2024-07-18 Racing Working Group Meeting #5013

pojenwang · 2024-07-18T14:32:21Z

pojenwang
Jul 18, 2024
Collaborator

Minutes: Po-Jen Wang

Autoware on F1tenth

f1tenth_humble branch updates: Now packages installed through autoware.repo are linked to the versions specified in autoware’s release/2024.06 branch. They link to specific commits so they would not cause issues when the packages get updated from their main branch.
Tested on a Jetson Orin Nano and a x86 computer. Successful build without an error. There were a lot of stderr warning messages that should be ignored. Simulation works well without a problem
Only on the Jetson Orin Nano, when sourcing the autoware environment, it couldn’t find local_setup.bash for three packages:
-- cuda_utils, tensorrt_common, tensorrt_classifier
-- F1tenth does not use these packages so no problem for now
Updated the documentation to include a new demo video: Recordreplay_demo which first builds a map with slam_toolbox, then manually record a trajectory, and finally playback the trajectory.

Autoware training vehicle

Currently developing a 1/5 scale off-road vehicle platform for training Autoware (in between the size of a go-kart and F1tenth car)
Based on the Traxxas X-Maxx RC car
Come with a better computing platform, sensors including a 3D Lidar instead of 2D Lidar on F1tenth, and potentially adding the Fixposition RTK2 sensor for positioning.
Vehicle design from another research team at Rahul's lab. We can reuse their design and make any necessary modifications/improvements
We are working on importing the vehicle into our simulator and adding environments with terrains for off-road vehicle
The current simulator already incorporated a full-size "Buggy" vehicle with off-road suspension for simulation. We can reuse this(scale it down?) for the Autoware training vehicle.