Skip to contents

Estimate orientation from IMU sensors using sensor fusion

Source: R/mocap-imu.R
estimateOrientation.Rd

Fuses accelerometer, gyroscope, and optionally magnetometer data to estimate 3D orientation over time. Returns both quaternion and Euler angle representations.

Usage

estimateOrientation(
  accel,
  gyro,
  mag = NULL,
  method = c("madgwick", "complementary"),
  beta = 0.1,
  sampling_rate
)

Arguments

accel

Numeric matrix (n x 3) of accelerometer readings in m/s^2. Columns correspond to x, y, z axes.

gyro

Numeric matrix (n x 3) of gyroscope readings in rad/s. Columns correspond to x, y, z axes.

mag

Numeric matrix (n x 3) of magnetometer readings, or NULL if unavailable. Default NULL.

method

Character. Sensor fusion method: "madgwick" (default) or "complementary".

beta

Numeric. Filter gain for the Madgwick filter (default 0.1). For the complementary filter this parameter sets the alpha coefficient (gyroscope trust weight). Lower values for Madgwick or higher values for complementary increase reliance on the gyroscope.

sampling_rate

Numeric. Sampling rate in Hz.

Value

A data.frame with columns:

time

Time in seconds from start.

roll

Roll angle (degrees), rotation about X axis.

pitch

Pitch angle (degrees), rotation about Y axis.

yaw

Yaw angle (degrees), rotation about Z axis.

q_w

Quaternion scalar component.

q_x

Quaternion x component.

q_y

Quaternion y component.

q_z

Quaternion z component.

Details

The Madgwick filter uses a gradient-descent algorithm to correct gyroscope drift using accelerometer (and optionally magnetometer) measurements. The beta parameter controls the correction strength.

The complementary filter blends gyroscope integration (high-pass) with accelerometer tilt estimation (low-pass). The beta parameter serves as the alpha coefficient, controlling the balance between gyroscope and accelerometer contributions.

References

Madgwick SOH, Harrison AJL, Vaidyanathan R (2011). "Estimation of IMU and MARG orientation using a gradient descent algorithm." IEEE International Conference on Rehabilitation Robotics.

See also

removeGravity(), calibrateIMU(), quaternionToEuler()

Examples

# Simulate static IMU data (sensor resting with gravity along -Z)
n <- 100
accel <- matrix(c(rep(0, n), rep(0, n), rep(-9.81, n)), ncol = 3)
gyro <- matrix(0, nrow = n, ncol = 3)
result <- estimateOrientation(accel, gyro, sampling_rate = 100)
head(result)
#>   time roll pitch yaw q_w q_x q_y q_z
#> 1 0.00    0     0   0   1   0   0   0
#> 2 0.01    0     0   0   1   0   0   0
#> 3 0.02    0     0   0   1   0   0   0
#> 4 0.03    0     0   0   1   0   0   0
#> 5 0.04    0     0   0   1   0   0   0
#> 6 0.05    0     0   0   1   0   0   0