Installation Success but Environment Loading Error with SKRL

[XinyuSong123]

Hi! I successfully installed this template and verified the installation using:
python scripts/rsl_rl/train.py --task=Template-Isaac-Velocity-Rough-Anymal-D-v0

However, when trying to use SKRL to test Template-Isaac-Velocity-Rough-Anymal-D-v0 with the following code:

# load and wrap the Isaac Lab environment
# env = load_isaaclab_env(task_name="Template-Isaac-Velocity-Rough-Anymal-D-v0")
env = load_isaaclab_env(task_name="Template-Isaac-Velocity-Rough-Anymal-D-v0")
env = wrap_env(env)

I encountered this error:

gymnasium.error.NameNotFound: Environment `Template-Isaac-Velocity-Rough-Anymal-D` doesn't exist. Did you mean: `Isaac-Velocity-Rough-Anymal-D`?

Could you please provide instructions for correctly setting up the template with SKRL? Thank you!!!

[virgilio96upm]

Hi, to use the SKRL library in the anymal_d example you need to follow these steps:

1. Copy the /skrl folder available in IsaacLab (https://github.com/isaac-sim/IsaacLab/tree/c4bec8fe01c2fd83a0a25da184494b37b3e3eb61/scripts/reinforcement_learning/skrl) to the /scripts folder in your extension template and modify both train.py and play.py files by substituing the tasks from IsaacLab to your example template extension tasks here: https://github.com/isaac-sim/IsaacLab/blob/c4bec8fe01c2fd83a0a25da184494b37b3e3eb61/scripts/reinforcement_learning/skrl/train.py#L100 (substitute this to import ext_template.tasks # noqa: F401)
2. In your anymal_d/agents directory copy the skrl configuration file for the rough terrain task (example from IsaacLab: https://github.com/isaac-sim/IsaacLab/blob/main/source/isaaclab_tasks/isaaclab_tasks/manager_based/locomotion/velocity/config/anymal_d/agents/skrl_rough_ppo_cfg.yaml)
3. In the anymal_d directory check the __innit__.py file where the tasks are registered and add the entrypoint to your skrl config file as shown here: https://github.com/isaac-sim/IsaacLab/blob/c4bec8fe01c2fd83a0a25da184494b37b3e3eb61/source/isaaclab_tasks/isaaclab_tasks/manager_based/locomotion/velocity/config/anymal_d/__init__.py#L43
4. test if this work by running in your terminal python scripts/skrl/train.py --task=Template-Isaac-Velocity-Rough-Anymal-D-v0

Let me know if it works for you!

[yizhengzhang1]

I met the same problem. I believe the problem lies in the docker-compose.yaml.

I don't know why this way let the gym can not find the registered custom environment.