WO2025214510A1 - 面向自动驾驶测试的动态场景仿真与场地测试的闭环方法 - Google Patents

面向自动驾驶测试的动态场景仿真与场地测试的闭环方法

Info

Publication number
WO2025214510A1
WO2025214510A1 PCT/CN2025/092353 CN2025092353W WO2025214510A1 WO 2025214510 A1 WO2025214510 A1 WO 2025214510A1 CN 2025092353 W CN2025092353 W CN 2025092353W WO 2025214510 A1 WO2025214510 A1 WO 2025214510A1
Authority
WO
WIPO (PCT)
Prior art keywords
scene
simulation
test
information file
site
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
PCT/CN2025/092353
Other languages
English (en)
French (fr)
Inventor
毕欣
李志超
翁才恩
仝盼盼
刘腾宇
张乾成
芮品君
王柞朝
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.)
Tongji University
Original Assignee
Tongji University
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 Tongji University filed Critical Tongji University
Publication of WO2025214510A1 publication Critical patent/WO2025214510A1/zh
Pending legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M17/00—Testing of vehicles
    • G01M17/007—Wheeled or endless-tracked vehicles
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Definitions

  • the present invention belongs to the field of autonomous driving testing, and in particular relates to a closed-loop method for dynamic scene simulation and field testing for autonomous driving testing.
  • the current method for generating autonomous driving test scenarios includes the following steps: first, a scenario modeling tool is needed to model static road environments such as road networks, traffic signs, traffic lights, and intersection structures. Map data or virtual modeling software is usually used for modeling, and standards such as OpenDrive are usually used; then, a dynamic scenario editing tool is used to generate various dynamic traffic participants, including vehicles, pedestrians, bicycles, etc., and the motion trajectories of traffic participants are generated through trajectory editing, including path planning, speed control, etc., to simulate their movement in the scene, usually using standards such as OpenScenario; finally, the generated dynamic and static scenarios are loaded into a simulation platform for testing and evaluation, which usually includes functions such as a physics engine and sensor simulation.
  • the purpose of this invention is to provide a closed-loop method for dynamic scene simulation and field testing systems for autonomous driving testing.
  • This method combines virtual simulation with closed-field testing to generate dynamic scenes containing dynamic traffic participants and achieves mutual conversion between simulation and field testing dynamic scenes.
  • the technical solutions adopted are:
  • a closed-loop method for dynamic scene simulation and field testing for autonomous driving testing comprising the following steps:
  • Step 1 Based on the dynamic scenario protocol FX for intelligent driving simulation testing, create a simulation scenario information file F1 for simulation testing and perform simulation testing.
  • the first simulation scene information file F1 is created by a dynamic scene editing tool, and the first simulation scene information file F1 includes road network information and traffic participant information;
  • Road network information includes: OpenDrive map, OpenSceneGraph scene model;
  • Traffic participant information includes: name, category, traffic participant rendering model, initialization position, initialization speed, trajectory information list, and speed change.
  • Step 2 Based on the field test dynamic scene protocol CX, a field test-oriented field scene information file No. 1 C1 is created, and the field test is performed.
  • the scene information file C1 of site 1 is created through the dynamic scene editing tool
  • the scene information file C1 of site 1 contains the following information: scene name, scene number, scene description, number of traffic participants, and traffic participant information;
  • Traffic participant information includes: OBU number, number of track points, and track information list;
  • Step 3 Convert the simulation test dynamic scene into a field test dynamic scene, and then conduct field testing, which specifically includes the following steps:
  • Step 3A Convert the first simulation scene information file F1 into the second site scene information file C2:
  • Step 3A1 Based on the field test dynamic scenario protocol CX, create a second field test scenario information file C2 for the field test;
  • the fields of the second site scene information file C2 are the same as those of the first site scene information file C1;
  • Step 3A2 Perform data conversion based on the dynamic scene protocol FX of the simulation test and the dynamic scene protocol CX of the field test, and convert the data into corresponding fields of the second field scene information file C2;
  • Step 3A3 Supplement data to the unfilled fields in the second site scene information file C2;
  • the fields that require additional information include: scene name, scene number, scene description, number of cloud-controlled simulated traffic participants, and number of trajectory points for each cloud-controlled simulated traffic participant;
  • Step 3B Store the second site scene information file C2 output in step 3A3 in the cloud database as a JSON file;
  • Step 3C Add the OBU numbers corresponding to the respective communication units for the cloud-controlled simulated traffic participants to ensure the correct distribution of trajectory information;
  • Step 3D Conduct field testing:
  • the cloud server reads the information of each traffic participant in the scene information file C2 of site No. 2 in the cloud database, and transmits the trajectory information list to the communication unit of the cloud-controlled simulated traffic participant via 5G according to the OBU number;
  • the cloud-controlled simulated traffic participants include human models and vehicle models placed in the actual test site; the cloud-controlled simulated traffic participants can all move in the actual test site and are all equipped with a communication unit OBU.
  • the step 3 further includes the following steps:
  • Step 4 Convert the field test dynamic scene into a simulation test dynamic scene, and then conduct a simulation test. This specifically includes the following steps:
  • Step 4A Convert the field test dynamic scene into the simulation test dynamic scene:
  • step 4A1 the communication unit of each cloud-controlled simulated traffic participant uploads the test data of the dynamic scene running in the field test in step 3D via 5G communication, including GPS positioning coordinate information, heading angle information, category, and OBU number.
  • the cloud receives the test data of each cloud-controlled simulated traffic participant.
  • the GPS positioning coordinate information is provided in real time by the GPS receiver, the heading angle information is provided in real time by the built-in inertial navigation unit, and the category and OBU number are obtained by reading the self-status information file pre-stored in the storage unit;
  • the site test dynamic scene protocol CX is used to generate the site scene information file C3 of the third site.
  • the site scene information file C3 of the third site there are fields, and the data in the fields is to be filled.
  • Step 4A2 Based on the dynamic scenario protocol FX for intelligent driving simulation testing, a second simulation scenario information file F2 for simulation testing is created;
  • the fields of the second simulation scenario information file F2 are the same as those of the first simulation scenario information file F1;
  • Step 4A3 According to the simulation test dynamic scene protocol FX and the field test dynamic scene protocol CX, the data of the third field scene information file C3 is converted into the corresponding fields of the second simulation scene information file F2;
  • Step 4A4 Supplement data to the unfilled fields in the second simulation scenario information file F2;
  • the fields that require additional information include: rendering models of each traffic participant, OpenDrive map, and OpenSceneGraph scene model;
  • Step 4B Load the second simulation scenario information file F2 containing the data stored in step 4A4 into simulation software supporting the OpenScenario protocol to perform a virtual simulation test.
  • step 3A2 specifically includes the following conversion steps:
  • the name is converted to OBU number
  • the coordinates in the initialization position are converted to the longitude and latitude at time 0
  • the attitude angle h in the initialization position is converted to the heading angle
  • the initialization speed is converted to the speed at time 0
  • the time t is converted
  • the current speed is converted.
  • step 4A2 specifically includes the following conversion steps:
  • the OBU number is converted into a name, the longitude and latitude at time 0 are converted into the coordinates of the initialization position, the heading angle at time 0 is converted into the attitude angle h at the initialization position, the speed at time 0 is converted into the initialization speed, the time t is converted, and the current speed is converted.
  • simulation visualization can be performed.
  • problems can be discovered and optimized in advance, reducing unnecessary repetitive testing during field testing, and significantly reducing the cost and time of field testing.
  • Figure 1 is a flow chart of a closed-loop method for dynamic scenario simulation and field testing for autonomous driving testing
  • Figure 2 shows a dynamic scenario simulation test protocol for multiple traffic participants based on the OpenScenario standard
  • Figure 3 shows the action elements of the multi-traffic participant simulation motion control method
  • Figure 4 shows the dynamic scenario protocol for field testing of cloud-controlled simulated traffic participants
  • Figure 5 shows the cloud-controlled simulation method for converting and generating trajectories of traffic participants
  • Figure 6 shows the cloud-controlled simulated traffic participant field test scenario regeneration method.
  • the closed-loop simulation and field testing method proposed in this invention focuses on achieving the transformation and regeneration of multi-target dynamic traffic scene simulation and real target trajectories in field testing, so as to further improve the authenticity, consistency and reproducibility of autonomous driving testing.
  • Step 1 Based on the dynamic scenario protocol FX for intelligent driving simulation testing, create a simulation scenario information file F1 for simulation testing and perform simulation testing.
  • the dynamic scenario protocol FX for intelligent driving simulation testing is developed based on the OpenScenario standard.
  • OpenScenario is an open protocol for describing autonomous driving scenarios. It primarily describes the synchronized actions of multiple entities within an autonomous driving scenario, such as a vehicle changing lanes or shifting speeds. It is commonly used to define dynamic scenarios in autonomous driving simulation tests.
  • the multi-traffic participant dynamic scenario protocol for simulation testing in this embodiment is developed based on the OpenScenario 1.0 protocol and provides a simple and standardized way to define complex traffic scenarios involving multiple traffic participants, including vehicles and pedestrians.
  • the dynamic scenario protocol FX for simulation testing in this embodiment carries all necessary dynamic scenario information through the OpenScenario 1.0 standard.
  • the movement trajectory of traffic participants is implemented by trajectory following actions, and the speed change of traffic participants is implemented by speed change actions.
  • Time triggers are used as the execution conditions of traffic participant actions, as shown in Figure 2. Among them, 1 to 3 are trajectory points.
  • the first simulation scene information file F1 is created by a dynamic scene editing tool.
  • the fields of the first simulation scene information file F1 and the second simulation scene information file F2 are the same.
  • the first simulation scenario information file F1 includes: road network information and traffic participant information, and its detailed composition is shown in FIG3 .
  • road network information includes: OpenDrive map and OpenSceneGraph scene model.
  • Traffic participant information includes: name, category, traffic participant rendering model, initialization position, initialization speed, trajectory information list, speed change (the speed change in time t is v).
  • Import high-precision maps in formats such as OpenDrive To maintain consistency between simulation and field testing, it is necessary to obtain in advance the OpenDrive map and OpenSceneGraph scene model of the closed test site produced by the high-precision map supplier.
  • the scenario information file F based on the protocol FX produced by the above method also meets the OpenScenario standard and is an xosc file. It can be loaded into autonomous driving simulation software such as Carla, Carmaker, VTD, and 51SimOne for virtual simulation testing.
  • the first simulation scene information file F1 and the second simulation scene information file F2 both belong to the scene information file F in FIG. 1 .
  • the simulation test step in step 1 is: loading the simulation scenario information file F1 No. 1 into the simulation software supporting the OpenScenario protocol to perform a virtual simulation test.
  • Step 2 Based on the field test dynamic scenario protocol CX (ie, the issuing protocol in FIG4 ), a field test-oriented field scenario information file No. 1 C1 is created, and the field test is performed.
  • CX the field test dynamic scenario protocol
  • the first site scene information file C1 and the second site scene information file C2 both belong to the scene information file C in FIG. 1 .
  • the No. 1 site scene information file C1 is created using a dynamic scene editing tool.
  • the field test dynamic scene protocol CX describes the movement information of cloud-controlled simulated traffic participants at different times in dynamic scenes.
  • Cloud-controlled simulated traffic participants include human models and vehicle models placed in the actual test site.
  • All cloud-controlled simulated traffic participants can move in the actual test site and are equipped with a communication unit OBU.
  • JSON syntax is a subset of JavaScript object representation syntax. JSON has two structures: objects and arrays. JSON syntax is an existing technology.
  • Objects are stored within curly braces ⁇ and are an unordered collection of name/value pairs.
  • An object begins with an open brace ⁇ and ends with a close brace ⁇ .
  • Each "key” is followed by a colon :, and name/value pairs are separated by commas, .
  • Arrays are stored in brackets [] and are an ordered collection of values.
  • An array begins with a left bracket [ and ends with a right bracket ], and values are separated by commas.
  • the No. 1 site scene information file C1 based on the site test dynamic scene protocol CX includes: scene name, scene number, scene description, number of traffic participants, and traffic participant information.
  • the detailed composition is shown in Figure 4.
  • Traffic participant information includes: OBU number (communication unit number), number of track points, and track information list.
  • Track points are discrete points, each of which includes time, longitude and latitude information, heading angle, and speed information.
  • step 2 Specifically, the site test steps in step 2 are:
  • the scene information file C1 of site 1 is stored in the cloud database as a JSON file;
  • the cloud server reads the information of each traffic participant in the scene information file C2 of site No. 2 in the cloud database, and transmits the trajectory information list to the communication unit of the cloud-controlled simulated traffic participant via 5G according to the OBU number;
  • the cloud-controlled simulated traffic participants include human models and vehicle models placed in the actual test site; all cloud-controlled simulated traffic participants can move in the actual test site and are equipped with a communication unit OBU.
  • Step 3 Convert the simulation test dynamic scene into a field test dynamic scene, and then conduct field testing. This specifically includes the following steps:
  • Step 3A Convert the first simulation scene information file F1 into the second site scene information file C2.
  • the first simulation scene information file F1 in step 1 is converted into the second site scene information file C2 .
  • Step 3A1 Based on the field test dynamic scenario protocol CX, create a second field test scenario information file C2 for the field test;
  • the fields of the second site scene information file C2 are the same as those of the first site scene information file C1, and the data in the fields are to be filled;
  • Step 3A2 Perform data conversion based on the dynamic scenario protocol FX for intelligent driving simulation testing and the dynamic scenario protocol CX for field testing, and convert the data into corresponding fields in the second field scenario information file C2;
  • the conversion program is deployed on the cloud server.
  • the specific process of converting the trajectory of traffic participants based on cloud control simulation is shown in Figure 5:
  • the name is converted into an OBU number.
  • the OBU number is "OBU-0001".
  • the OBU number is "OBU-0010".
  • the position (x, y) in the initialization position under the dynamic scene protocol FX of the simulation test is converted into the longitude and latitude at time 0 under the dynamic scene protocol CX of the field test.
  • the locations in the dynamic scene protocol FX of the simulation test use UTM coordinates, and when converted to the dynamic scene protocol CX of the field test, they need to be switched to WGS84 longitude and latitude coordinates.
  • the principle of coordinate transformation is based on existing technology. Specifically, both coordinates are based on international coordinate system standards and can be implemented using the to_latlon function of the Python utm library.
  • the z coordinate value of the UTM coordinate is not used, only the x and y coordinates are converted.
  • attitude angle h in the initial position under the dynamic scenario protocol FX in the simulation test corresponds to the heading angle under the dynamic scenario protocol CX in the field test. That is, the attitude angle corresponds to the heading angle.
  • “Corresponding” means that the numerical connotation is the same, but the unit, resolution, and syntax are different. The definition of "corresponding” below is the same as that used here.
  • the initialization speed v under the dynamic scene protocol FX of the simulation test corresponds to the speed v at time 0 under the dynamic scene protocol CX of the field test.
  • the time t for each track point and the time t for speed change in the trajectory information list correspond to the time t under the dynamic scenario protocol CX for field testing.
  • the two times t in the dynamic scenario protocol FX for simulation testing are the same value.
  • the current speed v under the dynamic scenario protocol FX of the simulation test corresponds to the current speed under the dynamic scenario protocol CX of the field test.
  • Unit conversion is as follows:
  • the time unit is milliseconds and is represented as an Int64 signed integer.
  • the time of 1.5s in the simulation test dynamic scenario protocol FX is converted to the time of 1500 in the field test dynamic scenario protocol CX.
  • the speed resolution is 0.02 m/s. Converted to a Uint32 unsigned integer, the valid range is [-8191, 8191]. For example, a speed of 1 m/s in a simulation using the dynamic scenario protocol FX converts to a speed of 50 in the field testing using the dynamic scenario protocol CX.
  • the heading angle resolution is 0.0125 degrees. Converted to a Uint16 unsigned integer, the valid range is [0, 28799].
  • the attitude angle h is 180°, which is converted to a heading angle of 14400 in the field test dynamic scenario protocol CX.
  • the unit of latitude and longitude is °, using Float64 floating point number.
  • Step 3A3 Supplement data to the unfilled fields in the second site scene information file C2.
  • the fields that require additional information include: scene name, scene number, scene description, number of cloud-controlled simulated traffic participants, and number of trajectory points for each cloud-controlled simulated traffic participant.
  • the number of cloud-controlled simulated traffic participants is obtained by counting all simulated traffic participants in the No. 1 simulation scenario information file F1.
  • the number of trajectory points of each cloud-controlled simulated traffic participant is obtained by counting the length of the trajectory information list of each simulated traffic participant in the No. 1 simulation scene information file F1.
  • the scene name, scene number, and scene description are used to distinguish the scene information files C in the cloud server scene library and need to be added manually.
  • Step 3B Store the second venue scene information file C2 output in step 3A3 in the cloud database as a JSON file.
  • Step 3C Add the OBU numbers corresponding to their respective communication units to the cloud-controlled simulated traffic participants to ensure the correct distribution of trajectory information.
  • Step 3D Conduct field testing.
  • the cloud server identifies and reads the information of each traffic participant in the scene information file C2 of site No. 2 in the cloud database based on the scene name, scene number, and scene description, and transmits the trajectory information list to the communication unit OBU of the cloud-controlled simulated traffic participants (human model and vehicle model) via 5G according to the OBU number.
  • control units of the human and vehicle models After receiving information from the communication unit, the control units of the human and vehicle models use a tracking control algorithm to control the drive motor and steering servo to achieve trajectory following.
  • the drive motor and steering servo constitute the driving and walking module, which both the human and vehicle models include.
  • the corresponding trajectory information list is issued in real time to the cloud-controlled simulated traffic participants who arrive at the designated starting point.
  • the specific process is as follows:
  • the cloud-controlled simulated traffic participant uploads its own coordinates.
  • the cloud retrieves the nearest neighbor trajectory point number based on the coordinates and sends a fixed number of subsequent trajectory points corresponding to the cloud-controlled simulated traffic participant through a sliding window method.
  • the cloud-controlled simulated traffic participants include a GPS module (GPS receiver), which can obtain real-time positioning and automatically upload positioning information through OBU and cloud communication.
  • GPS receiver GPS receiver
  • the method for judging whether the starting position has been reached is: the cloud-controlled simulated traffic participant uploads its own coordinates, and the cloud calculates the distance and heading angle error with the initial moment of the trajectory point based on the cloud-controlled simulated traffic participant coordinates. If it is less than the specified threshold, such as the distance is 1m and the heading angle error is 10 degrees, it is considered to have reached the starting position.
  • the number of cloud-controlled simulated traffic participants is used to confirm the number of cloud-controlled simulated traffic participants that need to be dispatched, and the number of trajectory points is used to avoid overflow when the trajectory points are indexed.
  • Steps 3A to 3D form a method for converting and generating trajectories of traffic participants based on cloud-controlled simulation.
  • Step 4 Convert the field test dynamic scene into a simulation test dynamic scene, and then conduct a simulation test. This specifically includes the following steps:
  • Step 4A Convert the field test dynamic scene into a simulation test dynamic scene.
  • step 4A1 the communication unit of each cloud-controlled simulated traffic participant uploads the test data of the dynamic scene operation in the field test in step 3D based on 5G communication, including GPS positioning coordinate information, heading angle information, category, OBU number, time and speed.
  • the cloud receives the test data of each cloud-controlled simulated traffic participant.
  • GPS positioning coordinate information is provided in real time by the GPS receiver
  • heading angle information is provided in real time by its own inertial navigation unit
  • category and OBU number are obtained by reading the self-status information file pre-stored in the storage unit.
  • step 4A1 the category and OBU number are pre-modified and stored in the storage unit of the cloud-controlled simulated traffic participant.
  • the time is the accumulated time of the control unit after the dynamic scene starts running, which is obtained by the control unit. That is, the time from the start of the scene is the current control unit time minus the control unit time at the start of the scene.
  • the angular velocity is provided in real time by the wheel speed sensor, which is installed on the motor shaft that drives the travel module.
  • the speed is calculated by the control unit in combination with the angular velocity.
  • the collected data is stored in the cloud database.
  • Reason Each cloud-controlled simulated traffic participant only communicates with the cloud, and all trajectory information is aggregated through the cloud.
  • the site test dynamic scene protocol CX is used to generate the site scene information file C3 No. 3; wherein, the site scene information file C3 No. 3 contains all the required fields of the protocol CX, and the data in the fields is to be filled;
  • Step 4A2 Based on the dynamic scenario protocol FX for intelligent driving simulation testing, a second simulation scenario information file F2 for simulation testing is created;
  • the fields of the second simulation scenario information file F2 are the same as those of the first simulation scenario information file F1;
  • All field information in the second simulation scene information file F2 is temporarily missing and needs to be read from the third site scene information file C3 and filled in through conversion.
  • Step 4A3 According to the dynamic scenario protocol FX for intelligent driving simulation testing and the dynamic scenario protocol CX for field testing, the data of the third field scenario information file C3 is converted into the corresponding fields of the second simulation scenario information file F2.
  • the conversion program is deployed on the cloud server.
  • the conversion process is shown in Figure 6:
  • the OBU number under the dynamic scenario protocol CX of the field test corresponds to the name under the dynamic scenario protocol FX of the simulation test.
  • the categories under the dynamic scenario protocol CX for field testing correspond to the categories under the dynamic scenario protocol FX for simulation testing.
  • the heading angle under the dynamic scene protocol CX in the field test corresponds to the attitude angle h in the initialization position under the dynamic scene protocol FX in the simulation test, and the pitch angle p and roll angle r are set to 0.
  • the speed v at time 0 under the dynamic scene protocol CX in the field test corresponds to the initialization speed v under the dynamic scene protocol FX in the simulation test.
  • the time t under the dynamic scenario protocol CX in the field test corresponds to the time t under the dynamic scenario protocol FX in the simulation test.
  • the current latitude and longitude under the field test dynamic scene protocol CX is converted into the current position (x, y) in the trajectory information list under the simulation test dynamic scene protocol FX.
  • the current speed under the dynamic scene protocol CX in the field test corresponds to the current speed v under the dynamic scene protocol FX in the simulation test.
  • the unit conversion is as follows:
  • the time unit is milliseconds and is represented as an Int64 signed integer.
  • the time 1500 in the field test dynamic scenario protocol CX is converted to the time 1.5s in the simulation test dynamic scenario protocol FX.
  • the velocity resolution is 0.02 m/s, represented by a Uint32 unsigned integer in the range [-8191, 8191].
  • a velocity of 50 in field testing using the dynamic scenario protocol CX translates to a velocity of 1 m/s in simulation testing using the dynamic scenario protocol FX.
  • the heading angle resolution is 0.0125 degrees. Converted to a Uint16 unsigned integer, the valid range is [0, 28799]. For example, a heading angle of 14400 in the field test dynamic scenario protocol CX converts to an attitude angle h of 180° in the simulation test dynamic scenario protocol FX.
  • Step 4A4 Supplement data to the unfilled fields in the second simulation scenario information file F2.
  • the fields that require additional information include: rendering models of each traffic participant, OpenDrive map, and OpenSceneGraph scene model;
  • Step 4B Load the second simulation scenario information file F2 containing the data stored in step 4A4 into simulation software supporting the OpenScenario protocol to perform a virtual simulation test.
  • Simulation software that supports the OpenScenario protocol, such as Carla, Carmaker, VTD, 51SimOne, etc.
  • Steps 4A to 4B form a method for regenerating a test scene based on cloud-controlled simulated traffic participant site.
  • this embodiment converts the virtually generated test scenario into a field-deployed scenario, and at the same time, through simulation and visual verification in a virtual environment, it can discover problems in advance and optimize them, reduce unnecessary repeated tests in field tests, and significantly reduce the cost and time of field tests.
  • the present invention has completed the deployment and verification of the program in the intelligent connected test field.
  • the OpenScenario standard simulation scenario can be converted into the field test scenario specified by the intelligent connected test field through a conversion program.
  • cloud control simulates traffic participants obtaining trajectory information from the cloud and performing tracking actions, enabling real-world smart driving field tests involving multiple traffic participants.
  • cloud-controlled simulated traffic participants After connecting to the cloud communication, cloud-controlled simulated traffic participants upload their movement trajectories, and OpenScenario standard simulation scenarios can be regenerated based on these movement trajectories.

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Evolutionary Computation (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本发明提出了一种面向自动驾驶测试的动态场景仿真与场地测试的闭环方法，包括以下步骤：步骤1、建立面向仿真测试的动态场景并进行仿真测试；步骤2、建立面向场地测试的动态场景并进行场地测试；步骤3、将仿真测试动态场景转化为场地测试动态场景，之后进行场地测试；步骤4、将场地测试动态场景转化为仿真测试动态场景，之后进行仿真测试。本发明结合虚拟仿真和封闭场地测试，生成包含动态交通参与者的动态场景，并实现仿真与场地测试的动态场景之间的互相转化。

Description

面向自动驾驶测试的动态场景仿真与场地测试的闭环方法 技术领域
本发明属于自动驾驶测试领域，尤其涉及一种面向自动驾驶测试的动态场景仿真与场地测试的闭环方法。
背景技术
目前的自动驾驶测试场景生成方法包括以下步骤：首先需要使用场景建模工具建模道路网络、交通标志、交通信号灯、路口结构等静态道路环境，通常采用地图数据或虚拟建模软件进行建模，通常采用OpenDrive等标准；然后使用动态场景编辑工具用于生成各种动态交通参与者，包括车辆、行人、自行车等，并通过轨迹编辑生成交通参与者的运动轨迹，包括路径规划、速度控制等，以模拟其在场景中的运动，通常使用OpenScenario等标准；最后再将生成的动静态场景加载到仿真平台中进行测试和评估，通常包括物理引擎、传感器模拟等功能。
由上，现有技术中存在仿真测试与场地测试割裂的缺陷，难以同时满足真实性和可再生性。具体为：在通过动静态场景设计满足场景生成与仿真可视化之后，不能在场地测试中实地部署，因此不能满足真实性要求；而传统的标准场地测试由于需要人为控制，又不能满足测试场景的一致性和可再生性。
发明内容
本发明的目的在于提供一种面向自动驾驶测试的动态场景仿真与场地测试系统闭环方法，结合虚拟仿真和封闭场地测试，生成包含动态交通参与者的动态场景，并实现仿真与场地测试的动态场景之间的互相转化。采用的技术方案为：
一种面向自动驾驶测试的动态场景仿真与场地测试的闭环方法，包括以下步骤：
步骤1、基于面向智能驾驶仿真测试的动态场景协议FX，建立面向仿真测试的一号仿真场景信息文件F1，并进行仿真测试。
其中，一号仿真场景信息文件F1通过动态场景编辑工具创建，一号仿真场景信息文件F1包含道路网络信息和交通参与者信息；
道路网络信息包括：OpenDrive地图、OpenSceneGraph场景模型；
交通参与者信息包括：名称、类别、交通参与者渲染模型、初始化位置、初始化速度、轨迹信息列表、速度变化。
步骤2、基于场地测试动态场景协议CX，建立面向场地测试的一号场地场景信息文件C1，并进行场地测试。
其中，一号场地场景信息文件C1通过动态场景编辑工具建立；
一号场地场景信息文件C1包含的信息：场景名称、场景编号、场景描述、交通参与者数量、交通参与者信息；
交通参与者信息包括：OBU编号、轨迹点数量、轨迹信息列表；
步骤3、将仿真测试动态场景转化为场地测试动态场景，之后进行场地测试，具体包括以下步骤：
步骤3A、将一号仿真场景信息文件F1转化为二号场地场景信息文件C2：
步骤3A1、基于场地测试动态场景协议CX，创建面向场地测试的二号场地场景信息文件C2；
其中，二号场地场景信息文件C2与一号场地场景信息文件C1的字段相同；
二号场地场景信息文件C2中所有字段，其中的信息为暂缺状态；
步骤3A2、根据仿真测试的动态场景协议FX和场地测试动态场景协议CX，进行数据转化，数据转化到二号场地场景信息文件C2的对应字段中；
步骤3A3、补充数据至二号场地场景信息文件C2中未填充的字段中；
其中，需要补充信息的字段包括：场景名称、场景编号、场景描述、云控模拟交通参与者数量、每个云控模拟交通参与者轨迹点数量；
步骤3B、将步骤3A3中输出的二号场地场景信息文件C2，以JSON文件存储在云端数据库中；
步骤3C、为云控模拟交通参与者添加与各自的通讯单元对应的OBU编号，以实现轨迹信息的正确下发；
步骤3D、进行场地测试：
云端服务器读取云端数据库中二号场地场景信息文件C2的各交通参与者信息，根据OBU编号，对应将轨迹信息列表通过5G传输至云控模拟交通参与者的通讯单元；
云控模拟交通参与者包括放置于实际测试场地中的人模型和车模型；云控模拟交通参与者均可在实际测试场地中移动，并均安装了通讯单元OBU。
优选地，步骤3之后还包括以下步骤：
步骤4、将场地测试动态场景转化为仿真测试动态场景，之后进行仿真测试。具体包括以下步骤：
步骤4A、将场地测试动态场景转化为仿真测试动态场景：
步骤4A1、各云控模拟交通参与者的通讯单元，基于5G通讯上传步骤3D中场地测试动态场景运行中的测试数据，包括GPS定位坐标信息、航向角信息、类别和OBU编号，由云端接收各云控模拟交通参与者测试数据；
GPS定位坐标信息由GPS接收机实时提供，航向角信息由自带的惯性导航单元实时提供，类别和OBU编号通过读取存储单元中预先存储的自身状态信息文件获得；
采用场地测试动态场景协议CX生成三号场地场景信息文件C3；其中，三号场地场景信息文件C3中，存在字段，字段中的数据待填充；
将采集的各云控模拟交通参与者测试数据填充至三号场地场景信息文件C3的对应字段中；
步骤4A2、基于面向智能驾驶仿真测试的动态场景协议FX，建立面向仿真测试的二号仿真场景信息文件F2；
其中，二号仿真场景信息文件F2与一号仿真场景信息文件F1的字段相同；
二号仿真场景信息文件F2中所有字段，其中的信息为暂缺状态；
步骤4A3、根据仿真测试的动态场景协议FX和场地测试动态场景协议CX，将三号场地场景信息文件C3的数据转化到二号仿真场景信息文件F2的对应字段中；
步骤4A4、补充数据至二号仿真场景信息文件F2中没有进行填充的字段中；
其中，需要补充信息的字段包括：各交通参与者渲染模型、OpenDrive地图、OpenSceneGraph场景模型；
步骤4B、将步骤4A4中存储数据的二号仿真场景信息文件F2加载到支持OpenScenario协议的仿真软件，进行虚拟仿真测试。
优选地，步骤3A2具体包括以下转化步骤：
名称转化为OBU编号、初始化位置中的坐标转化为0时刻的经纬度、初始化位置中的姿态角h转化为航向角、初始化速度转化为0时刻的速度、时间t转化、当前速度转化。
优选地，步骤4A2具体包括以下转化步骤：
OBU编号转化为名称、0时刻的经纬度转化为初始化位置中的坐标、0时刻的航向角转化为初始化位置中的姿态角h、0时刻的速度转化为初始化速度、时间t转化、当前速度转化。
与现有技术相比，本发明的优点为：
1、实现了自动驾驶动态场景在仿真测试与场地测试之间的转化与再生。具体的：
通过场景转化在场地测试中实地部署，实现了从虚拟环境测试到实际道路测试的转化，使得仿真测试场景能够直接应用于实际场地测试，保证了自动驾驶测试的真实性，提高了自动驾驶场地测试的一致性和可再生性。
2、实现了从虚拟场景到场地测试的全流程闭环，为自动驾驶系统的关键边缘测试提供了一种全面、高效的解决方案。
3、在动静态场景设计生成后，或场景再生之后，可进行仿真可视化。通过在虚拟环境中进行仿真可视化验证，可以提前发现问题并进行优化，减少实地测试中的不必要重复测试，可以大幅减少场地测试的成本和时间。
4、通过动态场景设计和转化、再生，可以更真实、低成本、快速地评估自动驾驶系统在各种复杂场景下的性能和安全性，推动自动驾驶测评技术的发展和应用。
附图说明
图1为面向自动驾驶测试的动态场景仿真与场地测试的闭环方法的流程图；
图2为基于OpenScenario标准承载多交通参与者动态场景仿真测试协议；
图3为多交通参与者仿真运动控制方法的动作要素；
图4为面向场地测试的云控模拟交通参与者场地测试动态场景协议；
图5为云控模拟交通参与者轨迹转化生成方法；
图6为云控模拟交通参与者场地测试场景再生方法。
具体实施方式
下面将结合示意图对本发明的面向自动驾驶测试的动态场景仿真与场地测试的闭环方法进行更详细的描述，其中表示了本发明的优选实施例，应该理解本领域技术人员可以修改在此描述的本发明，而仍然实现本发明的有利效果。因此，下列描述应当被理解为对于本领域技术人员的广泛知道，而并不作为对本发明的限制。
如图1～6，本发明提出的仿真与场地测试的闭环方法，着重于实现交通多目标动态场景仿真和场地测试真实目标轨迹的转化与再生，以进一步提高自动驾驶测试的真实性、一致性和可再生性。
步骤1、基于面向智能驾驶仿真测试的动态场景协议FX，建立面向仿真测试的一号仿真场景信息文件F1，并进行仿真测试。
其中，面向智能驾驶仿真测试的动态场景协议FX是基于OpenScenario标准开发。
OpenScenario是一种用于描述自动驾驶场景的开放式协议，主要用于描述自动驾驶场景中涉及多个实体的同步操作，例如车辆的换道或变速动作，通常用于自动驾驶仿真测试中动态场景的定义。本实施例中的面向仿真测试的多交通参与者动态场景协议基于OpenScenario 1.0协议开发，提供一种简单且标准化的方式来定义复杂的交通场景，包括车辆、行人等多种交通参与者。
本实施例中的面向仿真测试的动态场景协议FX通过OpenScenario 1.0标准承载所有必须的动态场景信息，交通参与者的运动轨迹由轨迹跟随动作来实现，交通参与者的速度变化由变速动作来实现，用时间触发器作为交通参与者动作的执行条件，如图2所示。其中，1～3为轨迹点。
其中，一号仿真场景信息文件F1通过动态场景编辑工具创建。在本实施例中，一号仿真场景信息文件F1、二号仿真场景信息文件F2的字段相同。
一号仿真场景信息文件F1包括：道路网络信息和交通参与者信息，详细组成如图3所示。
其中，道路网络信息包括：OpenDrive地图、OpenSceneGraph场景模型。
交通参与者信息包括：名称、类别、交通参与者渲染模型、初始化位置、初始化速度、轨迹信息列表、速度变化(t时间内速度变化为v)。
创建基于协议FX的一号仿真场景信息文件F1的详细步骤如下：
导入OpenDrive等格式的高精度地图，为了保持仿真模拟和场地测试的一致性，需要提前获取通过高精地图供应商制作的封闭测试场地OpenDrive地图和OpenSceneGraph场景模型。
采用支持OpenScenario标准的驾驶场景编辑器，建立动态场景，添加动态场景所需的交通参与者，为每个参与者设置名称，并建立连续轨迹，不仅需要在地图上选择轨迹点坐标，还需要调整不同轨迹点所对应的车辆速度和航向角，以及设置等待时间。
通过上述方法制作的基于协议FX的场景信息文件F，同时也满足OpenScenario标准，为xosc文件，可加载到如Carla、Carmaker、VTD、51SimOne等自动驾驶仿真软件进行虚拟仿真测试。
一号仿真场景信息文件F1、二号仿真场景信息文件F2均属于图1中的场景信息文件F。
具体的，步骤1中的仿真测试步骤：将一号仿真场景信息文件F1加载到支持OpenScenario协议的仿真软件，进行虚拟仿真测试。
步骤2、基于场地测试动态场景协议CX(即图4中的下发协议)，建立面向场地测试的一号场地场景信息文件C1，并进行场地测试。
在本实施例中，一号场地场景信息文件C1、二号场地场景信息文件C2均属于图1中的场景信息文件C。
一号场地场景信息文件C1通过动态场景编辑工具建立。
场地测试动态场景协议CX描述了云控模拟交通参与者在动态场景中不同时刻的运动信息。
云控模拟交通参与者包括放置于实际测试场地中的人模型和车模型。
云控模拟交通参与者均可在实际测试场地中移动，并均安装了通讯单元OBU。
其中，基于场地测试动态场景协议CX是基于JSON语法开发。JSON语法是JavaScript对象表示语法的子集，JSON的两种结构为对象和数组，JSON语法为现有技术。
对象是通过大括号{}保存，是一个无序的名称/值对集合。一个对象以左括号{开始，右括号}结束。每个"键"后跟一个冒号:，名称/值对使用逗号,分隔。
数组是通过中括号[]保存，是值(value)的有序集合。一个数组以左中括号[开始，右中括号]结束，值之间使用逗号,分隔。
其中，基于场地测试动态场景协议CX的一号场地场景信息文件C1中包含：场景名称、场景编号、场景描述、交通参与者数量、交通参与者信息，详细组成如图4所示。
交通参与者信息包括：OBU编号(通讯单元的编号)、轨迹点数量、轨迹信息列表。其中，轨迹点为离散点，每个点包括了时间、经纬度信息、航向角、速度信息。
具体的，步骤2中的场地测试步骤：
一号场地场景信息文件C1以JSON文件存储在云端数据库中；
为云控模拟交通参与者添加与各自的通讯单元对应的OBU编号，以实现轨迹信息的正确下发；
云端服务器读取云端数据库中二号场地场景信息文件C2的各交通参与者信息，根据OBU编号，对应将轨迹信息列表通过5G传输至云控模拟交通参与者的通讯单元；
云控模拟交通参与者包括放置于实际测试场地中的人模型和车模型；云控模拟交通参与者均可在实际测试场地中移动，并均安装通讯单元OBU。
步骤3、将仿真测试动态场景转化为场地测试动态场景，之后进行场地测试。具体包括以下步骤：
步骤3A、将一号仿真场景信息文件F1转化为二号场地场景信息文件C2。
如图5所示，将步骤1中的一号仿真场景信息文件F1转化为二号场地场景信息文件C2。
步骤3A1、基于场地测试动态场景协议CX，创建面向场地测试的二号场地场景信息文件C2；
其中，二号场地场景信息文件C2与一号场地场景信息文件C1的字段相同，字段中的数据待填充；
即二号场地场景信息文件C2中所有字段信息为暂缺状态，需要读取一号仿真场景信息文件F1并通过转化填充。
步骤3A2、根据面向智能驾驶仿真测试的动态场景协议FX和场地测试动态场景协议CX，进行数据转化，数据转化到二号场地场景信息文件C2的对应字段中；
转化程序部署在云端服务器，基于云控模拟交通参与者轨迹转化过程具体如图5所示：
1、名称转化为OBU编号。
对于仿真场景交通参与者的名称，为了使云控模拟交通参与者在场地测试部署并受控，在场地测试动态场景中需要进行一一映射操作。
如"fcar1"映射为一号云控模拟交通参与者，OBU编号为"OBU-0001"。
如"fhuman1"映射为十号云控模拟交通参与者，OBU编号为"OBU-0010"。
2、仿真测试的动态场景协议FX下的初始化位置中的位置(x,y)转化为场地测试动态场景协议CX下的0时刻的经纬度。
仿真测试的动态场景协议FX中位置采用UTM坐标，转化到场地测试动态场景协议CX需切换到WGS84经纬度坐标。
其中，坐标变换的原理为现有技术。具体的，两种坐标均为国际坐标系标准，可采用Python的utm库的to_latlon函数实现。
若UTM坐标为[x,y,z]，所在UTM分区符号记为utm_zone，所在南北半球符号分别"N"和"S"，则通过Python计算北半球经纬度方法为lat,lon＝utm.to_latlon(x,y,utm_zone,'N')。
UTM坐标的z坐标值不使用，只进行x坐标和y坐标的转化。
3、仿真测试的动态场景协议FX下的初始化位置中的姿态角h对应为场地测试动态场景协议CX下的航向角。即姿态角对应航向角。“对应”即数值内涵一致，但单位或分辨率及语法不一致，以下“对应”的定义与该处相同。
对于三轴姿态角[h,p,r]，由于常态下俯仰角p和翻滚角r均接近0°，故只保留仿真场景轨迹中的航向角h作为场地测试轨迹中的航向角heading。
4、仿真测试的动态场景协议FX下的初始化速度v对应为场地测试动态场景协议CX下的0时刻速度v。
5、仿真测试的动态场景协议FX下，轨迹信息列表中各轨迹点时间t、速度变化中的变化时间t均对应为场地测试动态场景协议CX下的时间t。其中，仿真测试的动态场景协议FX中的两个时间t为同一个数值。
6、仿真测试的动态场景协议FX下的当前速度v对应为场地测试动态场景协议CX下的当前速度。
7、单位转化。单位转化具体为：
据实际传输效率需要，仿真测试的动态场景协议FX到场地测试动态场景协议CX的转化过程中，单位需进行转换，部分数值还需从浮点数压缩为特定分辨率的无符号整数，详细如下。
场地测试动态场景协议CX下，时间单位为毫秒，采用Int64符号整数表示。如仿真测试的动态场景协议FX下的时间1.5s，转化为场地测试动态场景协议CX下的时间1500。
场地测试动态场景协议CX下，速度分辨率为0.02m/s，转换为Uint32无符号整数，有效范围为[-8191,8191]之间。如仿真测试的动态场景协议FX下的速度1m/s，转化为场地测试动态场景协议CX下的速度50。
场地测试动态场景协议CX下，航向角分辨率为0.0125度，转换为Uint16无符号整数，有效范围在[0,28799]之间。如仿真测试的动态场景协议FX下姿态角h为180°，转化为场地测试动态场景协议CX下的航向角14400。
经纬度单位为°，采用Float64浮点数。
步骤3A3、补充数据至二号场地场景信息文件C2中未填充的字段中。
需要补充信息的字段包括：场景名称、场景编号、场景描述、云控模拟交通参与者数量、每个云控模拟交通参与者轨迹点数量。
其中，计算云控模拟交通参与者数量、每个云控模拟交通参与者的轨迹点数量的过程如下：
云控模拟交通参与者数量：通过对一号仿真场景信息文件F1中所有仿真交通参与者进行计数得到。
每个云控模拟交通参与者的轨迹点数量：通过对一号仿真场景信息文件F1中各仿真交通参与者轨迹信息列表长度计数得到。
场景名称、场景编号、场景描述是为了对云端服务器场景库各场景信息文件C进行区分，需要手动添加。
步骤3B、将步骤3A3中输出的二号场地场景信息文件C2，以JSON文件存储在云端数据库中。
步骤3C、为云控模拟交通参与者添加与各自的通讯单元对应的OBU编号，以实现轨迹信息的正确下发。
步骤3D、进行场地测试。
云端服务器根据场景名称、场景编号、场景描述识别并读取云端数据库中二号场地场景信息文件C2的各交通参与者信息，根据OBU编号，对应将轨迹信息列表通过5G传输至云控模拟交通参与者(人模型和车模型)的通讯单元OBU。
人模型和车模型的控制单元接收通讯单元输出的信息后，通过循迹控制算法控制驱动电机和转向舵机实现轨迹跟随。其中，驱动电机和转向舵机组成驱动行走模块，人模型和车模型均包括驱动行走模块。
基于以上人模型和车模型自动循迹行驶，实现自动驾驶动态场景的自动化测试。
其中，对到达指定起点位置的云控模拟交通参与者实时下达对应的轨迹信息列表，具体过程为：
1.云控模拟交通参与者上传自身坐标，云端根据坐标检索最近邻轨迹点序号，通过滑动窗口的方式，下发云控模拟交通参与者对应轨迹的后续固定数量轨迹点。
其中，云控模拟交通参与者包含GPS模块(GPS接收机)，GPS模块可获得实时定位，通过OBU和云端通讯实现自动定位信息上传。
其中，到达起点位置的判断方法：云控模拟交通参与者上传自身坐标，云端根据云控模拟交通参与者坐标计算与轨迹点初始时刻距离和航向角误差，小于指定阈值，如距离1m，如航向角误差10度，则认为到达起点位置。
2.云控模拟交通参与者通过OBU接收轨迹点列表数据后，通过循迹控制算法实现预定路线行驶。
3.当云控模拟交通参与者到达轨迹终点后，云端对其下发停止信号。
此外，云控模拟交通参与者数量用于确认云控模拟交通参与者所需调度的数量，轨迹点数量用于避免轨迹点下发索引时越界溢出。
步骤3A～3D，形成基于云控模拟交通参与者轨迹转化生成方法。
步骤4、将场地测试动态场景转化为仿真测试动态场景，之后进行仿真测试。具体包括以下步骤：
步骤4A、将场地测试动态场景转化为仿真测试动态场景。
步骤4A1、各云控模拟交通参与者的通讯单元，基于5G通讯上传步骤3D中场地测试动态场景运行中的测试数据，包括GPS定位坐标信息、航向角信息、类别、OBU编号、时间和速度，由云端接收各云控模拟交通参与者测试数据。
其中，GPS定位坐标信息由GPS接收机实时提供，航向角信息由自带的惯性导航单元实时提供，类别和OBU编号通过读取存储单元中预先存储的自身状态信息文件获得。
其中，步骤4A1中，类别和OBU编号预先修改并存入云控模拟交通参与者的存储单元中。
时间为动态场景启动运行后控制单元的累计时间，由控制单元获取。即从场景启动开始的时间，需要当前时刻的控制单元时间减去场景开始时刻的控制单元时间。
角速度由轮速传感器实时提供，轮速传感器安装在驱动行走模块的电机轴上。速度由控制单元结合角速度计算得出。
将采集到的数据存储在云端数据库。原因：各云控模拟交通参与者只和云端进行通讯，通过云端汇总所有轨迹信息。
采用场地测试动态场景协议CX生成三号场地场景信息文件C3；其中，三号场地场景信息文件C3中，存在协议CX所有必须字段，字段中的数据待填充；
三号场地场景信息文件C3中所有字段信息为暂缺状态，需要读取上述采集数据进行填充；
将采集的数据填充至三号场地场景信息文件C3的对应字段中；
步骤4A2、基于面向智能驾驶仿真测试的动态场景协议FX，建立面向仿真测试的二号仿真场景信息文件F2；
二号仿真场景信息文件F2与一号仿真场景信息文件F1的字段相同；
二号仿真场景信息文件F2中所有字段信息为暂缺状态，需要读取三号场地场景信息文件C3并通过转化填充。
步骤4A3、根据面向智能驾驶仿真测试的动态场景协议FX和场地测试动态场景协议CX，将三号场地场景信息文件C3的数据转化到二号仿真场景信息文件F2的对应字段中。
转化程序部署在云端服务器，转化过程如图6所示：
1、场地测试动态场景协议CX下的OBU编号对应为仿真测试的动态场景协议FX下的名称。
2、场地测试动态场景协议CX下的类别对应为仿真测试的动态场景协议FX下的类别。
3、场地测试动态场景协议CX下的0时刻的经纬度转化为仿真测试的动态场景协议FX下的初始化位置中的位置(x,y)。
将场地测试动态场景协议CX的WGS84经纬度坐标切换至仿真测试的动态场景协议FX中位置采用UTM坐标。采用Python的utm库的from_latlon函数实现。若经纬度为[lat,lon]，则通过Python计算UTM坐标信息方法为[x,y,zone,band]＝utm.from_latlon(lat,lon)。其中zone为所在UTM分区符号，band为所在南北半球符号，分别"N"和"S"。
4、场地测试动态场景协议CX下的航向角对应为仿真测试的动态场景协议FX下的初始化位置中的姿态角h，俯仰角p和翻滚角r设置为0。
5、场地测试动态场景协议CX下的0时刻速度v对应为仿真测试的动态场景协议FX下的初始化速度v。
6、场地测试动态场景协议CX下的时间t对应为仿真测试的动态场景协议FX下的时间t。
7、场地测试动态场景协议CX下的当前经纬度转化为仿真测试的动态场景协议FX下的轨迹信息列表中的当前位置(x,y)。
采用Python的utm库的from_latlon函数实现。若经纬度为[lat,lon]，则通过Python计算UTM坐标信息方法为[x,y,zone,band]＝utm.from_latlon(lat,lon)。其中zone为所在UTM分区符号，band为所在南北半球符号，分别"N"和"S"。
8、场地测试动态场景协议CX下的当前速度对应为仿真测试的动态场景协议FX下的当前速度v。
9、单位转化。
单位转化具体为：
根据OpenScenario协议的要求，场地测试动态场景协议CX到仿真测试的动态场景协议FX的转化过程中，单位需进行转换，部分数值还需从特定分辨率的无符号整数还原为浮点数，详细如下。
场地测试动态场景协议CX下，时间单位为毫秒，采用Int64符号整数表示。如场地测试动态场景协议CX下的时间1500，转化为仿真测试的动态场景协议FX下的时间1.5s。
场地测试动态场景协议CX下，速度分辨率为0.02m/s，采用Uint32无符号整数表示，有效范围为[-8191,8191]之间。如场地测试动态场景协议CX下的速度50，转化为仿真测试的动态场景协议FX下的速度1m/s。
场地测试动态场景协议CX下，航向角分辨率为0.0125度，转换为Uint16无符号整数，有效范围在[0,28799]之间。如场地测试动态场景协议CX下的航向角14400，转化为仿真测试的动态场景协议FX下的姿态角h为180°。
步骤4A4、补充数据至二号仿真场景信息文件F2中没有进行填充的字段中。
其中，需要补充信息的字段包括：各交通参与者渲染模型、OpenDrive地图、OpenSceneGraph场景模型；
步骤4B、将步骤4A4中存储数据的二号仿真场景信息文件F2导加载到支持OpenScenario协议的仿真软件，进行虚拟仿真测试。
支持OpenScenario协议的仿真软件，如Carla、Carmaker、VTD、51SimOne等。
步骤4A～4B，形成基于云控模拟交通参与者场地测试场景再生方法。
综上，本实施例通过将虚拟生成的测试场景转化为实地部署的场景，同时通过在虚拟环境中进行仿真可视化验证，可以提前发现问题并进行优化，减少实地测试中的不必要重复测试，可以大幅减少实地测试的成本和时间。
本发明在智能网联测试场已完成了程序的部署与验证。
OpenScenario标准仿真场景可以通过转化程序变为智能网联测试场规定的场地测试场景。
通过云端运行场景开始，云控模拟交通参与者从云端获取轨迹信息并执行循迹动作，可真实运行包含多交通参与者的智能驾驶场地测试。
云控模拟交通参与者在接入云端通讯后上传运动轨迹，可以基于该运动轨迹再生出OpenScenario标准的仿真场景。
上述仅为本发明的优选实施例而已，并不对本发明起到任何限制作用。任何所属技术领域的技术人员，在不脱离本发明的技术方案的范围内，对本发明揭露的技术方案和技术内容做任何形式的等同替换或修改等变动，均属未脱离本发明的技术方案的内容，仍属于本发明的保护范围之内。

Claims (4)

  1. 一种面向自动驾驶测试的动态场景仿真与场地测试的闭环方法，其特征在于，包括以下步骤：
    步骤1、基于面向智能驾驶仿真测试的动态场景协议FX，建立面向仿真测试的一号仿真场景信息文件F1，并进行仿真测试；
    其中，一号仿真场景信息文件F1通过动态场景编辑工具创建，一号仿真场景信息文件F1包含道路网络信息和交通参与者信息；
    道路网络信息包括：OpenDrive地图、OpenSceneGraph场景模型；
    交通参与者信息包括：名称、类别、交通参与者渲染模型、初始化位置、初始化速度、轨迹信息列表、速度变化；
    步骤2、基于场地测试动态场景协议CX，建立面向场地测试的一号场地场景信息文件C1，并进行场地测试；
    其中，一号场地场景信息文件C1通过动态场景编辑工具建立；
    一号场地场景信息文件C1包含的信息：场景名称、场景编号、场景描述、交通参与者数量、交通参与者信息；
    交通参与者信息包括：OBU编号、轨迹点数量、轨迹信息列表；
    步骤3、将仿真测试动态场景转化为场地测试动态场景，之后进行场地测试，具体包括以下步骤：
    步骤3A、将一号仿真场景信息文件F1转化为二号场地场景信息文件C2：
    步骤3A1、基于场地测试动态场景协议CX，创建面向场地测试的二号场地场景信息文件C2；
    其中，二号场地场景信息文件C2与一号场地场景信息文件C1的字段相同；
    二号场地场景信息文件C2中所有字段，其中的信息为暂缺状态；
    步骤3A2、根据仿真测试的动态场景协议FX和场地测试动态场景协议CX，进行数据转化，数据转化到二号场地场景信息文件C2的对应字段中；
    步骤3A3、补充数据至二号场地场景信息文件C2中未填充的字段中；
    其中，需要补充信息的字段包括：场景名称、场景编号、场景描述、云控模拟交通参与者数量、每个云控模拟交通参与者轨迹点数量；
    步骤3B、将步骤3A3中输出的二号场地场景信息文件C2，以JSON文件存储在云端数据库中；
    步骤3C、为云控模拟交通参与者添加与各自的通讯单元对应的OBU编号，以实现轨迹信息的正确下发；
    步骤3D、进行场地测试：
    云端服务器读取云端数据库中二号场地场景信息文件C2的各交通参与者信息，根据OBU编号，对应将轨迹信息列表通过5G传输至云控模拟交通参与者的通讯单元；
    云控模拟交通参与者包括放置于实际测试场地中的人模型和车模型；云控模拟交通参与者均可在实际测试场地中移动，并均安装通讯单元OBU。
  2. 根据权利要求1所述的面向自动驾驶测试的动态场景仿真与场地测试的闭环方法，其特征在于，步骤3之后还包括以下步骤：
    步骤4、将场地测试动态场景转化为仿真测试动态场景，之后进行仿真测试。具体包括以下步骤：
    步骤4A、将场地测试动态场景转化为仿真测试动态场景：
    步骤4A1、各云控模拟交通参与者的通讯单元，基于5G通讯上传步骤3D中场地测试动态场景运行中的数据，包括GPS定位坐标信息、航向角信息、类别和OBU编号，由云端接收各云控模拟交通参与者测试数据；
    采用场地测试动态场景协议CX生成三号场地场景信息文件C3；其中，三号场地场景信息文件C3中，存在字段，字段中的数据待填充；
    将采集的各云控模拟交通参与者测试数据填充至三号场地场景信息文件C3的对应字段中；
    步骤4A2、基于面向智能驾驶仿真测试的动态场景协议FX，建立面向仿真测试的二号仿真场景信息文件F2；
    其中，二号仿真场景信息文件F2与一号仿真场景信息文件F1的字段相同；
    二号仿真场景信息文件F2中所有字段，其中的信息为暂缺状态；
    步骤4A3、根据仿真测试的动态场景协议FX和场地测试动态场景协议CX，将三号场地场景信息文件C3的数据转化到二号仿真场景信息文件F2的对应字段中；
    步骤4A4、补充数据至二号仿真场景信息文件F2中没有进行填充的字段中；
    其中，需要补充信息的字段包括：各交通参与者渲染模型、OpenDrive地图、OpenSceneGraph场景模型；
    步骤4B、将步骤4A4中存储数据的二号仿真场景信息文件F2加载到支持OpenScenario协议的仿真软件，进行虚拟仿真测试。
  3. 根据权利要求2所述的面向自动驾驶测试的动态场景仿真与场地测试的闭环方法，其特征在于，步骤3A2具体包括以下转化步骤：
    名称转化为OBU编号、初始化位置中的坐标转化为0时刻的经纬度、初始化位置中的姿态角h转化为航向角、初始化速度转化为0时刻的速度、时间t转化、当前速度转化。
  4. 根据权利要求2所述的面向自动驾驶测试的动态场景仿真与场地测试的闭环方法，其特征在于，步骤4A2具体包括以下转化步骤：
    OBU编号转化为名称、0时刻的经纬度转化为初始化位置中的坐标、0时刻的航向角转化为初始化位置中的姿态角h、0时刻的速度转化为初始化速度、时间t转化、当前速度转化。
PCT/CN2025/092353 2024-04-10 2025-04-30 面向自动驾驶测试的动态场景仿真与场地测试的闭环方法 Pending WO2025214510A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202410427820.2 2024-04-10
CN202410427820.2A CN118364617A (zh) 2024-04-10 2024-04-10 面向自动驾驶测试的动态场景仿真与场地测试的闭环方法

Publications (1)

Publication Number Publication Date
WO2025214510A1 true WO2025214510A1 (zh) 2025-10-16

Family

ID=91883381

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2025/092353 Pending WO2025214510A1 (zh) 2024-04-10 2025-04-30 面向自动驾驶测试的动态场景仿真与场地测试的闭环方法

Country Status (2)

Country Link
CN (1) CN118364617A (zh)
WO (1) WO2025214510A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118364617A (zh) * 2024-04-10 2024-07-19 同济大学 面向自动驾驶测试的动态场景仿真与场地测试的闭环方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107727411A (zh) * 2017-10-30 2018-02-23 青岛慧拓智能机器有限公司 一种自动驾驶车辆测评场景生成系统及方法
US20210270630A1 (en) * 2020-02-28 2021-09-02 International Business Machines Corporation Probe data generating system for simulator
CN116310159A (zh) * 2022-12-22 2023-06-23 嬴彻星创智能科技(上海)有限公司 自动驾驶场景提取方法及装置
CN117094182A (zh) * 2023-10-19 2023-11-21 中汽研（天津）汽车工程研究院有限公司 V2v交通场景构建方法及v2x虚实融合测试系统
CN118364617A (zh) * 2024-04-10 2024-07-19 同济大学 面向自动驾驶测试的动态场景仿真与场地测试的闭环方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107727411A (zh) * 2017-10-30 2018-02-23 青岛慧拓智能机器有限公司 一种自动驾驶车辆测评场景生成系统及方法
US20210270630A1 (en) * 2020-02-28 2021-09-02 International Business Machines Corporation Probe data generating system for simulator
CN116310159A (zh) * 2022-12-22 2023-06-23 嬴彻星创智能科技(上海)有限公司 自动驾驶场景提取方法及装置
CN117094182A (zh) * 2023-10-19 2023-11-21 中汽研（天津）汽车工程研究院有限公司 V2v交通场景构建方法及v2x虚实融合测试系统
CN118364617A (zh) * 2024-04-10 2024-07-19 同济大学 面向自动驾驶测试的动态场景仿真与场地测试的闭环方法

Also Published As

Publication number Publication date
CN118364617A (zh) 2024-07-19

Similar Documents

Publication Publication Date Title
CN112286206B (zh) 自动驾驶的模拟方法、系统、设备、可读存储介质及平台
CN109781431B (zh) 基于混合现实的自动驾驶测试方法及系统
CN111897305B (zh) 一种基于自动驾驶的数据处理方法、装置、设备及介质
CN111061167B (zh) 一种面向智能网联示范区的混合现实自动驾驶的测试方法及虚拟测试平台
CN109765803A (zh) 一种自动驾驶汽车多icu共时空的硬件仿真测试系统及方法
CN114296424B (zh) 仿真测试系统和方法
CN109931937A (zh) 高精导航信息模拟方法及系统
CN107844635B (zh) 实现bim信息与交通仿真信息集成系统及其集成方法
CN113378413B (zh) 基于Vissim构建仿真路网平台系统、方法、设备及存储介质
CN109993849A (zh) 一种自动驾驶测试场景再现模拟方法、装置及系统
CN110793548A (zh) 基于gnss接收机硬件在环的虚实结合的导航仿真测试系统
JP2024035047A (ja) Ｖｉｌシステムベースの自律走行機能の検証方法
CN114895574B (zh) 用于无人驾驶矿车的仿真系统、方法以及存储介质
CN103164579A (zh) 一种实现微观交通仿真器与驾驶模拟器交互式一体化实验平台的方法
Gruyer et al. From virtual to reality, how to prototype, test and evaluate new ADAS: Application to automatic car parking
CN112860575A (zh) 一种交通环境融合感知在环的自动驾驶算法仿真测试方法
CN111816022A (zh) 一种模拟场景的仿真方法、装置、存储介质以及电子设备
WO2025214510A1 (zh) 面向自动驾驶测试的动态场景仿真与场地测试的闭环方法
CN109060370A (zh) 对自动驾驶车辆进行车辆测试的方法及装置
CN109859562A (zh) 数据生成方法、装置、服务器及存储介质
CN115655748A (zh) 多目标运动事件实时测量方法及装置、设备、介质
CN117031979A (zh) 一种基于仿真环境的规划算法测试方法及系统
CN109376664A (zh) 机器学习训练方法、装置、服务器和介质
CN117872803A (zh) 一种数字孪生现实场景的智能网联仿真测试系统
Guvenc et al. Simulation Environment for Safety Assessment of CEAV Deployment in Linden

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 25785973

Country of ref document: EP

Kind code of ref document: A1