Add multirotor/thruster actuator, multirotor asset and manager-based ARL drone task

[grzemal]

Description

This PR introduces multirotor and thruster support and adds a manager-based example/task for the ARL drone. The change contains a new low-level thruster actuator model, a new Multirotor articulation asset class + configs, new thrust actions, and a manager-based drone task (ARL drone) with MDP configs and RL agent configs.

Motivation and context

• Provides a reusable multirotor abstraction and a parameterized thruster actuator model so we can simulate multirotor vehicles (quad/hex/other).
• Adds a manager-based ARL drone task and configuration files to enable repro and training workflows for the ARL drone platform.
• Consolidates drone-specific code and prepares the repo for future control/sensor improvements.

Type of change

• New feature (non-breaking addition of new functionality)
• Documentation update (added docs/comments where applicable)

Files changed (high-level summary)

• New/major files added:
  • source/isaaclab/isaaclab/actuators/thruster.py (new thruster actuator model)
  • source/isaaclab/isaaclab/assets/articulation/multirotor.py (new multirotor articulation)
  • source/isaaclab/isaaclab/assets/articulation/multirotor_cfg.py
  • source/isaaclab/isaaclab/assets/articulation/multirotor_data.py
  • source/isaaclab/isaaclab/envs/mdp/actions/thrust_actions.py
  • source/isaaclab_assets/isaaclab_assets/robots/arl_robot_1.py and ARL drone URDF + asset files as a submodule
  • source/isaaclab_tasks/isaaclab_tasks/manager_based/drone_ntnu/* (new task code, commands, observations, rewards, state-based control configs and agent configs)
• Modified:
  • source/isaaclab/isaaclab/actuators/actuator_cfg.py (register thruster config)
  • source/isaaclab/isaaclab/envs/mdp/actions/actions_cfg.py (register thrust actions)
  • small edits to various utils and types, and docs/make.bat
• Total diff (branch vs main when I checked): 33 files changed, ~2225 insertions, 65 deletions

Dependencies

• No new external top-level dependencies introduced. The branch adds assets (binary .zip) — ensure Git LFS is used if you want large assets tracked by LFS.
• The new drone task references standard repo-internal packages and Isaac Sim; no external pip packages required beyond the repo standard.

Checklist (status)

• [x] I have read and understood the contribution guidelines
• [x] I have run the pre-commit checks with ./isaaclab.sh --format
• [x] I have made corresponding changes to the documentation
• [x] My changes generate no new warnings
• [x] I have added tests that prove my fix is effective or that my feature works
• [ ] I have updated the changelog and the corresponding version in the extension's config/extension.toml file
• [x] I have added my name to the CONTRIBUTORS.md or my name already exists there

[pascal-roth]

@AntoineRichard @ooctipus would be great if you could give the PR another pass through and give your okay

[greptile-apps[bot]]

Greptile Overview

Greptile Summary

This PR introduces comprehensive multirotor support for Isaac Lab, adding a low-level thruster actuator model, a Multirotor articulation asset class, thrust actions, and a manager-based ARL drone task.

Key additions:

• Thruster actuator model (thruster.py): Implements first-order motor dynamics with asymmetric rise/fall time constants, RK4/Euler integration, and thrust-to-RPM conversion. Well-tested with edge case coverage.
• Multirotor asset (multirotor.py, multirotor_data.py): Extends Articulation to support thrusters with allocation matrices for control distribution. MultirotorData properly inherits from ArticulationData.
• Thrust actions (thrust_actions.py): Action term for processing thrust commands with scale/offset/clip transformations.
• ARL drone task: Manager-based RL environment with MDP configurations (commands, observations, rewards) and agent configs for multiple RL frameworks.

Critical issue found:

• state_based_control_env_cfg.py:63-69 - All command ranges are set to (0.0, 0.0), meaning the drone always targets the same position/orientation with no variation. This makes training completely ineffective since the policy never learns to reach different goals. Must be fixed before training.

Other observations:

• The yaw_aligned reward assumes target yaw is always 0, but the command config includes a yaw range (currently 0.0, 0.0). If you enable yaw variation, the reward should read the target yaw from commands instead of hardcoding 0.
• ARL_ROBOT_1_CFG has a TODO comment about motor/propeller specifications that should be completed.
• Comprehensive test coverage demonstrates robustness of the thruster implementation.

Confidence Score: 2/5

• This PR has solid implementation but contains a critical configuration bug that breaks training
• Score reflects one critical blocker: the task configuration has all command ranges set to (0.0, 0.0), making the training environment non-functional. The core thruster/multirotor implementation is well-designed and thoroughly tested, but the example task cannot train agents without fixing the command ranges. This is a simple config fix but essential for the PR's stated goal of enabling training workflows.
• Pay immediate attention to state_based_control_env_cfg.py - the zero command ranges must be fixed before any training attempts. The core implementation files (thruster.py, multirotor.py) are solid.

Important Files Changed

File Analysis

Filename	Score	Overview
source/isaaclab/isaaclab/actuators/thruster.py	4/5	Implements low-level thruster dynamics with first-order motor model, RK4/Euler integration, and asymmetric rise/fall time constants - comprehensive and well-tested
source/isaaclab/isaaclab/assets/articulation/multirotor.py	4/5	Extends Articulation for multirotor vehicles with thruster support, allocation matrix, and combined wrench application - properly inherits from parent
source/isaaclab/isaaclab/envs/mdp/actions/thrust_actions.py	4/5	Implements thrust action term with scale/offset/clip transformations - handles both float and dict configs
source/isaaclab_tasks/isaaclab_tasks/manager_based/drone_arl/state_based_control/config/arl_robot_1/state_based_control_env_cfg.py	1/5	Drone task config with CRITICAL ISSUE: all command ranges set to (0.0, 0.0) making training ineffective - drone always targets same pose
source/isaaclab_tasks/isaaclab_tasks/manager_based/drone_arl/mdp/rewards.py	3/5	Drone reward functions (distance, velocity penalties, yaw alignment) - yaw_aligned assumes target yaw is always 0 which may not match command
source/isaaclab_assets/isaaclab_assets/robots/arl_robot_1.py	4/5	ARL Robot 1 config with thruster parameters and allocation matrix for quadrotor - has TODO for motor/propeller specs

Sequence Diagram

sequenceDiagram
    participant Env as ManagerBasedRLEnv
    participant Action as ThrustAction
    participant Multirotor as Multirotor Asset
    participant Thruster as Thruster Actuator
    participant PhysX as PhysX Simulation

    Note over Env,PhysX: Simulation Step

    Env->>Action: process_actions(raw_actions)
    Action->>Action: Apply scale/offset/clip
    Note over Action: processed_actions = raw * scale + offset

    Env->>Action: apply_actions()
    Action->>Multirotor: set_thrust_target(processed_actions)
    Note over Multirotor: Store in _data.thrust_target

    Env->>Multirotor: write_data_to_sim()
    
    Multirotor->>Multirotor: _apply_actuator_model()
    loop For each thruster actuator
        Multirotor->>Thruster: compute(control_action)
        Note over Thruster: Read thrust_target
        Thruster->>Thruster: Clamp to thrust_range
        Thruster->>Thruster: Select tau (inc/dec)
        Thruster->>Thruster: Compute mixing factor
        Thruster->>Thruster: Integrate (Euler/RK4)
        Note over Thruster: curr_thrust updated
        Thruster-->>Multirotor: Return applied_thrust
        Multirotor->>Multirotor: Update computed_thrust, applied_thrust
    end

    Multirotor->>Multirotor: _apply_combined_wrench()
    Multirotor->>Multirotor: _combine_thrusts()
    Note over Multirotor: wrench = allocation_matrix @ thrusts
    Multirotor->>PhysX: apply_forces_and_torques_at_position()
    Note over PhysX: Apply force/torque to base link

    PhysX-->>Multirotor: Physics update
    Multirotor->>Multirotor: update(dt)
    Note over Multirotor: Read state from PhysX

[pascal-roth]

@grzemal when running the example task, I get a lot of following warning

2025-12-17 09:50:35 [102,769ms] [Warning] [omni.usd] Warning: in GetInstanceIndices at line 2190 of /builds/omniverse/usd-ci/USD/pxr/usdImaging/usdImaging/pointInstancerAdapter.cpp -- ProtoIndex 1 out of bounds (prototypes size = 1) for (/Visuals/Command/body_pose, /Visuals/Command/body_pose.proto1_mesh_id0)

did you also experience them?

[grzemal]

@grzemal when running the example task, I get a lot of following warning

2025-12-17 09:50:35 [102,769ms] [Warning] [omni.usd] Warning: in GetInstanceIndices at line 2190 of /builds/omniverse/usd-ci/USD/pxr/usdImaging/usdImaging/pointInstancerAdapter.cpp -- ProtoIndex 1 out of bounds (prototypes size = 1) for (/Visuals/Command/body_pose, /Visuals/Command/body_pose.proto1_mesh_id0)

did you also experience them?

@pascal-roth just double-checked and i do not get such a warning, the only warning which I get and it is related to the Visual/Command :

2025-12-17T11:44:32Z [12,876ms] [Warning] [omni.physx.fabric.plugin] FabricManager::initializePointInstancer mismatched prototypes on point instancer: /Visuals/Command/goal_pose

[pascal-roth]

@grzemal when running the example task, I get a lot of following warning

2025-12-17 09:50:35 [102,769ms] [Warning] [omni.usd] Warning: in GetInstanceIndices at line 2190 of /builds/omniverse/usd-ci/USD/pxr/usdImaging/usdImaging/pointInstancerAdapter.cpp -- ProtoIndex 1 out of bounds (prototypes size = 1) for (/Visuals/Command/body_pose, /Visuals/Command/body_pose.proto1_mesh_id0)

did you also experience them?

@pascal-roth just double-checked and i do not get such a warning, the only warning which I get and it is related to the Visual/Command :

2025-12-17T11:44:32Z [12,876ms] [Warning] [omni.physx.fabric.plugin] FabricManager::initializePointInstancer mismatched prototypes on point instancer: /Visuals/Command/goal_pose

hm weird, then that must have something to do with the asset. Could anyone try on their PC to check if the same happens when they init the submodule? @Mayankm96 @ooctipus @AntoineRichard