CN116753938A - Vehicle test scene generation method, device, storage medium and equipment - Google Patents

Vehicle test scene generation method, device, storage medium and equipment Download PDF

Info

Publication number
CN116753938A
CN116753938A CN202310698303.4A CN202310698303A CN116753938A CN 116753938 A CN116753938 A CN 116753938A CN 202310698303 A CN202310698303 A CN 202310698303A CN 116753938 A CN116753938 A CN 116753938A
Authority
CN
China
Prior art keywords
vehicle
scene
risk
dangerous
environment
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310698303.4A
Other languages
Chinese (zh)
Inventor
李玉峰
秦武韬
王鹏
张浪
王振凯
李娜
王亚军
陶士昌
沈毅
贾宏颖
孔繁亮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Network Communication and Security Zijinshan Laboratory
Original Assignee
Network Communication and Security Zijinshan Laboratory
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Network Communication and Security Zijinshan Laboratory filed Critical Network Communication and Security Zijinshan Laboratory
Priority to CN202310698303.4A priority Critical patent/CN116753938A/en
Publication of CN116753938A publication Critical patent/CN116753938A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
    • G01C21/34—Route searching; Route guidance
    • G01C21/3407—Route searching; Route guidance specially adapted for specific applications
    • G01C21/343—Calculating itineraries, i.e. routes leading from a starting point to a series of categorical destinations using a global route restraint, round trips, touristic trips
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
    • G01C21/34—Route searching; Route guidance
    • G01C21/3453—Special cost functions, i.e. other than distance or default speed limit of road segments
    • G01C21/3492—Special cost functions, i.e. other than distance or default speed limit of road segments employing speed data or traffic data, e.g. real-time or historical
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/38—Electronic maps specially adapted for navigation; Updating thereof
    • G01C21/3804—Creation or updating of map data
    • G01C21/3807—Creation or updating of map data characterised by the type of data
    • G01C21/3815—Road data
    • G01C21/3822—Road feature data, e.g. slope data
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/38—Electronic maps specially adapted for navigation; Updating thereof
    • G01C21/3804—Creation or updating of map data
    • G01C21/3833—Creation or updating of map data characterised by the source of data
    • G01C21/3841—Data obtained from two or more sources, e.g. probe vehicles
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D21/00—Measuring or testing not otherwise provided for
    • G01D21/02—Measuring two or more variables by means not covered by a single other subclass

Landscapes

  • Engineering & Computer Science (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Medical Informatics (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a vehicle test scene generation method, a device, a storage medium and equipment. The method comprises the following steps: acquiring a plurality of scene elements of an automatic driving vehicle, wherein the types of the scene elements comprise vehicle elements, road side elements and cloud elements; dividing the driving environment of the autonomous vehicle into a risk environment and a general environment when the risk element exists in the plurality of scene elements; determining a detailed measurement scene set corresponding to the risk environment and a general measurement scene set corresponding to the general environment; and generating a target test scene set of the automatic driving vehicle based on the detailed test scene set and the general test scene set. The method solves the technical problems that the existing vehicle test scene generation method lacks vehicle-road cooperative elements, has huge number of test elements and has lower test efficiency.

Description

Vehicle test scene generation method, device, storage medium and equipment
Technical Field
The invention relates to the technical field of automatic driving safety test, in particular to a vehicle test scene generation method, a device, a storage medium and equipment.
Background
With the deep development of autopilot technology, some autopilot products have been put into commercial use in specific situations and conditions. The construction of the vehicle test scene is an important basis for carrying out the automatic driving test, is also a key factor for determining whether the automatic driving test is sufficient, is limited by factors such as computing power, space position, sensor level, climate conditions and the like, and the problem of long tail effect encountered by high-level automatic driving is continuously emerging, so that automatic driving under a plurality of edge scenes is still very difficult.
The existing automatic driving test scene still takes a vehicle under 'intelligent isolated vehicle' as a test object, consideration factors are still limited to a sensor, an executing mechanism, a road environment and the like of a bicycle, influences of external scene elements such as cloud state, road side state and the like are not fully considered, but as the scene elements are increased, the number of scenes to be tested increases exponentially, even if a simulation test method is adopted, test workload and test time are difficult to estimate, and the requirement of high-efficiency intelligent automobile automatic driving test under the cooperation of a vehicle and a road cannot be met.
In view of the above problems, no effective solution has been proposed at present.
Disclosure of Invention
The embodiment of the invention provides a vehicle test scene generation method, a device, a storage medium and equipment, which at least solve the technical problems that the existing vehicle test scene generation method lacks vehicle-road cooperative elements, has huge number of test elements and has lower test efficiency.
According to an aspect of an embodiment of the present invention, there is provided a vehicle test scene generating method including: acquiring a plurality of scene elements of an automatic driving vehicle, wherein the types of the scene elements comprise vehicle elements, road side elements and cloud elements; dividing the driving environment of the autonomous vehicle into a risk environment and a general environment when the risk element exists in the plurality of scene elements; determining a detailed measurement scene set corresponding to the risk environment and a general measurement scene set corresponding to the general environment; and generating a target test scene set of the automatic driving vehicle based on the detailed test scene set and the general test scene set.
Optionally, the acquiring a plurality of scene elements of the autonomous vehicle includes: determining a plurality of scene element levels of a vehicle test scene; and respectively carrying out two-stage hierarchical refinement or multi-stage hierarchical refinement on the scene element levels to obtain refined scene elements respectively corresponding to the scene element levels, and taking the refined scene elements respectively corresponding to the scene element levels as the scene elements of the automatic driving vehicle.
Optionally, the classifying the driving environment of the autonomous vehicle into a risk environment and a general environment based on the risk factor includes: enumerating similar risk elements similar to the risk elements; and determining a driving environment including the risk element or the similar risk element as the risk environment, and a driving environment excluding the risk element and the similar risk element as the general environment.
Optionally, the determining the detailed test scene set corresponding to the risk environment and determining the general test scene set corresponding to the general environment includes: and determining a detailed scene set corresponding to the risk environment based on a first granularity, and determining a general scene set corresponding to the general environment based on a second granularity, wherein the first granularity is smaller than the second granularity.
Optionally, the determining the set of general measurement scenes corresponding to the general environment includes: determining a dangerous background vehicle of the automatic driving vehicle at the current moment, wherein the background vehicle of the automatic driving vehicle comprises a common background vehicle and the dangerous background vehicle, and the dangerous background vehicle is the background vehicle with the highest collision probability; acquiring an initial state of the dangerous background vehicle at the current moment; performing maneuver sampling on the dangerous background vehicle based on a natural-risk countermeasure hybrid distribution corresponding to the initial state of the dangerous background vehicle at the current moment to obtain the maneuver state of the dangerous background vehicle at the moment next to the current moment, wherein the probability of collision when maneuver sampling is performed based on the natural-risk countermeasure hybrid distribution is larger than the probability of collision when maneuver sampling is performed based on maneuver distribution in the natural state; based on the maneuver distribution in the natural state, maneuver sampling is carried out on the common background vehicle to obtain the maneuver state of the common background vehicle at the moment next to the current moment; and determining a general survey scene set corresponding to the general environment based on the maneuvering state of the dangerous background vehicle at the moment next to the current moment and the maneuvering state of the common background vehicle at the moment next to the current moment.
Optionally, the determining the general survey scene set corresponding to the general environment based on the maneuvering state of the dangerous background vehicle at the moment next to the current moment and the maneuvering state of the common background vehicle at the moment next to the current moment includes: repeatedly obtaining the maneuvering states of the dangerous background vehicle and the common background vehicle at a plurality of moments within a preset time period respectively, and determining a general survey scene corresponding to the common environment based on the maneuvering states at the plurality of moments within the preset time period; repeating the above steps for a plurality of times to obtain the general survey scene set corresponding to the general environment.
Optionally, before the dangerous background vehicle is maneuver sampled based on the natural-risk countermeasure hybrid distribution corresponding to the initial state of the dangerous background vehicle at the current time to obtain the maneuver state of the dangerous background vehicle at the next time of the current time, the method further includes: acquiring collision probability distribution of the dangerous background vehicle in a natural state in an initial state at the current moment; acquiring maneuvering distribution of the vehicle in a natural state; and determining a natural-risk countermeasure mixture distribution corresponding to the initial state of the dangerous background vehicle at the current moment based on the collision probability distribution in the natural state and the maneuvering distribution in the natural state.
Optionally, the acquiring the collision probability distribution of the dangerous background vehicle in the natural state in the initial state at the current time includes: acquiring the maneuvering-collision probability distribution of the dangerous background vehicle in the initial state at the current moment; determining the collision probability distribution in the natural state based on the maneuver-collision probability distribution and the maneuver-in-natural state distribution.
Optionally, the acquiring the maneuver-collision probability distribution of the dangerous background vehicle in the initial state at the current time includes: and inputting the initial state of the dangerous background vehicle at the current moment into an anti-driving intelligent agent model to obtain the maneuvering-collision probability distribution of the dangerous background vehicle at the initial state of the current moment, wherein the anti-driving intelligent agent model is obtained based on collision training between a sample tested vehicle and the sample background vehicle.
Optionally, the determining, based on the collision probability distribution in the natural state and the maneuver distribution in the natural state, a natural-risk countermeasure hybrid distribution corresponding to an initial state of the dangerous background vehicle at the current time includes: and mixing the collision probability distribution in the natural state and the maneuvering distribution in the natural state according to a preset proportion to obtain a natural-risk countermeasure mixing distribution corresponding to the initial state of the dangerous background vehicle at the current moment.
Optionally, the determining the dangerous background vehicle of the autopilot vehicle at the current moment includes: under the condition that a plurality of background vehicles of the automatic driving vehicle are provided, respectively acquiring collision probability distribution of the plurality of background vehicles in a natural state corresponding to the current moment; the probabilities of the plurality of background vehicles in various maneuvering states in the collision probability distribution in the natural state corresponding to the current moment are summed to obtain a summed probability; and determining the background vehicle with the highest summation probability among the plurality of background vehicles as the dangerous background vehicle.
Optionally, the generating the target test scene set of the automatic driving vehicle based on the detail test scene set and the general test scene set includes: performing first redundancy elimination processing on a plurality of detail measurement scenes in the detail measurement scene set to obtain a detail measurement scene set after the first redundancy elimination processing; performing second redundancy elimination processing on a plurality of census scenes in the census scene set to obtain a census scene set after the second redundancy elimination processing; and obtaining the target test scene set of the automatic driving vehicle based on the detailed test scene set after the first redundancy elimination processing and the general test scene set after the second redundancy elimination processing.
Optionally, the first redundancy removing process and the second redundancy removing process include: determining equivalent scenes in a plurality of scenes, and removing redundant scenes by combining the equivalent scenes; determining sub-scenes in the membership scenes with membership in the scenes, and removing redundant scenes by removing the sub-scenes.
According to another aspect of the embodiments of the present invention, there is also provided a non-volatile storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform any one of the above-described vehicle test scenario generation methods.
According to another aspect of an embodiment of the present invention, there is also provided an electronic device including a memory, in which a computer program is stored, and a processor configured to run the computer program to perform any one of the above-described vehicle test scenario generation methods.
In the embodiment of the invention, a plurality of scene elements of an automatic driving vehicle are acquired, wherein the types of the scene elements comprise vehicle elements, road side elements and cloud elements; dividing the driving environment of the autonomous vehicle into a risk environment and a general environment when the risk element exists in the plurality of scene elements; determining a detailed measurement scene set corresponding to the risk environment and a general measurement scene set corresponding to the general environment; based on the detailed test scene set and the general test scene set, the target test scene set of the automatic driving vehicle is generated, a plurality of driving environments are divided into a risk environment and a general environment, scene merging is carried out under two types of environments respectively, and the purpose of determining the detailed test scene set and the general test scene set is achieved, so that the purposes of reducing redundant scenes while considering road side elements and cloud elements are achieved, the technical effect of meeting the requirement of efficiently testing the automatic driving of an intelligent automobile under the cooperation of a vehicle and a road is achieved, and the technical problems that the existing vehicle test scene generation method lacks the cooperation elements of the vehicle and has huge quantity of test elements and lower test efficiency are solved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute a limitation on the application. In the drawings:
FIG. 1 is a flow chart of a vehicle test scenario generation method according to an embodiment of the present application;
FIG. 2 is a schematic diagram of a detailed security awareness-based scenario generation process in an alternative risk environment according to an embodiment of the present application;
FIG. 3 is a schematic diagram of an alternate classification of autonomous vehicle driving according to an embodiment of the application;
FIG. 4 is a schematic illustration of an alternative longitudinal maneuver state interval partitioning according to an embodiment of the present application;
FIG. 5 is a schematic diagram of an alternative survey scene generation flow chart according to an embodiment of the present application;
FIG. 6 is a schematic diagram of an alternative natural-risk challenge mix profile generation in accordance with an embodiment of the present application;
FIG. 7 is an alternative natural state collision probability distribution generation schematic diagram in accordance with an embodiment of the application;
FIG. 8 is a schematic diagram of an alternative equivalent scenario in accordance with an embodiment of the present application;
FIG. 9 is a schematic diagram of an alternative membership scenario in accordance with an embodiment of the present application;
Fig. 10 is a schematic structural view of a vehicle test scene generating apparatus according to an embodiment of the present invention.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Example 1
According to an embodiment of the present invention, there is provided an embodiment of a vehicle test scenario generation method, it being noted that the steps shown in the flowchart of the drawings may be performed in a computer system such as a set of computer executable instructions, and that although a logical order is shown in the flowchart, in some cases the steps shown or described may be performed in an order different from that herein.
Fig. 1 is a flowchart of a vehicle test scenario generation method according to an embodiment of the present invention, as shown in fig. 1, the method includes the steps of:
step S102, a plurality of scene elements of an automatic driving vehicle are obtained, wherein the types of the scene elements comprise vehicle elements, road side elements and cloud elements;
step S104, dividing the driving environment of the automatic driving vehicle into a risk environment and a general environment when the risk element exists in the plurality of scene elements;
step S106, determining a detailed measurement scene set corresponding to the risk environment and a general measurement scene set corresponding to the general environment;
step S108, a target test scene set of the automatic driving vehicle is generated based on the detailed test scene set and the general test scene set.
In the embodiment of the present invention, the execution subject of the vehicle test scene generating method provided in the steps S102 to S108 is a vehicle test scene generating system, and the system is adopted to obtain a plurality of scene elements of an automatic driving vehicle, where the types of the plurality of scene elements include a vehicle element, a road side element and a cloud end element; dividing the driving environment of the autonomous vehicle into a risk environment and a general environment when the risk element exists in the plurality of scene elements; determining a detailed measurement scene set corresponding to the risk environment and a general measurement scene set corresponding to the general environment; and generating a target test scene set of the automatic driving vehicle based on the detailed test scene set and the general test scene set.
As an optional embodiment, the vehicle-road cooperative data generated or collected during the driving process of the automatic driving vehicle can be obtained through various ways, the vehicle-road cooperative data can be roughly divided into a vehicle self element, a static environment element, a dynamic environment element, a traffic participant element, a weather element, a road side element, a cloud element, an intelligent network element and the like of the test vehicle, the elements are subjected to secondary hierarchical refinement to determine scene elements, the scene elements in various actual driving environment characteristics corresponding to the vehicle-road cooperative data can be determined after being processed through research, extraction, classification, induction and other modes, and driving scenes are generated according to the arrangement and combination of the scene elements, and the generated driving scenes can cover all the possible safety scenes, namely a driving scene set, of the automatic driving vehicle; the risk data is analyzed through the automobile expected functional safety (Safety Of The Intended Funct ional ity, SOTIF) cognitive technology, the driving scene set is further divided into a risk environment and a general environment by judging whether the risk data exists in the driving scene, expansion or training is carried out according to the dimensions of scene elements in the risk environment and the general environment respectively, a detail scene set and a general scene set are determined, and finally the detail scene set and the general scene set are combined to generate the target test scene set.
According to the embodiment of the application, the road side elements and cloud end elements which are missing in the existing intelligent network-connected automobile test scene are supplemented, but the number of the test scenes becomes huge due to the fact that the road side elements and the cloud end elements are added, so that the efficient scene generation method combining the general test scene set and the detail test scene set is also provided for the problem that the test quantity increases exponentially with the element dimension, and the test efficiency can be remarkably improved while the test accuracy is ensured.
In an alternative embodiment, the acquiring a plurality of scene elements of the autonomous vehicle includes: determining a plurality of scene element levels of a vehicle test scene; and respectively carrying out two-stage hierarchical refinement or multi-stage hierarchical refinement on the scene element levels to obtain refined scene elements respectively corresponding to the scene element levels, and taking the refined scene elements respectively corresponding to the scene element levels as the scene elements of the automatic driving vehicle.
As an optional embodiment, based on the analysis and processing of the vehicle-road cooperative data by using the vehicle expected functional safety (Safety Of The Intended Funct ional ity, SOTIF) cognition technology, determining a data category corresponding to a plurality of the vehicle-road cooperative data, that is, performing secondary hierarchical refinement on the vehicle-road cooperative data such as the test vehicle self element, the static environment element, the dynamic environment element, the traffic participant element, the meteorological element and the intelligent network element, and determining a plurality of scene elements may include, in a hierarchy 1: road elements such as road class, number of lanes, road type, road material, road shape, etc.; layering 2: traffic facility elements such as street lamps, signal lamps, speed limit signs, lane lines and the like; layering 3: road condition changing elements such as construction, road damage, accumulated water, snow and the like; layering 4: traffic participant elements, pedestrians, livestock, buses, trucks, military vehicles, police vehicles, battery cars, motorcycles, and the like; layering 5: weather environmental factors, weather, temperature, pollution conditions, illumination conditions and the like; layering 6: from the car state factor, tire pressure, speed, passenger carrying condition, cargo carrying condition, sensor condition, navigation positioning condition, car machine condition, etc.; layering 7: a road side status element, a road side computing device status, a road side sensing device condition, a road side communication device condition, a vehicle road communication status, and the like; layer 8: cloud state elements, cloud side communication state, cloud vehicle communication state, cloud operation state and the like.
In an alternative embodiment, the classifying the driving environment of the autonomous vehicle into a risk environment and a general environment based on the risk factor includes: enumerating similar risk elements similar to the risk elements; and determining a driving environment including the risk element or the similar risk element as the risk environment, and a driving environment excluding the risk element and the similar risk element as the general environment.
As an alternative embodiment, the eight-layer scene element may be processed layer by an exhaustive (enumerated) process, such as: the road layer sub-elements include road class, number of lanes, road type, road material, etc. The sub-element results may also be represented digitally, for example: highway grades include highway (denoted by "0"), primary highway (denoted by "1"), secondary highway (denoted by "2"), and the like. And permutation and combination processing is performed on the exhaustive results, for example: the method comprises the steps that a driving scene A is formed by a highway (0), a rainy day (01) and one traffic participant (001), a driving scene B is formed by the highway (0), a sunny day (02) and two traffic participants (002), and a plurality of initial driving environments can be obtained through multiple exhaustion, combination and other operations; and analyzing risk factors which cause the automatic driving automobile to generate driving risks in the initial driving scene based on the automobile expected functional safety cognition technology, for example: snowfall, server attack, reduced sensor performance, etc., and is used to determine whether there is risk data for the initial driving scenario to divide the driving scenario set into a risk scenario set and a general scenario set.
Optionally, after determining the risk element, enumeration processing may be further performed on the risk element, to determine a similar risk element similar to the risk element, and determine a driving environment including the risk element or the similar risk element as the risk environment, and a driving environment not including the risk element and the similar risk element as the general environment.
As an alternative embodiment, as shown in fig. 2, a detailed measurement scene generation flow diagram based on safety cognition in a risk environment is taken as an example, risk data is taken as a snowfall element, influence elements related to snowfall such as a highway grade, a highway shape, an air temperature condition, an illumination condition and the like can be respectively adjusted, and a large number of measurement scenes aiming at the snowfall environment are obtained through enumeration means. Taking risk data as an example of attack of a server, a large number of evaluation scenes aiming at the attack of the server can be obtained through enumeration means by adjusting elements related to the risk data such as the state of a self-vehicle sensor, the running state of road side equipment, the communication state, the state of information transmitted by the server and the like. Taking risk data as an example of sensor performance reduction, elements related to the sensor performance reduction such as illumination, rainfall, snowfall, vehicle operation state, road side equipment state, communication state and server state are adjusted, and a large number of evaluation scenes aiming at the sensor performance reduction are obtained through enumeration means. And removing the equivalent scene and the membership scene after determining a plurality of risk environments to obtain a detailed scene set.
Alternatively, the enumeration process may obtain similar vehicle-road coordination data, such as: enumerating the risk data of 'snowfall' in the risk scene to obtain similar risk data of 'rainfall', 'hail', keeping the cooperative data of other roads in the scene unchanged, and then adopting 'rainfall', 'hail', and the like to replace 'snowfall', so as to obtain a new risk scene. For another example: and (3) keeping the risk data 'snowfall' in the risk scene unchanged, enumerating other vehicle-road cooperative data 'common roads' in the scene, and then replacing the 'common roads' with the enumerated data 'expressways' to obtain a new risk scene.
It should be noted that, when the exhaustive results are processed in a permutation and combination manner, at least one of the vehicle-road cooperative data may be selected from each of the scene elements, and the selected vehicle-road cooperative data may be combined to obtain an initial driving scene.
It should be further noted that, based on the first granularity, a detailed scene set corresponding to the risk environment is determined, and based on the second granularity, a general scene set corresponding to the general environment is determined, where the first granularity is smaller than the second granularity, and because the risk element is focused relatively more, there is more scene generated for the risk element than for the non-risk element.
In an optional embodiment, the determining the set of general measurement scenes corresponding to the general environment includes: determining a dangerous background vehicle of the automatic driving vehicle at the current moment, wherein the background vehicle of the automatic driving vehicle comprises a common background vehicle and the dangerous background vehicle, and the dangerous background vehicle is the background vehicle with the highest collision probability; acquiring an initial state of the dangerous background vehicle at the current moment; performing maneuver sampling on the dangerous background vehicle based on a natural-risk countermeasure hybrid distribution corresponding to the initial state of the dangerous background vehicle at the current moment to obtain the maneuver state of the dangerous background vehicle at the moment next to the current moment, wherein the probability of collision when maneuver sampling is performed based on the natural-risk countermeasure hybrid distribution is larger than the probability of collision when maneuver sampling is performed based on maneuver distribution in the natural state; based on the maneuver distribution in the natural state, maneuver sampling is carried out on the common background vehicle to obtain the maneuver state of the common background vehicle at the moment next to the current moment; and determining a general survey scene set corresponding to the general environment based on the maneuvering state of the dangerous background vehicle at the moment next to the current moment and the maneuvering state of the common background vehicle at the moment next to the current moment.
As an alternative embodiment, the classification diagram of the driving situation of the automatic driving vehicle as shown in fig. 3 may be based on a natural driving database, or may be an acquired scene element, and the driving environment information is classified into the following seven categories according to the driving situation of the automatic driving vehicle: free driving (several meters in front without vehicle), following driving, cut-in driving, cut-out driving, lane change driving without adjacent vehicles, lane change driving with 1 adjacent vehicle and lane change driving with 2 adjacent vehicles; and determining maneuver data information based on the maneuver capability of the autonomous vehicle, such as: the range of the maneuvering acceleration is about [ -11,7]m/s 2 The lower acceleration limit can be calculated according to the hundred kilometers braking distance of 35 meters, the upper acceleration limit can be calculated according to the hundred kilometers acceleration of 4 seconds, the maneuvering distribution under the natural state is determined, the numerical values are only used as examples, and the numerical values can be adjusted according to actual conditions.
In the actual driving process, the situations such as rapid acceleration and rapid deceleration are generally less, so that the motorized acceleration value is discretized in a gradual change mode, namely, the acceleration intervals with high occurrence frequency in the natural driving are densely divided, and the acceleration intervals with low occurrence frequency are roughly divided.
In an alternative embodiment, the acquiring the natural dynamic distribution of the vehicle includes: discretizing the value of the maneuvering state in a gradual change mode to obtain maneuvering distribution of the vehicle in the natural state.
As an alternative embodiment, as shown in fig. 4, a longitudinal maneuver state interval dividing schematic diagram is provided with preset discrete acceleration values a 0 Representative of distribution in [ a ] 0 -ε,a 0 +ε) range of acceleration, namely:
wherein epsilon determines the thickness of the interval division, the value of epsilon can comprise 1,0.5 and 0.2, 25 longitudinal maneuvering states are divided totally, the method also comprises two lateral maneuvering states of left lane changing and right lane changing, and 27 maneuvering states in natural states can be obtained in total, for example: the acceleration is-10, -8, 6, 8 can be roughly divided less, the acceleration is-6.5 to-3.5, the acceleration is 3.5 to 4.5 can be roughly divided more, and the acceleration is-2.8 to 2.8 can be finely divided at most. The above values are merely exemplary and may be modified according to the actual circumstances.
As an alternative embodiment, according to the classified 7-class driving environment information, according to the information such as the automobile state and the maneuvering data information in the scene elements, the maneuvering states in different natural states obtained by calculation based on different thicknesses of the interval partition can be called natural distribution; that is, the above-described natural state maneuver states may be used to characterize the probability of occurrence of various maneuver states of the autonomous vehicle in the natural driving state.
As an alternative embodiment, as shown in fig. 5, a schematic flow chart of the general survey scene generation is shown, after the calculation of natural distribution is completed, a reinforcement learning training model is further adopted to train risk countermeasure driving agents, so as to obtain agents which can be used as a general Background Vehicle (BV) and a Dangerous Background Vehicle (DBV); in the process of generating a general survey scene, firstly randomly generating one or more intelligent agents and initial states of the intelligent agents at initial moments corresponding to the intelligent agents, and acquiring the initial states of the dangerous background vehicles at the current moments; the method comprises the steps of performing maneuver sampling on the dangerous background vehicle based on natural-risk countermeasure mixed distribution corresponding to the initial state of the dangerous background vehicle at the current moment to obtain the maneuver state of the dangerous background vehicle at the moment next to the current moment, and performing maneuver sampling on the common background vehicle based on maneuver distribution in the natural state to obtain the maneuver state of the common background vehicle at the moment next to the current moment; and determining continuous risk countermeasure scenes corresponding to the general environment, namely the plague scenes by utilizing a numerical integration method based on the maneuvering state of the dangerous background vehicle at the moment next to the current moment and the maneuvering state of the common background vehicle at the moment next to the current moment, and repeating the generation process of the plague scenes for a plurality of times to obtain a plague scene set.
In an optional embodiment, the determining the set of general measurement scenarios corresponding to the general environment based on the maneuver state of the dangerous background vehicle at the time next to the current time and the maneuver state of the common background vehicle at the time next to the current time includes: repeatedly obtaining the maneuvering states of the dangerous background vehicle and the common background vehicle at a plurality of moments within a preset time period respectively, and determining a general survey scene corresponding to the common environment based on the maneuvering states at the plurality of moments within the preset time period; repeating the above steps for a plurality of times to obtain the general survey scene set corresponding to the general environment.
As an optional embodiment, repeatedly obtaining the maneuvering states of the dangerous background vehicle and the common background vehicle at a plurality of continuous moments within a preset time period respectively, determining a continuous scene based on the maneuvering states of each moment within the preset time period, repeating the operation within the preset time period for a plurality of times to obtain a plurality of scenes, and constructing a general survey scene set based on the plurality of scenes; judging whether the number of the scenes in the general survey scene set is sufficient, if the number of the scenes does not reach the preset number, repeating the operation for a plurality of times within the preset time length again until the number of the scenes in the general survey scene set is sufficient; in addition, after the sufficient number of the scenes in the census scene set is ensured, the equivalent scenes and the membership scenes in the scene set are removed, and the redundancy-removed census scene set is obtained.
In an alternative embodiment, before the dangerous background vehicle is maneuver-sampled based on the natural-risk countermeasure hybrid distribution corresponding to the initial state of the dangerous background vehicle at the current time to obtain the maneuver state of the dangerous background vehicle at the next time of the current time, the method further includes: acquiring collision probability distribution of the dangerous background vehicle in a natural state in an initial state at the current moment; acquiring maneuvering distribution of the vehicle in a natural state; and determining a natural-risk countermeasure mixture distribution corresponding to the initial state of the dangerous background vehicle at the current moment based on the collision probability distribution in the natural state and the maneuvering distribution in the natural state.
Optionally, the collision probability distribution in the natural state and the maneuver distribution in the natural state are mixed according to a predetermined proportion, so as to obtain a natural-risk countermeasure mixed distribution corresponding to the initial state of the dangerous background vehicle at the current moment, a schematic diagram is generated according to the natural-risk countermeasure mixed distribution shown in fig. 6, and the natural-risk countermeasure mixed probability distribution is obtained based on the fact that the collision probability distribution in the natural state after normalization processing of the dangerous background vehicle at the initial state at the current moment and the maneuver distribution in the natural state are mixed according to a certain proportion.
In an alternative embodiment, the acquiring the collision probability distribution of the dangerous background vehicle in the natural state in the initial state at the current time includes: inputting the initial state of the dangerous background vehicle at the current moment into an anti-driving intelligent agent model to obtain the maneuvering-collision probability distribution of the dangerous background vehicle at the initial state of the current moment, and determining the collision probability distribution in the natural state based on the maneuvering-collision probability distribution and the maneuvering distribution in the natural state.
As an alternative embodiment, the training process of risk countermeasure driving agent by using reinforcement learning training model is as follows, wherein the countermeasure driving agent model is obtained based on collision training between a sample measured vehicle and a sample background vehicle, and the background is definedThe vehicle is an intelligent body, the intelligent driving model IDM is utilized for training, and the duration of each training data is half a minute. And defines a background vehicle speed v b Relative position r and relative velocity v between the vehicle and the vehicle to be measured r For the state, define the maneuver u of the background vehicle b For the action, the vehicle u is measured a The maneuver of (1) is sampled and simulated according to natural distribution.
The r value range was [0,150 ] ]m，v r The range of the value of (C) is [ -35,35]m/s，v b The range of the value of (C) is [ -2,35]m/s,u b The range of the value of (C) is [ -11,7]m/s 2 The numerical values are merely illustrative and may be adjusted according to the actual circumstances.
Optionally, determining rewards according to judging results of whether the tested vehicle collides with the background vehicle, giving rewards 1 to branches of the collided decision tree, and giving no rewards to non-collided decision trees. Training against background vehicles (driving agents) using reinforcement learning methods until the number of training times reaches a preset threshold (e.g., 1 x 10) 7 Secondary) or convergence (collision probability no longer changes), completing training of the anti-driving agent model.
As an alternative embodiment, as shown in fig. 7, a schematic diagram may be generated by randomly sampling the position and speed of the first trolley of each lane according to the natural distribution, determining the inter-vehicle distance and the relative speed based on the natural distribution, randomly generating the position and speed of the subsequent trolley, generating a group of vehicles in each lane, inputting the data of the position, speed, relative distance, relative speed, etc. of the vehicles into the trained anti-driving agent model, to obtain the maneuver-collision probability, and multiplying the maneuver-collision probability by the maneuver distribution in the natural state, to obtain the collision probability distribution in the natural state.
The above-mentioned maneuvering-collision probability is used to represent the probability of collision after substituting the training-completed driving-countermeasure agent model into the data such as vehicle position, speed, relative distance, and relative speed; however, the intelligent body always trains towards the collision situation in the training process, so that the maneuvering-collision probability is extremely high, and the maneuvering distribution in the natural state is needed to be multiplied for neutralization to obtain the collision probability distribution in the natural state; that is, the above-described maneuver-to-collision probability may represent the probability that a background vehicle (agent) collides with when a maneuver of some kind is employed, and the above-described natural-state collision probability distribution may represent the maneuver conditions that may occur in the natural driving state and the collision probability corresponding to each maneuver condition.
In an alternative embodiment, the determining the dangerous background vehicle of the autopilot vehicle at the current moment includes: under the condition that a plurality of background vehicles of the automatic driving vehicle are provided, respectively acquiring collision probability distribution of the plurality of background vehicles in a natural state corresponding to the current moment; the probabilities of the plurality of background vehicles in various maneuvering states in the collision probability distribution in the natural state corresponding to the current moment are summed to obtain a summed probability; and determining the background vehicle with the highest summation probability among the plurality of background vehicles as the dangerous background vehicle.
As an alternative embodiment, based on the collision probability distribution in the natural state, a plurality of background vehicles are determined, and the probability of each maneuver state in the collision probability distribution in the natural state corresponding to the current time is summed to determine the background vehicle with the highest probability of collision occurrence in all the background vehicles, namely, the dangerous background vehicle.
Optionally, as shown in fig. 6, the collision probability of the dangerous background vehicle in the natural state is normalized to obtain the normalized collision probability in the natural state, and the normalized collision probability is mixed with the maneuver distribution in the natural state according to a certain proportion to obtain the natural-risk countermeasure mixed probability distribution, wherein the mixed formula is as follows: p (P) mix ＝(1-e)*P accodent +e*P nature ，P mix Mixed probability distribution for natural-risk challenge, P accodent For normalizing the collision probability distribution in natural state, P nature For the dynamic distribution in the natural state, the value of e can be 0.5 in the normal case orSo as to be adjusted according to actual requirements. Performing maneuver sampling on dangerous background vehicles according to the mixed probability distribution, performing maneuver sampling on other background vehicles according to the maneuver probability distribution, obtaining the states of the position, the speed and the like of each vehicle at the next time through kinematic model integration, obtaining maneuver states of the dangerous background vehicles and the common background vehicles at a plurality of continuous moments within a preset time period respectively, repeatedly determining a continuous scene based on the maneuver states at each moment within the preset time period, and determining the general survey scene; and repeating the operation for a plurality of times within a preset time length to obtain a plurality of scenes, and constructing a general survey scene set based on the plurality of scenes.
In an alternative embodiment, the generating the target test scene set of the autonomous vehicle based on the detailed test scene set and the general test scene set includes: performing first redundancy elimination processing on a plurality of detail measurement scenes in the detail measurement scene set to obtain a detail measurement scene set after the first redundancy elimination processing; performing second redundancy elimination processing on a plurality of census scenes in the census scene set to obtain a census scene set after the second redundancy elimination processing; and obtaining the target test scene set of the automatic driving vehicle based on the detailed test scene set after the first redundancy elimination processing and the general test scene set after the second redundancy elimination processing.
As an alternative embodiment, the first redundancy elimination process and the second redundancy elimination process include: determining equivalent scenes in a plurality of scenes, and removing redundant scenes by combining the equivalent scenes; determining sub-scenes in the membership scenes with membership in the scenes, and removing redundant scenes by removing the sub-scenes.
Optionally, a scene equivalent relation can be used to remove redundant repeated scenes from a risk scene set formed by multiple risk evaluation scenes, as shown in an equivalent scene schematic diagram in fig. 8, the positions of the background vehicles of the scenes a and B are different, the positions of the vehicles on the left side are far away, but the vehicle lane is far away from the vehicle lane to be tested, and the vehicle to be tested is hardly influenced by the vehicle; the middle lane background vehicle has a certain position difference in the transverse direction, and the right lane background vehicle has a certain position in the longitudinal direction Difference in position, but difference in transverse distance epsilon 1 Less than a preset threshold gamma 1 Difference epsilon in longitudinal distance 2 Less than a preset threshold gamma 2 At this time, the equivalent similarity value of the scene B may be regarded as smaller than the first equivalent similarity threshold with the scene a, and the scene B may be regarded as an equivalent scene of the scene a to be combined.
Alternatively, the sub-scene membership may be used to remove redundant repeated scenes, such as the membership scene schematic diagram shown in fig. 9, where a background vehicle exists in the middle lane of scene a, and the rest elements of scene B are consistent with scene a except that there is no vehicle in the middle lane, i.e., the traffic participant information contained in scene B is a subset of scene a. When the tested vehicle is tested for lane change and the like, if the test of the scene A can be passed, the test of the scene B can be passed certainly, and the scene B is considered to be a sub-scene of the scene A at the moment and can be removed.
It should be noted that, the redundant repeated scenes can be removed by using the scene equivalence relation and the sub-scene membership relation for multiple times, so as to determine the target test scene set.
As an optional embodiment, the target test scene construction is completed based on the detailed test scene set after the first redundancy elimination process and the general test scene set after the second redundancy elimination process by combining the detailed test scene set and the general test scene set. Different granularity scene generation methods are respectively adopted for the risk scenes and the general scenes, so that the number of unnecessary test scenes is reduced, and the problem of dimension disaster is relieved. And finally, merging the detailed test scene set obtained according to the risk scene with the general test scene set obtained according to the general scene to obtain a target test scene set.
Example 2
According to an embodiment of the present invention, there is further provided an embodiment of an apparatus for implementing the above-mentioned vehicle test scene generating method, and fig. 10 is a schematic structural diagram of a vehicle test scene generating apparatus according to an embodiment of the present invention, as shown in fig. 10, where the apparatus includes: an acquisition module 100, a partitioning module 102, a determination module 104, and a generation module 106, wherein:
the acquiring module 100 is configured to acquire a plurality of scene elements of an autopilot vehicle, where types of the plurality of scene elements include a vehicle element, a road side element and a cloud element;
a dividing module 102, configured to divide a driving environment of the autonomous vehicle into a risk environment and a general environment when there are risk elements in the plurality of scene elements;
a determining module 104, configured to determine a detailed measurement scene set corresponding to the risk environment, and determine a general measurement scene set corresponding to the general environment;
and the generating module 106 is configured to generate a target test scene set of the automatic driving vehicle based on the detail test scene set and the general test scene set.
Here, the above-mentioned obtaining module 100, dividing module 102, determining module 104 and generating module 106 correspond to steps S102 to S108 in embodiment 1, and the four modules are the same as the examples and application scenarios implemented by the corresponding steps, but are not limited to those disclosed in embodiment 1 above.
It should be noted that, the preferred implementation manner of this embodiment may be referred to the related description in embodiment 1, and will not be repeated here.
According to an embodiment of the present invention, there is also provided an embodiment of a computer-readable storage medium. Alternatively, in the present embodiment, the above-described computer-readable storage medium may be used to store the program code executed by the vehicle test scenario generation method provided in the above-described embodiment 1.
Alternatively, in this embodiment, the above-mentioned computer readable storage medium may be located in any one of the computer terminals in the computer terminal group in the computer network, or in any one of the mobile terminals in the mobile terminal group.
Optionally, in the present embodiment, the computer readable storage medium is configured to store program code for performing the steps of: acquiring a plurality of scene elements of an automatic driving vehicle, wherein the types of the scene elements comprise vehicle elements, road side elements and cloud elements; dividing the driving environment of the autonomous vehicle into a risk environment and a general environment when the risk element exists in the plurality of scene elements; determining a detailed measurement scene set corresponding to the risk environment and a general measurement scene set corresponding to the general environment; and generating a target test scene set of the automatic driving vehicle based on the detailed test scene set and the general test scene set.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: determining a plurality of scene element levels of a vehicle test scene; and respectively carrying out two-stage hierarchical refinement or multi-stage hierarchical refinement on the scene element levels to obtain refined scene elements respectively corresponding to the scene element levels, and taking the refined scene elements respectively corresponding to the scene element levels as the scene elements of the automatic driving vehicle.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: determining initial risk data corresponding to each risk scene in the risk scene set; performing enumeration processing based on the initial risk data to determine a target risk data set, wherein the enumeration processing is used for determining a plurality of target risk data; traversing the target risk data in the target risk data set, and replacing the initial risk data by adopting the target risk data to generate the detailed measurement scene set.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: enumerating similar risk elements similar to the risk elements; and determining a driving environment including the risk element or the similar risk element as the risk environment, and a driving environment excluding the risk element and the similar risk element as the general environment.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: and determining a detailed scene set corresponding to the risk environment based on a first granularity, and determining a general scene set corresponding to the general environment based on a second granularity, wherein the first granularity is smaller than the second granularity.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: determining a dangerous background vehicle of the automatic driving vehicle at the current moment, wherein the background vehicle of the automatic driving vehicle comprises a common background vehicle and the dangerous background vehicle, and the dangerous background vehicle is the background vehicle with the highest collision probability; acquiring an initial state of the dangerous background vehicle at the current moment; performing maneuver sampling on the dangerous background vehicle based on a natural-risk countermeasure hybrid distribution corresponding to the initial state of the dangerous background vehicle at the current moment to obtain the maneuver state of the dangerous background vehicle at the moment next to the current moment, wherein the probability of collision when maneuver sampling is performed based on the natural-risk countermeasure hybrid distribution is larger than the probability of collision when maneuver sampling is performed based on maneuver distribution in the natural state; based on the maneuver distribution in the natural state, maneuver sampling is carried out on the common background vehicle to obtain the maneuver state of the common background vehicle at the moment next to the current moment; and determining a general survey scene set corresponding to the general environment based on the maneuvering state of the dangerous background vehicle at the moment next to the current moment and the maneuvering state of the common background vehicle at the moment next to the current moment.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: repeatedly obtaining the maneuvering states of the dangerous background vehicle and the common background vehicle at a plurality of moments within a preset time period respectively, and determining a general survey scene corresponding to the common environment based on the maneuvering states at the plurality of moments within the preset time period; repeating the above steps for a plurality of times to obtain the general survey scene set corresponding to the general environment.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: acquiring collision probability distribution of the dangerous background vehicle in a natural state in an initial state at the current moment; acquiring maneuvering distribution of the vehicle in a natural state; and determining a natural-risk countermeasure mixture distribution corresponding to the initial state of the dangerous background vehicle at the current moment based on the collision probability distribution in the natural state and the maneuvering distribution in the natural state.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: acquiring the maneuvering-collision probability distribution of the dangerous background vehicle in the initial state at the current moment; determining the collision probability distribution in the natural state based on the maneuver-collision probability distribution and the maneuver-in-natural state distribution.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: and inputting the initial state of the dangerous background vehicle at the current moment into an anti-driving intelligent agent model to obtain the maneuvering-collision probability distribution of the dangerous background vehicle at the initial state of the current moment, wherein the anti-driving intelligent agent model is obtained based on collision training between a sample tested vehicle and the sample background vehicle.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: and mixing the collision probability distribution in the natural state and the maneuvering distribution in the natural state according to a preset proportion to obtain a natural-risk countermeasure mixing distribution corresponding to the initial state of the dangerous background vehicle at the current moment.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: under the condition that a plurality of background vehicles of the automatic driving vehicle are provided, respectively acquiring collision probability distribution of the plurality of background vehicles in a natural state corresponding to the current moment; the probabilities of the plurality of background vehicles in various maneuvering states in the collision probability distribution in the natural state corresponding to the current moment are summed to obtain a summed probability; and determining the background vehicle with the highest summation probability among the plurality of background vehicles as the dangerous background vehicle.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: performing first redundancy elimination processing on a plurality of detail measurement scenes in the detail measurement scene set to obtain a detail measurement scene set after the first redundancy elimination processing; performing second redundancy elimination processing on a plurality of census scenes in the census scene set to obtain a census scene set after the second redundancy elimination processing; and obtaining the target test scene set of the automatic driving vehicle based on the detailed test scene set after the first redundancy elimination processing and the general test scene set after the second redundancy elimination processing.
Optionally, the above computer readable storage medium is configured to store program code for performing the steps of: determining equivalent scenes in a plurality of scenes, and removing redundant scenes by combining the equivalent scenes; determining sub-scenes in the membership scenes with membership in the scenes, and removing redundant scenes by removing the sub-scenes.
According to an embodiment of the present invention, there is also provided an embodiment of a processor. Alternatively, in the present embodiment, the above-described computer-readable storage medium may be used to store the program code executed by the vehicle test scenario generation method provided in the above-described embodiment 1.
The embodiment of the application provides an electronic device, which comprises a processor, a memory and a program stored on the memory and capable of running on the processor, wherein the processor realizes the following steps when executing the program: acquiring a plurality of scene elements of an automatic driving vehicle, wherein the types of the scene elements comprise vehicle elements, road side elements and cloud elements; dividing the driving environment of the autonomous vehicle into a risk environment and a general environment when the risk element exists in the plurality of scene elements; determining a detailed measurement scene set corresponding to the risk environment and a general measurement scene set corresponding to the general environment; and generating a target test scene set of the automatic driving vehicle based on the detailed test scene set and the general test scene set.
The application also provides a computer program product adapted to perform, when executed on a data processing device, a program initialized with the method steps of: acquiring a plurality of scene elements of an automatic driving vehicle, wherein the types of the scene elements comprise vehicle elements, road side elements and cloud elements; dividing the driving environment of the autonomous vehicle into a risk environment and a general environment when the risk element exists in the plurality of scene elements; determining a detailed measurement scene set corresponding to the risk environment and a general measurement scene set corresponding to the general environment; and generating a target test scene set of the automatic driving vehicle based on the detailed test scene set and the general test scene set.
The foregoing embodiment numbers of the present application are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
In the foregoing embodiments of the present application, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed technology may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of the units, for example, may be a logic function division, and may be implemented in another manner, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.

Claims (15)

1. A vehicle test scene generation method, characterized by comprising:
acquiring a plurality of scene elements of an automatic driving vehicle, wherein the types of the scene elements comprise vehicle elements, road side elements and cloud elements;
dividing a driving environment of the autonomous vehicle into a risk environment and a general environment in the case where a risk element exists among the plurality of scene elements;
determining a detailed measurement scene set corresponding to the risk environment and a general measurement scene set corresponding to the general environment;
and generating a target test scene set of the automatic driving vehicle based on the detailed test scene set and the general test scene set.
2. The method of claim 1, wherein the acquiring a plurality of scene elements of the autonomous vehicle comprises:
determining a plurality of scene element levels of a vehicle test scene;
and respectively carrying out two-stage hierarchical refinement or multi-stage hierarchical refinement on the scene element levels to obtain refined scene elements respectively corresponding to the scene element levels, and taking the refined scene elements respectively corresponding to the scene element levels as the scene elements of the automatic driving vehicle.
3. The method of claim 1, wherein the classifying the driving environment of the autonomous vehicle into a risk environment and a general environment based on the risk element comprises:
enumerating similar risk elements similar to the risk elements;
determining a driving environment including the risk element or the similar risk element as the risk environment, and a driving environment excluding the risk element and the similar risk element as the general environment.
4. The method of claim 1, wherein the determining the detailed set of scenes corresponding to the risk environment and the general set of scenes corresponding to the general environment comprises:
and determining a detailed scene set corresponding to the risk environment based on a first granularity, and determining a general scene set corresponding to the general environment based on a second granularity, wherein the first granularity is smaller than the second granularity.
5. The method of claim 1, wherein the determining the set of survey scenes corresponding to the general environment comprises:
determining a dangerous background vehicle of the automatic driving vehicle at the current moment, wherein the background vehicle of the automatic driving vehicle comprises a common background vehicle and the dangerous background vehicle, and the dangerous background vehicle is the background vehicle with the highest collision probability;
Acquiring an initial state of the dangerous background vehicle at the current moment;
performing maneuver sampling on the dangerous background vehicle based on natural-risk countermeasure mixed distribution corresponding to the initial state of the dangerous background vehicle at the current moment to obtain the maneuver state of the dangerous background vehicle at the moment next to the current moment, wherein the probability of collision when maneuver sampling is performed based on the natural-risk countermeasure mixed distribution is larger than the probability of collision when maneuver sampling is performed based on maneuver distribution in the natural state;
based on the maneuver distribution in the natural state, maneuver sampling is carried out on the common background vehicle to obtain the maneuver state of the common background vehicle at the moment next to the current moment;
and determining a general survey scene set corresponding to the general environment based on the maneuvering state of the dangerous background vehicle at the moment next to the current moment and the maneuvering state of the common background vehicle at the moment next to the current moment.
6. The method of claim 5, wherein the determining the set of plausible scenes corresponding to the general environment based on the maneuver state of the hazardous background vehicle at the time next to the current time and the maneuver state of the common background vehicle at the time next to the current time comprises:
Repeatedly obtaining the maneuvering states of the dangerous background vehicle and the common background vehicle at a plurality of moments within a preset time period respectively, and determining a general survey scene corresponding to the common environment based on the maneuvering states at the plurality of moments within the preset time period;
repeating the preset time for a plurality of times to obtain the general survey scene set corresponding to the general environment.
7. The method of claim 5, wherein prior to maneuver sampling the hazard-background vehicle based on a natural-risk challenge mix profile corresponding to an initial state of the hazard-background vehicle at the current time to obtain a maneuver state of the hazard-background vehicle at a time next to the current time, the method further comprises:
acquiring collision probability distribution of the dangerous background vehicle in a natural state in an initial state at the current moment;
acquiring maneuvering distribution of the dangerous background vehicle in a natural state;
and determining a natural-risk countermeasure hybrid distribution corresponding to the initial state of the dangerous background vehicle at the current moment based on the collision probability distribution in the natural state and the maneuvering distribution in the natural state.
8. The method according to claim 7, wherein the acquiring the collision probability distribution of the dangerous background vehicle in a natural state in an initial state at the current time includes:
Acquiring the maneuvering-collision probability distribution of the dangerous background vehicle in the initial state of the current moment;
determining a collision probability distribution in the natural state based on the maneuver-collision probability distribution and the maneuver-in-natural state distribution.
9. The method of claim 8, wherein the obtaining a maneuver-to-collision probability distribution of the hazard background vehicle in an initial state at the current time comprises:
inputting the initial state of the dangerous background vehicle at the current moment into an anti-driving intelligent agent model to obtain the maneuvering-collision probability distribution of the dangerous background vehicle at the initial state of the current moment, wherein the anti-driving intelligent agent model is obtained based on collision training between a sample measured vehicle and a sample background vehicle.
10. The method of claim 7, wherein the determining a natural-risk challenge mixture profile corresponding to an initial state of the hazard-background vehicle at the current time based on the natural-state collision probability profile and the natural-state maneuver profile comprises:
and mixing the collision probability distribution in the natural state and the maneuvering distribution in the natural state according to a preset proportion to obtain a natural-risk countermeasure mixing distribution corresponding to the initial state of the dangerous background vehicle at the current moment.
11. The method of claim 5, wherein the determining a dangerous background vehicle of the autonomous vehicle at a current time comprises:
under the condition that a plurality of background vehicles of the automatic driving vehicle are provided, respectively acquiring collision probability distribution of the plurality of background vehicles in a natural state corresponding to the current moment;
summing the probabilities of the plurality of background vehicles in various maneuvering states in the collision probability distribution in the natural state corresponding to the current moment respectively to obtain a summation probability;
and determining the background vehicle with the highest summation probability among the plurality of background vehicles as the dangerous background vehicle.
12. The method of any one of claims 1 to 11, wherein the generating a target test scenario set for the autonomous vehicle based on the detail scenario set and the plausible scenario set comprises:
performing first redundancy elimination processing on a plurality of detail scenes in the detail scene set to obtain a detail scene set after the first redundancy elimination processing;
performing second redundancy elimination processing on a plurality of census scenes in the census scene set to obtain a census scene set after the second redundancy elimination processing;
And obtaining the target test scene set of the automatic driving vehicle based on the detailed test scene set after the first redundancy elimination processing and the general test scene set after the second redundancy elimination processing.
13. The method of claim 12, wherein the first de-redundancy process and the second de-redundancy process comprise:
determining equivalent scenes in a plurality of scenes, and removing redundant scenes by combining the equivalent scenes;
determining sub-scenes in the membership scenes with membership in the scenes, and removing redundant scenes by removing the sub-scenes.
14. A non-volatile storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the vehicle test scenario generation method of any one of claims 1 to 13.
15. An electronic device comprising a memory and a processor, wherein the memory has stored therein a computer program, the processor being arranged to run the computer program to perform the vehicle test scenario generation method of any one of claims 1 to 13.
CN202310698303.4A 2023-06-13 2023-06-13 Vehicle test scene generation method, device, storage medium and equipment Pending CN116753938A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310698303.4A CN116753938A (en) 2023-06-13 2023-06-13 Vehicle test scene generation method, device, storage medium and equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310698303.4A CN116753938A (en) 2023-06-13 2023-06-13 Vehicle test scene generation method, device, storage medium and equipment

Publications (1)

Publication Number Publication Date
CN116753938A true CN116753938A (en) 2023-09-15

Family

ID=87947182

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310698303.4A Pending CN116753938A (en) 2023-06-13 2023-06-13 Vehicle test scene generation method, device, storage medium and equipment

Country Status (1)

Country Link
CN (1) CN116753938A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117041916A (en) * 2023-09-27 2023-11-10 创意信息技术股份有限公司 Mass data processing method, device, system and storage medium

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117041916A (en) * 2023-09-27 2023-11-10 创意信息技术股份有限公司 Mass data processing method, device, system and storage medium
CN117041916B (en) * 2023-09-27 2024-01-09 创意信息技术股份有限公司 Mass data processing method, device, system and storage medium

Similar Documents

Publication Publication Date Title
EP3533681B1 (en) Method for detecting safety of driving behavior, apparatus and storage medium
CN110400478B (en) Road condition notification method and device
CN113609016B (en) Method, device, equipment and medium for constructing automatic driving test scene of vehicle
CN109493566B (en) Fatigue driving prediction method based on GPS data
CN107038860B (en) A kind of user's driving behavior methods of marking based on ADAS technology and regression model
CN107784251A (en) The method evaluated based on image recognition technology driving behavior
CN107784587A (en) A kind of driving behavior evaluation system
CN110304068B (en) Method, device, equipment and storage medium for collecting automobile driving environment information
Kovaceva et al. Identification of aggressive driving from naturalistic data in car-following situations
CN110544373A (en) truck early warning information extraction and risk identification method based on Beidou Internet of vehicles
CN116753938A (en) Vehicle test scene generation method, device, storage medium and equipment
Wu et al. Clustering of several typical behavioral characteristics of commercial vehicle drivers based on GPS data mining: Case study of highways in China
Yu et al. Prediction of instantaneous driving safety in emergency scenarios based on connected vehicle basic safety messages
CN114912689A (en) Map grid index and XGBOST-based over-limit vehicle destination prediction method and system
CN116092037B (en) Vehicle type identification method integrating track space-semantic features
CN116740940A (en) Severe weather high-impact road section risk prediction and safety management method, device and equipment
CN116307699A (en) Road hidden trouble point segment grading method, device and storage medium based on multi-source data
CN116461546A (en) Vehicle early warning method, device, storage medium and processor
CN113192340B (en) Method, device, equipment and storage medium for identifying highway construction vehicles
Xu et al. Novel fast safety assessment method for the buffer section of maintenance work zone
CN115578856A (en) Method and device for evaluating safety risk of road test of automatic driving automobile
CN111341106B (en) Traffic early warning method, device and equipment
CN114756599A (en) Driver abnormal speed change identification method based on vehicle GPS data
CN114492544A (en) Model training method and device and traffic incident occurrence probability evaluation method and device
Ali et al. Employment of instrumented vehicles to identify real-time snowy weather conditions on freeways using supervised machine learning techniques–A naturalistic driving study

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination