[BUG FIX] Add angular damping to drone to prevent numerical instability

[yun-long]

Description

This PR fixes a bug where the drone simulation could become numerically unstable due to unbounded angular velocities.

🔧 What changed:

• Added small damping to the angular DOFs using self.drone.set_dofs_damping(torch.tensor([0.0, 0.0, 0.0, 1e-4, 1e-4, 1e-4]))
  • The first three values correspond to linear damping (kept at 0).
  • The last three apply small damping to the angular components (roll, pitch, yaw).
• This helps suppress high-frequency oscillations and prevents angular velocity from exploding under extreme control inputs.

No other functional or API changes were made.

Related Issue

Related to #909

Motivation and Context

How Has This Been / Can This Be Tested?

Yes. The training still works as before

cd examples/drone/
python hover_train.py

# after training
python hover_eval.py

Also tested with the following minimum example

import torch

import genesis as gs

from hover_env import HoverEnv


gs.init(backend=gs.cpu, precision="32", logging_level="error", seed=0)

env = HoverEnv(
    env_cfg={
        "num_actions": 4,
        # termination
        "termination_if_roll_greater_than": 180,  # degree
        "termination_if_pitch_greater_than": 180,
        "termination_if_close_to_ground": 0.1,
        "termination_if_x_greater_than": 3.0,
        "termination_if_y_greater_than": 3.0,
        "termination_if_z_greater_than": 2.0,
        # base pose
        "base_init_pos": [0.0, 0.0, 1.0],
        "base_init_quat": [1.0, 0.0, 0.0, 0.0],
        "episode_length_s": 15.0,
        "at_target_threshold": 0.1,
        "resampling_time_s": 3.0,
        "simulate_action_latency": True,
        "clip_actions": 1.0,
        # visualization
        "visualize_target": False,
        "visualize_camera": False,
        "max_visualize_FPS": 60,
    },
    obs_cfg={
        "num_obs": 17,
        "obs_scales": {
            "rel_pos": 1 / 3.0,
            "lin_vel": 1 / 3.0,
            "ang_vel": 1 / 3.14159,
        },
    },
    reward_cfg={
        "yaw_lambda": -10.0,
        "reward_scales": {
            "target": 10.0,
            "smooth": -1e-4,
            "yaw": 0.01,
            "angular": -2e-4,
            "crash": -10.0,
        },
    },
    command_cfg={
        "num_commands": 3,
        "pos_x_range": [-1.0, 1.0],
        "pos_y_range": [-1.0, 1.0],
        "pos_z_range": [1.0, 1.0],
    },
    show_viewer=False,
    num_envs=1,
)

action = torch.tensor([[-0.5, 0.77, -1.0, 0.99]], dtype=gs.tc_float, device=gs.device)

i = 0
iter_exploding = -1
env.reset()
while True:
    env.step(action)
    num_constraints = env.scene.rigid_solver.constraint_solver.n_constraints.to_torch().item()
    acc = env.scene.rigid_solver.dofs_state.acc.to_torch()[:, 0]
    print(f"[{i}] num constraints: {num_constraints}")
    print(f"[{i}] acc: {acc}")
    i += 1

    if iter_exploding != -1:
        if torch.isnan(acc).any():
            break
    else:
        if (torch.abs(acc) > 1e5).any():
            iter_exploding = i

Screenshots (if appropriate):

Checklist:

• [ x] I read the CONTRIBUTING document.
• [ x] I followed the Submitting Code Changes section of CONTRIBUTING document.
• [ x] I tagged the title correctly (including BUG FIX/FEATURE/MISC/BREAKING)
• [ x] I updated the documentation accordingly or no change is needed.
• [ x] I tested my changes and added instructions on how to test it for reviewers.

• I have added tests to cover my changes.
• All new and existing tests passed.

[duburcqa]

What about making this the default for Drone in general. Not just this example. Because I think this issue is more general. I can do it if you don’t want to spend time on it.

[duburcqa]

By the way, I changed “Resolve” for “Related”. Because I don’t think it is addressing the root cause of the issue, but it mitigates it successfully.

[YilingQiao]

I think this issue is more general.

Do you mean damping parameter for rotational DoFS?