[easy to understand and intensive learning] 1 Introduction

1.5 Construction of reinforcement learning simulation environment

1.5.1 gym Installation and simple demo Example

pip3 install gym

The simplest example

import gym	%  Import Gym modular 
env = gym.make('CartPole-v0')	%  Create a model of a trolley inverted pendulum 
env.reset()		%  Initialization environment 
env.render()	%  Refresh the current environment and display 

Through this 6 Step , You can get a trolley inverted pendulum system .

1.5.2 In depth analysis of gym Environment construction

1. reset() Function details
   Initialization function

2. render() Function details
   Play the role of image engine .

Actually , For reinforcement learning algorithms , There can be no render() function , however , In order to visually display the objects in the current environment , Image engine is still necessary .

3. step() Function details
   Play the role of physics engine .

In this function , General use of agent Kinematic model and The kinetic model Calculate the status of the next step and immediate return , And judge whether it reaches the termination state .