Construction Practice of High-precision Map Data Application Distribution Engine

1. What is a high-precision data distribution engine

1.1 Overview of

The main difference between High Definitation Map (HD MAP) and ordinary navigation electronic maps is higher accuracy and richer information. Higher accuracy is mainly reflected in the higher accuracy of the absolute coordinates of the high-precision map (refers to the accuracy between a target on the map and the location of an external real world thing), which can be accurate to the centimeter level; richer information is mainly reflected in The high-precision map not only contains road information, but also covers almost all surrounding static information related to traffic.

Compared with ordinary navigation electronic maps, the road traffic information contained in high-precision maps is richer and more accurate. In addition, in terms of application scenarios, ordinary navigation maps are mainly used by drivers, while high-precision maps are machine-oriented maps for autonomous vehicles.

accuracy is the biggest difference between high-precision maps and ordinary navigation electronic maps. The accuracy of ordinary on-board electronic navigation maps is generally about 10 meters. High-precision maps are used in the field of autonomous driving. needs to be accurately located on a specific lane , . You also need to know all the surrounding roads and traffic information that may participate in automatic driving decision-making. , the accuracy needs to reach 10~20 cm , this accuracy is basically the same as the width of a lane sideline, in order to ensure that the smart-driving car will not cross to other lanes and avoid the risk of side collisions with other vehicles.

Ordinary navigation electronic maps must depict the road (link) , while the high-precision map not only depicts the road, but also depicts how many lanes (lane) on a road, which truly reflects the actual style of the road.

The richer high-precision map information is mainly reflected in the following aspects:

Accurate road shape : The slope, curvature, heading, elevation, and roll data of each lane.

detailed lane line information : whether the lane line between the lanes is a dashed line, a solid line or a double yellow line, the color of the line, the road isolation zone, and the material of the isolation zone will be described.

In addition, pedestrian crossings, signs along the road, speed limit signs, traffic lights, roadside phone booths, etc., which are usually collectively referred to as the absolute geographic coordinates , physical dimensions and their characteristics will also appear in high-precision data. .

1.2 high-precision data distribution engine

ADAS (Advanced Driver Assistant System) applications need to use the road network and attribute data information in front of the vehicle for decision-making control and judgment. Ordinary digital map data is usually only used by the navigation system, but high-precision map data can be used in the vehicle. It is used by other ADAS applications, so it needs to rely on high-precision data and a high-precision data distribution engine for high-precision data broadcasting.

ADASIS (ADAS Interface Specification) defines the "ADAS Electronic Horizon" . "ADAS Electronic Horizon" expresses the road network and road network attribute information in front of the vehicle. In order to achieve this way of expression, we need to build the vehicle's position model and the possible road models of the road network in front of the vehicle. The passable roads can be expressed through a tree-like hierarchical structure. In addition, the geometric shape and related attributes of the road will also establish related attribute models to express. "ADAS Electronic Horizon" data is serialized and transmitted through the on-board Ethernet network.

1.3 Noun explanation

ADAS(Advanced DriverAssistance System)

That is, the advanced driving assistance system uses on-board sensors to perceive the vehicle environment and integrates calculations to allow the driver to perceive possible dangers in advance, effectively improving the safety, economy and comfort of driving.

ADASIS(Advanced DriverAssistance System Interface Specification)

The industry international standards developed by the ADAS Forum are used to standardize the standard interface protocol for exchanging map data between map data and vehicle ADAS applications.

AHP(ADAS Horizon Provider)

The high-precision data distribution engine provides ADAS applications with road ahead and data information beyond the visual range.

AHR(ADAS Horizon Reconstructor)

It is used to parse the message sent by AHP and reconstruct the map data for use by the terminal ADAS application module.

2. Why do you need a high-precision data distribution engine

high-precision data and ADAS applications, its value can be summarized in the following aspects:

• For the long-distance visual range of automatic driving, high-precision maps, as the map sensor for automatic driving, can provide more reliable beyond visual range and support more reliable decision-making judgments.
• The need for improved accuracy, the transition from guiding people to guiding vehicles has increased the requirements for accuracy.
• Standardization of the interface for high-precision map data distribution.

3. Construction of high-precision data distribution engine

3.1 The relationship between the high-precision data distribution engine and ADAS applications

The data distribution engine involves the following components and interactions:

• AHP
• AHR
• ADASIS V3 Protocol
• ADAS application, please refer to the terminal application department in the figure above for details

3.2 High-precision data distribution engine architecture

The high-precision data distribution engine is composed of multiple levels, including the engine layer, the protocol organization layer, and the system adaptation layer. The relevant platform and tool support are shown in the following figure:

• Engine layer: high-precision data loading, analysis and organization of road network data.
• Protocol layer: It mainly assembles the data provided by the engine layer into protocol messages, and transmits and distributes them to the adaptation layer.
• Adaptation layer: Mainly responsible for docking and interaction 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 the data distribution engine includes three levels of model abstraction. First, the real world model is abstracted into a high-precision road network model, and then the high-precision road network model is further organized and divided into a tree model expressed by Path and Offset.

• Representation of an abstract model of the real world

• Digital map model and the navigation path set by the user, the expression of map elements

• Vehicle position and road network expression in data map model

• The road network model near the location of the vehicle expresses the connection relationship between the road networks through links. In the digital map database, the road network is represented as a set of connections and nodes that define the links.

• From the perspective of ADAS applications, the road network behind the vehicle is not concerned, so the data distribution engine is composed of the road network in front of the vehicle.

• The road network in front of the vehicle is organized by Path, and each Path is a collection of a set of links. The road network data in front of the car can be expressed by two algorithms.

Simple Path mode, starting from the link where the car is located, each passable path is independently expressed as Path.

Optimize the route organization method, this method reduces data redundancy, and can also completely express the road network data in front of the vehicle.

Therefore, the data distribution engine is described as a collection of different paths and map data attributes according to the shape of the road network in front of the vehicle and its surrounding environment to form a prediction tree. This prediction tree is composed of multiple paths connected, each path represents a part of the road, and the intersection between the road and the road.

Once the vehicle moves and changes its position, the forecast view will also change. Some paths behind the vehicle may be deleted, or new paths in front of the vehicle may be added. The characteristics of the path are expressed as a set of attributes, such as the number of lanes, geometry, curvature, etc. included in the high-speed and urban express network itself. The position of the attribute on the path is represented by a set of offset values. The offset value is a distance marker that defines the absolute distance along the path itself, expressed in centimeters. The origin of a path is the zero offset value point, and the offset value of the attribute represents the distance between the attribute itself and the origin of the path. If the path is newly started and there is no parent path, the offset value 0 point is the starting position of the vehicle.

3.4.2 The attribute model of the high-precision data distribution engine

The attribute model data of the data distribution engine comes from the attribute information on the Gaojing Road network, which is defined as expressed along the Path, and defined as the position on the Path, expressed by Offset. For example, the speed limit attribute provides speed limit values for points on the path.

The attribute model can be divided into the following three different types according to the interpolation type, namely Spot, Step, Linear type

Spot type attributes are only valid at a given Offset position in Path, and the difference in attributes is expressed by different Offset positions. For example, a traffic light can be defined as an attribute of the Spot type, because it can be expressed as a point attribute in a certain position in the Path

The attribute of the Step type is defined to be valid until the Offset position of the next attribute. The attribute is expressed as a value in the range of Offset to EndOffset on Path.

In the example shown in the figure above, the Path length is 200. Speed limit 80 is effective as a whole, from Offset 0 to 200. There are two speed limit values starting from offset 50 and 100. Therefore, the distribution of attributes on the entire graph is as follows:

• Offset 0: Start speed limit value 80.
• Offset 50: Introduce the rainy day speed limit value of 60, and the attribute of the speed limit of 80 continues.
• Offset 100: Repeat the speed limit of 80, add a foggy limit of 50, and the speed limit of 60 for rainy days ends.
• Offset 150: Repeat the speed limit of 80, and end the speed limit of 50 in fog.

The attribute of the Linear type is defined as a linear difference expression between the given positions.

Linear interpolation attributes are not expressed continuously. At the same Offset, the value on the left is different from the value on the right. The attribute model uses the following way to express this discontinuous attribute value.

• At Offset, store an attribute, the value stores the attribute value on the left, and EndOffset is 0.
• Store an attribute in the same Offset, and store the value of the attribute on the right, but EndOffset> Offset

3.4.3 Vehicle location information model

In the data distribution engine, the position information of the car can be expressed through Path and Offset. In the case of uncertainty, the position of the car may exist on multiple Paths, so a set is needed to describe the position of the car. The following information can be expressed through the vehicle position information:

• Whether the car information is out of the data area.
• Whether the car information matches the data range of Path.
• Whether the car information matches multiple Paths.
• Whether the car information has entered and left the data area.

The TimeStamp value of the vehicle position information expresses the time and time value of receiving the sensor information.

The position information of the car can also express the Path path that is more likely to be chosen ahead.

As shown in the figure above, the path that may be selected on the left is P1, and the path on the right is P3.

3.4.4 The synchronization mechanism between the high-precision data distribution engine and the receiving end

The data distribution engine synchronizes the path data of the road network between AHP and AHR through pathControl messages.

• When the pathControl message does not contain a certain Path, the AHR deletes the Path in the road network after receiving the message.
• When the pathControl message remains unchanged from the previous time, the AHR maintains the current road network unchanged after receiving the message.
• When a path is added to a pathControl message, AHR adds Path information after receiving the message

Sync attribute data through profileControl.

3.4.5 The interaction mechanism between the high-precision data distribution engine and the receiving end

The data distribution engine (AHP) and the receiving end (AHR) have the following interaction mechanisms:

• Broadcasting method
• Request/Provide Method
• Subscription/publishing model

At present, the “request/provide” method is adopted when the high-precision data distribution engine is constructed. AHP sends ADAS messages to AHR, and AHR can request and feedback information.

3.4.6 Auxiliary AHP and ADAS application integration

3.4.6.1 Main AHP and auxiliary AHP

Not all data in the ADASIS protocol is provided by the data distribution engine, and auxiliary AHP engines can also be added. The auxiliary AHP engine can send sensor information or sensor fusion information.

The formed main data distribution engine and auxiliary AHP engine.

3.4.6.2 Two Convergence Methods of ADAS Application

According to the main AHP and auxiliary AHP engines, two ways of fusion of ADAS applications can be realized, namely downstream fusion and upstream fusion.

downstream fusion

There is no fusion processing on the AHP side, and each sensor data and high-precision map data are transmitted to the AHR side through communication, for fusion processing, and then transmitted to the ADAS function application.

upstream fusion

Perform fusion processing on the AHP side, and pass the fusion result to AHR through the protocol for processing, which directly affects the ADAS function.

4. Quality construction

In order to ensure the quality of the software, the following technical means are used in the construction of the high-precision data distribution engine:

• unit test
• function test
• Quality inspection tools

Visualization tool

• Visualization tool screenshot

5. Typical architecture application form

According to the high-precision data distribution engine architecture, it can be divided into the following integration forms:

5.1 Data distribution engine (ie EHP engine) integrated in the map box

map box concept

Car-level software and hardware integrated products used to carry "map data + high-precision positioning" capabilities are different from pure software product forms. (The following names refer to the high-precision map box: positioning box/MAP ECU/MAP BOX/HDLM...where L:Localization M:Module)

contains content

• Maps and related applications: HD data, AHP, positioning, OTA...
• Basic software: system, bottom driver, diagnosis...
• Basic hardware: system-on-chip (SoC), memory, storage, IMU (optional), protective casing...
• Network and communication interface: CAN/Ethernet input, Ethernet output, USB interface...

program features

clear division of tasks : Car companies can disassemble functions into small modules with this architecture, and put forward product requirements for management and control, so as to avoid all black box solutions. Suppliers can be replaced when encountering delivery risks.

Functional safety considerations : Chip selection, hardware design, network security, system diagnostics and other details can be handed over to professional suppliers; functional safety levels such as map quality, online updates, and backhaul have uncertainties, needs, and The AD ECU is isolated so that the AD ECU meets the functional safety requirements.

convenient for product management such as high configuration and low configuration. : different configuration products of the supplier can be selected.

reduces the computing power burden of the domain controller : It is convenient to find functional safety hardware that meets computing power requirements.

5.2 integrated in IHU

program features

reduces costs : No need to purchase additional hardware modules.

integrates the V2 solution to reduce uncertainty : AHP V2 is mostly on the car and machine side, and the solution has already run through, so the map and V3 can be used in a similar way to avoid the uncertainty of the new architecture.

car companies are easy to advance because of internal reasons: some car companies, high-precision map business planning navigation map department, if the box plan is promoted from the bottom up, the overall structure will be greatly changed and it is difficult to advance.

5.3 integrated in the domain controller

program features

reduces the occupation of vehicle network bandwidth by cross-domain communication : Most sensors used for perception are connected to the domain controller. If the map and positioning are placed on the domain controller, back-end applications can be used directly or indirectly without cross-domain communication. Map, reduce the occupation of the vehicle network bandwidth.

more suitable for auto companies that follow the self-developed route and choose the overall solution. : For auto companies that follow the self-developed route and choose a single solution provider to provide a complete solution, there is no need to deploy functional modules separately.

6. Scenario application example

6.1 High-precision positioning application

Combined with high-precision data to assist in horizontal and vertical positioning.

Longitudinal positioning is mostly combined with relevant information such as road signs obj and lane geometry, and horizontal positioning is mostly combined with relevant information such as lane lines and guardrails.

Active safety applications combine sensor (millimeter wave radar, camera) information and map data to match and correct deviations, thereby improving positioning accuracy.

6.2 High Speed Autonomous Driving (HWP)

function activation

The driving environment mainly relies on map judgments: (1) High speed city; (2) Clear lane lines; (3) Curvature gradient; (4) No objects or events triggering alarm or braking: including dynamic road environment; (5) Non At night, the weather is good (visibility above 200 meters).

realizes the function

Take the lateral cruise control of the lane and the autonomous parking in the lane in abnormal scenarios as an example:

• lane type : Autonomous driving relies on the lane type to divide the drivable area. If the wrong type will cause the vehicle to drive in the non-driving area, it will bring safety hazards to the self-car; at the same time, in the scene of autonomous safe parking, if the lane type is wrong Will directly lead to the autonomy and safety of autonomous safe parking.
• Lane line type : assist the camera to identify the lane line type; compare and check with the camera, and then carry out lane keeping.

6.3 Automatic cruise based on navigation route

function activation

The working environment depends on the map to judge:

• Road class: highway/city block.
• Whether PartOfcalculateRoute (navigation route identification) is continuous without disconnection.
• Weather type: In sunny/light rain/cloudy weather conditions, the function is allowed to activate.

realizes the function

• The up/down JCT will judge whether to go up/down JCT according to the navigation path identification and the road network ahead of the vehicle, and remind the side of the ramp in advance.
• When automatically changing lanes into JCT/joining high-speed, the lane line type recognition will be carried out according to the lane line type auxiliary camera, and the detection and comparison with the camera will be carried out. The virtual and real line types determine the timing of the vehicle lane change.

7. Future evolution

On the one hand, consider further integrating the architecture design of AHP V2 and V3 to better assist autonomous driving. In addition, as part of the data closed loop, enrich data provision and recovery capabilities.