Skip to contents

Model-based inverse kinematics from markers (planar)

Source: R/inverse-kinematics.R
inverseKinematicsMarkers.Rd

Solves the segment angles of a scaled planar chain that best fit observed marker positions, frame by frame, by least squares. Because the segment lengths are fixed by the model, the solution is consistent (unlike reading angles directly off noisy markers, which lets the limb stretch).

Usage

inverseKinematicsMarkers(markers, model, anchor_root = TRUE)

Arguments

markers

A [frame, node, 2] array of observed node positions (root and each joint/end), or a list of (n_segments + 1) x 2 matrices.

model

A planar_ik_model from scalePlanarModel().

anchor_root

If TRUE (default) the root is fixed to the observed root marker each frame; otherwise the root is also fitted.

Value

an ik_result: angles (frame x n_segments, absolute radians), rmse (per-frame marker RMS error, in position units), and n_frames.

References

Lu TW, O'Connor JJ (1999) global optimisation IK, J Biomech 32:129-134.

See also

scalePlanarModel(), forwardKinematics2D()

Examples

model <- scalePlanarModel(rbind(c(0, 0), c(1, 0), c(2, 0)))
truth <- forwardKinematics2D(c(0.3, 0.7), model$lengths)
arr <- array(0, c(1, 3, 2)); arr[1, , ] <- truth
inverseKinematicsMarkers(arr, model)$angles         # ~ c(0.3, 0.7)
#>      [,1] [,2]
#> [1,]  0.3  0.7