With the development of autopilot technology, the importance of simulation test to the safe landing of autopilot is more and more obvious in the field of development and verification. Standards such as opendrive, openscenario and OSI, drafted by ASAM, are important cornerstone in the field of integration and test evaluation of simulation system for autopilot, which has been paid more and more attention worldwide, and gradually become the vehicle factories Uniform data format standards used by suppliers and simulation tool providers.
Recently, the Chinese version of ASAM opendrive 1.6 and openscenario 1.0 hosted by c-asam and translated by Liangdao has been officially launched on the ASAM official website (https://www.asam.net/). The launch of opendrive and openscenario standard file versions is the first time ASAM has released standards for languages other than English, and is a milestone for ASAM and c-asam.
This article will bring you a detailed interpretation of the ASAM openscenario 1.0 standard.
1. Highlights of openscenario 1.0 Chinese version
It is the first time in the world to accurately define part of the scene content in Chinese, such as actions, triggers, etc.,Solved the problem of inconsistent language description of Chinese scenes and lack of clear vocabulary translation.There are more than 100 keywords in Chinese vocabulary, which cover the key definitions of scenes, computer languages and data description languages. By optimizing the reasonable collocation of the same words in different contexts, this paper achieves a perfect match between the full text and a single word, and gives corresponding suggestions for the use of corresponding words in Chinese in the future.
2. Interpretation of OpenSCENARIO 1.0 standard
Part 1: Concept of scene composition
The OpenSCENARIO data format mainly describes the dynamic information in the automated driving simulation test. This standard mainly attempts to solve the problem of objective scene information. The main elements of the scene are who (target object), where (relationship with the road or reference), what happened (the description of the scene action), and how long it lasted (the time the scene was running). The combination of these objective information can effectively cover the establishment of all test cases, realize high-coverage scenario testing, and enhance the value of the scenario. At present, the standard focuses on the standardized description of this information, and provides a mechanism and foundation for data interaction for many enterprises.
OpenSCENARIO 1.0 Each scene contains three basic components
In addition, the scene is given two additional concepts, namely 4. Catalogs and 5. parameter declaration, to allow multiple scenes to reuse scene elements.
1. Highway network
RoadNetwork is composed of entityEntityInstance，The entity here refers to:
These different entity instances interact through instructions contained in the Storyboard scene.
Storyboard It covers a complete description of the scene. The concept structure and naming of melodrama are similar to the narrative style of dramas and other dramas.Each storyboard scene contains an initialization element (init) and one or more scene content stories.
The storyboard builds the scene into a script with dynamic content in the virtual world:
▲Structural display of the scene script
▲Storyboard element structure
The final action taken by the actor Actor is triggered by the condition, where the actor actor refers to the entity instance that participates in the action. More generally speaking,Trigger Trigger will use the condition to start the action set Act and event Event, or stop the action set Act and the scenario storyboard.Therefore, Condition can be considered as the basic module used to define dynamic behavior and interaction.
PartⅡ: Scene creation example
This scenario is suitable for countries driving on the left. The scene is about: the externally controlled Ego vehicle (Ego) is driving on a city road and heading to an intersection (the right side of the vehicle faces the intersection), followed by two vehicles c1 and c2 that will affect it Vehicles (the movement of the two vehicles is controlled by the scene). The third vehicle (c3) that will have an impact waits for a right turn at the intersection. When the Ego vehicle (Ego) approaches the intersection, c1 and c2 begin to overtake. c3 then starts to turn right, which forces c1 and c2 to make an emergency brake. The following figure identifies the initial position of the vehicle.
Example: the initial position of the vehicle
1. Init initial segment
The following XML example shows how Action uses the global coordinate system to locate the vehicle Ego vehicle (Ego). A similar action (not shown) is used to illustrate the speed and position of other vehicles.
init XML code
2. Stories scene content
Two scene content Story instances are used to describe the scene:
The story describing overtaking and emergency braking, named AbortedOvertake
Contains two action sets Act:
Ø Control overtaking behavior AbortedOvertakeAct1
Ø Control emergency brake AbortedOvertakeAct2
Describe the story of turning right, named RightTurn
Contains a single action set Act:
Ø Control right turn RightTurnAct
Story XML code
3. Act action set
When the Ego vehicle approaches the intersection, the vehicle c3 is triggered to turn right action set Act (RightTurnAct)
Act XML code
4. Maneuver Groups operation group
Each operation group ManeuverGroup assigns an operation Maneuver to a car from the catalog. In this operation, Maneuver will direct the vehicle to change lanes, accelerate, and then return the vehicle to the original lane and drive in front of the Ego vehicle (Ego).
ManeuverGroup XML code
5. Maneuver operation
The operation Maneuver can be repeatedly referenced, so the overtake catalog can be generated by defining the catalog operation Catalog Maneuver—operate the Overtake Ego Vehicle to define the overtake action for the ego vehicle of the vehicle.
Maneuver XML code
After completing the catalog and operation Maneuver definitions, you can refer to the operation in the operation groups of vehicles c1 and c2 in this example.
Maneuver2 XML code
In this example, the lane change action Action should be started as soon as its parent action set Act is triggered. Since the event Events need to apply the trigger Trigger to the action Actions, a small condition condition will be used to trigger the immediate execution.
Event XML code
In summary, we can use OpenSCENARIO to build the expected test scenario. So how to use OpenSCENARIO to support simulation testing? How to use it with static scenes, design operation domains, data scene annotations, etc.? The ASAM OpenX series of standards provide a complete standard plan for automated driving simulation testing. Let's take a look at the standard system of ASAM OpenX.
Three, ASAM OpenX simulation standard
In order to solve the problems of scene-based simulation testing, actual development and application of intelligent networked vehicles, the interface between OEMs, suppliers and simulation platforms is not uniform, and there are many types of data formats. ASAM has launched the OpenX series of standards in the simulation field and promised to As a long-term standard to promote future industrial development, it has gained global attention.
The ASAM simulation format standard as a whole includes the following seven aspects:
▲ ASAM OpenX series standard system
The ASAM OpenX standard provides the industry with a complete set of unified interfaces and data format standards, making simulation development more effective. ASAM officially provides OpenSCENARIO1.0 standard documents and cases, and provides verification tools. Users can use the API (internal testing phase) provided by ASAM to verify whether the OSC 1.0 scene files generated by third-party simulation software comply with the standard specifications.
4. Join C-ASAM to participate in standard formulation
The driving scenes in China are very distinctive. Not only are the road structures, traffic signs, traffic lights and other forms different, but the mixed traffic conditions of people and vehicles also add a lot of difficulty to the dynamic scenes. In order to solve many problems related to scenarios with Chinese characteristics in a more targeted manner, C-ASAM actively participates in the formulation of the OpenX series of standards.
If you want to participate in the formulation and revision of the OpenX series of standards more extensively and in-depth, promote the research process of domestic simulation scenarios, and promote the integration of my country's simulation scenario technology with international frontiers, welcome to join C-ASAM. Please get in touch with the C-ASAM liaison:Wang Yunzhi, email@example.com
▲Member units that have joined the C-ASAM working group
The C-ASAM working group was jointly established by China Automotive Data and ASAM. As the only representative unit of the ASAM Association in China, it is responsible for the overall management of ASAM members in China, holding C-ASAM related meetings and training, and regularly updating members of the ASAM standard research progress, Promote international cooperation and other related matters. The standard fields involved are simulation, data management and analysis, measurement automation, software development, measurement and calibration, diagnosis, and ECU network.
Bright Road Intelligence
Liangdao Intelligence is a perception development and test verification service provider. Its core technology is the application development of lidar and multi-sensor fusion. It has R&D teams in both China and Germany. Liangdao Intelligent is a member unit of ASAM and C-ASAM. Since 2019, it has participated in promoting the formulation of ASAM OpenX series standards, and has undertaken the translation of the official Chinese version of ASAM OpenSCENARIO and OpenDRIVE standards.
Weltweit führend in der Anwendung von LiDAR
Dinge zu entdecken, die zuvor nie gesehen wurden und Wege zu beleuchten, die in die Zukunft führen
Aufrichtigkeit gegenüber Mitmenschen,
Sachlichkeit gegenüber Dingen und Gelassenheit,
Offen und inklusiv zu bleiben,
sich an Veränderungen anzupassen sowie schnell und agil zu reagieren,
Fokus auf Innovationen zu legen und dabei zu bleiben,
wertvolle Dinge weiterzumachen und zum Erfolg zu bestehen.