Reliability determination method and device for vehicle cluster driving, and server

Abstract

The present application discloses a method and device for determining the reliability of vehicle cluster driving, and a server, the method comprising: obtaining first data involved in the driving path of vehicles in the autonomous driving vehicle cluster; determining M deviation rates at a first moment based on the first data, where M is a positive integer; inputting the M deviation rates into a neural network model to obtain a system failure rate; determining a state transition probability matrix of vehicles in the autonomous driving vehicle cluster modeled by a Markov model at a first moment based on the system failure rate; determining the reliability of vehicles in the autonomous driving vehicle cluster at a second moment based on the state transition probability matrix, where the second moment is the next moment after the first moment, thereby utilizing the neural network model and the Markov model to achieve real-time, dynamic, and accurate reliability determination of the driving process of the autonomous driving vehicle cluster.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and specifically to a method and device for determining the reliability of vehicle cluster driving, and a server. Background Art

[0002] Autonomous driving vehicle swarms are an important application in the field of autonomous driving. They can effectively improve the efficiency of large-scale autonomous driving traffic operations and significantly alleviate traffic congestion.

[0003] Autonomous driving vehicle clusters include autonomous single vehicles and autonomous driving vehicle formations. Among them, autonomous single vehicles are single vehicles with autonomous driving functions, while autonomous driving vehicle formations are formed by rationally arranging multiple vehicles with autonomous driving functions.

[0004] The reliability of autonomous driving vehicle cluster driving is related to the driving efficiency and driving time of autonomous driving single vehicles and autonomous driving vehicle formations, and has an important impact on autonomous driving strategies, autonomous driving safety and stability. However, there are currently few studies on the reliability of autonomous driving vehicle cluster driving, and further research is needed. Summary of the invention

[0005] The embodiments of the present application provide a method and device for determining the reliability of vehicle cluster driving, and a server, in the hope of using a neural network model and a Markov model to achieve real-time, dynamic, and accurate reliability determination of the driving process of an autonomous driving vehicle cluster.

[0006] In a first aspect, a method for determining reliability of vehicle cluster driving of the present application includes:

[0007] Acquire first data involved in a driving path of a vehicle in a cluster of autonomous vehicles, wherein the first data includes at least one of the following: basic safety messages, roadside information, roadside safety messages, signal phase timing information, and map data messages;

[0008] Determine M deviation rates at a first moment according to the first data, where the deviation rates are used to represent probabilities that vehicles in the autonomous driving vehicle cluster will deviate from the driving path, and M is a positive integer;

[0009] Inputting the M deviation rates into a neural network model to obtain a system failure rate, wherein the system failure rate is used to represent a probability of a system failure occurring in a vehicle in the autonomous driving vehicle cluster at the first moment;

[0010] Determine, according to the system failure rate, a state transition probability matrix of vehicles in the autonomous driving vehicle cluster modeled by the Markov model traveling at the first moment;

[0011] The reliability of the vehicles in the autonomous driving vehicle cluster traveling at a second moment is determined according to the state transition probability matrix, where the second moment is a moment after the first moment.

[0012] It can be seen that the present application can start from the factors that affect the reliability of the driving of the autonomous driving vehicle cluster, use the data acquired in real time (i.e., the first data) to calculate at least one (i.e., M) deviation rate, input the at least one deviation rate into the neural network model to calculate the probability of a system failure of the vehicles in the autonomous driving vehicle cluster at a certain moment (i.e., the first moment) to obtain the system failure rate, and input the system failure rate into the Markov model to predict the state change trend of the vehicles in the autonomous driving vehicle cluster at the next moment (i.e., the second moment) to determine (evaluate / estimate / analyze, etc.) reliability.

[0013] In this way, by adopting the continuous learning, training ability and high prediction ability of the neural network model, it can be ensured that the obtained system failure rate has a higher accuracy, and by adopting the predictive ability of the Markov model without the need for analysis and deduction based on a large amount of historical data, the amount of data required to be collected is small, the amount of calculation is small, and the practicality is strong. Ultimately, the real-time, dynamic and accurate reliability determination of the driving process of the autonomous driving vehicle cluster can be achieved, which greatly shortens the reliability evaluation cycle, improves the reliability evaluation efficiency, and meets the current development needs of the automotive industry for reliability evaluation of autonomous driving vehicle cluster driving.

[0014] In a second aspect, a reliability determination device for vehicle cluster driving of the present application includes:

[0015] A data acquisition unit, configured to acquire first data involved in a driving path of a vehicle in a cluster of autonomous vehicles, wherein the first data includes at least one of the following: basic safety messages, roadside information, roadside safety messages, signal phase timing information, and map data messages;

[0016] a deviation rate determination unit, configured to determine M deviation rates at a first moment according to the first data, wherein the deviation rates are used to represent probabilities that vehicles in the autonomous driving vehicle cluster deviate from the driving path, and M is a positive integer;

[0017] a system failure rate determination unit, configured to input the M deviation rates into a neural network model to obtain a system failure rate, wherein the system failure rate is used to represent a probability of a system failure occurring in a vehicle in the autonomous driving vehicle cluster at the first moment;

[0018] A state transition probability matrix determining unit, used to determine the state transition probability matrix of the Markov model at the first moment according to the system failure rate;

[0019] A reliability determination unit is used to determine the reliability of the autonomous driving vehicle cluster traveling at a second moment according to the state transition probability matrix, where the second moment is the next moment of the first moment.

[0020] It can be seen that the present application can use the reliability determination device of vehicle cluster driving to start from the factors affecting the reliability of autonomous driving vehicle cluster driving, use the data acquired in real time (i.e., the first data) to calculate at least one (i.e., M) deviation rate, input the at least one deviation rate into the neural network model to calculate the probability of a system failure of the vehicles in the autonomous driving vehicle cluster at a certain moment (i.e., the first moment) to obtain the system failure rate, and input the system failure rate into the Markov model to predict the state change trend of the vehicles in the autonomous driving vehicle cluster at the next moment (i.e., the second moment) to determine the reliability.

[0021] In this way, by adopting the continuous learning, training ability and high prediction ability of the neural network model, it can be ensured that the obtained system failure rate has a higher accuracy, and by adopting the predictive ability of the Markov model without the need for analysis and deduction based on a large amount of historical data, the amount of data required to be collected is small, the amount of calculation is small, and the practicality is strong. Ultimately, the driving process of the autonomous driving vehicle cluster can be determined in real time, dynamically and accurately (evaluation / estimation / analysis, etc.), which greatly shortens the reliability evaluation cycle, improves the reliability evaluation efficiency, and meets the current development needs of the automotive industry for reliability evaluation of autonomous driving vehicle cluster driving.

[0022] The third aspect is a server of the present application, comprising a processor, a memory, and a computer program or instructions stored in the memory, wherein the processor executes the computer program or instructions to implement the steps in the method designed in the first aspect.

[0023] It can be seen that the present application can use the computing advantages of the server to achieve the above-mentioned first aspect of using the neural network model and the Markov model to perform real-time, dynamic and accurate reliability determination of the driving process of the autonomous driving vehicle cluster, and improve the computational efficiency of reliability determination.

[0024] The fourth aspect is a computer-readable storage medium of the present application, wherein a computer program or instruction is stored therein, and when the computer program or instruction is executed, the steps in the method designed in the first aspect are implemented.

[0025] The fifth aspect is a computer program product of the present application, comprising a computer program or instructions, wherein the computer program or instructions, when executed, implement the steps in the method designed in the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.

[0027] Figure 1 is a schematic diagram of a reliability determination system architecture for self-driving vehicle cluster driving according to an embodiment of the present application;

[0028] Figure 2 It is a structural schematic diagram of a reliability determination system for autonomous driving vehicle cluster driving according to an embodiment of the present application;

[0029] Figure 3 is a schematic diagram of a speed limit of a current driving path of a vehicle in an embodiment of the present application;

[0030] Figure 4 is a schematic diagram of the current driving direction of a vehicle in an embodiment of the present application;

[0031] Figure 5 is a schematic diagram of the current center position of a vehicle body in an embodiment of the present application;

[0032] Figure 6 is a schematic diagram of two vehicles in front and behind on the same driving path in an embodiment of the present application;

[0033] Figure 7 is a schematic diagram of a forward stop line at a traffic light intersection according to an embodiment of the present application;

[0034] Figure 8 is a schematic diagram of mutual transitions between five states in a Markov state in an embodiment of the present application;

[0035] Fig. 9 is a schematic diagram of an updating process of a state transition probability matrix in an embodiment of the present application;

[0036] Fig.10 It is a flow chart of a method for determining reliability of vehicle cluster driving according to an embodiment of the present application;

[0037] Fig.11 This is a block diagram of the functional units of a device for determining the reliability of vehicle cluster driving according to an embodiment of the present application;

[0038] Fig.12 It is a structural diagram of a server in an embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to better understand the technical solution of the present application by those skilled in the art, the technical solution in the embodiments of the present application is described below in conjunction with the drawings in the embodiments of the present application. It is obvious that the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. With respect to the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0040] It should be understood that the terms "first", "second", etc. involved in the embodiments of the present application are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, software, product, or device that includes a series of steps or units is not limited to the listed steps or units, but also includes steps or units that are not listed, or also includes other steps or units inherent to these processes, methods, products, or devices.

[0041] The "embodiment" involved in the embodiments of the present application means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0042] The "and / or" in the embodiments of the present application describes the association relationship of the associated objects, indicating that three relationships may exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist at the same time, and B exists alone. Among them, A and B can be singular or plural. The character " / " can indicate that the previous and next associated objects are in an "or" relationship. In addition, the symbol " / " can also represent a division sign, that is, performing a division operation.

[0043] In the embodiments of the present application, "at least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent the following seven situations: a, b, c, a and b, a and c, b and c, a, b, and c. Each of a, b, and c can be an element or a set containing one or more elements.

[0044] At present, the reliability of autonomous driving vehicle cluster driving is related to the driving efficiency and driving time of autonomous driving single vehicles and autonomous driving vehicle formations, and has an important impact on autonomous driving strategies, autonomous driving safety and stability. However, there are currently few research methods on the reliability of autonomous driving vehicle cluster driving at home and abroad, and there are certain shortcomings.

[0045] For example, one approach is to study the reliability of the hardware and software of a single autonomous vehicle, but this approach is a static evaluation method and is a post-statistical result. In the process of autonomous vehicle cluster operation, it is necessary to keep track of the current status of the single vehicle, the fleet, and each vehicle in the fleet at any time, requiring the reliability measurement value to be real-time and dynamic, so it is necessary to conduct a dynamic reliability evaluation for the autonomous vehicle cluster.

[0046] One way is to use a hierarchical statistical method to evaluate the reliability of autonomous driving vehicle cluster driving. Although this method can evaluate the dynamic reliability of autonomous driving vehicle cluster driving, this method has shortcomings such as complex calculations, long reliability evaluation cycle, and poor practicality.

[0047] Based on this, the embodiments of the present application can start from the factors that affect the reliability of the driving of the autonomous driving vehicle cluster, use the data acquired in real time to calculate at least one (i.e., M) deviation rate, input the at least one deviation rate into the neural network model to calculate the probability of a system failure occurring in the vehicles in the autonomous driving vehicle cluster at a certain moment to obtain the system failure rate, and input the system failure rate into the Markov model to predict the state change trend of the vehicles in the autonomous driving vehicle cluster at the next moment to determine the reliability.

[0048] In this way, by adopting the continuous learning, training ability and high prediction ability of the neural network model, it can be ensured that the obtained system failure rate has a higher accuracy, and by adopting the predictive ability of the Markov model without the need for analysis and deduction based on a large amount of historical data, the amount of data required to be collected is small, the amount of calculation is small, and the practicality is strong. Ultimately, the driving process of the autonomous driving vehicle cluster can be determined in real time and dynamically (evaluation / estimation / analysis, etc.), which greatly shortens the reliability evaluation cycle, improves the reliability evaluation efficiency, and meets the current development needs of the automotive industry for reliability evaluation of autonomous driving vehicle clusters.

[0049] The technical solutions, beneficial effects and related concepts involved in the embodiments of the present application are described in detail below.

[0050] 1. Reliability determination system architecture for autonomous vehicle swarm driving

[0051] 1) Self-driving vehicle swarm

[0052] The autonomous driving vehicle cluster of the embodiment of the present application is an important application in the field of autonomous driving. It can effectively improve the efficiency of large-scale traffic operation of autonomous driving and significantly alleviate traffic congestion.

[0053] An autonomous vehicle cluster can be a complex polymorphic system composed of multiple dynamic factors, and can include autonomous single vehicles and autonomous vehicle formations. Among them, an autonomous single vehicle is a single vehicle with autonomous driving functions, while an autonomous vehicle formation is formed by rationally arranging multiple vehicles with autonomous driving functions.

[0054] The individual vehicles in a cluster of autonomous vehicles can be interconnected through vehicle to everything (V2X).

[0055] 2) On-Board Unit (OBU)

[0056] The vehicle-mounted unit of the embodiment of the present application can be installed on each vehicle in an autonomous driving vehicle cluster, and can be a hardware unit for realizing vehicle-to-everything (V2X) communication and supporting V2X applications.

[0057] The on-board unit can obtain basic safety messages (BSM) such as vehicle identification information, driving speed, heading angle, steering wheel angle, four-axis acceleration, etc.

[0058] The on-board unit can communicate with the road side unit (RSU) through vehicle networking technologies such as dedicated short-range communications (DSRC), long term evolution-vehicle (LTE-V), new radio-V2X (NR-V2X), and vehicle to infrastructure (V2I).

[0059] It should be noted that in the embodiment of the present application, the on-board unit may be referred to as an on-board device.

[0060] 3) Roadside Unit

[0061] The roadside unit of the embodiment of the present application is installed in road areas such as intersections, accident-prone areas, narrow and dangerous roads, or highway entrances and exits, and is a hardware unit for realizing V2X communication and supporting V2X applications.

[0062] The roadside unit can perform V2X communication with the vehicle-mounted unit to obtain the BSM from the vehicle-mounted unit.

[0063] It should be noted that the roadside unit in the embodiment of the present application may be referred to as roadside equipment.

[0064] 4) Server

[0065] The server of the embodiment of the present application may be a software or hardware unit for providing functions such as reliability determination (estimation / evaluation / analysis, etc.) of vehicle cluster driving.

[0066] For example, the hardware and software unit may be an infrastructure as a service (IaaS), platform as a service (PaaS), software as a service (SAAS) platform, a database, etc.

[0067] The server can be a cloud platform, a cloud server, a hardware server, a software server, a hardware and software server, an Internet of Things server, a web server, an application server, a load balancer (Nginx), a data center network device, a personal computer (PC), a computing device, a network device in a 5G system, and a network device and a core network device in a future evolved public land mobile network (PLMN), etc., without any specific restrictions.

[0068] The server can perform V2X communication with the vehicle-mounted unit or with the roadside unit.

[0069] For example, the vehicle-to-network (V2N) technology of the on-board unit is connected to the server, so that the server can exchange data with the on-board unit, store and process data, and provide various application services required by the vehicle.

[0070] 5) Reliability determination system architecture for autonomous vehicle swarm driving

[0071] The technical solution of the embodiment of the present application can be applied to the reliability determination system architecture of the self-driving vehicle cluster driving. Among them, the reliability determination system architecture of the self-driving vehicle cluster driving can be composed of a self-driving vehicle cluster, a roadside unit and a server.

[0072] For example, Figure 1As shown, the reliability determination system architecture 10 of the self-driving vehicle cluster driving includes a server 110, a roadside unit 120 and an autonomous driving vehicle cluster 130, and the autonomous driving vehicle cluster 130 includes an autonomous driving single vehicle 1301, an autonomous driving single vehicle 1302, an autonomous driving single vehicle 1303, an autonomous driving vehicle formation 1304 and an autonomous driving vehicle formation 1305. Among them, the autonomous driving vehicle formation 1304 is composed of multiple vehicles in a reasonable formation, and the autonomous driving vehicle formation 1305 is composed of multiple vehicles in a reasonable formation.

[0073] V2X communication is established between the server 110 and the roadside unit 120. V2X communication is established between the server 110 and each vehicle in the autonomous driving vehicle cluster 130. V2X communication is established between each vehicle in the autonomous driving vehicle cluster 130.

[0074] certainly, Figure 1 The reliability determination system architecture 10 for the self-driving vehicle cluster travel shown may also include other numbers of servers, roadside units or self-driving vehicle clusters, without specific limitation.

[0075] 2. First Data

[0076] It should be noted that in the reliability determination system architecture of the autonomous driving vehicle cluster, each vehicle, roadside unit and server in the autonomous driving vehicle cluster needs to continuously interact and generate real-time and dynamic data.

[0077] In order to achieve real-time and dynamic reliability determination (estimation / evaluation / analysis, etc.) of the driving process of the autonomous driving vehicle cluster, the embodiment of the present application needs to obtain a large amount of data generated or required by the autonomous driving vehicle cluster during the driving process, and the obtained data can be collectively referred to as first data. In addition, the first data can also be described by other terms, which are not specifically limited.

[0078] Specifically, the first data may include at least one of the following: Basic Safety Message (BSM), Road Side Information (RSI), Road Side Safety Message (RSM), Signal Phase Timing Message (SPAT), and Map Data Message (MAP).

[0079] Specifically, the first data is time-sensitive. Therefore, the first data at different times may be different. The following specifically describes each piece of information that the first data may include.

[0080] (1) BSM

[0081] BSM is the most widely used application layer message and can be used to exchange safety status data between vehicles in an autonomous vehicle cluster.

[0082] Vehicles in an autonomous vehicle cluster can broadcast BSM to inform other vehicles in the cluster of their real-time status, thereby supporting a range of collaborative safety applications.

[0083] It should be noted that vehicles in an autonomous driving vehicle cluster can broadcast their own BSM to other nearby vehicles, roadside units, servers, etc. through the V2X network.

[0084] When a vehicle broadcasts its own BSM to a roadside unit, the roadside unit can upload the BSM to a server. In other words, the data collection method of the embodiment of the present application can collect the BSM of each vehicle in the autonomous driving vehicle cluster through the roadside unit and upload it to the server.

[0085] The BSM may include message number, vehicle identification information, time accuracy, vehicle driving position, vehicle driving speed, vehicle positioning system accuracy, vehicle current position accuracy, vehicle gear status, heading angle, steering wheel angle, vehicle running status accuracy, vehicle four-axis acceleration, vehicle brake system status, vehicle size, vehicle type, vehicle safety auxiliary information, and emergency vehicle additional information. For example, as shown in Table 1.

[0086] Table 1

[0087] parameter describe Vehicle driving position Position3D_t pos; Vehicle speed Speed_t speed; Heading Angle Heading_t heading; Steering wheel angle SteeringWheelAngle_t angle; Vehicle brake system status BrakeSystemStatus_t brakes; Vehicle four-axis acceleration AccelerationSet4Way_t accelSet; Vehicle safety auxiliary information struct VehicleSafetyExtensions safetyExt; … …

[0088] 1) Vehicle safety auxiliary information

[0089] In an embodiment of the present application, the vehicle safety auxiliary information may include at least one of the vehicle's special status bits (such as the emergency lights are on, the ABS system is triggered, the body stability system is triggered, the tire is blown, the airbag is deployed, the vehicle is faulty and cannot drive, etc.), the vehicle's historical driving trajectory, the vehicle's driving route, the vehicle's lighting information, etc.

[0090] 2) Vehicle identification information

[0091] In the embodiment of the present application, the vehicle identification information can be used to uniquely identify the vehicle, the driving behavior of the vehicle, and the vehicle's own data, etc.

[0092] Exemplarily, as shown in Table 2, the vehicle identification information includes at least one of a vehicle identifier, driving behavior information, position information in an autonomous driving vehicle formation, and a VID of a head vehicle in an autonomous driving vehicle formation.

[0093] The driving behavior information can be used to indicate whether the vehicle is a single autonomous vehicle or a platoon of autonomous vehicles. If the driving behavior information indicates 0, it means that the vehicle is a single autonomous vehicle. If the driving behavior information indicates 0, it means that the vehicle is a vehicle in a platoon of autonomous vehicles.

[0094] The position information in the autonomous driving vehicle formation can be used to indicate the position of the vehicle in the autonomous driving vehicle formation. If the position information in the autonomous driving vehicle formation is 0, it means that the vehicle is the head vehicle in the autonomous driving vehicle formation; if the position information in the autonomous driving vehicle formation is 1, it means that the vehicle is the second vehicle in the autonomous driving vehicle formation, and so on.

[0095] The VID of the head vehicle in the autonomous driving vehicle formation can be used to indicate the unique identification number of the head vehicle in the autonomous driving vehicle formation.

[0096] Table 2

[0097]

[0098] (2)RSI

[0099] RSI can be traffic event information and traffic sign information released by the roadside unit to vehicles in the autonomous driving vehicle cluster around itself.

[0100] RSI may include one or more traffic event information or traffic sign information, as well as the number of the roadside unit that sends the roadside information and the reference position coordinates, etc.

[0101] It should be noted that the server can obtain the RSI from the roadside unit, or can be configured with the RSI itself, and there is no specific limitation on this.

[0102] (3)RSM

[0103] RSM can be a roadside unit that obtains real-time status information of surrounding traffic participants (such as the roadside unit itself, surrounding vehicles, non-motor vehicles, pedestrians, etc.) through its own corresponding detection means, and broadcasts it to vehicles in the autonomous driving vehicle cluster around itself.

[0104] It should be noted that the server can obtain the RSM from the roadside unit, or it can be configured with the RSM itself, and there is no specific limitation on this.

[0105] (4)SPAT

[0106] SPAT can contain the current status information of one or more intersection information lights. Combining SPAT with MAP can provide real-time forward signal light phase information to vehicles in the autonomous vehicle cluster.

[0107] It should be noted that the server can obtain the SPAT from the roadside unit, or it can be configured with the SPAT itself, and there is no specific limitation on this.

[0108] For example, as shown in Table 3, SPAT includes state information, wherein the state information may include intersection location information, intersection traffic light phase information, and the like.

[0109] Table 3

[0110] parameter describe Status Information IntersectionStateList_t intersections;

[0111] (5)MAP

[0112] MAP can be broadcast by the roadside unit to deliver map information of the local area to vehicles in the autonomous driving vehicle cluster around itself.

[0113] It should be noted that the server may obtain the MAP from the roadside unit, or may be configured with the MAP itself, and there is no specific limitation on this.

[0114] MAP may include intersection information, road section information, lane information, connection relationships between roads, etc. in a local area.

[0115] In addition, the MAP may include map data for multiple intersections or areas.

[0116] For example, as shown in Table 4, MAP may include road information, wherein the road information includes road node locations, lane information, road speed limit information, and the like.

[0117] Table 4

[0118] parameter describe Road Information NodeList_t nodes;

[0119] (6) Application of First Data

[0120] The embodiments of the present application can use the first data to perform real-time and dynamic reliability determination (estimation / evaluation / analysis, etc.) of the autonomous driving vehicle cluster during driving.

[0121] For example, Figure 2 As shown, the embodiment of the present application can input the first data into the reliability determination system of the autonomous driving vehicle cluster driving.

[0122] The reliability determination system of the autonomous driving vehicle cluster driving can be used to evaluate the reliability of the autonomous driving vehicle cluster during driving.

[0123] For example, a reliability determination system for autonomous driving vehicle cluster driving can be implemented as follows: use the first data to calculate at least one (i.e., M, M is a positive integer) deviation rate, input the at least one deviation rate into a neural network model to calculate the probability of a system failure occurring in a vehicle in the autonomous driving vehicle cluster at a certain moment to obtain a system failure rate, and input the system failure rate into a Markov model to predict the state change trend of the vehicles in the autonomous driving vehicle cluster at the next moment to determine the reliability.

[0124] The reliability determination system for autonomous driving vehicle cluster driving may include a reliability determination subsystem for autonomous driving single vehicle and a reliability determination subsystem for autonomous driving vehicle platoon driving.

[0125] Among them, the reliability determination subsystem of the autonomous driving vehicle can be used to evaluate the reliability of the autonomous driving vehicle during driving.

[0126] The reliability determination subsystem of the autonomous driving vehicle platoon can be used to evaluate the reliability of the autonomous driving vehicle platoon during driving.

[0127] The result output unit can be used to visualize the results of the reliability determination system of the autonomous driving vehicle cluster driving.

[0128] For example, the result output unit can output data reports, or dynamically display the output results in real time on a monitoring screen of the server.

[0129] 3. Deviation rate

[0130] It should be noted that the embodiments of the present application can start from the factors that affect the reliability of the driving of the autonomous driving vehicle cluster, and use the first data acquired in real time to calculate at least one (i.e., M (M≥1)) deviation rate at different times. The deviation rate can be used to represent the probability of vehicles in the autonomous driving vehicle cluster deviating from the driving path.

[0131] That is, the deviation rate can be a dynamic factor that affects the reliability of the driving of the autonomous driving vehicle cluster, and the autonomous driving vehicle cluster can be a complex polymorphic system composed of multiple dynamic factors.

[0132] At the same time, all vehicles in the autonomous driving vehicle cluster have M corresponding deviation rates at a certain moment.

[0133] For example, when the first data is included in the time period [t i ,t j ], the embodiment of the present application can use the data to calculate the vehicle k in the autonomous driving vehicle cluster at time t j There are M deviation rates corresponding to vehicle k under .

[0134] In some possible implementations, the M deviation rates may include at least one of a speed deviation rate, a steering angle deviation rate, a lateral position offset rate, a first collision time deviation rate, a second collision time deviation rate, and the like.

[0135] (1) Speed ​​deviation rate

[0136] In an embodiment of the present application, the speed deviation rate can be used to indicate the probability of a speed deviation occurring between the driving speed of a vehicle in a cluster of autonomous driving vehicles and the speed limit of the driving path.

[0137] Specifically, the speed deviation rate is time-dependent, so the speed deviation rate at different times may be different.

[0138] It should be noted that the driving route can be planned and determined by the test items of the autonomous driving vehicle cluster, can be in the navigation system sent to the vehicles in the autonomous driving vehicle cluster through the server, or can be in the navigation system pre-configured in the vehicles in the autonomous driving vehicle cluster, and the test items may include vehicle driving distance, vehicle driving speed, driving speed limit, target vehicle driving speed, headway between adjacent vehicles, road condition recognition, direction indicator light recognition, stop and yield sign recognition, road traffic sign recognition, etc.

[0139] In addition, there may be multiple speed limits on a driving route. Therefore, when a vehicle in a cluster of autonomous vehicles is driving on a certain driving route, the current speed of the vehicle may be greater than, equal to, or less than the speed limit on the driving route. Figure 3 As shown, the speed limit of the vehicle's current route is 60km / h.

[0140] Since the embodiment of the present application can obtain the current driving speed of each vehicle in the autonomous driving vehicle cluster through the real-time reported BSM, it is possible to determine whether the current driving speed has a speed deviation from the driving limit through these driving speeds and the driving speed limit in the driving path, so as to evaluate the reliability of the vehicles in the autonomous driving vehicle cluster during driving.

[0141] For example, if vehicle k is at time t j The speed limit of the route below is v l (t), and vehicle k is at time t j The driving speed is v h (t), then the speed deviation rate for:

[0142]

[0143] (2) Steering angle deviation rate

[0144] In an embodiment of the present application, the steering angle deviation rate can be used to indicate the probability of a steering angle deviation occurring between the driving direction and the driving path of a vehicle in a cluster of autonomous driving vehicles.

[0145] In addition, the steering angle deviation rate has a time characteristic, so the steering angle deviation rate at different times may be different.

[0146] It should be noted that when a vehicle in an autonomous driving vehicle cluster is traveling along a driving path, there may be a certain steering angle deviation between the driving direction of the vehicle and the driving path.

[0147] Since the embodiments of the present application can obtain the current heading angle, steering wheel angle, etc. of each vehicle in the autonomous driving vehicle cluster through the real-time reported BSM in order to determine the driving direction of each vehicle, it is possible to determine whether the autonomous driving vehicle cluster has a steering angle deviation from the driving path through these driving directions and driving paths, so as to evaluate the reliability of the vehicles in the autonomous driving vehicle cluster during driving.

[0148] For example, Figure 4 As shown, if vehicle k is at time t j The driving direction is represented by the dashed line direction 410, the driving route is represented by the dashed line 420, and the angle between the dashed line direction 410 and the dashed line 420 is a(t), then the steering angle deviation rate for

[0149]

[0150] Here, π represents radians.

[0151] (3) Lateral position deviation rate

[0152] In an embodiment of the present application, the lateral position offset rate can be used to indicate the probability of a lateral position deviation occurring between the body center position and the driving path of a vehicle in a cluster of autonomous driving vehicles.

[0153] In addition, the lateral position deviation rate has time characteristics, so the lateral position deviation rate at different times may be different.

[0154] It should be noted that when a vehicle in an autonomous driving vehicle cluster is traveling along a driving path, there may be a certain lateral position deviation between the center position of the vehicle body and the driving path.

[0155] Since the embodiments of the present application can obtain the current vehicle driving position, vehicle size, etc. of each vehicle in the autonomous driving vehicle cluster through the real-time reported BSM in order to determine the body center position of each vehicle, it is possible to determine whether the autonomous driving vehicle cluster has a lateral position deviation from the driving path through these body center positions and driving paths, so as to evaluate the reliability of the vehicles in the autonomous driving vehicle cluster during driving.

[0156] For example, Figure 5 As shown, if at time t j The driving route under is represented by the dashed line 510. The vehicle k at time t j The vertical distance between the center position of the vehicle body under and the dotted line 510 is d(t), and the vehicle k at time t j The lane width of the lane is w(t), then the lateral position deviation rate for

[0157]

[0158] (4) First collision time deviation rate

[0159] In an embodiment of the present application, the first collision time deviation rate can be used to indicate the probability of a collision time deviation between the forward collision warning (FCW) collision time (TTC) of the front and rear vehicles in the autonomous driving vehicle cluster and the first preset forward collision warning collision time.

[0160] In addition, the first collision time deviation rate has time characteristics, so the first collision time deviation rate at different times may be different.

[0161] It should be noted that when the vehicles in the autonomous driving vehicle cluster are traveling along the driving path, there may be two vehicles in front and behind on the same driving path. Figure 6 As shown, vehicle 610 and vehicle 620 are on the same driving path 630 .

[0162] To this end, it is necessary to calculate the forward collision warning collision time of the two vehicles in front and behind. The forward collision warning collision time T 1 (t) is:

[0163]

[0164] Among them, D 1 (t) represents the distance between the two vehicles in front and behind; v r (t) represents the relative speed between the two vehicles in front and behind.

[0165] Since the embodiment of the present application can obtain the current vehicle driving position, driving speed, etc. of each vehicle in the autonomous driving vehicle cluster through the real-time reported BSM, so as to determine the vehicle-to-vehicle distance between the front and rear vehicles on the same driving path, the forward collision warning collision time can be determined through these driving speeds and vehicle-to-vehicle distances.

[0166] The first preset forward collision warning collision time may be a preset value specified by the standard, the preset value may be the latest forward collision warning collision time corresponding to the current vehicle speed, or the earliest forward collision warning collision time corresponding to the current vehicle speed.

[0167] Table 5

[0168]

[0169] For example, as shown in Table 5, if the current vehicle's speed is 40 km / h, the vehicle in front of the current vehicle's speed is 20 km / h, and the overlap rate of the two vehicles when they collide is 50%, then the latest forward collision warning vehicle distance is 10.51 m, the earliest forward collision warning vehicle distance is 17.29 m, the latest forward collision warning collision time is 1.89 s, and the earliest forward collision warning collision time is 3.11 s; the same logic can be applied.

[0170] In addition, the first preset forward collision warning collision time can be broadcasted by the roadside unit to the vehicles or servers in the autonomous driving vehicle cluster in real time. In this regard, the vehicles or servers in the autonomous driving vehicle cluster can learn the first preset forward collision warning collision time. For example, the first preset forward collision warning collision time can be carried by the RSI or RSM.

[0171] If the first preset forward collision warning collision time is T C (t), and the forward collision warning collision time of the front and rear vehicles in the autonomous driving vehicle cluster is T 1 (t), then vehicle k at time t j The first collision time deviation rate E under j k , rT1 (t) is:

[0172]

[0173] (5) Second collision time deviation rate

[0174] In an embodiment of the present application, the second collision time deviation rate can be used to indicate the probability of a collision time deviation between the forward collision warning collision time from a vehicle in the autonomous driving vehicle cluster to the collision warning target and the second preset forward collision warning collision time.

[0175] In addition, the second collision time deviation rate has time characteristics, so the second collision time deviation rate at different times may be different.

[0176] In some possible implementations, the collision warning target may be a forward stop line at a traffic light intersection in the driving path, or may be any collision warning target object in the driving path, etc.

[0177] In addition, the location information of the collision warning target can be broadcasted by the roadside unit to the vehicles or servers in the autonomous driving vehicle cluster in real time. In this regard, the vehicles or servers in the autonomous driving vehicle cluster can know the distance between the vehicle and the collision warning target. For example, the location information of the collision warning target can be carried by RSI or RSM.

[0178] Similarly, the location information of the collision warning target can be reported to the server in real time by the roadside unit. In this regard, the server can determine the distance between the vehicle in the autonomous driving vehicle cluster and the collision warning target based on the location information of the collision warning target and the BSM reported in real time.

[0179] For example, the collision warning target is the forward stop line of the traffic light intersection in the driving path. Figure 7 As shown, line segment 720 represents the forward stop line at the traffic light intersection. Roadside unit 730 can broadcast the position information of the forward stop line at the traffic light intersection to vehicle 710 in real time.

[0180] To this end, it is necessary to calculate the forward collision warning collision time from the vehicle to the collision warning target. The forward collision warning collision time T 2 (t) is:

[0181]

[0182] Among them, D 2 (t) represents the distance between the vehicle and the collision warning target; v(t) represents the vehicle's speed.

[0183] Since the embodiment of the present application can obtain the current vehicle driving position, driving speed, etc. of each vehicle in the autonomous driving vehicle cluster through the real-time reported BSM, and obtain the position information of the collision warning target through RSI or RSM, the forward collision warning collision time can be determined through these driving speeds, vehicle driving positions and the position information of the collision warning target.

[0184] The second preset forward collision warning collision time may be a preset value specified by the standard. The preset value may be the latest forward collision warning collision time corresponding to the current vehicle speed, or the earliest forward collision warning collision time corresponding to the current vehicle speed.

[0185] For example, as shown in Table 6, if the current vehicle speed is 20 km / h, the latest forward collision warning distance is 10.51 m, the earliest forward collision warning vehicle distance is 17.29 m, the latest forward collision warning collision time is 1.89 s, and the earliest forward collision warning collision time is 3.11 s; the same logic can be applied.

[0186] Table 6

[0187]

[0188] In addition, the second preset forward collision warning collision time can be broadcasted by the roadside unit to the vehicles or servers in the autonomous driving vehicle cluster in real time. In this regard, the vehicles or servers in the autonomous driving vehicle cluster can learn the second preset forward collision warning collision time. For example, the second preset forward collision warning collision time can be carried by the RSI or RSM.

[0189] If the second preset forward collision warning collision time is T S (t), and the forward collision warning collision time from the vehicle in the autonomous driving vehicle cluster to the collision warning target is T 2 (t), then vehicle k at time t j The second collision time deviation rate under for:

[0190]

[0191] 4. Using neural network model to calculate system failure rate

[0192] Combined with the content in "3. Deviation rate" above, it can be known that all vehicles in the autonomous driving vehicle cluster each have M deviation rates at a certain moment. Since there are a large number of vehicles in the autonomous driving vehicle cluster (such as P vehicles), the embodiment of the present application can calculate a large number of deviation rates (such as P*M deviation rates), and the deviation rates corresponding to different vehicles may be different, so that when determining whether a system failure occurs in the autonomous driving vehicle cluster based on the deviation rate, there are situations such as large processing volume and high processing complexity.

[0193] In order to improve processing efficiency and accuracy, the embodiments of the present application can input these deviation rates into the neural network model to output the probability of a system failure occurring in a vehicle in the autonomous driving vehicle cluster at a certain moment, that is, the system failure rate, so as to improve the accuracy of the system failure rate by adopting the continuous learning and training ability of the neural network model and the higher predictive ability, and input the system failure rate into the Markov model to predict the state change trend of the vehicles in the autonomous driving vehicle cluster at the next moment to determine the reliability.

[0194] (1) Neural network model

[0195] In the embodiment of the present application, the neural network model can be various types of network models, for example, one of a feedforward neural network (FNN) model, a feedback neural network (FNN) model, etc.

[0196] The feedforward neural network model may include a back propagation (BP) neural network model, a convolutional neural network (CNN) model, a deep convolutional neural network (DCNN) model, a residual network (ResNet) model, a generative adversarial network (GAN) model, a fully convolutional neural network (FCN) model, and the like.

[0197] The feedback neural network model may include one of a recurrent neural network (RNN) model, a long short term memory network (LSTM) model, a Hopfield network model, a Boltzmann machine, and the like.

[0198] (2) System failure rate

[0199] In an embodiment of the present application, the system failure rate can be used to indicate the probability of a system failure occurring in a vehicle in a cluster of autonomous driving vehicles at a certain moment.

[0200] Specifically, the system failure rate presents continuous values ​​in the time dimension. In other words, the values ​​of the system failure rate at different times can form a continuous curve.

[0201] (3) Confidence value of system failure rate

[0202] It should be noted that, since the neural network model needs to continuously use certain expected values ​​for learning and training in order to ensure the accuracy of the output, the present application introduces a confidence value for the system failure rate.

[0203] The confidence value of the system failure rate can be used to represent the expected value required for learning and training of the neural network model.

[0204] In addition, different deviation rates among the M deviation rates each correspond to a confidence value of the system failure rate.

[0205] For example, the speed deviation rate Corresponding to a confidence value of system failure rate, steering angle deviation rate Corresponding to a confidence value of a system failure rate, etc.

[0206] (4) Impact range

[0207] In order to determine the confidence value of the system failure rate, the embodiment of the present application introduces an impact interval.

[0208] The influence degree interval can be used to represent the threshold interval of the influence degree of the deviation rate on the system failure rate. The influence degree interval is composed of the influence degree threshold, and the influence degree threshold is used to represent the influence degree of the deviation rate on the confidence value of the system failure rate.

[0209] For example, the deviation rate is the speed deviation rate For example, if the speed deviation rate The threshold interval of the impact on the system failure rate is The speed deviation rate The corresponding influence interval is in, and are two influence thresholds.

[0210] In some possible implementations, the influence interval may include at least one of the following:

[0211] ◆ Each of the M deviation rates corresponds to L (L is a positive integer) states in the Markov state, and different states in the L states may correspond to different influence intervals; wherein the Markov state may be the state of the vehicles in the autonomous driving vehicle cluster modeled by the Markov model at different times, which will be described in detail later;

[0212] ◆The influence intervals corresponding to different deviation rates among the M deviation rates may be different;

[0213] ◆The influence threshold in the influence interval can be determined through big data statistics;

[0214] ◆The larger the impact threshold in the impact interval, the greater the impact of the deviation rate on the system failure rate;

[0215] The influence threshold in the influence interval does not change over time, that is, the influence threshold is a fixed value;

[0216] ◆The confidence value of the system failure rate can be a random value in the impact interval.

[0217] For example, the Markov state includes "normal state", "general state", "abnormal state" and "fault state", that is, L = 4. Among them, vehicle k in the autonomous driving vehicle cluster is at time t j Speed ​​deviation rate under The corresponding influence interval of "normal state" is Speed ​​deviation rate The corresponding influence interval of the "general state" is Speed ​​deviation rate The corresponding influence range of "abnormal state" is Speed ​​deviation rate The corresponding impact interval of the “fault state” is in, is the impact threshold, used to indicate the speed deviation rate The corresponding Markov state has an influence on the confidence value of the system failure rate. The larger it is, the greater the impact.

[0218] Similarly, vehicle k in the autonomous vehicle cluster at time t j Steering angle deviation rate under The corresponding normal state influence interval is and Different, the rest can be understood by the same logic.

[0219] (5) Sensitivity threshold corresponding to the impact interval

[0220] In order to determine the confidence value of the system failure rate, the embodiment of the present application introduces a sensitivity threshold corresponding to the impact interval.

[0221] The sensitivity threshold can be used to indicate the sensitivity of the deviation rate to the system failure rate.

[0222] It should be noted that, if the sensitivity threshold is larger, the deviation rate has a greater sensitivity to the system failure rate, which means that even a small change in the deviation rate will seriously affect the system failure rate.

[0223] In some possible implementations, the sensitivity threshold may include at least one of the following:

[0224] ◆The sensitivity thresholds corresponding to different impact intervals may be different;

[0225] ◆Different deviation rates among the M deviation rates may have different corresponding sensitivity thresholds;

[0226] ◆The sensitivity threshold can be determined through big data statistics;

[0227] ◆The larger the sensitivity threshold, the greater the sensitivity of the deviation rate to the system failure rate, that is, a small change in the deviation rate will seriously affect the system failure rate;

[0228] ◆The sensitivity threshold does not change over time, that is, the sensitivity threshold is a fixed value.

[0229] For example, if vehicle k in the autonomous vehicle cluster is at time t j Speed ​​deviation rate under The corresponding influence interval of "normal state" is Speed ​​deviation rate The corresponding influence interval of the "general state" is Speed ​​deviation rate The corresponding influence range of "abnormal state" is Speed ​​deviation rate The corresponding impact interval of the “fault state” is The above four influence areas correspond to three sensitivity thresholds, namely in,

[0230] (6) Determine the confidence value of the system failure rate corresponding to the deviation rate

[0231] It should be noted that, in the embodiment of the present application, the confidence value of the system failure rate corresponding to each of the M deviation rates can be determined based on the M deviation rates, the impact interval and the sensitivity threshold.

[0232] For the sake of distinction, the confidence value of the system failure rate corresponding to the deviation rate can be called the first confidence value, that is, the first confidence value is used to represent the confidence value of the system failure rate. Currently, the first confidence value can also be described by other terms, which are not specifically limited.

[0233] For example, if vehicle k in the autonomous vehicle cluster is at time t j Speed ​​deviation rate under The corresponding normal state influence interval is Speed ​​deviation rate The corresponding general state influence interval is Speed ​​deviation rate The corresponding abnormal state influence interval is Speed ​​deviation rate The corresponding fault state influence interval is The above four influence intervals correspond to three sensitivity thresholds, namely The speed deviation rate The corresponding confidence value of the system failure rate is

[0234]

[0235] in, Indicates the influence interval The random value of , and the rest can be known in the same way.

[0236] (7) Determine the confidence value of the system failure rate corresponding to the vehicle

[0237] It should be noted that, in the embodiment of the present application, the confidence value of the system failure rate corresponding to the vehicle can be determined based on the confidence values ​​of the system failure rates corresponding to the M deviation rates of the vehicle.

[0238] In addition, since the confidence value of the system failure rate corresponding to the vehicle is a value at a certain moment, it presents a discrete state in the time dimension.

[0239] For the sake of distinction, the confidence value of the system failure rate corresponding to the vehicle may be referred to as a second confidence value. Currently, the second confidence value may also be described using other terms, which are not specifically limited.

[0240] In some possible implementations, the confidence value of the system failure rate corresponding to the vehicle may be the maximum value among the confidence values ​​of the system failure rates corresponding to the M deviation rates of the vehicle. That is, the second confidence value is the maximum value among the confidence values ​​of the system failure rates corresponding to the M deviation rates.

[0241] For example, when vehicle k in the autonomous vehicle cluster is at time t j The M (M=5) deviation rates under this condition include the speed deviation rate Steering angle deviation rate Lateral position deviation rate First collision time deviation rate Second collision time deviation rate If the speed deviation rate The corresponding confidence value of the system failure rate is Steering angle deviation rate The corresponding confidence value of the system failure rate is Lateral position deviation rate The corresponding confidence value of the system failure rate is First collision time deviation rate The corresponding confidence value of the system failure rate is Second collision time deviation rate The corresponding confidence value of the system failure rate is Then vehicle k at time t j The confidence value of the system failure rate corresponding to

[0242]

[0243] (8) Output of the neural network model

[0244] It should be noted that, in combination with the above, the embodiment of the present application can input M deviation rates into the neural network model, and use the confidence value of the system failure rate corresponding to the vehicle as the expected value for learning and training to output the system failure rate.

[0245] The confidence value of the system failure rate corresponding to the vehicle is discrete in the time dimension, while the system failure rate is continuous in the time dimension. Therefore, the embodiment of the present application can fit the confidence value of the system failure rate that is discrete in the time dimension into a continuous system failure rate curve through a neural network model, which meets the time continuity in the real scene and ensures that the system failure rate is more real, reliable and accurate.

[0246] For example, if vehicle k is at time t j The M (M = 5) deviation rates under the above conditions, namely the speed deviation rate Steering angle deviation rate Lateral position deviation rate First collision time deviation rate Second collision time deviation rate Input the neural network model and take vehicle k at time t j The confidence value of the system failure rate corresponding to Learn and train as expected value to output system failure rate

[0247] (9) Exemplary Description

[0248] In summary, the following is an exemplary description of the process of calculating the system failure rate using a neural network model.

[0249] Step 1: Obtain N influence intervals and the sensitivity thresholds corresponding to the influence intervals.

[0250] Among them, N is the product of M and L, M is the number of deviation rates corresponding to the vehicles in the autonomous driving vehicle cluster at the first moment, and L is the number of states in the Markov state corresponding to the deviation rate.

[0251] For example, vehicle k in the autonomous vehicle cluster is at time t j The following corresponds to M (M = 5) deviation rates, namely, the speed deviation rate Steering angle deviation rate Lateral position deviation rate First collision time deviation rate Second collision time deviation rate

[0252] Each deviation rate corresponds to L (L=4) states in the Markov state, namely, a normal state, a general state, an abnormal state, and a fault state.

[0253] In this regard, the embodiment of the present application can obtain 20 influence intervals. Among them, the speed deviation rate The corresponding normal state influence interval is Steering angle deviation rate The corresponding normal state influence interval is The rest can be understood in the same way.

[0254] In addition, the speed deviation rate The sensitivity thresholds corresponding to the influence intervals of the corresponding four states are The rest can be understood in the same way.

[0255] Step 2: Determine the confidence value of the system failure rate corresponding to each of the M deviation rates according to the M deviation rates, the N impact intervals and the sensitivity thresholds of the impact intervals.

[0256] It should be noted that the embodiment of the present application can determine one influence interval among N influence intervals (for the sake of ease of distinction, the influence interval is called the first influence interval) based on the size relationship between the deviation rate and the sensitivity threshold, and use the random value in the influence interval as the confidence value of the system failure rate corresponding to the deviation rate.

[0257] For example, combined with the above description, vehicle k in the autonomous driving vehicle cluster is at time t j Speed ​​deviation rate under The corresponding confidence value of the system failure rate is

[0258]

[0259] The confidence values ​​of the system failure rates corresponding to the other deviation rates can be known in the same way.

[0260] Step 3: Input the M deviation rates into the neural network model, and use the confidence values ​​of the system failure rates corresponding to the M deviation rates as expected values ​​for learning and training to output the system failure rates.

[0261] It should be noted that since the neural network model needs to continuously use certain expected values ​​for learning and training in order to ensure the accuracy of the output, the embodiment of the present application uses the confidence values ​​of the system failure rate corresponding to each of the M deviation rates as the expected values ​​for learning and training, so that the confidence values ​​of the system failure rate discrete in the time dimension are fitted into a continuous system failure rate curve through the neural network model, which meets the time continuity in the real scene and ensures that the system failure rate is more real, reliable and accurate.

[0262] In some possible implementations, the embodiments of the present application may also use the confidence value of the system failure rate corresponding to the vehicle as the expected value for learning and training, so as to reduce the scale of learning and training.

[0263] For example, an embodiment of the present application can determine the maximum value of the confidence values ​​of the system failure rate corresponding to each of the M deviation rates to obtain the confidence value of the system failure rate corresponding to the vehicle, and input the M deviation rates into a neural network model, and use the confidence value of the system failure rate corresponding to the vehicle as the expected value for learning and training to output the system failure rate.

[0264] 5. Using the Markov model to predict the reliability of autonomous vehicle cluster driving

[0265] It should be noted that the Markov model of the embodiment of the present application can model the transition of the driving of a cluster of autonomous driving vehicles from one state to another through a state transition probability matrix. It does not need to be analyzed and derived based on a large amount of data. The amount of data required to be collected is small, the amount of calculation is small, and the practicality is strong. It greatly shortens the reliability evaluation cycle and improves the reliability evaluation efficiency, meeting the current development needs of the automotive industry for reliability determination (estimation / evaluation / analysis, etc.) of autonomous driving vehicle cluster driving.

[0266] In an embodiment of the present application, the driving process of an autonomous driving vehicle cluster can be regarded as a continuously changing process in the time dimension. Without affecting the significance of reliability assessment, in order to reduce the computational complexity, shorten the reliability assessment cycle, and improve the reliability assessment efficiency, the embodiment of the present application can model the continuously changing process as a discrete state in the time dimension, namely, the Markov state, through a Markov model.

[0267] (1) Markov state

[0268] In an embodiment of the present application, the Markov state may be the state of vehicles in a cluster of autonomous driving vehicles modeled by a Markov model at different times.

[0269] In some possible implementations, the Markov state may include at least one of the following:

[0270] Initial state (for the sake of ease of distinction, it may also be called the first state or other terms), normal state (for the sake of ease of distinction, it may also be called the second state or other terms), general state (for the sake of ease of distinction, it may also be called the third state or other terms), abnormal state (for the sake of ease of distinction, it may also be called the fourth state or other terms), and fault state (for the sake of ease of distinction, it may also be called the fifth state or other terms).

[0271] Among them, the initial state can be used to represent the state in which the vehicles in the autonomous driving vehicle cluster are not driving, which is modeled using the Markov model.

[0272] The normal state can be used to represent the state in which the vehicles in the autonomous driving vehicle cluster are modeled in normal driving using the Markov model;

[0273] The general state can be used to represent the state in which the vehicles in the autonomous driving vehicle cluster are in general driving mode using the Markov model;

[0274] The abnormal state can be used to represent the state in which the vehicles in the autonomous driving vehicle cluster are in abnormal driving mode modeled by the Markov model;

[0275] The fault state can be used to represent the state in which the vehicle in the autonomous driving vehicle cluster is in fault driving modeled by using the Markov model.

[0276] It should be noted that the initial state will be irreversibly transferred to other states except the initial state, while other states can be transferred to or maintained between each other.

[0277] For example, Figure 8 As shown, the initial state will be irreversibly transferred to the normal state, the general state, the abnormal state or the fault state, while the normal state, the general state, the abnormal state and the fault state can be transferred or maintained to each other.

[0278] In addition, whether the vehicles in the autonomous driving vehicle cluster are in normal driving, general driving, abnormal driving or faulty driving can be determined based on the system failure rate in the above-mentioned "using the neural network model to calculate the system failure rate".

[0279] For example, combined with the above example, vehicle k at time t j The system failure rate output by the neural network model is At this point, the following exists:

[0280] Normal driving: That is, vehicle k in the autonomous driving vehicle cluster modeled by the Markov model is at time t j The driving status is "normal status".

[0281] General driving: That is, vehicle k in the autonomous driving vehicle cluster modeled by the Markov model is at time t j The state of the next trip is "normal state".

[0282] Abnormal driving: That is, vehicle k in the autonomous driving vehicle cluster modeled by the Markov model is at time tj The driving status is "abnormal status".

[0283] Fault driving: That is, vehicle k in the autonomous driving vehicle cluster modeled by the Markov model is at time t j The state of driving down is "fault state".

[0284] (2) State transition probability matrix

[0285] The state transition probability matrix, which can be called the state migration probability matrix, can be used to represent the probability of a cluster of autonomous driving vehicles transitioning from one state to another.

[0286] It should be noted that after a period of sample data learning and without sample data input, the state transition probability matrix will tend to a stable state, that is, the state transition probability matrix will no longer change with time. Only when the statistical probability results of new sample data change significantly, the stable state will be broken, that is, the state transition probability matrix will change again with time.

[0287] The following is a specific description using Markov states including an initial state, a normal state, a general state, an abnormal state, and a fault state as examples.

[0288] (3) State transition probability matrix in the initial state

[0289] It should be noted that the transition probabilities among the normal state, general state, abnormal state, and fault state in the initial state are the same.

[0290] For example, the state transition probability matrix in the initial state is:

[0291]

[0292] Among them, the transition probability from the normal state to the normal state is 0.25, the transition probability from the normal state to the general state is 0.25, the transition probability from the normal state to the abnormal state is 0.25, and the transition probability from the normal state to the fault state is 0.25. The rest can be understood in the same way.

[0293] (4) State probability corresponding to the initial state

[0294] It should be noted that, in the initial state, the probability of the normal state, general state, abnormal state, and fault state occurring is the same, that is, the values ​​of the state probabilities corresponding to the initial state are the same.

[0295] For example, the state probability corresponding to the initial state is:

[0296] P0 =[0.25 0.25 0.25 0.25].

[0297] (5) Update of state transition probability matrix

[0298] It should be noted that, since the driving conditions of vehicles in an autonomous driving vehicle cluster may change at different times, the embodiment of the present application needs to update the state transition probability matrix through data statistics and state changes.

[0299] For example, in the time period [t i ,t j ], vehicle k at time t i The state of driving under is the normal state. If vehicle k is at time t j The system failure rate output by the neural network model is and Then vehicle k at time t j The state of driving under the condition of t will change from the normal state to the general state. i The state of driving under and at time t j The state of the next trip is used to update the state transition probability matrix.

[0300] For another example, if a state transition process at different times is: "initial state" → "normal state" → "normal state" → "general state" → "general state" → "abnormal state", then the update process of the state transition probability matrix is ​​as follows: Fig. 9 shown.

[0301] exist Fig. 9 In the initial state, the state transition probability matrix is:

[0302]

[0303] Among them, the transition probabilities between the normal state, general state, abnormal state, and fault state are the same, all 0.25.

[0304] When "normal state" → "normal state", that is, the normal state remains unchanged, the state transition probability matrix needs to be updated, and the updated state transition probability matrix is:

[0305]

[0306] According to the statistical results, the probability of transition from the normal state to the normal state is 1. Of course, other values ​​may also be used, and there is no specific limitation on this.

[0307] When "normal state" → "general state", that is, changing from normal state to general state, the state transition probability matrix needs to be updated, and the updated state transition probability matrix is:

[0308]

[0309] According to the statistical results, the transition probability from the normal state to the general state is 0.5, and the transition probability from the normal state to the normal state is 0.5. Of course, other values ​​may be used, and there is no specific limitation on this.

[0310] When "general state" → "general state", that is, the general state remains unchanged, the state transition probability matrix needs to be updated, and the updated state transition probability matrix is:

[0311]

[0312] According to the statistical results, the probability of transition from the general state to the general state is 1. Of course, other values ​​may also be used, and there is no specific limitation on this.

[0313] When "normal state" → "abnormal state", that is, changing from normal state to abnormal state, the state transition probability matrix needs to be updated, and the updated state transition probability matrix is:

[0314]

[0315] According to the statistical results, the transition probability from the normal state to the abnormal state is 0.5, and the transition probability from the normal state to the normal state is 0.5. Of course, other values ​​may also be used, and there is no specific limitation on this.

[0316] (6) Predict the state probability of the vehicles in the autonomous driving vehicle cluster at the next moment

[0317] It should be noted that the embodiments of the present application can predict the state probability of driving at the next moment based on the state probability corresponding to the driving state of the vehicles in the autonomous driving vehicle cluster at the current moment and the updated state transition probability matrix.

[0318] For example, vehicle k in the autonomous vehicle cluster is at time t j The state of driving down is the normal state, and the state probability corresponding to the normal state is And the updated state transition probability matrix is ​​PM j Therefore, according to the state probability And the state transition probability matrix PM j To predict the vehicle k at the next time t j+1The state probability corresponding to the state of driving down

[0319]

[0320] Among them, the state probability The value a in represents the next moment t j+1 The probability that the driving state is normal; state probability The value b in represents the next moment t j+1 The probability of the state of the next driving being the general state; state probability The value c in represents the next moment t j+1 The probability that the driving state is abnormal; state probability The value d in represents the next moment t j+1 The probability of the next driving state being a fault state occurring.

[0321] (7) Predicting the reliability of the vehicles in the autonomous vehicle cluster at the next moment

[0322] It should be noted that the embodiments of the present application can determine the reliability of the vehicles in the autonomous driving vehicle cluster traveling at the next moment based on the above-mentioned predicted state probability of the vehicles in the autonomous driving vehicle cluster traveling at the next moment.

[0323] In the embodiment of the present application, when the vehicles in the autonomous driving vehicle cluster are in normal driving (i.e., normal state), or when the vehicles in the autonomous driving vehicle cluster are in normal driving (i.e., normal state) and general driving (i.e., general state), the driving of the vehicles in the autonomous driving vehicle cluster can be considered reliable. Therefore, the reliability of the driving of the vehicles in the autonomous driving vehicle cluster at the next moment can be as follows:

[0324] ◆The reliability of the vehicles in the autonomous driving vehicle cluster driving at the next moment is the probability of the normal state occurring in the state probability of the vehicles in the autonomous driving vehicle cluster driving at the next moment predicted above;

[0325] For example, if the vehicle k is predicted to be at the next time t j+1 The state probability corresponding to the state of driving down Then vehicle k at the next time t j+1 The reliability of driving is

[0326] ◆The reliability of the vehicles in the autonomous driving vehicle cluster traveling at the next moment is the sum of the probability of the normal state occurring and the probability of the general state occurring in the state probability of the vehicles in the autonomous driving vehicle cluster traveling at the next moment predicted above.

[0327] For example, if the vehicle k is predicted to be at the next time t j+1 The state probability corresponding to the state of driving down Then vehicle k at the next time t j+1 The reliability of driving is

[0328] (8) Feedback the predicted state probability of the vehicles in the autonomous driving vehicle cluster at the next moment to the neural network model

[0329] It should be noted that the embodiment of the present application can use the state probability corresponding to the predicted driving state of the vehicles in the autonomous driving vehicle cluster at the next moment as the expected value of the neural network model output to feed back to the neural network model, thereby realizing learning and training of the neural network model, so as to improve the accuracy of the failure rate of the neural network model output system.

[0330] In an embodiment of the present application, when a vehicle in an autonomous driving vehicle cluster is in a faulty driving state (i.e., a faulty state), or a vehicle in an autonomous driving vehicle cluster is in an abnormal driving state (i.e., an abnormal state) and a faulty driving state (i.e., a faulty state), the driving of the vehicles in the autonomous driving vehicle cluster can be considered to have a system fault. Therefore, the system fault rate of the vehicles in the autonomous driving vehicle cluster at the next moment is used as the expected value output by the neural network model to be fed back to the neural network model. Among them, the system fault rate of the vehicles in the autonomous driving vehicle cluster at the next moment can be as follows:

[0331] ◆The system failure rate of the vehicles in the autonomous driving vehicle cluster at the next moment is the probability of the failure state occurring in the state probability of the vehicles in the autonomous driving vehicle cluster at the next moment predicted above;

[0332] For example, if the vehicle k is predicted to be at the next time t j+1 The state probability corresponding to the state of driving down Then vehicle k at the next time t j+1 The system failure rate is

[0333] ◆The system failure rate of the vehicles in the autonomous driving vehicle cluster at the next moment is the sum of the probability of the abnormal state occurring and the probability of the failure state occurring in the state probability of the vehicles in the autonomous driving vehicle cluster at the next moment predicted above.

[0334] For example, if the vehicle k is predicted to be at the next time t j+1 The state probability corresponding to the state of driving down Then vehicle k at the next time t j+1 The system failure rate is

[0335] (9) Predicting the reliability of the autonomous vehicle cluster’s driving at the next moment

[0336] It should be noted that, through the content of “(7) Predicting the reliability of vehicles in an autonomous driving vehicle cluster at the next moment”, it can be seen that the embodiment of the present application can predict the reliability of each vehicle in an autonomous driving vehicle cluster at the next moment.

[0337] 1) The reliability of the autonomous vehicle in the next moment

[0338] If the predicted vehicle is an autonomous bicycle, then the reliability of the vehicle's driving at the next moment is the reliability of the autonomous bicycle's driving at the next moment.

[0339] 2) The reliability of the autonomous vehicle platoon in the next moment

[0340] If the predicted vehicle is a vehicle in an autonomous driving vehicle formation, the reliability of the autonomous driving vehicle formation traveling at the next moment may be the average of the sum of the reliabilities of all vehicles in the autonomous driving vehicle formation traveling at the next moment.

[0341] For example, the nth autonomous driving vehicle formation in the autonomous driving cluster will j+1 Reliability of driving

[0342]

[0343] Where m represents the total number of vehicles in the nth autonomous vehicle formation; Indicates that the first vehicle (i.e., the head vehicle) in the nth autonomous driving vehicle formation will j+1 Reliability of driving; Indicates that the second vehicle in the nth autonomous driving vehicle formation will be j+1 Reliability of driving; It indicates that the mth vehicle (i.e. the rear vehicle) in the nth autonomous driving vehicle formation will be j+1 The reliability of driving can be understood by the same logic as the rest.

[0344] 3) The reliability of the autonomous driving vehicle cluster in the next moment

[0345] It should be noted that the reliability of the driving of the autonomous driving vehicle cluster at the next moment can be the average of the sum of the reliability of all autonomous driving single vehicles at the next moment and the reliability of all autonomous driving vehicle formations at the next moment.

[0346] For example, the autonomous driving cluster at the next moment t j+1 Reliability of driving j+1 :

[0347]

[0348] Wherein, N represents the total number of autonomous driving single vehicles and autonomous driving vehicle formations in the autonomous driving vehicle cluster.

[0349] 6. An exemplary description of a method for determining the reliability of vehicle cluster driving

[0350] In combination with the above description, the following will provide an exemplary description of the reliability determination method of vehicle cluster driving from the perspective of method example, please refer to Fig.10 . Fig.10 : is a flow chart of a method for determining reliability of vehicle cluster driving according to an embodiment of the present application. The method can be applied to a server. The method may include:

[0351] S1010. Obtain first data related to a driving path of a vehicle in a cluster of autonomous driving vehicles.

[0352] The first data may include at least one of the following: basic safety message (BSM), roadside information (RSI), roadside safety message (RSM), signal phase timing information (SPAT), and map data message (MAP).

[0353] It should be noted that the description of the “first data” can be found in the content description in the above “2. First data”, which will not be repeated here.

[0354] S1020. Determine M deviation rates at a first moment according to the first data.

[0355] The deviation rate is used to indicate the probability that a vehicle in a cluster of autonomous driving vehicles will deviate from its driving path, and M is a positive integer.

[0356] It should be noted that the “first moment” can be understood as a certain moment or the current moment, without any specific limitation.

[0357] The explanation of “deviation rate” can be found in the description of “3. Deviation rate” above, which will not be elaborated on here.

[0358] In some possible implementations, the M deviation rates include at least one of the following: a speed deviation rate, a steering angle deviation rate, a lateral position offset rate, a first collision time deviation rate, and a second collision time deviation rate;

[0359] The speed deviation rate is used to indicate the probability of a speed deviation between the driving speed of the vehicles in the autonomous vehicle cluster and the speed limit of the driving path;

[0360] The steering angle deviation rate is used to indicate the probability of a steering angle deviation between the driving direction and the driving path of the vehicles in the autonomous driving vehicle cluster;

[0361] The lateral position deviation rate is used to indicate the probability of a lateral position deviation between the body center position of a vehicle in the autonomous vehicle cluster and the driving path;

[0362] A first collision time deviation rate, used to indicate the probability of a collision time deviation between the forward collision warning collision time of the front and rear vehicles in the autonomous driving vehicle cluster and the first preset forward collision warning collision time;

[0363] The second collision time deviation rate is used to indicate the probability of a collision time deviation between the forward collision warning collision time from a vehicle in the autonomous driving vehicle cluster to the collision warning target and the second preset forward collision warning collision time.

[0364] S1030, inputting M deviation rates into a neural network model to obtain a system failure rate.

[0365] Among them, the system failure rate is used to indicate the probability of a system failure occurring in a vehicle in a cluster of autonomous vehicles at the first moment.

[0366] It should be noted that the description of “system failure rate” can be found in the above “4. Using neural network model to calculate system failure rate”, which will not be repeated here.

[0367] In some possible implementations, inputting the M deviation rates into the neural network model to obtain the system failure rate in S1030 may include:

[0368] Obtain N influence intervals and sensitivity thresholds corresponding to the influence intervals; wherein the influence interval is used to represent the threshold interval of the influence of the deviation rate on the system failure rate; the sensitivity threshold corresponding to the influence interval is used to represent the threshold of the sensitivity of the deviation rate to the system failure rate; N is the product of M and L, and L is the number of states in the Markov state corresponding to the deviation rate;

[0369] Determine a first confidence value corresponding to each of the M deviation rates according to the M deviation rates, the N impact intervals, and the sensitivity threshold of the impact interval, where the first confidence value is used to represent a confidence value of a system failure rate;

[0370] The M deviation rates are input into the neural network model, and the first confidence value is used as the expected value for learning and training to output the system failure rate.

[0371] It can be seen that the embodiment of the present application can determine the confidence value of the system failure rate based on M deviation rates, N influence intervals and the sensitivity thresholds of the influence intervals, and use the confidence value of the system failure rate as the expected value required for learning and training the neural network model, which is beneficial to improve the accuracy of the system failure rate output by the neural network model.

[0372] Furthermore, inputting the M deviation rates into the neural network model, and using the first confidence value as the expected value for learning and training to output the system failure rate may include:

[0373] Determine the maximum value of the first confidence values ​​corresponding to the M deviation rates to obtain a second confidence value;

[0374] The M deviation rates are input into the neural network model, and the second confidence value is used as the expected value for learning and training to output the system failure rate.

[0375] It can be seen that the embodiment of the present application can use the maximum value of the confidence values ​​of the system failure rate corresponding to each of the M deviation rates as the expected value required for learning and training the neural network model, so as to avoid using the confidence values ​​of the system failure rate corresponding to each of the M deviation rates as the expected value, which is beneficial to simplify the learning and training process of the neural network model and improve the learning and training efficiency.

[0376] Furthermore, the Markov state may include at least one of the following:

[0377] first state, second state, third state, fourth state, fifth state;

[0378] The first state is used to represent a state where a vehicle in a cluster of autonomous vehicles is not in motion, modeled using a Markov model;

[0379] The second state is used to represent the state in which the vehicles in the autonomous driving vehicle cluster are modeled in normal driving using the Markov model;

[0380] The third state is used to represent the state in which the vehicles in the autonomous driving vehicle cluster are modeled in general driving using the Markov model;

[0381] The fourth state is used to represent a state where the vehicles in the autonomous driving vehicle cluster are in abnormal driving mode modeled by the Markov model;

[0382] The fifth state is used to represent the state in which the vehicles in the autonomous driving vehicle cluster are modeled in a faulty driving state using a Markov model.

[0383] It can be seen that the embodiment of the present application can use the Markov model to model the driving status of vehicles in an autonomous driving vehicle cluster at different times.

[0384] S1040. Determine, based on the system failure rate, a state transition probability matrix of vehicles in the autonomous driving vehicle cluster modeled by the Markov model traveling at the first moment.

[0385] It should be noted that the description of the “Markov model” and the “state transition probability matrix” can be found in the above “5. Using the Markov model to predict the reliability of autonomous driving vehicle cluster driving”, which will not be repeated here.

[0386] In some possible implementations, determining the state transition probability matrix of vehicles in the autonomous driving vehicle cluster modeled by the Markov model at the first moment according to the system failure rate in S1040 may include:

[0387] Determine the driving state of the vehicles in the autonomous driving vehicle cluster modeled by the Markov model at the first moment according to the system failure rate;

[0388] The state transfer probability matrix of the Markov model at the first moment is determined according to the driving state of the vehicles in the autonomous driving vehicle cluster at the first moment and the driving state at the third moment, and the third moment is the previous moment of the first moment.

[0389] It can be seen that the embodiments of the present application can determine the driving state of the vehicle at the first moment through the system failure rate, and determine the state transition probability matrix through the driving state of the vehicle at the first moment and the driving state at the previous moment before the first moment, thereby updating the state transition probability matrix through the state changes at different moments.

[0390] S1050. Determine the reliability of vehicles in the autonomous driving vehicle cluster traveling at a second moment according to the state transition probability matrix, where the second moment is the next moment of the first moment.

[0391] In some possible implementations, determining the reliability of a vehicle in the autonomous driving vehicle cluster traveling at a second moment according to the state transition probability matrix may include:

[0392] Determine the state probability corresponding to the state in which the vehicles in the autonomous driving vehicle cluster are traveling at the second moment according to the state probability corresponding to the state in which the vehicles in the autonomous driving vehicle cluster are traveling at the first moment and the state transition probability matrix;

[0393] The reliability of the vehicles in the autonomous driving vehicle cluster traveling at the second moment is determined based on the state probability corresponding to the state in which the vehicles in the autonomous driving vehicle cluster are traveling at the second moment.

[0394] It can be seen that the embodiments of the present application can predict the state probability of the vehicle traveling at the next moment after the first moment through the state probability corresponding to the state of the vehicle traveling at the first moment and the state transition probability matrix, and determine the reliability of the vehicle traveling at the next moment through the predicted state probability of the vehicle traveling at the next moment, thereby realizing the prediction of the reliability of the vehicle traveling at the next moment.

[0395] Furthermore, the method may also include:

[0396] The state probability corresponding to the driving state of the vehicles in the autonomous driving vehicle cluster at the second moment is used as the expected value of the output of the neural network model to learn and train the neural network model.

[0397] It can be seen that the embodiment of the present application can use the state probability corresponding to the predicted driving state of the vehicle at the next moment as the expected value of the neural network model output to feed back to the neural network model, thereby realizing learning and training of the neural network model in order to improve the accuracy of the failure rate of the neural network model output system.

[0398] It can be seen that in Fig.10 In the described method, the embodiment of the present application can start from the factors that affect the reliability of the driving of the autonomous driving vehicle cluster, use the data acquired in real time to calculate at least one (i.e., M) deviation rate, input the at least one deviation rate into the neural network model to calculate the probability of a system failure of the vehicles in the autonomous driving vehicle cluster at a certain moment (i.e., the first moment) to obtain the system failure rate, and input the system failure rate into the Markov model to predict the state change trend of the vehicles in the autonomous driving vehicle cluster at the next moment (i.e., the second moment) to determine the reliability.

[0399] In this way, by adopting the continuous learning, training ability and high prediction ability of the neural network model, it can be ensured that the obtained system failure rate has a higher accuracy, and by adopting the predictive ability of the Markov model without the need for analysis and deduction based on a large amount of historical data, the amount of data required to be collected is small, the amount of calculation is small, and the practicality is strong. Ultimately, the driving process of the autonomous driving vehicle cluster can be determined in real time, dynamically and accurately (evaluation / estimation / analysis, etc.), which greatly shortens the reliability evaluation cycle, improves the reliability evaluation efficiency, and meets the current development needs of the automotive industry for reliability evaluation of autonomous driving vehicle cluster driving.

[0400] 7. Exemplary description of a device for determining the reliability of vehicle cluster driving

[0401] 1) Description

[0402] The above mainly introduces the solution of the embodiment of the present application from the perspective of the execution process of the method side. It can be understood that in order to realize the above functions, the server includes the hardware structure and / or software modules corresponding to the execution of each function.

[0403] Those skilled in the art should be aware that, in combination with the methods, functions, modules, units or steps of each example described in the embodiments provided herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a method, function, module, unit or step is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described methods, functions, modules, units or steps for each specific application, but such implementation should not be considered to be beyond the scope of the present application.

[0404] The embodiment of the present application can divide the server into functional units / modules according to the above method example. For example, each functional unit / module can be divided according to each function, or two or more functions can be integrated into one functional unit / module. The above integrated functional unit / module can be implemented in hardware or in software program. It should be noted that the division of functional units / modules in the embodiment of the present application is schematic, which is only a logical functional division, and there may be other division methods in actual implementation.

[0405] In the case of integrated units / modules, Fig.11 The present invention is a block diagram of the functional units of a vehicle cluster driving reliability determination device according to an embodiment of the present invention. The vehicle cluster driving reliability determination device 1100 specifically includes: a data acquisition unit 1101, a deviation rate determination unit 1102, a system failure rate determination unit 1103, a state transition probability matrix determination unit 1104, and a reliability determination unit 1105.

[0406] In some possible implementations, the data acquisition unit 1101, the deviation rate determination unit 1102, the system failure rate determination unit 1103, the state transition probability matrix determination unit 1104, and the reliability determination unit 1105 may be separate units or may be integrated together.

[0407] For example, the data acquisition unit 1101, the deviation rate determination unit 1102, the system failure rate determination unit 1103, the state transition probability matrix determination unit 1104, and the reliability determination unit 1105 are integrated in the processing unit.

[0408] The processing unit can be a processor or a controller, such as a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof.

[0409] The processing unit can implement or execute various exemplary logic blocks, units, modules or circuits described in combination with the contents disclosed in this application.

[0410] The processing unit may be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0411] In some possible designs, the vehicle cluster driving reliability determination device 1100 may further include a storage unit for storing a computer program or instruction executed by the vehicle cluster driving reliability determination device 1100. The storage unit may be a memory.

[0412] In some possible designs, the vehicle cluster driving reliability determination device 1100 may be a chip / chip module / processor / server / operating system.

[0413] In specific implementation, the data acquisition unit 1101, the deviation rate determination unit 1102, the system failure rate determination unit 1103, the state transition probability matrix determination unit 1104, and the reliability determination unit 1105 can be used to execute the steps described in the above method embodiment.

[0414] The data acquisition unit 1101 is used to acquire first data involved in the driving path of the vehicles in the autonomous driving vehicle cluster, where the first data includes at least one of the following: basic safety message, roadside information, roadside safety message, signal phase timing information, and map data message;

[0415] A deviation rate determination unit 1102, configured to determine M deviation rates at a first moment according to the first data, wherein the deviation rate is used to represent a probability that a vehicle in the autonomous driving vehicle cluster deviates from a driving path, and M is a positive integer;

[0416] A system failure rate determination unit 1103, used for inputting the M deviation rates into the neural network model to obtain a system failure rate, where the system failure rate is used to represent the probability of a system failure occurring in a vehicle in the autonomous driving vehicle cluster at a first moment;

[0417] A state transition probability matrix determining unit 1104 is used to determine a state transition probability matrix of the Markov model at a first moment according to the system failure rate;

[0418] The reliability determination unit 1105 is used to determine the reliability of the autonomous driving vehicle cluster traveling at a second moment according to the state transition probability matrix, where the second moment is the next moment of the first moment.

[0419] It can be seen that the embodiments of the present application can start from the factors that affect the reliability of the driving of the autonomous driving vehicle cluster, use the data acquired in real time to calculate at least one (i.e., M) deviation rate, input the at least one deviation rate into the neural network model to calculate the probability of a system failure of the vehicles in the autonomous driving vehicle cluster at a certain moment to obtain the system failure rate, and input the system failure rate into the Markov model to predict the state change trend of the vehicles in the autonomous driving vehicle cluster at the next moment to determine the reliability.

[0420] In this way, by adopting the continuous learning, training ability and high prediction ability of the neural network model, it can be ensured that the obtained system failure rate has a higher accuracy, and by adopting the predictive ability of the Markov model without the need for analysis and deduction based on a large amount of historical data, the amount of data required to be collected is small, the amount of calculation is small, and the practicality is strong. Ultimately, the driving process of the autonomous driving vehicle cluster can be determined in real time and dynamically (evaluation / estimation / analysis, etc.), which greatly shortens the reliability evaluation cycle, improves the reliability evaluation efficiency, and meets the current development needs of the automotive industry for reliability evaluation of autonomous driving vehicle clusters.

[0421] It should be noted that the specific implementation of each operation performed by the vehicle cluster driving reliability determination device 1100 can refer to the above Fig.10 The corresponding description of the method embodiment shown will not be repeated here.

[0422] 2) Some specific implementation methods

[0423] In some possible implementations, in inputting the M deviation rates into the neural network model to obtain the system failure rate, the system failure rate determination unit 1103 is used to:

[0424] Obtain N influence intervals and sensitivity thresholds corresponding to the influence intervals, where the influence intervals are used to represent the threshold interval of the influence of the deviation rate on the system failure rate; wherein the sensitivity threshold corresponding to the influence interval is used to represent the threshold of the sensitivity of the deviation rate to the system failure rate; N is the product of M and L, and L is the number of states in the Markov state corresponding to the deviation rate;

[0425] Determine a first confidence value corresponding to each of the M deviation rates according to the M deviation rates, the N impact intervals, and the sensitivity threshold of the impact interval, where the first confidence value is used to represent a confidence value of a system failure rate;

[0426] The M deviation rates are input into the neural network model, and the first confidence value is used as the expected value for learning and training to output the system failure rate.

[0427] In some possible implementations, in inputting the M deviation rates into the neural network model and performing learning and training with the first confidence value as the expected value to output the system failure rate, the system failure rate determination unit 1103 is used to:

[0428] Determine the maximum value of the first confidence values ​​corresponding to the M deviation rates to obtain a second confidence value;

[0429] The M deviation rates are input into the neural network model, and the second confidence value is used as the expected value for learning and training to output the system failure rate.

[0430] In some possible implementations, the Markov state includes at least one of the following:

[0431] first state, second state, third state, fourth state, fifth state;

[0432] The first state is used to represent a state where a vehicle in a cluster of autonomous vehicles is not in motion, modeled using a Markov model;

[0433] The second state is used to represent the state in which the vehicles in the autonomous driving vehicle cluster are modeled in normal driving using the Markov model;

[0434] The third state is used to represent the state in which the vehicles in the autonomous driving vehicle cluster are modeled in general driving using the Markov model;

[0435] The fourth state is used to represent a state where the vehicles in the autonomous driving vehicle cluster are in abnormal driving mode modeled by the Markov model;

[0436] The fifth state is used to represent the state in which the vehicles in the autonomous driving vehicle cluster are modeled in a faulty driving state using a Markov model.

[0437] In some possible implementations, in determining the state transition probability matrix of vehicles in the autonomous driving vehicle cluster modeled by the Markov model traveling at a first moment according to the system failure rate, the state transition probability matrix determination unit 1104 is used to:

[0438] Determine the driving state of the vehicles in the autonomous driving vehicle cluster modeled by the Markov model at the first moment according to the system failure rate;

[0439] The state transfer probability matrix of the Markov model at the first moment is determined according to the driving state of the vehicles in the autonomous driving vehicle cluster at the first moment and the driving state at the third moment, and the third moment is the previous moment of the first moment.

[0440] In some possible implementations, in determining the reliability of the vehicles in the autonomous driving vehicle cluster traveling at the second moment according to the state transition probability matrix, the reliability determination unit 1105 is used to:

[0441] Determine the state probability corresponding to the state in which the vehicles in the autonomous driving vehicle cluster are traveling at the second moment according to the state probability corresponding to the state in which the vehicles in the autonomous driving vehicle cluster are traveling at the first moment and the state transition probability matrix;

[0442] The reliability of the vehicles in the autonomous driving vehicle cluster traveling at the second moment is determined based on the state probability corresponding to the state in which the vehicles in the autonomous driving vehicle cluster are traveling at the second moment.

[0443] In some possible implementations, the vehicle cluster driving reliability determination device 1100 further includes:

[0444] The state probability feedback unit is used to use the state probability corresponding to the driving state of the vehicles in the autonomous driving vehicle cluster at the second moment as the expected value of the output of the neural network model to learn and train the neural network model.

[0445] In some possible implementations, the M deviation rates include at least one of the following:

[0446] Speed ​​deviation rate, steering angle deviation rate, lateral position deviation rate, first collision time deviation rate, second collision time deviation rate;

[0447] The speed deviation rate is used to indicate the probability of a speed deviation between the driving speed of the vehicles in the autonomous vehicle cluster and the speed limit of the driving path;

[0448] The steering angle deviation rate is used to indicate the probability of a steering angle deviation between the driving direction and the driving path of the vehicles in the autonomous driving vehicle cluster;

[0449] The lateral position deviation rate is used to indicate the probability of a lateral position deviation between the body center position of a vehicle in the autonomous vehicle cluster and the driving path;

[0450] A first collision time deviation rate, used to indicate the probability of a collision time deviation between the forward collision warning collision time of the front and rear vehicles in the autonomous driving vehicle cluster and the first preset forward collision warning collision time;

[0451] The second collision time deviation rate is used to indicate the probability of a collision time deviation between the forward collision warning collision time from a vehicle in the autonomous driving vehicle cluster to the collision warning target and the second preset forward collision warning collision time.

[0452] 8. An exemplary description of a server

[0453] The following is a schematic diagram of the structure of a server in an embodiment of the present application. Fig.12 The server 1200 includes a processor 1210 , a memory 1220 , and a communication bus for connecting the processor 1210 and the memory 1220 .

[0454] The processor 1210 may be one or more central processing units (CPUs). In the case where the processor 1210 is a CPU, the CPU may be a single-core CPU or a multi-core CPU. The memory 1220 includes, but is not limited to, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM) or a portable read-only memory (CD-ROM), and the memory 1220 is used to store relevant instructions and data.

[0455] In some possible implementations, the server 1200 may further include a communication interface for receiving and sending data.

[0456] In a specific implementation, the processor 1210 executes a computer program or instruction 1221 stored in the memory 1220 to implement the following steps: obtaining first data involved in a driving path of a vehicle in the autonomous driving vehicle cluster, the first data including at least one of the following: basic safety message, roadside information, roadside safety message, signal phase timing information, and map data message;

[0457] Determine M deviation rates at a first moment according to the first data, where the deviation rate is used to represent the probability that a vehicle in the autonomous driving vehicle cluster deviates from a driving path, and M is a positive integer;

[0458] The M deviation rates are input into the neural network model to obtain the system failure rate, which is used to represent the probability of a system failure occurring in a vehicle in the autonomous driving vehicle cluster at the first moment;

[0459] Determine a state transition probability matrix of vehicles in the autonomous driving vehicle cluster modeled by the Markov model at the first moment according to the system failure rate;

[0460] The reliability of the vehicles in the autonomous driving vehicle cluster traveling at a second moment is determined according to the state transition probability matrix, where the second moment is the next moment of the first moment.

[0461] It can be seen that the embodiments of the present application can start from the factors that affect the reliability of the driving of the autonomous driving vehicle cluster, use the data acquired in real time to calculate at least one (i.e., M) deviation rate, input the at least one deviation rate into the neural network model to calculate the probability of a system failure of the vehicles in the autonomous driving vehicle cluster at a certain moment to obtain the system failure rate, and input the system failure rate into the Markov model to predict the state change trend of the vehicles in the autonomous driving vehicle cluster at the next moment to determine the reliability.

[0462] In this way, by adopting the continuous learning, training ability and high prediction ability of the neural network model, it can be ensured that the obtained system failure rate has a higher accuracy, and by adopting the predictive ability of the Markov model without the need for analysis and deduction based on a large amount of historical data, the amount of data required to be collected is small, the amount of calculation is small, and the practicality is strong. Ultimately, the driving process of the autonomous driving vehicle cluster can be determined in real time and dynamically (evaluation / estimation / analysis, etc.), which greatly shortens the reliability evaluation cycle, improves the reliability evaluation efficiency, and meets the current development needs of the automotive industry for reliability evaluation of autonomous driving vehicle clusters.

[0463] It should be noted that the specific implementation of each operation performed by the server 1200 can refer to the above Fig.10 The corresponding description of the method embodiment shown will not be repeated here.

[0464] 9. Other exemplary explanations

[0465] An embodiment of the present application also provides a computer-readable storage medium, wherein a computer program or instructions are stored on the computer-readable storage medium, and when the computer program or instructions are executed by a processor, the steps in the method designed in the above embodiment are implemented.

[0466] The present application also provides a computer program product, including a computer program or instructions, wherein the computer program or instructions are executed by a processor to implement the steps in the method designed in the above embodiment. Exemplarily, the computer program product can be a software installation package.

[0467] It should be noted that, for the above-mentioned various embodiments, for the sake of simple description, they are all expressed as a series of action combinations. Those skilled in the art should be aware that the present application is not limited by the described order of actions, because some steps in the embodiments of the present application can be performed in other orders or simultaneously. In addition, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions, steps, modules or units involved are not necessarily required by the embodiments of the present application.

[0468] In the above embodiments, the embodiments of the present application have different focuses on the description of each embodiment. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0469] The various devices and products described in the above embodiments include modules / units, which may be software modules / units or hardware modules / units, or may be partially software modules / units and partially hardware modules / units. For example, for each device or product that applies to or integrates a chip, each module / unit contained therein can be implemented in the form of hardware such as circuits, or at least some modules / units can be implemented in the form of software programs, which run on the integrated processor inside the chip, and the remaining modules / units can be implemented in the form of hardware such as circuits; for each device or product that applies to or integrates a chip module, each module / unit contained therein can be implemented in the form of hardware such as circuits, and different modules / units can be located in the same part of the chip module (for example, a chip, a circuit module, etc.) or in different components, at least some / units can be implemented in the form of software programs, which run on the integrated processor inside the chip module, and the remaining modules / units can be implemented in the form of hardware such as circuits; for each device or product that applies to or integrates a terminal, the modules / units contained therein can be implemented in the form of hardware such as circuits, and different modules / units can be located in the same component (for example, a chip, a circuit module, etc.) or in different components in the terminal, or at least some modules / units can be implemented in the form of software programs, which run on the integrated processor inside the terminal, and the remaining modules / units can be implemented in the form of hardware such as circuits.

[0470] Those skilled in the art should be aware that the methods, steps or functions of related modules / units described in the embodiments of the present application can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product, or it can be implemented by a processor executing a computer program instruction. Wherein, the computer program product includes at least one computer program instruction, and the computer program instruction can be composed of corresponding software modules, and the software module can be stored in RAM, flash memory, ROM, EPROM, EEPROM, register, hard disk, mobile hard disk, read-only compact disk (CD-ROM) or any other form of storage medium known in the art. The computer program instruction can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer program instruction can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrations. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium, or a semiconductor medium (eg, an SSD), etc.

[0471] The modules / units included in the devices or products described in the above embodiments may be software modules / units, hardware modules / units, or may be partially software modules / units and partially hardware modules / units. For example, for each device or product applied to or integrated in a chip, each module / unit included therein may be implemented in the form of hardware such as circuits; or, a portion of the modules / units included therein may be implemented in the form of a software program, which runs on a processor integrated inside the chip, while a portion of the modules / units of another portion (if any) may be implemented in the form of hardware such as circuits. The same is true for each device or product applied to or integrated in a chip module, or each device or product applied to or integrated in a terminal.

[0472] The specific implementation methods described above further describe the purpose, technical solutions and beneficial effects of the embodiments of the present application in detail. It should be understood that the above description is only the specific implementation method of the embodiments of the present application and is not intended to limit the protection scope of the embodiments of the present application. Any modification, equivalent replacement, improvement, etc. made on the basis of the technical solution of the embodiments of the present application shall be included in the protection scope of the embodiments of the present application.

Claims

1. A method for determining the reliability of vehicle cluster driving, It is characterized in that include: Acquire first data involved in a driving path of a vehicle in a cluster of autonomous vehicles, wherein the first data includes at least one of the following: basic safety messages, roadside information, roadside safety messages, signal phase timing information, and map data messages; Determine M deviation rates at a first moment according to the first data, where the deviation rates are used to represent probabilities that vehicles in the autonomous driving vehicle cluster will deviate from the driving path, and M is a positive integer; Inputting the M deviation rates into a neural network model to obtain a system failure rate, wherein the system failure rate is used to represent a probability of a system failure occurring in a vehicle in the autonomous driving vehicle cluster at the first moment; Determine, according to the system failure rate, the driving state of the vehicles in the autonomous driving vehicle cluster modeled by the Markov model at the first moment; Determining a state transition probability matrix of the Markov model at the first moment according to the driving state of the vehicles in the autonomous driving vehicle cluster at the first moment and the driving state at the third moment, wherein the third moment is a moment before the first moment; The reliability of the vehicles in the autonomous driving vehicle cluster traveling at a second moment is determined according to the state transition probability matrix, where the second moment is a moment after the first moment.

2. The method according to claim 1, in, The step of inputting the M deviation rates into a neural network model to obtain a system failure rate comprises: Obtain N influence intervals and sensitivity thresholds corresponding to the influence intervals, wherein the influence intervals are used to represent the threshold intervals of the influence of the deviation rate on the system failure rate; wherein the sensitivity threshold corresponding to the influence intervals is used to represent the threshold of the sensitivity of the deviation rate to the system failure rate; N is the product of M and L, and L is the number of states in the Markov state corresponding to the deviation rate; Determine, according to the M deviation rates, the N impact intervals and the sensitivity thresholds of the impact intervals, a first confidence value corresponding to each of the M deviation rates, wherein the first confidence value is used to represent a confidence value of the system failure rate; The M deviation rates are input into the neural network model, and the first confidence value is used as the expected value for learning and training to output the system failure rate.

3. The method according to claim 2, in, The step of inputting the M deviation rates into the neural network model and using the first confidence value as an expected value for learning and training to output the system failure rate comprises: Determine the maximum value of the first confidence values ​​corresponding to each of the M deviation rates to obtain a second confidence value; The M deviation rates are input into the neural network model, and the second confidence value is used as the expected value for learning and training to output the system failure rate.

4. The method according to claim 2, in, The Markov state includes at least one of the following: first state, second state, third state, fourth state, fifth state; The first state is used to represent a state in which the vehicles in the autonomous driving vehicle cluster are not driving, modeled using the Markov model; The second state is used to represent a state in which the vehicles in the autonomous driving vehicle cluster are modeled in normal driving using the Markov model; The third state is used to represent a state in which the vehicles in the autonomous driving vehicle cluster are modeled in general driving using the Markov model; The fourth state is used to represent a state in which the vehicles in the autonomous driving vehicle cluster are modeled in abnormal driving using the Markov model; The fifth state is used to represent the state in which the vehicles in the autonomous driving vehicle cluster are modeled in a faulty driving state using the Markov model.

5. The method according to claim 1, in, The determining, according to the state transition probability matrix, the reliability of the vehicles in the autonomous driving vehicle cluster traveling at the second moment comprises: Determining, based on the state probabilities corresponding to the states in which the vehicles in the autonomous driving vehicle cluster are traveling at the first moment and the state transition probability matrix, the state probabilities corresponding to the states in which the vehicles in the autonomous driving vehicle cluster are traveling at the second moment; The reliability of the vehicles in the autonomous driving vehicle cluster traveling at the second moment is determined based on the state probability corresponding to the state of the vehicles in the autonomous driving vehicle cluster traveling at the second moment.

6. The method according to claim 5, in, Also includes: The state probability corresponding to the driving state of the vehicles in the autonomous driving vehicle cluster at the second moment is used as the expected value of the output of the neural network model to learn and train the neural network model.

7. The method according to claim 1, in, The M deviation rates include at least one of the following: Speed ​​deviation rate, steering angle deviation rate, lateral position deviation rate, first collision time deviation rate, second collision time deviation rate; The speed deviation rate is used to indicate the probability of a speed deviation between the driving speed of the vehicles in the autonomous driving vehicle cluster and the driving speed limit of the driving path; The steering angle deviation rate is used to indicate the probability of a steering angle deviation between the driving direction of the vehicles in the autonomous driving vehicle cluster and the driving path; The lateral position deviation rate is used to indicate the probability of a lateral position deviation between the body center position of the vehicle in the autonomous driving vehicle cluster and the driving path; The first collision time deviation rate is used to indicate the probability of a collision time deviation between the forward collision warning collision time of the front and rear vehicles in the autonomous driving vehicle cluster and the first preset forward collision warning collision time; The second collision time deviation rate is used to indicate the probability of a collision time deviation between the forward collision warning collision time from the vehicle in the autonomous driving vehicle cluster to the collision warning target and the second preset forward collision warning collision time.

8. A device for determining the reliability of vehicle cluster driving, It is characterized in that include: A data acquisition unit, configured to acquire first data involved in a driving path of a vehicle in a cluster of autonomous vehicles, wherein the first data includes at least one of the following: basic safety messages, roadside information, roadside safety messages, signal phase timing information, and map data messages; a deviation rate determination unit, configured to determine M deviation rates at a first moment according to the first data, wherein the deviation rate is used to represent a probability that a vehicle in the autonomous driving vehicle cluster deviates from the driving path, and M is a positive integer; a system failure rate determination unit, configured to input the M deviation rates into a neural network model to obtain a system failure rate, wherein the system failure rate is used to represent a probability of a system failure occurring in a vehicle in the autonomous driving vehicle cluster at the first moment; a state transition probability matrix determination unit, configured to determine the driving state of the vehicles in the autonomous driving vehicle cluster modeled by the Markov model at the first moment according to the system failure rate; and determine the state transition probability matrix of the Markov model at the first moment according to the driving state of the vehicles in the autonomous driving vehicle cluster at the first moment and the driving state at a third moment, wherein the third moment is a moment before the first moment; A reliability determination unit is used to determine the reliability of the autonomous driving vehicle cluster traveling at a second moment according to the state transition probability matrix, where the second moment is the next moment of the first moment.

9. A server comprising a processor, a memory and a computer program or instruction stored in the memory, It is characterized in that The processor executes the computer program or instructions to implement the steps of the method according to any one of claims 1 to 7.

Images