madgwick.md

Madgwick Filter API

Module: pyIMU.madgwick

Class

Madgwick(**kwargs)

Madgwick gradient-descent AHRS filter (NED convention).

Initialization Parameters

• dt: float, default 0.01
  • Default time step used when update(..., dt<=0).
• gain: float, optional
  • Convenience gain applied to both IMU and MARG unless overridden.
• gain_imu: float, default 0.033 (or gain if provided)
  • Gain used for IMU updates (gyr+acc).
• gain_marg: float, default 0.041 (or gain if provided)
  • Gain used for MARG updates (gyr+acc+mag).
• acc_in_g: bool, default True
  • True: accelerometer input expected in g.
  • False: input treated as m/s^2 and converted to g.
• gyr_in_dps: bool, default False
  • True: gyroscope input expected in deg/s and converted to rad/s.
• convention: str, default "NED"
  • Only NED is currently supported.

Methods

• update(gyr, acc, mag=None, dt=-1) -> Quaternion
  • gyr: Vector3D gyro sample.
  • acc: Vector3D accel sample.
  • mag: optional Vector3D magnetometer sample.
  • dt: time step in seconds. If <=0, uses instance dt.

Behavior:

• First call initializes orientation from accel (accel2q) or accel+mag (accelmag2q).
• Later calls run IMU or MARG update equations.

Outputs (instance fields updated on each update(...) call):

• q: Quaternion
  • Current attitude estimate.
• azero: Vector3D
  • Gravity-removed acceleration in sensor/body frame, in g.
• aglobal: Vector3D
  • Gravity-removed acceleration in earth/NED frame, in g.

Functional API

• updateIMU(q, gyr, acc, dt, gain) -> Quaternion
• updateMARG(q, gyr, acc, mag, dt, gain) -> Quaternion
• updateIMU_inplace(...) -> Quaternion
• updateMARG_inplace(...) -> Quaternion

Use inplace variants to mutate and reuse an existing quaternion object.