Update drone-controllers to 0.2.0

[amacati]

No description provided.

[Copilot]

Pull request overview

This PR updates the project to use drone-controllers==0.2.0 and adapts Crazyflow’s Mellinger control integration, simulation pipeline calls, and tests to the updated controller APIs/parameter-loading approach.

Changes:

• Bump drone-controllers dependency to 0.2.0 and refresh the Pixi lockfile accordingly.
• Update Mellinger controller wiring to load controller params via drone_controllers.core.load_params and pass params as dictionaries.
• Adjust simulation/controller call sites and tests to match the updated controller function signatures, and add a fused-model smoke test.

Reviewed changes

Copilot reviewed 9 out of 10 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
tests/unit/test_sim.py	Adds dict handling in commit-check helper and introduces a fused model smoke test.
tests/integration/test_interfaces.py	Updates state2attitude invocation to the new signature in the attitude interface test.
pyproject.toml	Bumps drone-controllers to 0.2.0.
pixi.lock	Lockfile refresh reflecting updated dependency graph (incl. drone-controllers 0.2.0).
crazyflow/utils.py	Removes named_tuple2device helper now superseded by controller param loading.
crazyflow/sim/sim.py	Updates controller step functions to pass updated inputs and unpack dict params.
crazyflow/envs/drone_env.py	Updates attitude action space bounds to use load_params(...) and dict keys.
crazyflow/control/mellinger.py	Switches controller param storage/loading to load_params(...) with dict-based params.

Comments suppressed due to low confidence (1)

tests/integration/test_interfaces.py:54

• These integration tests create Sim(...) instances but never call sim.close(). This can leave MuJoCo/JAX resources alive across tests and cause interference/flakiness. Consider adding sim.close() in a finally block (or using a fixture) to ensure cleanup even if assertions fail.

def test_attitude_interface(physics: Physics):
    sim = Sim(physics=physics, control=Control.attitude)
    target_pos = np.array([0.0, 0.0, 1.0])
    jit_state2attitude = jax.jit(parametrize(state2attitude, drone_model="cf2x_L250"))

    i_error = np.zeros((1, 1, 3))
    cmd = np.zeros((1, 1, 13))
    cmd[0, 0, 2] = 1.0  # Set z position target to 1.0

    for _ in range(int(2 * sim.control_freq)):  # Run simulation for 2 seconds
        pos, vel, quat = sim.data.states.pos, sim.data.states.vel, sim.data.states.quat
        rpyt, i_error = jit_state2attitude(pos, quat, vel, cmd, (i_error,), ctrl_freq=100)
        sim.attitude_control(rpyt)
        sim.step(sim.freq // sim.control_freq)

    # Check if drone maintained hover position
    dpos = sim.data.states.pos[0, 0] - target_pos
    distance = np.linalg.norm(dpos)
    assert distance < 0.05, f"Failed to maintain hover with {physics} ({dpos})"