Skip to content

Latest commit

 

History

History
22 lines (17 loc) · 854 Bytes

README.md

File metadata and controls

22 lines (17 loc) · 854 Bytes

DDPG_D2C

Project to evaluate our D2C approach and compare it with DDPG

Contributers: Karthikeya S Parunandi and Ran Wang.

'karthik_branch' has DDPG [1] setup (Python3) (adapted from Keras-rl[2]'s implementation) and 'ran_branch' has the implementation of D2C (C++). Further, 'karthik_branch' also has the implementation of D2C in Python3. The following systems are considered as of now:

  • Pendulum
  • Cartpole
  • Swimmer (3-link)
  • Swimmer (6-link)
  • Fish
  • Hopper
  • Cheetah

The models are taken from OpenAI gym [3] and Deepmind-Control suite[4] and then modified according to our problem.

References:

  1. Continuous control with deep reinforcement learning, https://arxiv.org/abs/1509.02971
  2. Keras-rl, https://github.com/keras-rl/keras-rl
  3. OpenAI gym, https://github.com/openai/gym
  4. Deepmind dm_control, https://github.com/deepmind/dm_control