Esteve Fernandez

According to https://forums.developer.nvidia.com/t/mounting-of-img-file/128257/8 it should be possible to mount the Jetson image as a loopback device, but so far I haven't succeeded. `fdisk` reports an offset of `1187840`, but neither...

More progress, I managed to use `kpartx` to mount the first partition in the image and `chroot`ed in to inspect the `/` filesystem. However, running Docker from inside the `chroot`...

According to https://github.com/orgs/autowarefoundation/discussions/255#discussioncomment-2664151 upgrading to Jetpack 5.0 fixes any CUDA-related issues. @jason914 could you try upgrading to Jetpack 5.0 and see if you can run Autoware? Thanks.

@kenji-miyake what do you think of having an `embree_vendor` package and have `scenario_simulator` depend on it? If `libembree-dev` is installed on the system, that'll be used, and if not, build...

I'll look into this as well

@yukkysaito how are you launching the sensing module? Separately via `ros2 launch tier4_sensing_launch sensing.launch.xml` ?

@yukkysaito I've been able to launch autoware via `ros2 launch autoware_launch logging_simulator.launch.xml ...`, with both modules loaded.

Here's a summary of what I learned about TVM (thanks @ambroise-arm for your help, please let me know if there's anything incorrect): - TVM aims to be backend agnostic when...

For a very similar problem, see https://gitlab.com/autowarefoundation/autoware.auto/AutowareAuto/-/merge_requests/978 and how it was solved in Autoware.Auto

@kenji-miyake sorry, this is still not ready, there are some more places in `euclidean_cluster` where I need to migrate to `point_cloud_msg_wrapper`