5G-based Driving Assistance for Autonomous Vehicles CMCC ZENGFENG 2018.6 1
2
R15 first version of the
5Gstandard
R14 Pre5G
Strive to achieve commercial use of 5G
in 2020
R16 complete 5G standard
Key Technology Verification
2013 2014 2015 2016 2017 2018 2019 2020
Large scale Trial & Pre-commercial Networks
Basic Needs and Objective
Key Technology Research
Establish Central and Regional Laboratory to Conduct Cross-industry Innovation
System Verification
Candidate Standard Scheme Study
Enhanced Technology Research
Technology Development Strategy
Technology Research
Strategy
Standard
Trial Verification
Ecological
Network Construction strategy
China Mobile 5G Innovation Center
China Mobile's 5G development layout and plan
CMCC 5G Trial Plan and Scale
5G Network -
5G Devices -
17 Cities 1100+ 5G eNBs
2000+ Terminals
Scale-up trials in 5 cities
Application showcase in 12 cities
Exploring converged innovation on 5G services
Launching 5G terminal forerunner plan
Speeding up 5G Development
2020~ 2019 2018
5G Pre-commercial Trial
5G E2E Commercialization
Scale-up Trial Application Showcase
Carry out 5G Application Demonstration and Study in 7 Areas
Major Projects in 2018
4
Smart Transportation
Autonomous driving
AR/VR 4K Live Streaming
Video Fusion
HD Cloud Gaming Smart Factory Smart Grid
Cloud Robot AI
Mobile Remote Medical
Transportation
Video & VR/AR Entertainment Manufacturing Energy
Artificial Intelligence Healthcare
5
Paving the path to more autonomous driving
Network Oriented
Intelligent Oriented
Driver Assistance Partial Automation Conditional Automation Full Automation
Informatio
n
interaction
Collaborat
ive
Sensing
Network
collaborati
ve decision
& control
5G
LTE-V
4G
The Roadmap of Evolution of Intelligent Connected Vehicle
The roadmap was released by
Ministry of Industry and
Information Technology of the
People’s Republic of China.
By 2025,China will establish a
ICV standard system that can
support full automatic driving.
The goal of “networked oriented
” is to support network
collaborative decision & control .
There are three stages of network
oriented ICV.
6
5G Enhances Information Interaction
Autonomous vehicles will become a
mobile space, which can be used for
entertainment,work and telematics.
AR/VR、Game、Movie…
Onboard Information
Download local 3D HD map imme
diately
Model and analysis of dangerous
situation
Download HD Map
Autonomous vehicles of Level 3: The cars
can handle almost all the scenario by itself,
but if they meet some unsolvable problem ,
remote drivers will be informed and intervene.
Remote drivers will control several
unmanned vehicles remotely.
Enables Commercial Operation
7
5G Enhances Collaborative Sensing
Perception is the key factor of autonomous driving. Sensors such as lidar,radar and camera can only ‘see’ the surrounding objects by processing echo signals , but processing results will be affected and limited by some extreme bad environment, such as foggy, rainy and sunlight. The network-assisted vehicles can communicate with surrounding vehicles and road side infrastructure directly:
To exchange information such as velocity,position , acceleration and trajectory in future. To overcome the limit of perception range and environment. To evolve from intelligent car to intelligent connected car.
8
5G Enhances Collaborative Sensing LTE-V2X defines 27 application scenarios for assisted driving, and demand of each scene is defined in terms of time delay, movement speed, reliability and other aspects.
5G-V2X defines a number of scenarios for autonomous driving requirements, and the requirements of each scene vary greatly. Different technical combinations are needed to meet different scene requirements.
Source:3GPP SA1 TS22.185 Source:3GPP SA1 TS22.186
peripheral message collection.
Traffic status alarms
Platoon
Remote control
maximum latency:5~10ms reliability : >99.999% End-to-End latency: 100ms
9
5G Enhances Collaborative Control & Decision
Drawbacks of Intelligent vehicles without network: Computational complexity : Driverless car needs variety sensors
to adapt complex and variant environment, so it needs powerful
CPU/GPUs to process signals ;
Expensive: The price of intelligent vehicles is very high,
because they needs expensive sensors and processors;
Limitation of on board perception: From spatial dimension, on
board perception is so limited, and it can only perceive the area of
a certain range centered on the vehicle. In some special areas,
such as street corner and intersection, there will be certain blind
area.
Hierarchical decision architecture for connected car: The data is processed at different layers according to safety
requirements and latency requirements.
Perceptions of peripheral environment will be enhanced by road
side units which are composed of cameras or radars, and also by
collecting more road information.
Some data may be processed in MEC server , that will reduce
computational complexity on board.
10
Hierarchical decision architecture of connected car
Source:3GPP SA1 TS22.185
……
Vehicle control
Get Message from
instant trends of
driving space
Decision on
board
Arbitration between
regional decision and
on-board decision
5G Gateway
Perception on
board
MEC Service(Storage、compute,..)
Vehicle state message
(position、velocity 、
acceleration …)
Road side
auxiliary
sensing
system
Original
sensor
message
Instant Trends of Driving Space
OEM A
Lane Level
planning,Splicing of
ITDS
Regional
Traffic
planning
Arbitration of
different
dispatchers
Regional decision Region Control
Regional Traffic planning
(Traffic light、
Lane、Traffic signal)
Vehicle control To
terminal
Vehicle control
results to RCL
Vehicle big
data
Climate
(Snowy 、rainy、
foggy )
Traffic big data
Data perception
Vehicle data
management
High level decision
HD map
management
Macro control
Management of
MEC hand over
Traffic
dispersion
Traffic
planning
Path
planning
Regional Control Layer(RCL)
Terminal Layer
Road cooperative Control Layer
Perception Decision Control
Network architecture of hierarchical autonomous driving
Operator 1's Core Network
Operator 2's Core Network
Path Planning Transport Traffic Control
Road Side unit
High-definition Map Database
Road Side unitRoad Side unit
Cloud
Traffic Light Control
Lane Level Planning
Multi-sensor Data Fusion
High accuracy positioning
Operator 1's Access Network
MEC Host
Traffic Light Control
Lane Level Planning
Multi-sensor Data Fusion
High accuracy positioning
MEC Host
Traffic Light Control
Lane Level Planning
Multi-sensor Data Fusion
High accuracy positioning
MEC Host
Operator 1's Access Network Operator 2's Access Network
Regional Control Layer(RCL)
Terminal Layer
Road cooperative Control Layer
MEC host deployed on the
access network can operate
multiple autonomous driving
applications. To be specific, it
completes the analysis and
processing of the data from
connected vehicles and road
side sensors.
12
Terminal Layer
Terminal Layer
Regional Control
Layer(RCL)
Road cooperative Control Layer
Upload: Integrated map from vehicles Original sensor message
Download Integrated regional map(Instant
Trends of Driving Space) Decisions from regional control layer
16
In summary
5G paves the path to more autonomous driving
5G enhances information interaction 5G enhances collaborative sensing 5G enhances collaborative control & decision
The architecture of hierarchical autonomous driving is established Terminal layer Regional control layer Road cooperative control layer