2021/07/14 22:00 UTC
Attendees
- Ryohsuke Mitsudome (TSC Chair)
- Shinpei Kato (AWF)
- Yang Zhang (AWF)
- Bobby Hambrick (AWF)
- Jim Liu (AWF)
- Girish Shirasat (AWF)
- Daisuke Tanaka (AWF)
- Christian John (SPC)
- Josh Whitly (AWF)
- Bonolo Mathibela (AWF)
- Hiroshi Igata (ODD WG)
- Kerem Par (Adastec)
- Joe Speed (ADLINK)
- Filipe Rinaldi (Arm)
- Cheng Chen (AutoCore)
- David Van Guyen (AutonomouStuff)
- Jaime Martin Losa (eProsima)
- Akihiko Tsukuda (eSOL)
- Rahul Mangharam (F1 Tenth Foundation)
- Michael Li (ITRI)
- M. Fatih Cırıt (Leo Drive)
- Seonman Kim (LGE)
- Tsutomu Otake (Macnica)
- Pascal Lecuyot (Navya)
- Adam Dabrowski (Robotec.ai)
- Jilada Eccleston (Tier IV)
- Yao Zhai (Ziiko)
Minutes:
Action items
Board Passdowns
- Autoware workshop held last Sunday with IV2021
- https://www.autoware.org/event-details/autoware-workshop-at-iv2021
- https://2021.ieee-iv.org/workshops/
- Recording will be available soon
Confirmation of previous minutes
Action items review
- Voting results for Autoware Core/Universe architecture
- TSC agreed on the proposal as there were no votes against the proposal
- Next Actions: Jilada, Josh, Mitsudome + any one interested would discuss further details
- Implementation will start after the ODD demo
Working group report: Autonomy Software
- Delays on vehicle setup from Zing
- Earliest we can get ROS data would be mid-August
- Significant issue receiving map data from PolyExplore
- Josh, Bonolo, Zing have been contacting Jens from PolyExplore, but Jens is no longer in PolyExplore. Now taking contact with the director of products and marketing, who disagrees with previously decided deliverables. They can only provide pointcloud map of the outside environment + semantic map in Shape format.
- PolyExplore is now refusing to use the sensor they promised to use for data collection
- Discussing with BoD to seek another map provider.
- We can first ask for Outdoor map to PolyExplore
- Tier IV can create indoor map in August and combine two together
Working group report: Operational Design Domain
After the brief communication with ASWG, ODD WG narrowed the additional scenarios (addressable by the scenario testing framework) by the analysis of the videos from the human-driven cargo vehicle in SeaTac provided by Zing Robotics down to four from the list of fifteen scenes of interest.
- Expected debris (boxes, cones, suitcases)
- Very narrow way/areas with objects protruding
- Articulated vehicles with multiple links
- Unusual-shaped work vehicles
(pending)
- Automated Rolling Gate
The forty-three cargo delivery scenarios ODD WG has created so far are being tested with the scenario testing framework and Autoware.auto with help from Robotec.ai.
Continuing discussion about the map technology, including 3D ‘ceiling’ map of Polyexplore’s Polymapper.
The ODD definition of racing in comparison with other ODDs (Cargo and AVP) has been finalized. Reaching out to Racing WG members to get the advice for the preliminary list of use cases for racing.
Introduced the ODD WG activities in the Autoware Workshop of IV21 conference.
Coming milestone:
- Create natural language & machine readable scenarios by the end of July.
Working group report: Hardware
- Some progress but much work to be done. Some discussion of lidar and f1tenth cost. Took an action to review and clean up older stale information.
- July 6th WG meeting minutes (June 22nd meeting was cancelled)
https://docs.google.com/document/d/1vKcIpyg1w2Tri4d3kN_x2-lWDVufONq-MWZfgw6sVA0/edit?usp=sharing
Working group report: Simulation
- Discussion about architecture of the entire E2E simulation pipeline. Initial architecture defined, consisting of the following modules:
- Scenario editor/generator for creation of domain specific scenarios (for now, scenarios created by ODD group using TierIV Scenario Editor tool)
- Scenario Runner for experiments execution
- Metrics Calculation module
- Simulation software
- Discussion about Scenario Runner - review of available solutions still in progress. As a starting point, TierIV will integrate their scenario_simulator_v2 with Autoware CI/CD
- Metrics calculation discussion - initial set of analyzed metrics defined. Discussion about the implementation of metrics calculation module still in progress
- Supported scenario formats discussion - list of potentially interesting formats created. Needs further analysis which formats should be supported in validation pipeline of Autoware
Working group report: Racing
- Working towards hybrid virtual/in-person hackathon event at Circuit Launch in Oakland, CA on August 12-13.
- Plan is to release baseline Autoware.Auto build running on F1TENTH hardware mid July (Josh to confirm) as starting point of hackathon development tasks.
- Stretch goal of competing in August 14 DYI Robocar event also at Circuit Launch. Event will be sponsored by AWF and Racing WG leads will give a presentation about Autoware and F1TENTH.
- LG has released update to SVL which includes Circuit Launch track asset and is working on sensor plugins for F1TENTH hardware (2D-lidar, stereo camera)
- New release will be available in few weeks with support of 2D-LiDAR and depth camera
- Josh and Joe submitting papers for ROSCon on IAC
Working group report: Open AD Kit
- Work group has been kicked-off and work is focused on achieving v0.5 release around end of July and defining roadmap to v1.0 in Sept timeframe.
- Matt Spencer from Arm has self-nominated for chair position (to be ratified today). Still investigating who will take the co-chair position.
- Action Item: Mitsudome will add access to the Autoware Calendar + TSC Working Folder
FTE activity reports review
FTE activity reports 2021_07_14.pdf (74.4 KB)
Autoware CI/CD Pipeline Proposal
- Presentation from Yoshinori Tsutake
- Raise discussion about the handling of CI/CD infrastructure cost
- Question:
- Is the cost assuming GPU cluster? → needs to be checked
- Is it using SVL simulator?
- First implementation will be using Tier IV, but API between the scenario simulator and SVL simulator
- Decision Made:
- TSC agrees that Scenario Simulation Pipeline is mandatory and should be implemented
- Frequency of the simulation execution should be determined.
- Check how current AWS is handled (Josh, Daisuke, Mitsudome to check)