Implementation of genetic algorithms and Deep Deterministic Policy Gradients (DDPG) trained on Modified Rewards of Gym's Bipedal Walker. For visualizations, see demo.
DDPG-Default Reward:
DDPG-RunFaster Reward:
This is the portion of a group project done by me.
git clone https://github.com/BatmanofZuhandArrgh/BipedalWalker-ModifiedReward.git
cd BipedalWalker-ModifiedReward
Install an appropriate version of pytorch, then run:
pip install -r requirements.txt
To install and see the run results of genetic algorithms see genetic_alg/README.md, and ddpg and modified_rewards, see ddpg/README.md