CN108304986B

CN108304986B - Evaluation method for behavior safety of automatic driving vehicle

Info

Publication number: CN108304986B
Application number: CN201711348568.2A
Authority: CN
Inventors: 宋娟; 何承坤; 王荣; 薛晓卿; 刘大鹏; 丁文龙; 朱科屹; 李京泰
Original assignee: Computer And Microelectronics Development Research Center Ministry Of Industry And Information Technology (china Software Testing Center)
Current assignee: China Software Evaluation Center
Priority date: 2017-12-15
Filing date: 2017-12-15
Publication date: 2021-06-29
Anticipated expiration: 2037-12-15
Also published as: CN108304986A

Abstract

The invention discloses an evaluation method for behavior safety of an automatic driving vehicle, which comprises the following steps: acquiring environment perception parameters, network communication parameters, judgment decision parameters, control execution parameters and network connection degree coefficients of an automatic driving vehicle; obtaining an evaluation score of the behavior safety of the automatic driving vehicle according to the environment perception parameter, the network communication parameter, the judgment decision parameter, the control execution parameter and the network connection degree coefficient; and obtaining the evaluation result of the behavior safety of the automatic driving vehicle according to the evaluation score of the behavior safety of the automatic driving vehicle. The method for evaluating the behavior safety of the automatic driving vehicle can comprehensively consider the behavior safety of the automatic driving vehicle from multiple angles, and enables the evaluation result to be accurate and reliable, thereby providing a feasible method for the behavior safety test and evaluation of the automatic driving vehicle, perfecting the relevant test standards and promoting the development of the industry of the automatic driving vehicle in China.

Description

Evaluation method for behavior safety of automatic driving vehicle

Technical Field

The invention relates to the technical field of driving behavior safety evaluation of intelligent networked automobiles and automatically driven automobiles, in particular to an evaluation method for behavior safety of automatically driven automobiles.

Background

The intelligent internet automobile is a new generation automobile which is provided with advanced vehicle-mounted sensors, controllers, actuators and the like, integrates modern communication and network technologies, realizes intelligent information exchange and sharing between vehicles and people, between vehicles and vehicles, between vehicles and roads, between vehicles and backstage and the like, has the functions of complex environment perception, intelligent decision, cooperative control, execution and the like, can realize safe, comfortable, energy-saving and efficient driving, and can finally replace people to operate.

The currently industrialized intelligent networked automobile technology is mainly an auxiliary driving system. The conventional driving assistance system includes: a front collision alarm system, a lane change auxiliary system, an automatic emergency brake system and the like. The core of evaluating these driving assistance systems is the standard conformity evaluation of a single function, which is relatively easy to develop. Furthermore, the complexity of the intelligent networked automobile automatic driving system is not inferior to that of the automobile, and the possible harm to the personal and property safety is also not inferior to that of the automobile. Compared with the evaluation requirement for guaranteeing the safety of the on-road automobiles, the existing evaluation scheme has insufficient integrity and systematicness for guaranteeing and supporting the safe and reliable use of the civil automatic driving system in a real road environment (especially under urban working conditions) and cannot serve as the safety baseline requirement of 'on-road admission' of the automatic driving system.

The domestic and foreign automobile industry is also aware of this problem, and although the safety baseline requirement for how to give "permit to go to the road" of the automatic driving system is still in the process of research and discussion, it also has a basic consensus on the evaluation of the automatic driving system, namely that the evaluation of the automatic driving system requires the support of a comprehensive test verification site. For example, the united states, europe, japan, china, and the like are all in experimental verification demonstration areas where autonomous driving systems or intelligent networked automobiles are established. In this respect, the united states walks in front of the united states, establishes laws and regulations for popularization of the automatic driving technology, establishes a relevant admission system for automobiles which need to be tested and operated on line, and sets a uniform examination point. An examination can generally be divided into three phases: (1) checking the data; (2) road examination; (3) and comprehensively evaluating, issuing license plates and allowing specified road section testing.

However, at present, in terms of how to systematically evaluate the safety of the behavior of the autonomous vehicle and approve the final driving of the autonomous vehicle on the road, countries in the world are still in the starting stage, and no systematic evaluation method is formed. Therefore, the research of the evaluation method is beneficial to the development of the automatic driving vehicle industry in China, and the improvement of the related test standards is promoted.

Disclosure of Invention

The technical problem to be solved by the invention is as follows: at present, in the aspects of how to systematically evaluate the safety of the behavior of the automatically-driven vehicle and approve the final driving of the automatically-driven vehicle on the road, all countries in the world are still in a starting stage, and a systematic evaluation method is not formed, so that the related test standards are not perfect, and the development of the industry of the automatically-driven vehicle in China is not facilitated.

In order to solve the technical problem, the invention provides an evaluation method for behavior safety of an automatic driving vehicle.

According to an embodiment of the present invention, there is provided an evaluation method of behavior safety of an autonomous vehicle, including:

acquiring environment perception parameters, network communication parameters, judgment decision parameters, control execution parameters and network connection degree coefficients of an automatic driving vehicle;

obtaining an evaluation score of the behavior safety of the automatic driving vehicle according to the environment perception parameter, the network communication parameter, the judgment decision parameter, the control execution parameter and the network connection degree coefficient;

and obtaining the evaluation result of the behavior safety of the automatic driving vehicle according to the evaluation score of the behavior safety of the automatic driving vehicle.

Preferably, the evaluation score is obtained according to the following expression:

wherein Z is the evaluation score, H is the environment sensing parameter, L is the network connectivity parameter, P is the decision parameter, K is the control execution parameter, a and b are the network connectivity degree coefficients, and a + b is 1.

Preferably, the evaluation result of the behavior safety of the automatic driving vehicle is obtained according to the size relation between the evaluation result and a preset score threshold value.

Preferably, the environmental perception parameter H of the autonomous vehicle is derived according to the following expression:

H＝HN×0.4+HW_W×0.4+HW_A×0.2

wherein HN is an environmental perception accuracy evaluation parameter, HW_WEvaluation of parameters for environmental perception of stability, HW_AAnd evaluating parameters for environmental perception and timeliness.

Preferably, the environment sensing accuracy evaluation parameter, the environment sensing stability evaluation parameter, and the environment sensing timeliness evaluation parameter are respectively determined by a static environment sensing evaluation parameter and a dynamic environment sensing evaluation parameter corresponding to each of them, where the dynamic environment sensing evaluation parameter includes: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters.

Preferably, the network connectivity parameter L of the autonomous vehicle is obtained according to the following expression:

L＝LN×0.4+LW_W×0.4+LW_A×0.2

LN is network connection accuracy evaluation parameter LW_WEvaluation of stability parameters for network connectivity, LW_AAnd evaluating the parameters of the timeliness of network communication.

Preferably, the network connectivity accuracy evaluation parameter, the network connectivity stability evaluation parameter, and the network connectivity timeliness evaluation parameter are respectively determined by a static network connectivity evaluation parameter and a dynamic network connectivity evaluation parameter corresponding to each of the parameters, where the dynamic network connectivity evaluation parameter includes: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters.

Preferably, the decision parameter P for the autonomous vehicle is derived according to the following expression:

P＝PN×0.2+PW_W×0.6+PW_A×0.2

wherein, PN is evaluation parameter for judging decision accuracy, PW_WFor determining decision stability evaluation parameters, PW_ATo determine the decision-making timeliness assessment parameters.

Preferably, the decision accuracy evaluation parameter, the decision stability evaluation parameter and the decision timeliness evaluation parameter are determined by a static decision evaluation parameter and a dynamic decision evaluation parameter respectively, where the dynamic decision evaluation parameter includes: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters.

Preferably, the control execution parameter K of the autonomous vehicle is derived from the following expression:

K＝KN×0.4+KW_W×0.4+KW_A×0.2

wherein KN is a control execution accuracy evaluation parameter, KW_WFor controlling the execution of stability evaluation parameters, KW_AAnd executing a timeliness evaluation parameter for control.

Preferably, the control execution accuracy evaluation parameter, the control execution stability evaluation parameter and the control execution timeliness evaluation parameter are respectively determined by a static control execution evaluation parameter and a dynamic control execution evaluation parameter corresponding to each of the parameters, wherein the dynamic control execution evaluation parameter includes: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters.

Compared with the prior art, one or more embodiments in the above scheme can have the following advantages or beneficial effects:

by applying the evaluation method for the behavior safety of the automatic driving vehicle provided by the embodiment of the invention, the evaluation score of the behavior safety of the automatic driving vehicle is obtained through the acquired environment perception parameter, network communication parameter, judgment decision parameter, control execution parameter and network connection degree coefficient of the automatic driving vehicle, and the evaluation result of the behavior safety of the automatic driving vehicle is obtained according to the evaluation score. Therefore, the method for evaluating the behavior safety of the automatic driving vehicle can comprehensively consider the behavior safety of the automatic driving vehicle from multiple angles, and the evaluation result is accurate and reliable, so that a feasible method is provided for the behavior safety test and evaluation of the automatic driving vehicle, the related test standards are perfect, and the development of the automatic driving vehicle industry in China is promoted.

In addition, in the evaluation method for automatically evaluating the behavior safety of the vehicle, the acquired environment perception parameter, the network communication parameter, the judgment decision parameter and the control execution parameter are obtained by respectively combining the characteristics of the static evaluation parameter and the dynamic evaluation parameter which correspond to each other. Therefore, the method and the device can comprehensively consider the behavior safety of the automatic driving vehicle from multiple angles, and effectively improve the accuracy and reliability of the evaluation result.

Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by the practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:

fig. 1 is a flowchart illustrating an evaluation method for behavioral safety of an autonomous vehicle according to an embodiment of the present invention.

Detailed Description

The following detailed description of the embodiments of the present invention will be provided with reference to the drawings and examples, so that how to apply the technical means to solve the technical problems and achieve the technical effects can be fully understood and implemented. It should be noted that, as long as there is no conflict, the embodiments and the features of the embodiments of the present invention may be combined with each other, and the technical solutions formed are within the scope of the present invention.

In order to solve the technical problems that in the prior art, in terms of how to systematically evaluate the safety of the behavior of the automatic driving vehicle and approve the final driving of the automatic driving vehicle on the road, all countries in the world are still in a starting stage, and a systematic evaluation method is not formed, so that the related test standards are not complete, and the development of the automatic driving vehicle industry in China is not facilitated, the embodiment of the invention provides the evaluation method for the behavior safety of the automatic driving vehicle.

The method for evaluating the behavior safety of the automatic driving vehicle adopts an evaluation score mode to evaluate the safety of the automatic driving vehicle.

The evaluation score of the behavior safety of the automatic driving vehicle mainly depends on an environment perception parameter H, a network communication parameter L, a judgment decision parameter P, a control execution parameter K and network connection degree coefficients a and b of the automatic driving vehicle.

As shown in fig. 1, the method for evaluating the behavior safety of an autonomous vehicle according to the embodiment of the present invention mainly includes the following steps S101 to S103.

Step S101, obtaining an environment perception parameter H, a network communication parameter L, a judgment decision parameter P, a control execution parameter K and network connection degree coefficients a and b of the automatic driving vehicle.

And S102, obtaining an evaluation score of the behavior safety of the automatic driving vehicle according to the environment perception parameter H, the network communication parameter L, the judgment decision parameter P, the control execution parameter K and the network connection degree coefficients a and b.

And step S103, obtaining an evaluation result of the behavior safety of the automatic driving vehicle according to the evaluation score of the behavior safety of the automatic driving vehicle.

Specifically, step S101 is first executed to acquire the environmental awareness parameter H of the autonomous vehicle.

Here, the environment awareness parameter of the autonomous vehicle is used to evaluate a degree of awareness of the autonomous vehicle during autonomous driving behavior of the associated physical environment, the physical environment comprising: road surface conditions, traffic information, road markings, etc.

Before determining the environmental awareness parameters H of the autonomous vehicle, the following parameters need to be determined: environmental perception accuracy evaluation parameter HN and environmental perception stability evaluation parameter HW_WAnd environmental awareness timeliness assessment parameter HW_A. Wherein, the environmental perception accuracy evaluation parameter HN and the environmental perception stability evaluation parameter HW_WAnd environmental awareness timeliness assessment parameter HW_ARespectively determined by the static environment perception evaluation parameter and the dynamic environment perception evaluation parameter which respectively correspond to the static environment perception evaluation parameter and the dynamic environment perception evaluation parameter. Here, the dynamic environment perception evaluation parameter includes, among others: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters. Specifically, the static environment-aware evaluation parameters include: static environmental perception accuracy HJ_CStatic environmental perception stability HJ_RAnd static context awareness timeliness HJ_T. The dynamic context-aware evaluation parameters include: normal driving condition dynamic environment perception accuracy HD_CNDynamic environment perception stability HD under normal driving condition_RNDynamic environment sensing timeliness HD under normal driving condition_TN(ii) a Dangerous working condition dynamic environment perception accuracy HD_CWWDynamic environment perception stability HD for dangerous working condition_RWWDynamic environment perception timeliness HD for dangerous working condition_TWW(ii) a Accuracy HD for sensing dynamic environment under dangerous release conditions_CWADynamic environment perception stability HD for dangerous release working condition_RWADynamic environment sensing timeliness HD for danger release working condition_TWA。

Therefore, the method for evaluating the behavior safety of the automatic driving vehicle, provided by the embodiment of the invention, can be used for calculating the environment perception parameter H of the automatic driving vehicle by combining the static environment perception evaluation parameter and the dynamic environment perception evaluation parameter, comprehensively considering the behavior safety of the automatic driving vehicle from multiple angles, and effectively improving the accuracy and reliability of an evaluation result.

After obtaining the above parameters, the environmental perception parameter H of the autonomous vehicle is obtained according to the following expression:

H＝HN×0.4+HW_W×0.4+HW_A×0.2 (1)

the parameters in formula (1) are obtained using the following formulae:

HN＝HC_N×0.5+HR_N×0.3+HT_N×0.2

HW_W＝HC_WW×0.5+HR_WW×0.3+HT_WW×0.2

HW_A＝HC_WA×0.5+HR_WA×0.3+HT_WA×0.2

wherein HC_NEvaluating the parameter, HR, for normal driving conditions and environmental perception accuracy_NEvaluation of environmental perception accuracy for dangerous conditions, HT_NEvaluation of environmental perception accuracy for hazardous release conditions, HC_WWEvaluating the parameter, HR, for the environmental perception stability under normal driving conditions_WWEvaluation of environmental perception stability for dangerous conditions, HT_WWEvaluation parameters for environmental perception stability of dangerous release conditions, HC_WAFor normal driving conditions, environmental perception and timeliness evaluation of parameters, HR_WAEnvironmental awareness and timeliness assessment parameters for dangerous conditions, HT_WAAnd (4) evaluating parameters for sensing environment and timeliness under dangerous release working conditions.

The parameters in the above expression are obtained using the following formulae:

the specific acquisition process of the parameters is as follows:

accuracy HJ of static environment perception_CStatic environmental perception stability HJ_RStatic context awareness timeliness HJ_TAre set to 100 minutes, respectively.

Static context awarenessThe accuracy refers to: the accuracy with which a stationary autonomous vehicle identifies an object. Static environmental perception accuracy HJ_CScoring rules: a stationary evaluation vehicle comes forward a certain distance (20 m) respectively: 150 perception objects such as 3 typical signal lamps (conventional natural light irradiation), 100 traffic signs (forward direction and gray direction), 20 automobiles (forward direction) and 21 pedestrians (forward direction) require that the vehicles recognize within a specified time, each object is recognized wrongly or not recognized and deducted by 1 point, the lowest point is 0 point, and a character result is output after the recognition is finished.

The static environment perception stability is that: when a recognition object parameter (e.g., signal light intensity, traffic sign angle, color, etc.) changes, the stationary autonomous vehicle accurately recognizes the stability of the object. Static environmental perception stability HJ_RScoring rules: a stationary evaluation vehicle comes forward a certain distance (20 m) respectively: 291 sensing objects such as 3 typical signal lamps (strong light irradiation), 100 traffic signs (15 degrees, white or blue), 20 automobiles (side and back) and 21 pedestrians (side and back) require the vehicles to be recognized within a specified time, each object is recognized wrongly or not recognized and deducted for 0.5 minutes, the lowest point is 0 minute, and a character result is output after the recognition is finished.

It should be noted that, in the present invention, the "lateral" direction of the car or the pedestrian refers to the meeting or meeting with the static (or dynamic) car or pedestrian on the side, and the angle is not limited.

Static context awareness timeliness refers to: stationary autonomous vehicles accurately identify the timeliness of an object. Static context awareness timeliness HJ_TScoring rules: a stationary evaluation vehicle comes forward a certain distance (20 m) respectively: 150 perception objects such as 3 typical signal lamps (conventional natural light irradiation), 100 traffic signs (forward direction, gray), 20 automobiles (forward direction) and 21 pedestrians (forward direction) require that the vehicles recognize within a specified time (100ms), each object is recognized wrongly or not recognized and deducted for 1 point, and the lowest point is 0 point, and a character result is output after the recognition is finished.

Sensing accuracy HD of normal driving condition dynamic environment_CNNormal driving conditionAttitude environment perception stability HD_RNDynamic environment sensing timeliness HD under normal driving condition_TNAre set to 100 minutes, respectively.

The normal driving condition dynamic environment perception accuracy is as follows: the degree of accuracy to which an autonomous vehicle recognizes objects (e.g., traffic lights, static traffic signs, other traffic participants) during normal driving. Normal driving condition dynamic environment perception accuracy HD_CNScoring rules: the normal running evaluation vehicle respectively appears in front of the vehicle for a certain distance (20 meters): 9 the general normal driving conditions of 9 categories total 45 perception objects, the perception objects are in normal state, the vehicle is required to be identified within the specified time, each object is identified wrongly or not identified for 3 points, the lowest point is 0 point, and the character result is output after the identification is finished.

It should be noted that, in the present invention, the traffic sign can have two variations: angle (positive and 15 degree angle) and background (white, gray and blue) etc. changes; testing cars and pedestrians can have a variation: orientation changes (forward, lateral and back); the test signal lamp may have a variant: light intensity (normal natural light irradiation, intense light irradiation). The "normal state" means 3 typical signal lights (normal natural light irradiation), 100 traffic signs (forward, gray), 20 cars (forward), 21 pedestrians (forward), and the like. Under normal conditions, the autonomous vehicle does not encounter dangerous scenes in the driving process. Except for the normal state, the other states are abnormal states.

The normal driving condition dynamic environment perception stability is as follows: the automatic driving vehicle can identify the object in the normal driving process, and when the parameters of the identification object (such as the brightness intensity of a signal lamp, the angle of a traffic sign, the color and the like) are changed or abnormal, the automatic driving vehicle can still accurately identify the stability of the object. Dynamic environment perception stability HD under normal driving condition_RNScoring rules: the normal running evaluation vehicle respectively appears in front of the vehicle for a certain distance (20 meters): the 9 major normal driving conditions total 45 sensing objects, the sensing objects are in abnormal states (the illumination, orientation, background and the like are different from the normal states), and the vehicle is required to move in the specified timeAnd line identification, wherein each object is wrongly identified or not identified and deducted for 3 points, the lowest point is 0 point, and a character result is output after the identification is finished.

The dynamic environment perception timeliness under the normal driving condition refers to that: the degree of timeliness of an object is accurately identified in the normal driving process of an automatic driving vehicle. Normal driving condition dynamic environment perception timeliness HD_TNScoring rules: the normal running evaluation vehicle respectively appears in front of the vehicle for a certain distance (20 meters): the 9 major types of normal driving conditions total 45 sensing objects, the vehicle is required to be identified within the specified time (0.5s), each object is identified wrongly or not identified for 3 points, the lowest point is 0 point, and a character result is output after the identification is finished.

Dynamic environment sensing accuracy HD for dangerous working condition_CWWDynamic environment perception stability HD for dangerous working condition_RWWDynamic environment perception timeliness HD for dangerous working condition_TWWAre set to 100 minutes, respectively.

The sensing accuracy of the dangerous working condition dynamic environment refers to: the accuracy of identifying objects (e.g., traffic lights, static traffic signs, other traffic participants) during normal driving of an autonomous vehicle under dangerous operating conditions. Dangerous working condition dynamic environment perception accuracy HD_CWWScoring rules: the normal running evaluation vehicle respectively appears in front of the vehicle for a certain distance (20 meters): the 32 major dangers are 230 sensing objects in total, the sensing objects are in normal states, the vehicle is required to be identified within the specified time, each object is identified wrongly or not identified by 0.5 point, the lowest point is 0 point, and the character result is output after the identification is finished.

The dynamic environment perception stability of the dangerous working condition refers to: the automatic driving vehicle identifies the object in the normal driving process, and under the working condition that danger occurs, when the parameters (such as the brightness intensity of a signal lamp, the angle of a traffic sign, the color and the like) of the identification object change or are abnormal, the automatic driving vehicle can still accurately identify the stability of the object. Dynamic environment perception stability HD for dangerous working condition_RWWScoring rules: the normal running evaluation vehicle respectively appears in front of the vehicle for a certain distance (20 meters): 32 major dangersThe risk occurrence working condition is 230 sensing objects in total, the sensing objects are in abnormal states (the illumination, the orientation, the background and the like are different from normal states), the vehicle is required to be identified within the specified time, each object is identified wrongly or not identified by 0.5 point, the lowest point is 0 point, and the character result is output after the identification is finished.

For ease of understanding, the non-motor vehicle cross-over is described herein as an example.

Dangerous working conditions occur: the non-motor vehicle traverses.

Environment: during the day, the weather is clear, supplemented with appropriate logos, etc., in addition to the main participants, for a total of 5 perception points.

Road scene: a T-shaped intersection or a crossroad in an urban area is provided with a large number of buildings at the intersection.

The vehicle state: the main vehicle turns right at a T-shaped intersection or a crossroad in an urban area, a rider crossing the road suddenly appears at the front right, the sight of a camera of the main vehicle and a driver is shielded by a building, and the two vehicles collide at the crossroad if the moving state of the two vehicles is kept unchanged.

The evaluation method comprises the following steps: the length of a scene road section is 10 meters, a curtain covers the scene 50 meters away, the vehicle runs towards the scene direction at the speed of 40km/h, timing is started after the curtain, and the number of correct identification signal lamps, marks, other traffic participants and the like of the vehicle is recorded.

The dynamic environment perception timeliness of the dangerous working condition refers to the following steps: during normal driving of an autonomous vehicle, the timeliness of objects (e.g., traffic lights, static traffic signs, other traffic participants) is accurately identified under dangerous conditions. Dangerous working condition dynamic environment perception timeliness HD_TWWScoring rules: the normal running evaluation vehicle respectively appears in front of the vehicle for a certain distance (20 meters): the 32 major dangerous working conditions total 230 perception objects, the vehicle is required to be identified within the specified time (0.5s), each object is identified wrongly or not identified by 0.5 point, the lowest point is 0 point, and the character result is output after the identification is finished.

Accuracy HD for sensing dynamic environment under danger release condition_CWADynamic environment perception stability HD for dangerous release working condition_RWADynamic environment sensing timeliness HD for danger release working condition_TWAAre set to 100 minutes, respectively.

The danger release working condition dynamic environment perception accuracy is as follows: the accuracy with which an autonomous vehicle identifies objects (e.g., traffic lights, static traffic signs, other traffic participants) during normal driving, in the event of a released hazard. Accuracy HD for sensing dynamic environment under dangerous release conditions_CWAScoring rules: the normal running evaluation vehicle respectively appears in front of the vehicle for a certain distance (20 meters): and 5, 25 sensing objects are calculated in the 5 major dangerous release working conditions, the sensing objects are in normal states, the vehicle is required to be identified within the specified time, each object is identified wrongly or not identified for 4 minutes, the lowest score is 0, and a character result is output after identification is finished.

In the present invention, the "dangerous release condition" refers to a situation of releasing danger, and the total of 5 categories are: collision with pedestrians, animals, straight-driving vehicles, vehicles at intersections, and cyclists.

For the sake of easy understanding, the description will be given taking the case of collision with a pedestrian as an example.

Dangerous release working condition: a collision occurs with a pedestrian.

Road scene: various roads.

The vehicle state: the main vehicle moves straight on the road at a certain speed (V), a static vehicle is arranged beside the road at the crossroad in front, the sight of the driver is shielded by the static background vehicle beside the road, and the main vehicle continues to move straight.

The pedestrian state: when the main vehicle reaches the crossroad, the pedestrian passes through the road, and the main vehicle collides with the pedestrian at the crossroad.

The evaluation method comprises the following steps: the length of a scene road section is 10 meters, a curtain covers the scene 50 meters away, the vehicle runs towards the scene direction at the speed of 40km/h, timing is started after collision, and the number of correct identification signal lamps, marks, other traffic participants and the like of the vehicle is recorded.

The dynamic environment perception stability of the dangerous release working condition is as follows: the automatic driving vehicle can identify the object in the normal driving process, and can still accurately identify the stability degree of the object when the parameters (such as the brightness intensity of a signal lamp, the angle of a traffic sign, the color and the like) of the identification object are changed or abnormal under the condition of releasing danger. Dynamic environment perception stability HD under dangerous release working condition_RWAScoring rules: the normal running evaluation vehicle respectively appears in front of the vehicle for a certain distance (20 meters): the 5 major dangerous release working conditions total 25 sensing objects, the sensing objects are in abnormal states (different from normal states in illumination, orientation, background and the like), the vehicle is required to be identified within a specified time, each object is mistakenly identified or not identified, 4 points are deducted, the lowest point is 0 point, and a character result is output after identification is finished.

The dynamic environment perception timeliness of the danger release working condition refers to that: autonomous vehicles accurately identify the timeliness of objects (e.g., traffic lights, static traffic signs, other traffic participants) during normal driving, in the event of a released hazard. Dynamic environment awareness timeliness HD for dangerous release conditions_TWAScoring rules: the normal running evaluation vehicle respectively appears in front of the vehicle for a certain distance (20 meters): the 5 major dangerous release working conditions total 25 sensing objects, the vehicle is required to be identified within the specified time (0.5s), each object is identified wrongly or not identified for 4 points, the lowest point is 0 point, and a character result is output after the identification is finished.

And acquiring a network communication parameter L of the automatic driving vehicle.

In order to enable the automatic driving vehicle to be better communicated with other vehicles and road side equipment, the control center can better provide service for the vehicle, and the intelligent networked automobile has good network communication capacity. The network connectivity parameters of the autonomous vehicle are used to evaluate the ability of the autonomous vehicle to communicate with other vehicles and roadside devices.

The network connection parameters of the automatic driving vehicle comprise static network connection evaluation parameters and dynamic network connection evaluation parameters.

The static network connection evaluation parameters mainly evaluate network connection aiming at the ad hoc network capability, the communication range, the bandwidth and the adaptive capacity of signal strength of the automatic driving vehicle. During evaluation, 1000 test response inputs are required to be generated according to the full operation range of the system in the four directions, no other interference signals are ensured, 1000 test response inputs are generated and matched with the interference signals, and evaluation is performed according to the communication receiving condition of the system.

The dynamic network communication evaluation parameter key evaluates the dynamic network communication capability of the intelligent networked automobile. During evaluation, the network communication capacity of different working conditions (such as constantly changing weather and communication states (such as signal strength, distance and the like) under the road) is evaluated through the vehicle in a moving state.

Before determining the network connectivity parameter L of the autonomous vehicle, the following parameters need to be determined: network connection accuracy evaluation parameter LN and network connection stability evaluation parameter LW_WTimeliness evaluation parameter LW communicated with network_A. Wherein, the network connection accuracy evaluation parameter LN and the network connection stability evaluation parameter LW_WTimeliness evaluation parameter LW communicated with network_ARespectively determined by the static network connection evaluation parameter and the dynamic network connection evaluation parameter which respectively correspond to the static network connection evaluation parameter and the dynamic network connection evaluation parameter. Here, the dynamic network connectivity evaluation parameter includes: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters. Specifically, the static network connectivity evaluation parameters include: static network connectivity accuracy LJ_CStability of static network link_RTimeliness LJ for communicating with static network_T. The dynamic network communication evaluation parameters comprise: normal driving condition dynamic network communication accuracy LD_CNDynamic network communication stability LD under normal driving condition_RNDynamic network communication timeliness LD under normal driving condition_TN(ii) a Dangerous working condition dynamic network communication accuracy LD_CWWDynamic network communication stability LD for dangerous working condition_RWWDynamic network communication timeliness LD for dangerous working condition_TWW(ii) a Dangerous release working condition dynamic network communication accuracy LD_CWAThe communication of the dangerous release working condition dynamic network is stableFixed-scale LD_RWADynamic network communication timeliness LD for dangerous release working condition_TWA。

Therefore, the method for evaluating the behavior safety of the automatic driving vehicle, provided by the embodiment of the invention, can be used for calculating the network communication parameter L of the automatic driving vehicle by combining the static network communication evaluation parameter and the dynamic network communication evaluation parameter, comprehensively considering the behavior safety of the automatic driving vehicle from multiple angles, and effectively improving the accuracy and reliability of an evaluation result.

After obtaining the parameters, obtaining a network connectivity parameter L of the autonomous vehicle according to the following expression:

L＝LN×0.4+LW_W×0.4+LW_A×0.2 (2)

the parameters in formula (2) are obtained using the following formulae:

LN＝LC_N×0.4+LR_N×0.3+LT_N×0.3

LW_W＝LC_WW×0.4+LR_WW×0.3+LT_WW×0.3

LW_A＝LC_WA×0.4+LR_WA×0.3+LT_WA×0.3

wherein LC_NEvaluation of accuracy of network connectivity parameters, LR, for normal driving conditions_NEvaluation of network connectivity accuracy parameters for dangerous conditions, LT_NEvaluation of network connectivity accuracy parameters, LC, for hazardous release conditions_WWEvaluation of stability of network connectivity for normal driving conditions, LR_WWEvaluation of the stability of network connections for dangerous conditions, LT_WWEvaluation of network connectivity stability parameters, LC, for hazardous release conditions_WANetwork connectivity timeliness assessment parameter, LR, for normal driving conditions_WAEvaluation of parameters for network connectivity timeliness, LT, for conditions of dangerous occurrence_WAAnd evaluating parameters for the network connectivity timeliness under the condition of dangerous release.

the specific acquisition process of the parameters is as follows:

communicating static networks to an accuracy of LJ_CStability of static network link_RStatic network connectivity timeliness LJ_TAre set to 100 minutes, respectively.

The static network communication accuracy refers to: the accuracy with which a stationary autonomous vehicle receives a communication. Static network connectivity accuracy LJ_CScoring rules: a stationary evaluation vehicle occurs at a distance (20 m) in front of it: and (4) responding to the input of 1000 scenes without other interference signals by testing, requiring the vehicle to receive in a specified time, deducting 0.1 point for each scene when the information is received incorrectly or not received, and outputting a character result after the receiving is finished, wherein the lowest point is 0 point.

The static network communication stability is as follows: the stationary autonomous vehicle accurately receives the stability of the communication. Stability LJ of static network link_RScoring rules: a stationary evaluation vehicle occurs at a distance (20 m) in front of it: and (4) responding to the input scenes with interference signals by 1000 tests, requiring the vehicle to receive within a specified time, deducting 0.1 point from the fact that the information received by each scene is wrong or not received, and outputting a character result after the receiving is finished, wherein the lowest point is 0 point.

The timeliness of the static network connection refers to: the degree of timeliness to which a stationary autonomous vehicle accurately receives communications. Static network connectivity timeliness LJ_TScoring rules: a stationary evaluation vehicle occurs at a distance (20 m) in front of it: the 1000 scenes with test response input and no other interference signals require the vehicle to receive within the specified time (100ms), each scene receives information in error or receives no information by 0.1 point, the lowest point is 0 point, and a character result is output after the receiving is finished.

LD for dynamic network communication accuracy under normal driving condition_CNDynamic network communication stability LD under normal driving condition_RNDynamic network communication timeliness LD under normal driving condition_TNAre set to 100 minutes, respectively.

The accuracy of the dynamic network communication under the normal driving condition is as follows: during normal driving of an autonomous vehicle, the vehicle receives an accurate level of communication under different conditions (e.g., changing weather, communication conditions (e.g., signal strength, distance, etc.) on the road). Normal driving condition dynamic network communication accuracy LD_CNScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: 9 major types of normal driving conditions total 9 network communication scenes, no signal interference exists, the vehicle is required to receive the information within the specified time, 15 points are deducted for each scene receiving information error or not receiving the information, the lowest point is 0 point, and a character result is output after the receiving is finished.

For ease of understanding, the straight-ahead section is described here as an example.

Normal driving conditions: a straight road section.

Description of the environment: the road side of the straight line section of the two-way two-lane is provided with traffic signs and other traffic participants (with communication capability), and the road side is provided with communication facilities and can be provided with interference equipment.

The evaluation method comprises the following steps: the length of a scene road section is 10 meters, a curtain covers the scene 50 meters away, the vehicle runs towards the scene direction at the speed of 40km/h, the timing is started after the curtain, and the condition that the vehicle receives information is recorded.

The dynamic network communication stability under normal driving conditions refers to: during normal driving of an autonomous vehicle, the vehicle accurately receives a stable degree of communication under different conditions (e.g., changing weather, communication conditions (e.g., signal strength, distance, etc.) on the road). Dynamic network communication stability LD under normal driving condition_RNScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: 9 general normal driving conditions total 9 network communication scenes, signal interference exists, the vehicle is required to receive within a specified time, and each scene receivesAnd (4) if the information is wrong or the information is not received, deducting 15 points, wherein the lowest point is 0 point, and outputting a character result after the information is received.

The timeliness of the dynamic network communication under the normal driving condition refers to that: during normal driving of an autonomous vehicle, the vehicle accurately receives the timeliness of communications under different conditions (e.g., changing weather, communication conditions (e.g., signal strength, distance, etc.) under the road). Dynamic network communication timeliness LD under normal driving condition_TNScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: 9 general normal driving conditions total 9 network communication scenes, no signal interference exists, the vehicle is required to receive within the specified time (0.5s), the information received by each scene is wrong or 15 points are deducted if the information is not received, the lowest point is 0 point, and a character result is output after the receiving is finished.

Dynamic network communication accuracy LD for dangerous working condition_CWWDynamic network communication stability LD for dangerous working condition_RWWDynamic network communication timeliness LD for dangerous working condition_TWWAre set to 100 minutes, respectively.

The accuracy of the communication of the dynamic network under the dangerous working condition refers to that: the automatic driving vehicle has signal interference and the accuracy of the vehicle receiving communication under dangerous working conditions in the normal driving process. Dangerous working condition dynamic network communication accuracy LD_CWWScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 32 major dangerous working conditions total 46 network communication scenes, no signal interference exists, the vehicle is required to receive in the specified time, the information received by each scene is wrong or not received, the lowest score is 3, and the character result is output after the receiving is finished.

The dynamic network communication stability of the dangerous working condition refers to: the automatic driving vehicle has signal interference and the stability of accurate receiving communication of the vehicle under dangerous working conditions in the normal driving process. Dynamic network communication stability LD for dangerous working condition_RWWScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 32 major dangerous working conditions total 46 network communication scenes with signal interference and vehicle requirementAnd (4) receiving the information in a specified time by the vehicle, deducting 3 points when the information received by each scene is wrong or not received, and outputting a character result after the receiving is finished, wherein the lowest point is 0 point.

The timeliness of the dynamic network communication under the dangerous working condition refers to: the automatic driving vehicle has signal interference and the timely degree of the vehicle for accurately receiving communication under dangerous working conditions in the normal driving process. Dynamic network communication timeliness LD for dangerous working condition_TWWScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 32 major dangerous working conditions total 46 network communication scenes, no signal interference exists, the vehicle is required to receive within the specified time (0.5s), the information received by each scene is wrong or not received, the minimum is 0 minute, and a character result is output after the receiving is finished.

LD for dynamic network communication accuracy under dangerous release condition_CWADynamic network communication stability LD for dangerous release working condition_RWADynamic network communication timeliness LD for dangerous release working condition_TWAAre set to 100 minutes, respectively.

The accuracy of the dynamic network communication under the danger release condition is as follows: in the normal driving process of the automatic driving vehicle, under the condition of releasing danger, signal interference exists, and the accuracy of the vehicle receiving communication is high. Dangerous release working condition dynamic network communication accuracy LD_CWAScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 5 major dangerous release working conditions total 5 network communication scenes, no signal interference exists, the vehicle is required to receive in the specified time, the information received by each scene is wrong or not received, 20 points are deducted, the lowest point is 0 point, and the character result is output after the receiving is finished.

The dynamic network communication stability under the dangerous release working condition is as follows: in the normal driving process of the automatic driving vehicle, under the condition of releasing danger, signal interference exists, and the vehicle accurately receives the stability of communication. Dynamic network communication stability LD under dangerous release working condition_RWAScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 5 major dangerous release working conditions total 25 network communication scenes, have signal interference and require the vehicle to be in the specified timeAnd (4) receiving within the time, wherein the received information of each scene is wrong or is not received, the lowest score is 0, and the character result is output after the receiving is finished.

The timeliness of the dynamic network communication under the danger release working condition refers to that: in the normal driving process of the automatic driving vehicle, under the condition of releasing danger, signal interference exists, and the vehicle accurately receives the timely degree of communication. Dynamic network communication timeliness LD under dangerous release working condition_TWAScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 5 major dangerous release working conditions total 5 network communication scenes, no signal interference exists, the vehicle is required to receive within the specified time (0.5s), the information received by each scene is wrong or not received, 20 points are deducted, the lowest point is 0 point, and the character result is output after the receiving is finished.

And acquiring a judgment decision parameter P of the automatic driving vehicle.

Under the condition that the automatic driving vehicle has the environment sensing capability and the network communication capability, the automatic driving vehicle also has the capability of accurately judging and making a decision. The judgment decision parameters of the automatic driving vehicle are used for evaluating the accuracy of judgment decisions made on different working conditions in the automatic driving behavior process of the automatic driving vehicle.

The decision-making parameters for automatic driving of the vehicle include static decision-making evaluation parameters and dynamic decision-making evaluation parameters.

The static judgment decision evaluation parameter mainly evaluates the vehicle running condition judgment capability, the road condition judgment capability and the decision capability. By evaluating the ability of the vehicle to determine that control should be taken (acceleration, deceleration, uniform speed, steering, etc.), the ability to determine the stopping position of the vehicle, the ability to determine the trajectory of the vehicle, the ability to determine what condition the road is in, etc. During evaluation, 1000 scenes that other traffic participants keep the original driving intention and 1000 scenes that other traffic participants can change the original driving intention are generated in a laboratory environment for the automobile to judge and make a decision, and whether the evaluation is carried out safely or correctly according to a decision result of the automobile.

The dynamic judgment decision evaluation parameter mainly evaluates the dynamic judgment decision capability of the intelligent networked automobile, and evaluates the judgment decision capability of different working conditions through the motion state of the automobile.

Before determining the decision parameter P for an autonomous vehicle, the following parameters need to be determined: judging decision accuracy evaluation parameter PN and judging decision stability evaluation parameter PW_WAnd judging a decision timeliness evaluation parameter PW_A. Judging decision accuracy evaluation parameter PN and judging decision stability evaluation parameter PW_WAnd judging a decision timeliness evaluation parameter PW_ARespectively determined by the static judgment decision evaluation parameter and the dynamic judgment decision evaluation parameter which respectively correspond to the static judgment decision evaluation parameter and the dynamic judgment decision evaluation parameter, wherein the dynamic judgment decision evaluation parameter comprises the following components: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters. Specifically, the static judgment decision evaluation parameters include: static judgment decision accuracy PJ_CStatic decision stability PJ_RAnd static decision timeliness PJ_T. The dynamic judgment of the decision evaluation parameters comprises the following steps: PD (proportion differentiation) for dynamically judging decision accuracy under normal driving condition_CNDynamic judgment decision stability PD of normal driving condition_RNDynamic judgment decision timeliness PD of normal driving condition_TN(ii) a PD (probability distribution) for dynamically judging decision accuracy of dangerous working condition_CWWDynamic judgment decision stability PD of dangerous working condition_RWWDynamic judgment decision timeliness PD of dangerous occurrence working condition_TWW(ii) a PD (probability distribution) for dynamically judging decision accuracy of dangerous release working condition_CWADynamic judgment decision stability PD of dangerous release working condition_RWADynamic judgment decision timeliness PD of dangerous release working condition_TWA。

Therefore, the evaluation method for automatically driving the vehicle behavior safety in the embodiment of the invention can calculate the judgment decision parameter P of the automatically driving vehicle by combining the static judgment decision evaluation parameter and the dynamic judgment decision evaluation parameter, comprehensively consider the behavior safety of the automatically driving vehicle from multiple angles, and effectively improve the accuracy and reliability of the evaluation result.

After obtaining the above parameters, a decision parameter P for the autonomous vehicle is obtained according to the following expression:

P＝PN×0.2+PW_W×0.6+PW_A×0.2 (3)

the parameters in formula (3) are obtained using the following formulae:

PN＝PC_N×0.5+PR_N×0.3+PT_N×0.2

PW_W＝PC_WW×0.5+PR_WW×0.3+PT_WW×0.2

PW_A＝PC_WA×0.5+PR_WA×0.3+PT_WA×0.2

wherein, PC_NDetermining a decision accuracy evaluation parameter, PR, for normal driving conditions_NJudging a decision accuracy evaluation parameter, PT, for a dangerous occurrence condition_NEvaluation parameters, PC, for determining the decision accuracy for the conditions of danger release_WWDetermining a decision stability evaluation parameter, PR, for normal driving conditions_WWJudging a decision stability evaluation parameter, PT, for a dangerous occurrence condition_WWEvaluation parameters, PC, for determining the stability of a decision for a hazardous release situation_WADecision-making timeliness evaluation parameter, PR, for normal driving conditions_WADeciding a decision-making timeliness evaluation parameter, PT, for a dangerous occurrence condition_WAAnd judging and deciding timeliness evaluation parameters for the dangerous release working condition.

the specific acquisition process of the parameters is as follows:

determine accuracy PJ of the decision statically_CStatic decision stability PJ_RStatic decision timeliness PJ_TRespectively of initial value ofSet to 100 minutes.

The static judgment decision accuracy refers to: the accuracy of judgment and decision making of the static automatic driving vehicle under different working conditions is improved. Static judgment decision accuracy PJ_CScoring rules: a stationary evaluation vehicle occurs at a distance (20 m) in front of it: 1000 other traffic participants keep the original scene of the driving intention, the vehicle is required to make a judgment decision within a specified time, the judgment decision of each scene is wrong or not, 0.1 point is deducted, the lowest point is 0 point, and a character result is output after the judgment decision is made.

It should be noted that the setting of the other traffic participants herein is to keep the original driving intention, so as to avoid the influence of the traffic participants' own factors on the decision making of the automatic driving vehicle.

The static judgment decision stability is as follows: when the driving intention scene changes, the stability degree of the accurate judgment decision is made by the static automatic driving vehicle. Static decision stability PJ_RScoring rules: a stationary evaluation vehicle occurs at a distance (20 m) in front of it: 1000 other traffic participants change the scene of the original driving intention, the vehicle is required to make a judgment decision within a specified time, the judgment decision of each scene is wrong or not, 0.1 point is deducted, the lowest point is 0 point, and a character result is output after the judgment decision is made.

It should be noted that, it is set here that other traffic participants change the original driving intention, so as to avoid the influence of the traffic participants' own factors on the decision making of the automatic driving vehicle.

The static judgment decision timeliness refers to: the degree of timeliness of the accurate judgment decision of the static automatic driving vehicle on different working conditions is made. Static decision timeliness PJ_TScoring rules: a stationary evaluation vehicle occurs at a distance (20 m) in front of it: 1000 other traffic participants keep the original scene of the driving intention, the vehicle is required to make a judgment decision within the specified time (100ms), each scene is wrong in judgment decision or is not decided, 0.1 point is deducted, the lowest point is 0 point, and a character result is output after the judgment decision is made.

Dynamically judging decision accuracy PD of normal driving condition_CNDynamic judgment decision stability PD of normal driving condition_RNDynamic judgment decision timeliness PD of normal driving condition_TNAre set to 100 minutes, respectively.

The decision accuracy of the dynamic judgment of the normal driving condition is as follows: and the accuracy of judgment and decision making is carried out on different scenes in the normal running process of the automatic driving vehicle. PD (proportion differentiation) for dynamically judging decision accuracy under normal driving condition_CNScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 9 kinds of normal driving conditions total 9 judgment decision scenes, other traffic participants keep the original driving intentions and require the vehicle to make judgment decisions within the specified time, 15 points are deducted if the judgment decision of each scene is wrong or not, the lowest point is 0 point, and character results are output after the judgment decision is made.

The dynamic judgment of the decision stability under the normal driving condition refers to the following steps: and in the normal running process of the automatic driving vehicle, when the scene changes, the automatic driving vehicle makes the stability of accurate judgment decision. Dynamic decision stability PD (proportion differentiation) judgment under normal driving condition_RNScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 9 kinds of normal driving conditions total 9 judgment decision scenes, other traffic participants change the original driving intentions and require the vehicle to make judgment decisions within the specified time, 15 points are deducted if the judgment decision of each scene is wrong or not, the lowest point is 0 point, and character results are output after the judgment decision is made.

The dynamic judgment decision timeliness of the normal driving condition refers to: and in the normal running process of the automatic driving vehicle, the timeliness of making accurate judgment decisions on different scenes. Dynamic judgment decision timeliness PD (probability distribution) of normal driving condition_TNScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: 9 general normal driving conditions total 9 judgment decision scenes, other traffic participants keep the original driving intention and require vehiclesAnd (4) performing judgment decision within the specified time (0.5s), deducting 15 points when each scene is judged to be wrong or not judged, and outputting a character result after the judgment decision is made, wherein the lowest point is 0 point.

PD for dynamically judging decision accuracy of dangerous working condition_CWWDynamic judgment decision stability PD of dangerous working condition_RWWDynamic judgment decision timeliness PD of dangerous occurrence working condition_TWWAre set to 100 minutes, respectively.

The dynamic judgment decision accuracy of the dangerous working condition refers to: and in the normal running process of the automatic driving vehicle, under the working condition that danger occurs, the accuracy of judgment decision is made. PD (probability distribution) for dynamically judging decision accuracy of dangerous working condition_CWWScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the total of 46 judgment decision scenes exist in the 32 major dangerous working conditions, other traffic participants keep the original driving intentions and require the vehicle to make judgment decisions within the specified time, the judgment decision of each scene is wrong or not, the minimum is 0 point, and a character result is output after the judgment decision is made.

The dynamic judgment of the decision stability of the dangerous working condition refers to the following steps: and in the normal running process of the automatic driving vehicle, making a stable degree of an accurate judgment decision under the working condition of danger. Dynamic decision stability PD (probability distribution) judgment of dangerous working condition_RWWScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the total of 46 judgment decision scenes exist in the 32 major dangerous working conditions, other traffic participants change the original driving intentions and require the vehicle to make judgment decisions within the specified time, the judgment decision of each scene is wrong or not, the minimum is 0 point, and a character result is output after the judgment decision is made.

The dynamic judgment decision timeliness of the dangerous working condition refers to: the timely degree of an accurate judgment decision is made under the working condition that danger occurs in the normal running process of the automatic driving vehicle. Dynamic judgment decision timeliness PD for dangerous working condition_TWWScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: 46 judgment decision fields in total for 32 major dangerous working conditionsAnd (3) other traffic participants keep the original driving intention, the vehicle is required to make a judgment decision within the specified time (0.5s), the judgment decision of each scene is wrong or not is deducted by 3 points, the lowest point is 0 point, and a character result is output after the judgment decision is made.

Dynamic judgment decision accuracy PD of danger release working condition_CWADynamic judgment decision stability PD of dangerous release working condition_RWADynamic judgment decision timeliness PD of dangerous release working condition_TWAAre set to 100 minutes, respectively.

The decision accuracy of the dynamic judgment of the danger release working condition is as follows: and in the normal running process of the automatic driving vehicle, under the condition of releasing danger, making the accuracy of judgment decision. PD (probability distribution) for dynamically judging decision accuracy of dangerous release working condition_CWAScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 5 major dangerous release working conditions total 5 judgment decision scenes, other traffic participants keep the original driving intentions and require the vehicle to make judgment decisions within the specified time, the judgment decision of each scene is wrong or not, 20 points are deducted, the lowest point is 0 point, and character results are output after the judgment decisions are made.

The dynamic judgment of the decision stability of the dangerous release working condition refers to that: and in the normal running process of the automatic driving vehicle, making a stable degree of an accurate judgment decision under the condition of releasing danger. Dynamic decision stability PD (probability distribution) judgment under dangerous release working condition_RWAScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 5 major dangerous release working conditions total 5 judgment decision scenes, other traffic participants change the original driving intentions and require the vehicle to make judgment decisions within the specified time, the judgment decision of each scene is wrong or not, 20 points are deducted, the lowest point is 0 point, and character results are output after the judgment decisions are made.

The dynamic judgment decision timeliness of the danger release working condition refers to that: in the normal running process of the automatic driving vehicle, under the condition of releasing danger, the timeliness degree of an accurate judgment decision is made. Dynamic judgment decision timeliness PD (probability distribution) of dangerous release working condition_TWAScoring rules: evaluating a distance ahead of the vehicle for normal travel (20)Rice) appeared: the 5 major dangerous release working conditions total 5 judgment decision scenes, other traffic participants keep the original driving intention and require the vehicle to make a judgment decision within the specified time (0.5s), 20 points are deducted if the judgment decision of each scene is wrong or not, the lowest point is 0 point, and a character result is output after the judgment decision is made.

A control execution parameter K of an autonomous vehicle is acquired.

The autonomous vehicle must have accurate control execution capability to convert environment perception, network connectivity, judgment decisions into accurate and safe driving behaviors of the vehicle on the road. The control execution parameter of the autonomous vehicle is used to evaluate the control execution performance of the autonomous vehicle.

The control execution parameters of the autonomous vehicle include a static control execution evaluation parameter and a dynamic control execution evaluation parameter.

The static control execution evaluation parameters mainly comprise: the response speed and the accuracy of the actuating mechanism and the total reaction time of the system are mainly evaluated aiming at the control execution performance of a brake-by-wire system, a drive-by-wire system and a steer-by-wire system of an intelligent networked automobile. During evaluation, 1000 test response inputs are required to be generated uniformly according to the full operation range of each system and the load is kept constant, and in addition, 1000 test response inputs are generated and matched with the change of the load, and whether the control execution result of the system is suitable or correct is evaluated.

The dynamic control execution evaluation parameters mainly evaluate the dynamic control execution capacity of the intelligent networked automobile, and the control execution capacities under different working conditions are evaluated in a moving state of the automobile. By evaluating the response time and accuracy of the vehicle motion state (e.g., the position and time of stopping, the time of acceleration, the accuracy of steering, etc.), the accuracy of trajectory maintenance, the rate of completion of the maneuver, the time taken from sensing to performing, etc.

Before determining the control execution parameter K of the autonomous vehicle, the following parameters need to be determined: control execution accuracy evaluation parameter KN and control execution stability evaluation parameter KW_WAnd controlling execution of timeliness evaluation parameter KW_A. Control execution accuracyEvaluation parameter KN, control execution stability evaluation parameter KW_WAnd controlling execution of timeliness evaluation parameter KW_ARespectively determining by the static control execution evaluation parameters and the dynamic control execution evaluation parameters which respectively correspond to the static control execution evaluation parameters and the dynamic control execution evaluation parameters, wherein the dynamic control execution evaluation parameters comprise: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters. Specifically, the static control execution evaluation parameters include: static control execution accuracy KJ_CStatic control execution stability KJ_RAnd static control execution timeliness KJ_T. The dynamic control execution evaluation parameters include: accuracy KD of dynamic control execution under normal driving condition_CNDynamic control execution stability KD under normal driving condition_RNDynamic control execution timeliness KD of normal driving condition_TN(ii) a Dynamic control execution accuracy KD of dangerous working condition_CWWDynamic control execution stability KD of dangerous working condition_RWWDynamic control execution timeliness KD for dangerous working condition_TWW(ii) a Dynamic control execution accuracy KD of dangerous release working condition_CWADynamic control execution stability KD of dangerous release working condition_RWADynamic control execution timeliness KD of dangerous release working condition_TWA。

Therefore, the method for evaluating the behavior safety of the automatic driving vehicle, provided by the embodiment of the invention, can be used for calculating the control execution parameter K of the automatic driving vehicle by combining the static control execution evaluation parameter and the dynamic control execution evaluation parameter, comprehensively considering the behavior safety of the automatic driving vehicle from multiple angles, and effectively improving the accuracy and reliability of an evaluation result.

After obtaining the above parameters, a control execution parameter K of the autonomous vehicle is obtained according to the following expression:

K＝KN×0.4+KW_W×0.4+KW_A×0.2 (4)

the parameters in formula (4) are obtained using the following formulae:

KN＝KC_N×0.4+KR_N×0.3+KT_N×0.3

KW_W＝KC_WW×0.4+KR_WW×0.3+KT_WW×0.3

KW_A＝KC_WA×0.4+KR_WA×0.3+KT_WA×0.3

wherein, KC_NFor controlling the execution of an accuracy evaluation parameter, KR, for normal driving conditions_NControl of the execution of an accuracy evaluation parameter, KT, for dangerous conditions_NEvaluation parameter, KC, for controlling execution accuracy for dangerous release conditions_WWPerforming stability evaluation parameters for normal driving condition control, KR_WWControl of the execution of stability evaluation parameters, KT, for dangerous conditions_WWControl of the execution stability evaluation parameter, KC, for hazardous release conditions_WAPerforming timeliness assessment parameters for normal driving condition control, KR_WAFor the control of the execution of timeliness evaluation parameters, KT, for dangerous conditions_WAAnd controlling and executing the timeliness evaluation parameters for the dangerous release working condition.

the specific acquisition process of the parameters is as follows:

execute static control with accuracy KJ_CStatic control execution stability KJ_RStatic control execution timeliness KJ_TAre set to 100 minutes, respectively.

The static control execution accuracy means: the accuracy with which a stationary autonomous vehicle performs scene control. Static control execution accuracy KJ_CScoring rules: a stationary evaluation vehicle occurs at a distance (20 m) in front of it: 1000 tests responding to the input and keeping the load constant, require the vehicle to be at the prescribedAnd controlling and executing within time, deducting 0.1 point from the control execution error or non-execution of each scene, and outputting a character result after the control execution, wherein the minimum is 0 point.

The static control execution stability means: when a scene change occurs, the stationary autonomous vehicle performs the correct degree of stability on the scene control. Static control execution stability KJ_RScoring rules: a stationary evaluation vehicle occurs at a distance (20 m) in front of it: the 1000 tests respond to the input and are matched with the change of the load, the vehicle is required to be controlled and executed within the specified time, the control execution error or non-execution of each scene is deducted by 0.1 minute, the lowest is 0 minute, and the character result is output after the control execution.

The static control execution timeliness refers to: a stationary autonomous vehicle performs the correct degree of timeliness for scene control. Static control execution timeliness KJ_TScoring rules: a stationary evaluation vehicle occurs at a distance (20 m) in front of it: the 1000 test responses to the input and maintains the constant load scene, the vehicle is required to be controlled and executed within the specified time (100ms), each scene is controlled and executed wrongly or not executed for 0.1 minute, the lowest time is 0 minute, and the character result is output after the control is executed.

Dynamic control of normal driving condition to execute accuracy KD_CNDynamic control execution stability KD under normal driving condition_RNDynamic control execution timeliness KD of normal driving condition_TNAre set to 100 minutes, respectively.

The accuracy of executing the dynamic control under the normal driving condition refers to that: and controlling the execution accuracy of different scenes in the normal running process of the automatic driving vehicle. Accuracy KD of dynamic control execution under normal driving condition_CNScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: 9 general normal driving conditions total 9 control execution scenes, the road surface condition is not changed, the vehicle is required to be controlled and executed within a specified time, the control execution error or non-execution of each scene is deducted for 15 minutes, the lowest point is 0 minute, and a character result is output after the control execution.

The dynamic control execution stability of the normal driving condition refers to that: the autonomous vehicle is running normallyIn the process, when the scene changes, the control executes the correct stability degree. Dynamic control execution stability KD under normal driving condition_RNScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: 9 general normal driving conditions total 9 control execution scenes, the road surface condition has sudden change, the vehicle is required to be controlled and executed within a specified time, the control execution error or non-execution of each scene is deducted for 15 minutes, the lowest point is 0 minute, and a character result is output after the control execution.

The timeliness of the dynamic control execution of the normal driving condition refers to that: the automatic driving vehicle controls and executes correct timeliness degree on different scenes in the normal driving process. Dynamic control execution timeliness KD under normal driving condition_TNScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: 9 general normal driving conditions total 9 control execution scenes, the road surface condition is not changed, the vehicle is required to be controlled and executed within the specified time (0.5s), the control execution error or non-execution of each scene is deducted for 15 minutes, the lowest time is 0 minute, and the character result is output after the control execution.

Dynamic control execution accuracy KD for dangerous working condition_CWWDynamic control execution stability KD of dangerous working condition_RWWDynamic control execution timeliness KD for dangerous working condition_TWWAre set to 100 minutes, respectively.

The execution accuracy of the dynamic control of the dangerous working condition refers to that: the accuracy of execution is controlled under dangerous working conditions in the normal running process of the automatic driving vehicle. Dynamic control execution accuracy KD of dangerous working condition_CWWScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 32 major dangerous working conditions are 46 control execution scenes in total, the road surface condition is not changed, the vehicle is required to be controlled and executed within the specified time, the control execution error or non-execution of each scene is deducted for 3 minutes, the lowest point is 0 minute, and the character result is output after the control execution.

The dynamic control execution stability of the dangerous working condition refers to: and controlling and executing the correct stability degree of the automatic driving vehicle under the dangerous working condition in the normal driving process.Dynamic control execution stability KD for dangerous working condition_RWWScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the total of 46 control execution scenes under the working conditions of 32 major dangers, the road surface condition has sudden change, the vehicle is required to be controlled and executed within the specified time, the control execution error or non-execution of each scene is deducted by 3 points, the lowest point is 0 point, and the character result is output after the control execution.

The timeliness of the dynamic control execution of the dangerous working condition refers to: the automatic driving vehicle controls and executes correct timely degree under dangerous working conditions in the normal driving process. Dynamic control execution timeliness KD for dangerous working condition_TWWScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 32 major dangerous working conditions are 46 control execution scenes in total, the road surface condition is not changed, the vehicle is required to be controlled and executed within the specified time (0.5s), the control execution error or non-execution of each scene is deducted for 3 minutes, the lowest point is 0 minute, and the character result is output after the control execution.

Dynamic control execution accuracy KD of danger release working condition_CWADynamic control execution stability KD of dangerous release working condition_RWADynamic control execution timeliness KD of dangerous release working condition_TWAAre set to 100 minutes, respectively.

The execution accuracy of the dynamic control of the danger release working condition is as follows: the accuracy of the execution of the automated driving vehicle is controlled during normal driving, in the case of a released danger. Dynamic control execution accuracy KD of dangerous release working condition_CWAScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 5 major dangerous release working conditions total 5 control execution scenes, the road surface condition is not changed, the vehicle is required to be controlled and executed within the specified time, the control execution error or non-execution of each scene is deducted for 20 minutes, the lowest point is 0 minute, and the character result is output after the control execution.

The dynamic control execution stability of the dangerous release working condition refers to that: during normal driving of the autonomous vehicle, the control executes a correct degree of stability in the event of a released risk. Dynamic control execution stability KD of dangerous release working condition_RWAScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 5 major dangerous release working conditions total 5 control execution scenes, the road surface condition has sudden change, the vehicle is required to be controlled and executed within the specified time, the control execution error or non-execution of each scene is deducted for 20 minutes, the lowest point is 0 minute, and the character result is output after the control execution.

The timeliness of the dynamic control execution of the danger release working condition refers to that: in the normal running process of the automatic driving vehicle, the correct timely degree is controlled and executed under the condition of releasing danger. Dynamic control execution timeliness KD for dangerous release working condition_TWAScoring rules: the normal running evaluation occurs a certain distance (20 meters) ahead of the vehicle: the 5 major dangerous release working conditions total 5 control execution scenes, the road surface condition is not changed, the vehicle is required to be controlled and executed within the specified time (0.2s), the control execution error or non-execution of each scene is deducted for 20 minutes, the lowest point is 0 minute, and the character result is output after the control execution.

And determining the networking degree coefficients a and b according to the networking degree of the intelligent networking automobile.

Then, step S102 is executed. Based on the environment perception parameter H, the network communication parameter L, the judgment decision parameter P, the control execution parameter K and the network connection degree coefficients a and b, obtaining an evaluation score Z of the behavior safety of the automatic driving vehicle according to the following expression:

in the expression (5), the specific value of the networking degree coefficient a is determined by the networking degree of the smart networked automobile, and a + b is equal to 1.

Finally, step S103 is performed. And obtaining the evaluation result of the behavior safety of the automatic driving vehicle according to the evaluation score of the behavior safety of the automatic driving vehicle.

Preferably, the evaluation result of the behavior safety of the automatic driving vehicle is obtained according to the magnitude relation between the evaluation score of the behavior safety of the automatic driving vehicle and a preset score threshold value.

Specifically, when the evaluation score Z of the automated driving vehicle behavior safety is equal to or greater than a1, it is determined that the automated driving vehicle behavior has better safety; when the evaluation score Z of the behavior safety of the automatic driving vehicle is greater than or equal to B1 and less than A1, the behavior of the automatic driving vehicle is judged to have certain safety; when the evaluation score Z of the automated driving vehicle behavior safety is less than B1, it is determined that the automated driving vehicle behavior has poor safety.

Although the embodiments of the present invention have been described above, the above description is only for the convenience of understanding the present invention, and is not intended to limit the present invention. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

1. An evaluation method for behavioral safety of an autonomous vehicle, comprising:

an assessment score for the behavioral safety of the autonomous vehicle is obtained according to the following expression:

wherein Z is the evaluation score, H is the environment sensing parameter, L is the network connectivity parameter, P is the decision-making parameter, K is the control execution parameter, a and b are the network connectivity degree coefficients, and a + b is 1;

2. The method of claim 1, wherein the evaluation result of the automated driving vehicle behavior safety is obtained according to a magnitude relationship between the evaluation score and a preset score threshold.

3. The automated driving vehicle behavior safety assessment method according to any one of claims 1 to 2, characterized in that the environmental awareness parameter H of the automated driving vehicle is obtained according to the following expression:

H＝HN×0.4+HW_W×0.4+HW_A×0.2

4. The method of claim 3, wherein the environmental awareness accuracy evaluation parameter, the environmental awareness stability evaluation parameter, and the environmental awareness timeliness evaluation parameter are determined by a static environmental awareness evaluation parameter and a dynamic environmental awareness evaluation parameter, respectively, corresponding thereto, wherein the dynamic environmental awareness evaluation parameter comprises: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters.

5. The method for assessing behavioral safety of an autonomous vehicle according to any one of claims 1 to 2, characterized in that the network connectivity parameter L of the autonomous vehicle is obtained according to the following expression:

L＝LN×0.4+LW_W×0.4+LW_A×0.2

6. The method of claim 5, wherein the network connectivity accuracy evaluation parameter, the network connectivity stability evaluation parameter, and the network connectivity timeliness evaluation parameter are determined by a static network connectivity evaluation parameter and a dynamic network connectivity evaluation parameter respectively corresponding thereto, wherein the dynamic network connectivity evaluation parameter comprises: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters.

7. The evaluation method for behavioral safety of an autonomous vehicle according to any one of claims 1 to 2, characterized in that the judgment decision parameter P of the autonomous vehicle is obtained according to the following expression:

P＝PN×0.2+PW_W×0.6+PW_A×0.2

8. The method of claim 7, wherein the decision accuracy evaluation parameter, the decision stability evaluation parameter, and the decision timeliness evaluation parameter are determined by a static decision evaluation parameter and a dynamic decision evaluation parameter respectively corresponding thereto, wherein the dynamic decision evaluation parameter comprises: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters.

9. The evaluation method for automated driving vehicle behavior safety according to any one of claims 1 to 2, characterized in that the control execution parameter K of the automated driving vehicle is obtained according to the following expression:

K＝KN×0.4+KW_W×0.4+KW_A×0.2

10. The method of assessing autonomous-vehicle behavioral safety according to claim 9, wherein the control execution accuracy assessment parameter, the control execution stability assessment parameter and the control execution timeliness assessment parameter are respectively determined by a static control execution assessment parameter and a dynamic control execution assessment parameter corresponding thereto, wherein the dynamic control execution assessment parameter includes: normal driving condition evaluation parameters, danger occurrence condition evaluation parameters and danger release condition evaluation parameters.