moveit2
The MoveIt Motion Planning Framework for ROS 2.
Loading...
Searching...
No Matches
kdl_kinematics_plugin::internal Namespace Reference

Functions

Eigen::VectorXd solvePseudoinverse (const Eigen::Ref< const Eigen::MatrixXd > &jacobian, const Eigen::Ref< const Eigen::VectorXd > &input, double threshold)

Function Documentation

◆ solvePseudoinverse()

Eigen::VectorXd kdl_kinematics_plugin::internal::solvePseudoinverse ( const Eigen::Ref< const Eigen::MatrixXd > & jacobian,
const Eigen::Ref< const Eigen::VectorXd > & input,
double threshold )
inline

Solve the minimum-norm system $x = J^T(JJ^T)^\#b$.

IK Jacobians have at most six rows and usually more columns. Decomposing the small symmetric $JJ^T$ matrix is therefore significantly faster than computing a thin SVD of $J$ on every Newton iteration. The cutoff below applies the solver's relative singular-value threshold to the eigenvalues.

Definition at line 18 of file pseudoinverse.hpp.

Here is the caller graph for this function: