rickstaa/stable-learning-control

Replicate results of Han et al. 2019

Closed this issue · 2 comments

User story

As a developer I want to replicate the results of the LAC algorithm in Han et al. 2019 such that I am sure the LAC algorithm works before trying to use it on the Panda robot environment.

Considerations

To do this we translate add the oscilator environment and LAC algorithm code from Actor-critic-with-stability-guarantee. While doing this the LAC will be changed such that pytorch is used instead of tensorflow.

Acceptance criteria

  • The results of the trained LAC algorithm should be nearly identical to those depicted in Han et al. 2019 for the swimmer and oscillator environments.

This feature has been implemented on the add_lac_algorithm branch and will be pulled into main with pull request #37.