# Dynamic motion primitives

The {mod}`robotblockset.dmp` module provides tools for encoding demonstrated
motions as dynamic motion primitives (DMPs) and reproducing them with temporal
scaling. It supports ordinary vector-valued signals, quaternion orientations,
and Cartesian pose trajectories.

The main encoding and decoding functions are:

- {func}`robotblockset.dmp.encodeDMP` and
  {func}`robotblockset.dmp.decodeDMP` for vector-valued trajectories
- {func}`robotblockset.dmp.encodeQuaternionDMP` and
  {func}`robotblockset.dmp.decodeQuaternionDMP` for orientations
- {func}`robotblockset.dmp.encodeCartesianDMP` and
  {func}`robotblockset.dmp.decodeCartesianDMP` for Cartesian poses

For time-parameterized paths, {func}`robotblockset.dmp.Path2DMP` and
{func}`robotblockset.dmp.DMP2Path` provide convenient conversions. The
`integrateStep*DMP` functions expose individual integration steps for custom
execution loops.

See the
[`tutorial_dmp` notebook](https://github.com/abr-ijs/robotblockset_python/blob/master/src/robotblockset/tutorials/tutorial_dmp.ipynb)
for a complete workflow and the {mod}`robotblockset.dmp` API reference for all
functions and parameter details.