Semi-Tied Gaussian Mixture Models/Hidden Semi-Markov Models

This package contains the algorithms, experiments and datasets for semi-tied GMMs/HMMs/HSMMs, included in

Tanwani, A. K., Calinon, S., "Learning Robot Manipulation Tasks with Task-Parameterized Semi-Tied    
Hidden Semi-Markov Model", IEEE Robotics and Automation Letters (RA-L), vol. 1 (1), pp. 235-242, 2016.  

Algorithms

EM_STGMM.m - semi-tied Gaussian mixture model (GMM)
EM_tensorHSMM.m - task-parameterized hidden semi-Markov model
EM_tensorSTHSMM.m - task-parameterized semi-tied hidden semi-Markov model
TV_LQR_continuous_ff.m - time varying finite horizon linear quadratic tracking controller with feedforward term
HSMM_LQR_controller.m - sampling the sequence of states from a hidden semi-Markov model and following the desired step-wise reference trajectory with a linear quadratic tracking controller

Examples

Main_ST_GMM_Zshape3D.m
Main_ST_GMM_ChickenDance.m
Main_TP_ST_HSMM_BaxterPickPlace.m
Main_TP_ST_HSMM_BaxterValve.m

Datasets

ChickenDance.mat - Chicken Dance Movement: The dataset is taken from http://mocap.cs.cmu.edu/
Zshape3D.mat - Zshape3D : Synthetic 3-dimensional z-shaped movement
BaxterPickPlace.mat - Pick and Place Task: Demonstrations collected to teach Baxter manipulation task of picking an object from different initial configurations and placing it on the target by avoiding an obstace of varying height
BaxterValveDemos2.mat - Valve Opening Task: Demonstrations collected to teach Baxter manipulation task of opening a valve from different initial configurations