Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
JUNE 21-25, 2021
REALSIM
DEAN DETERPrinciple Investigator Vehicle Systems Integration Research GroupOak Ridge National Laboratory
PROJECT ID: EEMS101
PROJECT OVERVIEW
2
Timeline Barriers
Budget Partners
• Project Start Date: Oct. 2020
• Project End Date: Sept. 2023
• Percent Complete: 10%
• Total Funding: $3.58M
• Funding per Year:
• FY 21: 1.12M
• FY 22: 1.23M (Planned)
• FY 23: 1.23M (Planned)
• Lead
• Oak Ridge National Laboratory
• Partner
• Argonne National Laboratory
• Collaborations
• American Center for Mobility (ACM)
• IPG Automotive
• Modeling and simulation environments are not
all inclusive for all scenarios.
• Lack of standard co-simulation tools or hooks
across vehicle and traffic environments.
• Computational requirements of complex
environment simulation.
OBJECTIVES AND RELEVANCE▪ Relevance:
1. A system of systems approach is required to properly manage vehicles equipped with advanced
driver-assistance systems (ADAS), connected and autonomous vehicle (CAVs) technologies, their
interactions with the surrounding environments, traffic networks, traffic control, and the various
scenarios that greatly effect each one of these.
2. Expensive in-field experimentation generally captures ‘focused’ datasets- not providing sufficient
detail to produce a digital twin or explain unique events during future experimental analysis.
Current on-road connectivity implementations are built upon large infrastructure investments at
permanent locations. (Corridors / test tracks / etc.)
▪ Objective: 1. Expand the capabilities of the VPPG Core-tools project feedback loop to include capabilities of
digital twins as well as greater real-world correlation by utilizing on-road/track data in a closed-loop
process for validation of the performance of controls and simulated scenarios/environments.
2. Incorporating joint vehicle and infrastructure experimental platforms can provide a “full picture” of
costly experiments for further analysis OR replication in a digital environment where more
variables may be controlled. Smaller scale, portable systems built on ‘nodes’ offer flexibility and
control of research activities, enabling experiments on controlled roadways, or public roads.
3
MILESTONES
4
Date Milestone Status
June 2021
Demonstration of sensor simulation and integration into
ORNL/ANL ROS/Autoware based perception stack including
camera, radar, and lidar systems.
(Task 1.1 ORNL)
On Track
June 2021
Demonstration of microtraffic simulation’s traffic control
functionality and integration into Real-Sim including
SPaT control and higher-level traffic control.
(Task 1.2 ORNL)
DelayedDue to COVID
shutdowns and
limited NRTL
inspection
resources.
Sept. 2021
Demonstration of ANL prototype platform with vehicle
and infrastructure components.
(Task 1.3 ANL)
On Track
APPROACH
5
Direct support through data for multiple
SMART 2.0 efforts
Develop a mobile connectivity-enabled
platform to coordinate and capture on-
road experimentation
Vehicle Component
Infrastructure Component
Capturing ‘full picture’
vehicle operation AND
environment during
experimentation
Providing control and
data capture of traffic
flows and infrastructure
components
APPROACH
▪ Subtask 1.1: Sensors - XIL and
virtual environment 2.0.
▪ Subtask 1.2: VPPG microtraffic
2.0, enhancements to include
SPaT, V2X, and traffic control
XIL applications.
▪ Subtask 1.3: Development of
portable platform for V2X
testing and data collection to
enable XIL validation.
6
Task 1 Task 2 Task 3
▪ Task 2 – Real-Sim platform and
digital twins
▪ Digital Twins to be created for
support of this project and
SMART 2.0
– Multiple areas of
Chattanooga, TN
– ACM (improvement of digital
twin)
– Lake County, IL
▪ Task 3 – Validation of the Real-
Sim Platform using Current
On-Road/Track EEMS Projects
▪ Projects that will support this
effort.
– Regional Mobility Project
– ACM Model Validation
Project
– EEMS096
– EEMS095
– EEMS061
TECHNICAL ACCOMPLISHMENTS
▪ IPG Carmaker has a ROS interface that
bridges communication between the
simulated sensors and a specific ROS topic
▪ This interface allows the user to build up a
message and stream information to a
specific topic over the ROS network
▪ The initial interface had a lidar and camera
streaming data over ROS. The interface
was extended to include a GPS and IMU
as well.
Task 1.1: Adding sensors to ROS data streaming from IPG Carmaker
8
Msg: gps_common/GPSFix
Topic: /fix
Msg: sensor_msgs/Imu
Topic: /imu
TECHNICAL PROGRESS
▪ The digital twin of the ORNL main
campus in IPG Carmaker imported to
this Carmaker project. Also, a visual
model of the ORNL RAV4 mule
vehicle was imported to Carmaker
▪ This setup deploys the ORNL mule
vehicle onto the digital twin of the
main campus running a defined
scenario. This scenario can match
the real mule vehicle data for better
correlation.
Task 1.1: Scenario and vehicle development in IPG Carmaker
9
TECHNICAL PROGRESS
▪ The sensor setup on the vehicle was
modified to match a subset of the
mule vehicle sensors (1 lidar, 1
camera w/ object list, gps and imu).
▪ The sensor streams can then be
visualized in tools that listen to ROS
messages (Rviz, command line tools,
etc.)
Task 1.1: Modification of vehicle sensor streams from IPG Carmaker
10
▪ Determined key signal list from collaborative efforts
– Experiment focused data collection
• Raw sensor data collection for digital twin
development from focused in-field experiments
• Condensed vehicle and object data for vehicle
and traffic impact analysis, reducing dataset size
▪ Development of platforms for vehicle component
– Vehicle selection based on collaborative studies
– Decoded vehicle data from CAN / FlexRay
– Hardware modifications for sensor array
– Development of ROS-based platform for time
aligned data collection of vehicle and sensor data
▪ Infrastructure component development - Q3/Q4
AMTL Mobile DAQ
TECHNICAL ACCOMPLISHMENTS
2020 Tesla Model 3AutoPilot / FSD
2019 Cadillac CT6SuperCruise
Transferable Mobile DAQ – independent sensor /
environment data
• Video (multi camera)
• LiDAR (Velodyne VLP-16 x2 / VLP32C x2 )
• Communication- (DSRC)
• Plugins for time alignment with recorded vehicle data
RESPONSE TO PREVIOUS YEARREVIEWERS’ COMMENTS
▪ This is the first year that the project has been reviewed.
13
PARTNERSHIPS ANDCOLLABORATIONS
14
Primary project partner and contributor for Task 1.3 focused
on the prototype DAQ unit for vehicles and infrastructure.
This system is key for the completion of Task 2 and 3.
EEMS096, EEMS095, EEMS067, EEMS061
Digital twin of ACM facility was provided to ORNL to allow for
a digital twin to be generated and utilized. Additionally, data
collected from ACM-led EEMS082 is provided to ORNL
IPG is collaborating with ORNL via software support for
sensor modeling and emulation, as well as being a key
partner is working with ROS and Autoware for XIL
applications.
REMAINING CHALLENGES AND BARRIERS
▪ Though research facilities are open for experimentation, controls are in place to ensure
safety during COVID-19 pandemic which impact worker productivity and operations.
Reduced staff available onsite, in addition to the needs for social distancing, can make
hands-on research and troubleshooting efforts challenging and more time consuming.
Though no milestones or objectives are expected to be missed at this time, the possibility
exists due to resource constraints. Creative methods for on-site and remote work will
continue throughout limited, or reduced, site operations and the risk of impacts to project
timing remain.
▪ Determining the computational requirements for these various applications is a major
challenge as well as a barrier to some applications. To run the most demanding sensor
simulations (i.e. lidar and radar) on HIL applications requires an extremely powerful PC to
apply ray tracing techniques utilizing GPU calculation and a powerful real-time node.
15
PROPOSED FUTURE RESEARCH
▪ Remainder of 2021– Continue work on sensor simulation and data stream
emulation to vehicle sensors.
– Demonstration of ANL prototype platform with vehicle
and infrastructure components functional.
▪ Future FY 2022-2023– Initial trial, refinement, and further deployment of field
components for AMTL Mobile DAQ
– Construct the digital twin of the Lake County intersection
in Illinois from scratch
– Design scenarios based on tests from 1591, 1562, and
SMART Mobility Validation at ACM on the Real-Sim
platform to generate comparison data for correlation with
road/track data.
Any proposed future work is subject to change based on funding levels.
16
SUMMARY
17
ApproachRelevance
Future Work Technical Accomplishments
• The importance of the system of systems approach for
understanding and managing vehicles equipped with
ADAS, CAVs technologies, and their interactions with the
surrounding traffic environments.
• Field experimentation captures ‘focused’ datasets lacking
detail for detailed experimental analysis. Current
connectivity implementations rely on large, fixed
infrastructure investments.
• Task 1:
• Sensors - XIL and virtual environment 2.0.
• VPPG microtraffic 2.0
• AMTL Mobile DAQ
• Task 2: Real-Sim platform and digital twins
• Task 3: Validation using Current On-Road/Track EEMS
Projects
• FY 2021:
• Sensor simulation and data stream emulation
• Demonstration of AMTL Mobile DAQ platform
• FY 2022/FY2023:
• AMTL Mobile DAQ initial trial, refinement, and
further deployment
• Lake County digital twin construction
• EEMS scenario execution and correlation
• Sensor data streaming using ROS
• Vehicle and scenario development in IPG Carmaker
• Modification of vehicle sensor streams
• Determined key signal list from collaborative efforts
• Development of platforms for vehicle component
MOBILITY FOR OPPORTUNITY
FOR MORE INFORMATION
Dean DeterR&D Staff
Vehicle Systems Integration Research
National Transportation Research Center
Oak Ridge National Laboratory
18