CartPole: CartPole Object

In XiaoqiLu/PhD-Thesis: Regularized Q-Learning

Description

The function CartPole() creates a CartPole object. The CartPole class is inherited from the Env class.

Usage

CartPole(
  internal_state = vector("numeric", 4),
  rng_state = .Random.seed,
  setting = list()
)

## S3 method for class 'CartPole'
Step(x, action)

## S3 method for class 'CartPole'
Reset(x)

Arguments

internal_state	vector of size 4, representing cart position, cart velocity, pole angle, and pole angular velocity
rng_state	the RNG state for non-deterministic evolution
setting	a list of constants that controls the setting of the CartPole object, see details.
x	an environment-like object
action	an action from agent

Details

The internal state of a CartPole object includes four elements (following OpenAI Gym convention):

Index	Observation
1	Cart Position
2	Cart Velocity
3	Pole Angle
4	Pole Angular Velocity

Note that the θ (index 3 and 4) has the opposite sign to OpenAI Gym. In our definition, anti-clockwise is positive direction.

Value

a CartPole object (class = c("CartPole", "Env"))

Methods (by generic)

• Step: evolve given left (-1) or right (+1) action

• Reset: reset by uniformly sample from U[-0.05, +0.05]

Examples

cp <- CartPole()

XiaoqiLu/PhD-Thesis documentation built on March 1, 2021, 10:49 a.m.