory of each other traffic participant according to the road state, the motion state of each other traffic participant and the set of preset acceleration strategies;

[0114] For each other traffic participant, the following is performed: according to the sampling time of the first drivable trajectory, sequentially connecting the trajectory points on each second drivable trajectory at the same sampling time to obtain an occupied area.

[0115] For each sampling time of the first drivable trajectory, the following is performed: determining whether the trajectory point on the first drivable trajectory at the sampling time is located in the occupied area of the other traffic participant at the sampling time; if yes, it is determined that the first drivable trajectory may collide at the sampling time.

[0116] It should be noted that other road users include pedestrians, non-motor vehicles and motor vehicles, and for pedestrians, the second drivable trajectory is the movable trajectory of the pedestrian. The second drivable trajectory is obtained according to the method of steps 100 and 102.

[0117] As described in the foregoing example, the first drivable trajectory has 50 sampling time points, and for the other road user B having 5 second drivable trajectories, for each drivable trajectory, the trajectory points of B at the 50 sampling time points are obtained, and then the 5 trajectory points obtained at the first sampling time point (0.1s) are sequentially connected to form an occupied area of the other road user B at the first sampling time point. It is judged whether there is at least one trajectory point in the trajectory points of the target vehicle at each first sampling time point in all drivable trajectories located in the occupied area, and if so, a collision may occur at the trajectory point.

[0118] In the present application, the collision detection adopts the method of the occupied area, and whether a collision occurs is judged according to whether an intersection is generated between the coordinate information in the trajectory sampling information and the occupied area, and if an intersection is generated, the earliest sampling time at which the intersection is generated is recorded as the trajectory termination time, and the corresponding first drivable trajectory is determined as the possible collision trajectory. Specifically, according to the sampling time sequence, the trajectory sampling information and the occupied area of the other road user are compared in sequence, and if it is calculated that there is a trajectory point coordinate in the trajectory sampling information located in the occupied area at the same sampling time, the earliest sampling time at which the intersection is generated is recorded as the trajectory termination time.

[0119] In step 108, according to the classification result, the risk level of the target vehicle in the target automatic driving level in the scene is determined, including:

[0120] According to the classification result, the trajectory termination time of the first drivable trajectory and the number of first drivable trajectories, the risk score of the target vehicle in the target automatic driving level in the scene is calculated;

[0121] According to the risk score, the risk level is determined;

[0122] Wherein, the risk score is determined by the following formula:

[0123]

[0124] Wherein, R is used to represent the risk score; M is used to represent the number of first drivable trajectories; t z is used to represent the trajectory termination time of the first drivable trajectory; t i is used to represent the trajectory termination time of the first drivable trajectory; t

[0125] It should be noted that the higher the risk score, the higher the risk level. The trajectory termination time of the first drivable trajectory tz are all the same.

[0126] In the present application, based on the trajectory termination time in the trajectory sampling information hierarchical result, the overall risk of the test scene is calculated, and the risk of the scene to the corresponding automatic driving level is evaluated as the test scene risk grading standard. The present application proposes an automatic driving test scene risk grading method based on sampling trajectories, which reflects the ability of the automatic driving vehicle to cope with risks in a way based on a set of drivable trajectories, proposes risk grading rules for test vehicles for automatic driving test scenes, differentiates the test targets of test scenes, speeds up the test process of automatic driving vehicles, supplements the evaluation of the risk coping ability of test vehicles in the test scene risk grading, and improves the rationality of the test scene risk grading method.

[0127] In a more specific embodiment, according to the risk score, the risk level is determined, including: according to the numerical value of the risk score, the scene is graded, when 0≤R<0.33, the risk level of the scene is considered to be low; when 0.33≤R<0.67, the risk level of the scene is considered to be medium; when 0.67≤R<1, the risk level of the scene is considered to be high.

[0128] As shown in Figure 2 , Figure 3 , the present embodiment provides an automatic driving test scene risk grading device. The device embodiment can be realized by software, or realized by hardware or a combination of software and hardware. From the hardware layer, as shown in Figure 2 , it is a hardware architecture diagram of a computing device where the automatic driving test scene risk grading device provided by the present embodiment is located. In addition to the processor, memory, network interface, and non-volatile memory shown in Figure 2 , the computing device where the device in the embodiment is usually also includes other hardware, such as a forwarding chip responsible for processing packets, etc. Taking the software implementation as an example, as shown in Figure 3 , as a logically meaningful device, it is formed by the CPU of the computing device where it is located to read the corresponding computer program in the non-volatile memory into the memory for running. The automatic driving test scene risk grading device provided by the present embodiment, the device includes:

[0129] The perception module 300 is configured to acquire the road state in the scene, the driving state of the target vehicle under the target automatic driving level, and a set of preset acceleration strategies.

[0130] The trajectory determination module 302 is configured to determine a first drivable trajectory of the target vehicle according to the road state, the driving state, and the set of preset acceleration strategies.

[0131] The sampling module 304 is configured to sample the first drivable trajectory to obtain trajectory sampling information.

[0132] The risk grading module 306 is configured to grade the trajectory sampling information of each first drivable trajectory to obtain a grading result of the first drivable trajectory, and determine a risk level of the target vehicle in the target automatic driving level in the scene according to the grading result.

[0133] In some specific embodiments, the perception module 300 can be configured to perform the step 100, the trajectory determination module 302 can be configured to perform the step 102, the sampling module 304 can be configured to perform the step 104, and the risk grading module 306 can be configured to perform the steps 106 and 108.

[0134] In an embodiment of the present application, the road state includes road adhesion, road structure, road width, lane number and road curvature; the driving state includes the road position of the vehicle, the driving speed of the vehicle, the driving speed direction of the vehicle and the vehicle body orientation; and the preset acceleration strategy set is a set of accelerations that can be reached by the target vehicle.

[0135] In an embodiment of the present application, the trajectory determination module 302 is further configured to perform the following operations:

[0136] determine the trajectory endpoint coordinates of the target vehicle according to the road state, the driving state, the preset acceleration strategy set and the planning time; the trajectory endpoint coordinates are determined by the following formula:

[0137] ο goal,n×m = {(s goal,n , l goal,m )α n ∈ [α min , α max ], m∈ [1, k]}

[0138] wherein n is used to represent the nth acceleration, α min and α max are used to represent the minimum acceleration and the maximum acceleration that can be reached by the preset acceleration strategy set respectively; m is used to represent the lane number, m∈ [1, k]; o goal,n×m is used to represent the trajectory endpoint coordinates of the target vehicle when using the acceleration α n in the mth lane; s goal,n is used to represent the longitudinal coordinate of the trajectory endpoint coordinates o goal,n×m , l goal,n is used to represent the transverse coordinate of the trajectory endpoint coordinates o goal,n×m , and k is the total number of lanes.

[0139] obtain the current position of the target vehicle and take the current position as the coordinate starting point;

[0140] The displacement curve of the first drivable trajectory and the speed curve of the first drivable trajectory are obtained through curve fitting based on the coordinate origin and the trajectory end point coordinate;

[0141] The displacement curve is determined through the following formula:

[0142] f p (t)=a0+a1t+a2t 2 +a3t 3 +a4t 4 +a5t 4

[0143] The speed curve is determined through the following formula:

[0144] f v (t)=a1+2a2t+3a3t 2 +4a4t 3 +4a5t 3

[0145] Wherein, f p (t) is used to represent the displacement curve; f v (t) is used to represent the speed curve; a0, a1, a2, a3, a4, a5 are all trajectory fitting parameters.

[0146] In an embodiment of the present application, the sampling module 304 is further configured to perform the following operations:

[0147] For each first drivable trajectory, the following operations are performed:

[0148] The sampling time of the first drivable trajectory is determined according to a preset time interval;

[0149] The trajectory information of the first drivable trajectory at each sampling time is obtained; wherein, the trajectory information includes coordinates, longitudinal speed, lateral speed and trajectory curvature;

[0150] The trajectory information at the sampling time arranged in time sequence is taken as trajectory sampling information.

[0151] In an embodiment of the present application, the risk grading module 306 is further configured to perform the following operations:

[0152] S1: For each sampling time in the trajectory sampling information, the following operations are performed: judging whether the curvature at the sampling time is greater than a preset curvature, and judging whether the slip angle at the sampling time is greater than a preset slip angle; if the judgment results of the sampling time are all no, the trajectory point at the sampling time is a feasible trajectory point; otherwise, it is determined that the trajectory point at the sampling time is an infeasible trajectory point;

[0153] S2: determining whether there is an infeasible trajectory point in each piece of trajectory sampling information; if yes, performing step S3; otherwise, performing step S4;

[0154] S3: determining the first drivable trajectory to which the trajectory sampling information belongs as an infeasible trajectory, and determining the sampling time corresponding to the first infeasible trajectory point in the trajectory sampling information as the trajectory termination time;

[0155] S4: performing collision detection on each sampling time in the trajectory sampling information to determine whether there is a sampling time at which a collision is likely to occur; if yes, performing step S5; otherwise, performing step S6;

[0156] S5: determining the first drivable trajectory to which the trajectory sampling information belongs as a possible collision trajectory, and determining the first sampling time at which a collision is likely to occur in the trajectory sampling information as the trajectory termination time in time sequence;

[0157] S6: determining the first drivable trajectory to which the trajectory sampling information belongs as a collision-free trajectory, and taking the trajectory termination time of the first drivable trajectory as the trajectory termination time; wherein the classification result comprises the trajectory termination time.

[0158] In an embodiment of the present application, the risk classification module 306 is further configured to perform the following operations:

[0159] S41: obtaining the motion state of other traffic participants in the scene and a set of preset acceleration strategies;

[0160] S42: determining the second drivable trajectory of other traffic participants according to the road state, the motion state of other traffic participants and the set of preset acceleration strategies;

[0161] S43: for each other traffic participant, performing the following operations: connecting the trajectory points on each second drivable trajectory at the same sampling time in sequence to obtain an occupied area according to the sampling time of the first drivable trajectory;

[0162] S44: for each sampling time of the first drivable trajectory, performing the following operations: determining whether the trajectory point on the first drivable trajectory at the sampling time is located in the occupied area of other traffic participants at the sampling time; if yes, determining that the first drivable trajectory is likely to collide at the sampling time.

[0163] In an embodiment of the present application, the risk classification module 306 is further configured to perform the following operations:

[0164] According to the classification result, the trajectory termination time of the first drivable trajectory and the number of first drivable trajectories, calculating the risk score of the target vehicle in the scene under the target automatic driving level;

[0165] determining a risk level according to the risk score;

[0166] wherein the risk score is determined by the following formula:

[0167]

[0168] wherein R is used to represent the risk score; M is used to represent the number of first drivable trajectories; t z is used to represent the trajectory termination time of the first drivable trajectory; t i is used to represent the trajectory suspension time of the i-th first drivable trajectory.

[0169] It can be understood that the structure shown in the embodiments of the present application does not constitute a specific limitation on the automatic driving test scene risk grading device. In other embodiments of the present application, an automatic driving test scene risk grading device can include more or fewer components than the illustration, or combine certain components, or split certain components, or different component arrangements. The illustrated components can be implemented in hardware, software, or a combination of software and hardware.

[0170] The information interaction and execution process between the modules in the above device are based on the same concept as the method embodiments of the present application, and the specific content can be referred to the description in the method embodiments of the present application, which will not be described here.

[0171] The embodiments of the present application also provide a computing device, comprising a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the automatic driving test scene risk grading method in any of the embodiments of the present application.

[0172] The embodiments of the present application also provide a computer readable storage medium, which stores a computer program, and the computer program, when executed by a processor, causes the processor to execute the automatic driving test scene risk grading method in any of the embodiments of the present application.

[0173] The embodiments of the present application also provide a computer program product, which comprises a computer program, and a processor of a computer device reads the computer program from a computer readable storage medium, and the processor executes the computer program to cause the computer device to execute the automatic driving test scene risk grading method in any of the above embodiments.

[0174] Specifically, a system or device equipped with a storage medium can be provided, and the storage medium stores a software program code for realizing the functions of any of the above embodiments, and the computer (or CPU or MPU) of the system or device reads and executes the program code stored in the storage medium.

[0175] In this case, the program code read from the storage medium can itself implement the functions of any of the above-described embodiments, and thus the program code and the storage medium that stores the program code can constitute the present application.

[0176] Embodiments of storage media that can be used to provide the program code include floppy disks, hard disks, magneto-optical disks, CD-ROMs, CD-Rs, CD-RWs, DVDs, flash memories, nonvolatile memories, or any other storage media. Alternatively, the program code can be downloaded from a network, such as the Internet, or a server computer.

[0177] In addition, it should be understood that the functions of the above-described embodiments can be implemented by one or more modules, and that each of the modules can be implemented by a hardware component such as a processor or a processor core, or by any suitable combination of hardware and software. Further, it should be appreciated that software implementations can be realized on any programming language, such as C, C++, Java, or any other suitable programming language.

[0178] In addition, it should be understood that the functions of the above-described embodiments can be implemented by one or more modules, and that each of the modules can be implemented by a hardware component such as a processor or a processor core, or by any suitable combination of hardware and software. Further, it should be appreciated that software implementations can be realized on any programming language, such as C, C++, Java, or any other suitable programming language.

[0179] It should be noted that the terms "first" and "second" and the like are used merely to distinguish one entity or action from another, and do not necessarily require or imply any actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. In other words, any of the steps of a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0180] It should be understood by those of ordinary skill in the art that all or part of the steps of the above-described method embodiments can be completed by program instruction-related hardware, and the aforementioned program can be stored in a computer-readable storage medium, and the program, when executed, performs steps including the above-described method embodiments; and the aforementioned storage medium includes ROM, RAM, magnetic disk or optical disk, and various media that can store program codes.

[0181] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit the same; and although the present application has been described in detail with reference to the foregoing embodiments, it should be appreciated by those skilled in the art that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features thereof can be replaced equivalently; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An automatic driving test scene risk grading method, characterized in that, The method comprises the following steps: acquiring a road state in a scene, a driving state of a target vehicle under a target automatic driving level, and a preset acceleration strategy set; determining a first drivable trajectory of the target vehicle according to the road state, the driving state, and the preset acceleration strategy set; for each of the first drivable trajectories, performing the following steps: determining a sampling time of the first drivable trajectory according to a preset time interval; acquiring trajectory information of the first drivable trajectory at each of the sampling times; wherein the trajectory information comprises coordinates, longitudinal speed, lateral speed, and trajectory curvature; and arranging the trajectory information at the sampling times in time sequence as trajectory sampling information; grading the trajectory sampling information of each of the first drivable trajectories to obtain a grading result of the first drivable trajectory; determining a risk level of the target vehicle in the scene under the target automatic driving level according to the grading result; the grading of the trajectory sampling information of each of the first drivable trajectories to obtain the grading result of the first drivable trajectory comprises: S1: for each of the sampling times in the trajectory sampling information, performing the following steps: determining whether the curvature at the sampling time is greater than a preset curvature, and determining whether the slip angle at the sampling time is greater than a preset slip angle; if the determination results of the sampling time are all no, the trajectory point at the sampling time is a feasible trajectory point; otherwise, it is determined that the trajectory point at the sampling time is an infeasible trajectory point; S2: determining whether there is an infeasible trajectory point in each of the trajectory sampling information; if there is, performing step S3; otherwise, performing step S4; S3: determining the first drivable trajectory to which the trajectory sampling information belongs as an infeasible trajectory, and determining the sampling time corresponding to the first infeasible trajectory point in the trajectory sampling information as a trajectory termination time; S4: performing collision detection on each of the sampling times in the trajectory sampling information to determine whether there is a sampling time at which a collision may occur; if there is, performing step S5; otherwise, performing step S6; S5: determining the first drivable trajectory to which the trajectory sampling information belongs as a possible collision trajectory, and determining the first sampling time at which a collision may occur in the trajectory sampling information as a trajectory termination time in time sequence; S6: determining the first drivable trajectory to which the trajectory sampling information belongs as a collision-free trajectory, and taking the trajectory termination time of the first drivable trajectory as a trajectory termination time; wherein the grading result comprises the trajectory termination time.

2. The method of claim 1, wherein, The road state comprises road adhesion, road structure, road width, lane number, and road curvature; the driving state comprises the road position of the vehicle, the driving speed of the vehicle, the driving speed direction of the vehicle, and the orientation of the vehicle body; and the preset acceleration strategy set is a set of accelerations that can be achieved by the target vehicle; the determination of the first drivable trajectory of the target vehicle according to the road state, the driving state, and the preset acceleration strategy set comprises: determine a trajectory endpoint coordinate of the target vehicle according to the road state, the driving state, the preset acceleration strategy set and a planning time; obtain a current position of the target vehicle and take the current position as a coordinate starting point; perform curve fitting based on the coordinate starting point and the trajectory endpoint coordinate to obtain a displacement curve and a speed curve of the first drivable trajectory.

3. The method of claim 2, wherein the trajectory endpoint coordinate is determined by the following formula:

4. The method of claim 2, wherein the displacement curve is determined by the following formula: wherein n is used to represent the nth acceleration, a min , max are respectively used to represent the minimum acceleration and the maximum acceleration that can be reached by the set of preset acceleration strategies; m is used to represent the number of lanes, ; o goal , n×m is used to represent the trajectory endpoint coordinate of the target vehicle when adopting the acceleration a n in the mth lane; s goal , n is used to represent the longitudinal coordinate of the trajectory endpoint coordinate o goal , n×m ; and l goal , n is used to represent the lateral coordinate of the trajectory endpoint coordinate o goal , n×m . the speed curve is determined by the following formula: the collision checking of each sampling time in the trajectory sampling information to determine whether there is a sampling time in which a collision is likely to occur, comprises: obtaining a motion state and a preset acceleration strategy set of other traffic participants in the scene; wherein f p (t) is used to characterize the displacement curve; f v (t) is used to characterize the velocity curve; a0, a1, a2, a3, a4, a5 are all trajectory fitting parameters.

5. The method of claim 1, wherein, determining a second drivable trajectory of the other traffic participants according to the road state, the motion state of the other traffic participants and the preset acceleration strategy set; for each of the other traffic participants, performing: sequentially connecting trajectory points on each of the second drivable trajectories at the same sampling time according to the sampling time of the first drivable trajectory to obtain an occupied area; for each sampling time of the first drivable trajectory, performing: determining whether a trajectory point on the first drivable trajectory at the sampling time is located in the occupied area of the other traffic participants at the sampling time; if yes, determining that the first drivable trajectory is likely to collide at the sampling time. determining the risk level of the target vehicle in the scene at the target automatic driving level according to the classification result, comprises: calculating a risk score of the target vehicle in the scene at the target automatic driving level according to the classification result, a trajectory termination time of the first drivable trajectory and a number of the first drivable trajectories; 6. The method of claim 1, wherein, determining the risk level according to the risk score; wherein the risk score is determined by the following formula: for implementing the method of any one of claims 1-6, comprising: a perception module configured to obtain a road state in a scene and a driving state and a preset acceleration strategy set of a target vehicle at a target automatic driving level; wherein R is used to represent the risk score; M is used to represent the number of the first drivable trajectories; t z is used to represent the trajectory termination time of the first drivable trajectory; t i is used to represent the trajectory suspension time of the i-th first drivable trajectory.

7. An automatic driving test scene risk grading apparatus characterized by comprising: a trajectory determination module configured to determine a first drivable trajectory of the target vehicle according to the road state, the driving state and the preset acceleration strategy set; a sampling module configured to sample the first drivable trajectory to obtain trajectory sampling information; a risk classification module configured to classify trajectory sampling information of each of the first drivable trajectories to obtain a classification result of the first drivable trajectory, and determine a risk level of the target vehicle in the scene at the target automatic driving level according to the classification result.

8. A computing device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the method of any one of claims 1-6 when executing the computer program. ​ ​ 9. A computer-readable storage medium having stored thereon a computer program, which, when executed in a computer, causes the computer to perform the method of any one of claims 1-6.

Images