[Proposal] Tune and Validate Joint Gains for Supported Robots in Simulation

[makolon]

Proposal

Tune the joint gains for the supported robots to ensure optimal performance. Verify that gain values specified in scripts are correctly applied in the simulation environment, and use a Gain Tuner to validate the appropriateness of the settings. Additionally, ensure that comprehensive documentation on gain tuning is available.

Motivation

The primary motivation for this proposal is to optimize the control and stability of supported robots by fine-tuning their joint gains. Proper gain settings are essential for achieving accurate and stable movement within the simulation. By confirming the application of gain values and validating with a Gain Tuner, we aim to enhance the simulation's overall realism and effectiveness.

Alternatives

An alternative would be to use the default gain settings provided by the simulation platform. However, these may not be optimal for all robot models, and customizing gain values can result in significantly improved performance.

Additional context

Using a Gain Tuner ensures that the applied gains are appropriate for each robot's specific characteristics. Detailed documentation will also be developed to guide users on how to conduct gain tuning effectively within the simulation environment.

Checklist

• I have checked that there is no similar issue in the repo (required)

Acceptance Criteria

• Joint gains for supported robots are appropriately tuned.
• Gain values specified in scripts are verified to reflect in the simulation.
• Gain Tuner has been used to confirm reasonable gain settings.
• Documentation on gain tuning is complete and accessible.

References

• https://docs.omniverse.nvidia.com/isaacsim/latest/features/robots_simulation/ext_omni_isaac_gain_tuner.html
• https://isaac-sim.github.io/IsaacLab/source/api/lab/omni.isaac.lab.actuators.html#omni.isaac.lab.actuators.ActuatorBaseCfg

[makolon]

Progress

• I verified that changing the gains via the Python script works, but it remains unclear whether the applied gains are appropriate. Therefore, we need to tune the gains using the Gain Tuner first, and then apply the tuned values in the Python script.
• When simulating a gripper with a closed chain, abnormal forces are being applied to the robot arm, causing the simulation to malfunction.

Possible Causes

• The gripper’s gain must be properly configured when simulating closed-chain grippers (e.g., xArm7, UR5e with Robotiq 2f 85, UR10e with Robotiq 2f 140).

Procedures

• Set stiffness, damping, and velocity limits in the <robot_name>.py file under omni.isaac.extended_assets.

FRANKA_PANDA_CFG = ArticulationCfg(
    spawn=sim_utils.UsdFileCfg(
        usd_path=f"{ISAACLAB_EXTENDED_ASSETS_DATA_DIR}/Robots/USD/Franka/franka_panda/franka_panda.usd",
        activate_contact_sensors=False,
        rigid_props=sim_utils.RigidBodyPropertiesCfg(
            disable_gravity=False,
            max_depenetration_velocity=5.0,
        ),
        articulation_props=sim_utils.ArticulationRootPropertiesCfg(
            enabled_self_collisions=True,
            solver_position_iteration_count=8,
            solver_velocity_iteration_count=0,
        ),
    ),
    init_state=ArticulationCfg.InitialStateCfg(
        joint_pos={
            "panda_joint1": 0.0,
            "panda_joint2": -0.569,
            "panda_joint3": 0.0,
            "panda_joint4": -2.810,
            "panda_joint5": 0.0,
            "panda_joint6": 3.037,
            "panda_joint7": 0.741,
            "panda_finger_joint.*": 0.04,
        },
    ),
    actuators={
        "panda_shoulder": ImplicitActuatorCfg(
            joint_names_expr=["panda_joint[1-4]"],
            effort_limit=87.0,
            velocity_limit=2.175,
            stiffness=80.0,
            damping=4.0,
        ),
        "panda_forearm": ImplicitActuatorCfg(
            joint_names_expr=["panda_joint[5-7]"],
            effort_limit=12.0,
            velocity_limit=2.61,
            stiffness=80.0,
            damping=4.0,
        ),
        "panda_hand": ImplicitActuatorCfg(
            joint_names_expr=["panda_finger_joint.*"],
            effort_limit=200.0,
            velocity_limit=0.2,
            stiffness=2e3,
            damping=1e2,
        ),
    },
    soft_joint_pos_limit_factor=1.0,
)