Extend deepbots to support stable-baselines and implement gym-style default enviroments about deepbots

[sanketsans]

Hi @ManosMagnus I am interested to work on this problem for this gsoc.
Contribution

• Develop RL based environments which are based on openAI gym scenarios (​ https://gym.openai.com/envs/#classic_control​ ). I also want to develop other testbeds which are different from the gym based environments.

• The environments should be enhanced to support following categories :

  ○ continuous state, discrete action space.
  ○ Continuous state - action space.
  ○ Discrete space - action

• Implement and integrate DQN, CEM and REINFORCE algorithms in the codebase. Currently, deepbots support DDPG and PPO which are only based on continuous state-action pairs. Other algorithms supporting an enhanced environment catalog will help beginners and researchers to understand much better about the RL algorithms and how they work on different sets of state-action pairs.

[tsampazk]

Assigned to @NickKok

[SidharajYadav]

how i can contribute
please guide