Skip to content

Controlled Gaussian Process Dynamical Models (CGPDMs) Python library.

License

Notifications You must be signed in to change notification settings

fabio-amadio/cgpdm_lib

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

21 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

cgpdm-lib

Python library implementing Controlled Gaussian Process Dynamical Models (CGPDMs).

Class GPDM() implements original model from Wang et al. (2005) "Gaussian process dynamical models".

Class CGPDM() extends the main class to take into account also the presence of control inputs.

Installation

To install cgpdm_lib on your system, clone the repository, open it in a terminal and run the following command:

pip install .

Instead if you want to install the package in editable mode run the following command:

pip install -e .

Dependencies

Usage

Open a terminal inside example/ folder.

  • Run $ python train_cgpdm.py to train CGPDM on a cloth movement dataset (stored inside folder example/DATA/) and save the resulting model (inside example/ROLLOUT/ folder).

    train_cgpdm.py takes the following command line arguments:

    • seed: select the random seed
    • num_data: select the number of trajectories used for training
    • deg: select the oscillation angle used in data collection (5, 10 or 15)
    • d: select the latent space dimension
    • num_opt_steps: select the number of optimization steps
    • lr: select the optimization learning rate
    • flg_show: set to 'True' for showing model results
  • Run $ python load_cgpdm.py to load a trained CGPDM and show rollouts on training data.

    load_cgpdm.py takes the following command line argument:

    • model_name: model label inside the example/ROLLOUT/ folder.

(train_gpdm.py and load_gpdm.py scripts works analogously but applying the original GPDM on the same some cloth movement dataset)

Citing

If you use this package for any academic work, please cite our original paper.

@article{amadio2023controlled,
  title={Controlled gaussian process dynamical models with application to robotic cloth manipulation},
  author={Amadio, Fabio and Delgado-Guerrero, Juan Antonio and Colom{\'e}, Adria and Torras, Carme},
  journal={International Journal of Dynamics and Control},
  pages={1--11},
  year={2023},
  publisher={Springer},
  doi={10.1007/s40435-023-01205-6}
}

About

Controlled Gaussian Process Dynamical Models (CGPDMs) Python library.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages