1. What is a high precision data distribution engine
1.1 Overview of high precision maps
The main differences between High Definitation Map (HD Map) and normal navigation digital Map are higher accuracy and richer information. The higher accuracy is mainly reflected in the higher precision of the absolute coordinates of the map (refers to the accuracy between the location of a target on the map and the external real world objects), can be accurate to the centimeter level; The rich information is mainly reflected in the high precision map not only contains road information, but also covers almost all the surrounding static information related to traffic.
Compared with common navigation electronic map, high precision map contains more abundant and accurate road traffic information. In addition, in terms of application scenarios, ordinary navigation maps are mainly used by drivers, while high-precision maps are machine-oriented and used by autonomous vehicles.
Precision is the biggest difference between high precision map and common navigation electronic map. The precision of the ordinary vehicle electronic navigation map is in commonly 10 meters, high precision mapping applications in the field of automatic driving, need accurate positioning on the specific lanes, also need to know all may participate in automated driving around the decision of road and traffic information, accuracy need to be 10 to 20 centimeters, the precision is basically similar to a lane edge width, In order to ensure that intelligent driving cars do not cross into other lanes and avoid the risk of side collision with other vehicles.
Whereas a navigational digital map maps the roads, a high-precision map maps not only the roads, but also how many lanes there are on a road, a true reflection of the actual road pattern.
More abundant high-precision map information is mainly reflected in the following aspects:
Accurate road shape: slope, curvature, heading, elevation, roll data for each lane.
Detailed lane line information: the lane lines between lanes are dotted, solid or double yellow lines, the color of the line, road isolation belt, isolation belt material will be described.
In addition, the absolute geographic coordinates, physical dimensions and characteristics of crosswalks, roadside billboards, speed limit signs, traffic lights, roadside telephone booths, etc., commonly referred to collectively as LandMark objects, also appear in the high-precision data.
**1.2 ** High precision data distribution engine
Advanced Driver Assistant System (ADAS) applications need to use the road network and attribute data information in front of the vehicle for decision control and judgment. Ordinary digital map data is usually only used by the navigation System, but high-precision map data can be used by other ADAS applications in the vehicle. Therefore, it is necessary to rely on high precision data and the high precision data distribution engine for high precision data transmission.
ADASIS(ADAS Interface Specification) defines the concept of ** “ADAS electronic Horizon” **, which represents the road network and road attribute information ahead of the vehicle. In order to achieve this expression, we need to build the location model of the vehicle and each possible road model of the road network in front of the vehicle, which can express the passable road through a tree-like hierarchy. In addition, the geometric shape and related attributes of the road will also build a related attribute model to express. ADAS Electronic Horizon data is serialized and transmitted over the on-board Ethernet.
1.3 Explanation of Nouns
ADAS(Advanced DriverAssistance System)
The advanced driving assistance system uses on-board sensors to perceive the vehicle environment and integrates computing to make drivers aware of possible dangers in advance, effectively improving the safety, economy and comfort of vehicle driving.
ADASIS(Advanced DriverAssistance System Interface Specification)
An industry international standard developed by the ADAS Forum to regulate the standard interface protocol for the exchange of map data between map data and vehicle ADAS applications.
AHP(ADAS Horizon Provider)
A high-precision data distribution engine that provides over-the-horizon road and data information for ADAS applications.
AHR(ADAS Horizon Reconstructor)
It is used to analyze the message sent by AHP and reconstruct the map data for the terminal ADAS application module.
2. Why do we need a high-precision data distribution engine
As a bridge between high precision data and ADAS applications, the value of high precision data distribution engine can be summarized in the following aspects:
- As the map sensor of autonomous driving, high-precision map can provide more reliable beyond-line range and support more reliable decision making.
- The need to improve the accuracy, from guide to guide the transformation of the car to improve the accuracy requirements.
- Interface standardization of high precision map data distribution.
3. Construction of high-precision data distribution engine
3.1 Relationship between high precision data distribution engine and ADAS application
The data distribution engine involves the following components and interactions:
- AHP
- AHR
- ADASIS V3 Protocol
- ADAS applications, see terminal applications in the figure above
3.2 High precision data distribution engine architecture
The high-precision data distribution engine is composed of multiple layers, including the engine layer, protocol organization layer and system adaptation layer. The related platform and tool support are shown in the following figure:
- Engine layer: high precision data loading, analysis and lane network data organization.
- Protocol layer: the data provided by the engine layer is assembled into protocol messages, which are delivered and distributed to the adaptation layer.
- The adaptation layer is mainly responsible for docking and interacting with the system and distributing the organization’s protocol data to ADAS applications.
3.4 Model expression of high precision data distribution engine
3.4.1 Abstraction and expression of road network model
The road network model of data distribution engine consists of three layers of model abstraction. First, the road network model is abstracted into a high-precision road network model through the real world model, and then the road network model is further organized and abstracted into a tree model expressed by Path and Offset.
- Representation of an abstract model of the real world
- Digital map model and user set navigation path, map elements expression
- Vehicle location and road network expression in data map model
- In the road network model near the location of vehicles, links are used to express the connection relationship between road networks. In a digital map database, a road network is represented as a set of connections and nodes between defined links.
- From the ADAS application perspective, the road network behind the vehicle is not a concern, so the data distribution engine consists of the road network in front of the vehicle.
- The road network in front of the vehicle is organized according to Path, and each Path is a set of links. The road network data in front of the vehicle can be expressed by two algorithms.
In simple Path mode, each passable Path is independently expressed as Path starting from the link where the car is located.
Optimize the path organization, which reduces data redundancy and can fully express the road network data in front of the vehicle.
Therefore, the data distribution engine forms a prediction tree according to the road network shape in front of the vehicle and its surroundings, which is described as a collection of different path and map data attributes. The prediction tree is made up of multiple paths, with each path representing a portion of the road and its intersections.
As the vehicle moves to change its position, the predictive view changes, and some paths behind the vehicle may be removed, or new paths in front of the vehicle may be added. The characteristics of the path are expressed as a group of attributes, such as the number of lanes, geometric shape and curvature of the expressway and urban expressway network itself. The position of the property on the path is represented by a set of offset values, which are distance markers that define the absolute distance, in centimeters, along the path itself. The origin of a path is the zero offset value point, and the offset value of the property represents the distance between the property itself and the origin of the path. If the path is newly started and there is no parent path, the offset point 0 is the starting position of the vehicle.
3.4.2 Attribute model of high precision data distribution engine
The attribute model data of the data distribution engine is derived from the attribute information on the high-precision road network, which is expressed along the Path and defined at the position on the Path, expressed by Offset. For example, a speed limit property provides a speed limit value for points on a path.
According to interpolation types, attribute models can be divided into the following three types, namely Spot, Step and Linear
A Spot type attribute is valid only at a given Offset position in the Path. The difference between the attributes is expressed by the different Offset positions. For example, a traffic light can be defined as a Spot property because it can be expressed as a point property at a location within the Path
An attribute of type Step is defined to remain valid until Offset of the next attribute. Attribute is expressed as the value in the range of Offset to EndOffset on Path.
In the example above, the Path length is 200. Speed limit 80 is in full effect, from Offset 0 to 200. There are two speed limits starting with offsets 50 and 100. Therefore, the attribute distribution on the whole graph is as follows:
- Offset 0: indicates the start rate limiting value 80.
- Offset 50: Introduced the rainy day speed limit value of 60, speed limit 80 attributes continue.
- Offset 100: Repeat speed limit 80, added fog limit 50, end of rainy speed limit 60.
- Offset 150: Repeat speed limit 80, end of fog speed limit 50.
Attributes of type Linear are defined to express Linear differences between given positions.
Linear interpolation attributes are not expressed continuously. At the same Offset, the value on the left and the value on the right are different. The attribute model expresses such discontinuous attribute values in the following way.
- At Offset, an attribute is stored. The value stores the attribute value on the left, and EndOffset is 0.
- Store an attribute in the same Offset, the value stores the attribute value to the right, but EndOffset > Offset
3.4.3 Vehicle location information model
In the data distribution engine, the location information of the vehicle can be expressed by Path, Offset. In the case of uncertainty, the location of the car may exist on multiple paths, so a collection is needed to describe the location information of the car. The following information can be expressed through vehicle location information:
- Whether the vehicle information is out of the data area.
- Whether the vehicle information matches the Path data range.
- Whether the vehicle information matches more than one Path.
- Whether the vehicle information enters and leaves the data area.
The TimeStamp value of vehicle position information expresses the time and moment value of receiving sensor information.
The location of the car can also express the more likely Path ahead.
In the figure above, the possible path to choose is P1 on the left and P3 on the right.
3.4.4 Synchronization mechanism between the high-precision data distribution engine and the receiver
The data distribution engine synchronizes road network Path data between AHP and AHR through pathControl messages.
- When the pathControl message does not contain a Path, the AHR receives the message and deletes the Path in the road network.
- When the pathControl message remains unchanged from the last time, AHR receives the message and leaves the current road network unchanged.
- When a Path is added to a pathControl message, the AHR receives the message and adds Path information
Synchronize property data through profileControl.
3.4.5 Interaction mechanism between the high-precision data distribution engine and the receiver
The data distribution engine (AHP) and the receiving end (AHR) have the following interaction mechanisms:
- Broadcast mode
- Request/offer mode
- Subscribe/publish model
At present, the construction of high-precision data distribution engine adopts the “request/offer” method. AHP sends ADAS message to AHR, and AHR can request and feedback information.
3.4.6 Assisting the application fusion of AHP and ADAS
3.4.6.1 Primary AHP and auxiliary AHP
Not all data in the ADASIS protocol is provided by the data distribution engine, and an auxiliary AHP engine can also be added. The auxiliary AHP engine can send sensor information or sensor fusion information.
The main data distribution engine and the auxiliary AHP engine are formed.
3.4.6.2 Two fusion modes of ADAS Applications
According to the main AHP and auxiliary AHP engine, two ADAS application fusion modes can be realized, namely downstream fusion and upstream fusion.
Downstream fusion
The AHP end does not do fusion processing, through communication to each sensor data and high-precision map data to AHR end for fusion processing, and then to ADAS functional application.
The upstream integration
The fusion process is carried out at the AHP end, and the fusion result is transmitted to AHR for processing through the protocol, which directly affects the ADAS function.
4. Quality construction
In order to ensure the quality of software, the following technical means are adopted in the construction of high-precision data distribution engine:
- Unit testing
- A functional test
- Quality inspection tools
Visualization tool
- Visualization tool Screenshot
5. Typical architecture application patterns
The high precision data distribution engine architecture can be divided into the following integration patterns:
5.1 The data distribution engine (EHP engine) is integrated in the map box
Map box concept
It is used to carry the capability of “map data + high precision positioning”, which is different from pure software products. MAP BOX/ MAP ECU/MAP BOX/HDLM… L: Localization M: Module)
Contains the content
- Mapping and related applications: HD data, AHP, location, OTA…
- Basic software: system, underlying drivers, diagnostics…
- Basic hardware: System-on-chip (SoC), memory, storage, IMU(optional), protective housing…
- Network and communication interface: CAN/ Ethernet input, Ethernet output, USB interface…
Program features
Clear division of tasks: Automobile enterprises can disassemble functions into small modules based on this architecture, and put forward product requirements for control respectively, so as to avoid being unable to start the all-black box scheme. Alternate suppliers when delivery risks are encountered.
Functional security considerations: chip selection, hardware design, network security, system diagnosis and other details can be handed over to professional suppliers; The AD ECU must be separated from the AD ECU to ensure that the AD ECU meets the functional security requirements.
It is convenient for high and low distribution and other product management: products with different configurations from suppliers can be selected.
Reduce the computing power burden on the domain controller: It facilitates the search for functional security hardware that meets the computing power requirements.
5.2 Integration in IHU
Program features
Cost reduction: No need to purchase additional hardware modules.
Integrate V2 scheme and reduce uncertainty: AHP V2 is mostly at the vehicle and machine end, and the scheme has already been run. Therefore, the uncertainty of the new architecture can be avoided by using the map and V3 in a similar way.
Internal reasons of automobile enterprises are easy to promote: Some automobile enterprises and navigation map departments of high-precision map business planning will change the overall structure greatly if they promote the box scheme from bottom to top.
5.3 Integration into a Domain Controller
Program features
Reduce cross-domain communication’s occupation of vehicle-mounted network bandwidth: Most sensors used for perception are connected to the domain controller. If the map and positioning are placed in the domain controller, the back-end application can directly or indirectly use the map without cross-domain communication, thus reducing the occupation of vehicle-mounted network bandwidth.
It is more suitable for the automobile enterprises that take the self-research route and choose the overall scheme: for the automobile enterprises that take the self-research route and choose a single scheme provider to provide the complete scheme, it is not necessary to deploy the functional modules separately.
6. Scenario Application Example
6.1 High-precision positioning applications
Combined with high precision data to assist the horizontal positioning and vertical positioning.
Vertical positioning combined with road sign OBJ, how many lane information, horizontal positioning combined with lane line, guardrail and other related information.
Active security applications often combine sensor (millimeter wave radar, camera) information and map data to match and correct deviations, thus improving positioning accuracy.
6.2 High-speed Automatic Driving (HWP)
Functional activation
Driving environment mainly depends on map judgment :(1) high-speed city; (2) Clear lane lines; (3) curvature slope; (4) Objects or events without triggering alarm or braking: including dynamic road environment; (5) It is not night, the weather condition is good (visibility above 200m).
Realize the function
Take horizontal control of lane cruise and autonomous parking of lane in abnormal scenarios as examples:
- Lane type: Automatic driving depends on the lane type to divide the drivable area, if the wrong type will lead to the vehicle driving in the non-driving area, will bring safety hazards to the car; Meanwhile, in the scenario of autonomous safe parking, the wrong lane type will directly lead to the autonomy and safety of autonomous safe parking.
- Lane line type: assist the camera to identify lane line; Check it against the camera, and we’ll do lane keeping.
6.3 Automatic cruise based on navigation route
Functional activation
Working environment depends on map judgment:
- Road grade: High speed/city block.
- PartOfcalculateRoute(navigation path identifier) Whether the connection is uninterrupted.
- Weather type: the function can be activated under sunny/rainy/cloudy weather conditions.
Realize the function
- On/off JCT, it will judge whether to go on/off JCT according to the navigation path identifier and road network in front of the vehicle, and remind the ramp side of lane change in advance.
- The automatic lane change enters the JCT/ merging expressway, and the lane line identification will be performed by the auxiliary camera according to the lane line type, and the detection and comparison will be made with the camera. The time of lane change can be determined by the virtual or real condition of the line type.
7. Future evolution
On the one hand, further integration of AHP V2 and V3 architecture design is considered to better assist automatic driving. In addition, as part of the data loop, it enriches data supply and recovery capabilities.