Upgrade to IsaacSim 4.5 and IsaacLab 2.0.2

[mingxin-zheng]

Fixes #86

Description

This PR upgrades the simulation framework to work with IsaacSim 4.5 and IsaacLab 2.0.2 by updating all import paths and dependencies from the old omni.isaac.lab modules to their new counterparts in isaaclab and isaacsim.

It also fixes the pydantic error by versioning to 2.10.6

[Copilot]

Pull Request Overview

This PR upgrades the simulation framework to work with IsaacSim 4.5 and IsaacLab 2.0.2 by updating all import paths and dependencies from the old omni.isaac.lab modules to their new counterparts in isaaclab and isaacsim.

• Update import paths across simulation tasks, configuration files, environment modules, and annotators.
• Replace deprecated package references with the new IsaacLab/IsaacSim modules.
• Adjust default parameters in command-line arguments where applicable.

Reviewed Changes

Copilot reviewed 24 out of 25 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
workflows/robotic_ultrasound/scripts/simulation/exts/robotic_us_ext/robotic_us_ext/tasks/ultrasound/approach/mdp/events.py	Updated package imports from omni.isaac.lab to isaaclab.
workflows/robotic_ultrasound/scripts/simulation/exts/robotic_us_ext/robotic_us_ext/tasks/ultrasound/approach/mdp/init.py	Updated imports to new IsaacLab modules.
workflows/robotic_ultrasound/scripts/simulation/exts/robotic_us_ext/robotic_us_ext/tasks/ultrasound/approach/config/teleop/ik_rel_env_cfg.py	Updated configuration imports from omni.isaac.lab to isaaclab.
workflows/robotic_ultrasound/scripts/simulation/exts/robotic_us_ext/robotic_us_ext/tasks/ultrasound/approach/config/franka/franka_manager_rl_env_cfg.py	Replaced omni.isaac.lab references with updated references in isaaclab.
workflows/robotic_ultrasound/scripts/simulation/exts/robotic_us_ext/robotic_us_ext/tasks/ultrasound/approach/config/franka/init.py	Adjusted the environment entry points to reference isaaclab.
workflows/robotic_ultrasound/scripts/simulation/exts/robotic_us_ext/robotic_us_ext/tasks/init.py	Updated task module import from omni.isaac.lab_tasks to isaaclab_tasks.
workflows/robotic_ultrasound/scripts/simulation/exts/robotic_us_ext/robotic_us_ext/lab_assets/franka.py	Updated asset and actuator imports with the new package names.
workflows/robotic_ultrasound/scripts/simulation/examples/sim_with_dds.py	Updated application launcher and config parser imports.
workflows/robotic_ultrasound/scripts/simulation/environments/teleoperation/teleop_se3_agent.py	Replaced multiple omni.isaac.lab imports with their isaaclab equivalents.
workflows/robotic_ultrasound/scripts/simulation/environments/state_machine/utils.py	Updated math utility function imports.
workflows/robotic_ultrasound/scripts/simulation/environments/state_machine/pi0_policy/eval.py	Modified command-line argument defaults and updated config parser import.
workflows/robotic_ultrasound/scripts/simulation/environments/state_machine/modules/path_planning_module.py	Updated math utility import as per new package structure.
workflows/robotic_ultrasound/scripts/simulation/environments/state_machine/modules/orientation_module.py	Updated math utility function imports.
workflows/robotic_ultrasound/scripts/simulation/environments/state_machine/meta_state_machine/ultrasound_state_machine.py	Updated prim view import to align with new isaacsim paths.
workflows/robotic_ultrasound/scripts/simulation/environments/state_machine/liver_scan_sm.py	Updated application launcher and config parser imports.
workflows/robotic_ultrasound/scripts/simulation/annotators/ultrasound.py	Updated prim import from omni.isaac.lab to isaacsim.
workflows/robotic_ultrasound/scripts/simulation/annotators/target.py	Updated prim import from omni.isaac.lab to isaacsim.
workflows/robotic_ultrasound/scripts/simulation/annotators/franka.py	Updated robot and controller imports, and adjusted type annotations.
workflows/robotic_ultrasound/scripts/simulation/annotators/base.py	Updated world import path to use isaacsim.

Files not reviewed (1)

• tools/env_setup_robot_us.sh: Language not supported

Comments suppressed due to low confidence (2)

workflows/robotic_ultrasound/scripts/simulation/annotators/franka.py:46

• [nitpick] Consider adding a type annotation for the 'franka' parameter for consistency and improved clarity across the codebase.

def __init__(self, franka, ik: bool, prim_path: str, topic: str, period: float, domain_id):

workflows/robotic_ultrasound/scripts/simulation/environments/state_machine/pi0_policy/eval.py:40

• Verify that updating the default task parameter from None to the specified environment name is intentional, as this may affect environment instantiation.

parser.add_argument("--task", type=str, default="Isaac-Teleop-Torso-FrankaUsRs-IK-RL-Rel-v0", help="Name of the task.")

[mingxin-zheng]

Working for eval.py

[Nic-Ma]

@KumoLiu Could you please help modify the training part accordingly?
I think you can refer to the IsaacSim doc: https://docs.isaacsim.omniverse.nvidia.com/latest/overview/extensions_renaming.html
CC @NigelNelson , @masoudmoghani .

Thanks in advance.

[KumoLiu]

@KumoLiu Could you please help modify the training part accordingly?

I tested the training component in this branch, and it works well. Several minor import issues mentioned above have been addressed. While I only tested robotic_ultrasound_lora in terms of memory usage, I also verified the code for robotic_ultrasound. The latter should work fine since the only difference is that robotic_ultrasound_lora involves fine-tuning a subset of the model parameters.

BTW, 30000 training steps need to train over 40 hours, does it expected? @NigelNelson

[NigelNelson]

@KumoLiu Could you please help modify the training part accordingly?

I tested the training component in this branch, and it works well. Several minor import issues mentioned above have been addressed. While I only tested robotic_ultrasound_lora in terms of memory usage, I also verified the code for robotic_ultrasound. The latter should work fine since the only difference is that robotic_ultrasound_lora involves fine-tuning a subset of the model parameters.

BTW, 30000 training steps need to train over 40 hours, does it expected? @NigelNelson

Yes this seems reasonable, it is a relatively large model to be training on a single consumer GPU.

[KumoLiu]

Yes this seems reasonable, it is a relatively large model to be training on a single consumer GPU.

Thanks @NigelNelson for the confirmation!

[mingxin-zheng]

integration.log
unit.log