Skip to content

Apply major Reinforcement Learning algorithms (DQN,PPO,A2C) to CarRacing-v0 from GymAI environment.

Notifications You must be signed in to change notification settings

AGiannoutsos/car_racer_gym

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CarRacing-v0

  1. Here we have an assignment in course: Reinforcement Learning, where we have been experimented with three major algorithms, so as to solve Car_Racing_v0 problem from Gym.AI environment. Note that Car_Racing_v0 belongs to Box2D family of popular RL problems.

  2. You can check for detailed information about these three RL algorithms here Report, where we analyzed them each one seperately. Also, it includes information about our experiments and setup and conclusions of them.

  3. In folder GIFs you can select whichever RL algorithm you prefer in order to observe all evaluation performance's videos (which we convert them to GIFs), that we managed to make with different number of timesteps.

Deep Q-Network (DQN)

DQN_Experiments

DQN_GIFs

Click here to open in Colab

alt te

Proximal Policy Optimization (PPO)

PPO_Experiments

PPO_GIFs

Click here to open in Colab

alt te

Actor-Critic (A2C)

A2C_Experiments

A2C_GIFs

Click here to open in Colab

alt te

Contributors

  1. Alexandra Apostolopoulou
  2. Andreas Giannoutsos
  3. Spyros Briakos