-
Follow the instructions here [https://isaac-sim.github.io/IsaacLab/main/source/setup/installation/pip_installation.html] to install IsaacSim and IsaacLab.
-
Install local versions of required packages by running the command below. This will install our copies of IsaacLab packages from
source/.
./isaaclab.sh -i- still figuring this out since robovision seems to be having issues
- running in docker will allow visualization streaming, but this is probably not necessary yet
- Make an account at https://wandb.ai/site
- Set the environment variable to enable
wandbfor logging. (NOTE: the field listed in User > username is not the username. The true username is listed in User > Teams)
export WANDB_USERNAME=<user-name>
- Follow the link printed in the terminal to view runs.
Choose an unused GPU. Please don't take too many at a time.
export CUDA_VISIBLE_DEVICES=<gpu id>
For baseline funcitonality with rsl_rl
./isaaclab.sh -p scripts/rsl_rl/train.py \
--task T1-Baseline-v0 \
--headless \
--videoThe following command will take you through the steps of pulling a policy from wandb and playing it in a single environment. If running on a lab server, use the --headless --video flags.
./isaaclab.sh -p scripts/rsl_rl/play.py \
--task T1-Baseline-Play-v0 \
--num_envs 1 \
--wandb \
--headless \
--video
[https://booster.feishu.cn/wiki/XY6Kwrq1bizif4kq7X9c14twnle]
Changes cannot be made directory to main by non-admins. Instead, make changes on a feature branch and make a pull request.
This repo is configured with a format check, which PRs must pass.
Install code formatter and linter: pip install black ruff.
Run code formatter on source and script directories:
black source scripts
Run linter on source and script directories:
ruff check --fix source scripts