Prison

environment gif

This environment is part of the butterfly environments. Please read that page first for general information.

Name Value
Actions Either
Agents 8
Parallel API True
Manual Control Yes
Action Shape (1,)
Action Values [0, 2]
Observation Shape (100, 300, 3) or (1,)
Observation Values (0, 255) or (-300, 300)
Import from pettingzoo.butterfly import prison_v1
Agents agents= ['prisoner_0', 'prisoner_1', ..., 'prisoner_7']
Average Total Reward 1.456

Agent Environment Cycle

environment aec diagram

Prison

In prison, 8 aliens locked in identical prison cells are controlled by the user. They cannot communicate with each other in any way, and can only pace in their cell. Every time they touch one end of the cell and then the other, they get a reward of 1. Due to the fully independent nature of these agents and the simplicity of the task this is an environment primarily intended for debugging purposes - its multiple individual purely single agent tasks. To make this debugging tool as compatible with as many methods as possible it can accept both discrete and continuous actions and the observation can be automatically turned into a number representing position of the alien from the left of its cell instead of the normal graphical output.

Manual Control

Select different aliens with ‘W’, ‘A’, ‘S’ or ‘D’. Move the selected alien left with ‘J’ and right with ‘K’.

Arguments

prison.env(vector_observation=False, continuous=False, synchronized_start=False,
identical_aliens=False, max_frames=900, num_floors=4, random_aliens=False)

vector_observation: If set to False an image of the prisoner’s cell is observed. If set to True, the distance to the left side wall is returned.

continuous: If False, each agent action is a discrete value indicating whether to move left or right one unit. If True, each agent action represents a real value that is added to the agent’s x position

synchronized_start: If set to true, all aliens will start in the same x position, relative to their cells. Otherwise, their position is random.

num_floors: A floor contains two aliens. Add more or fewer floor to increase or decrease the number of agents in the environment.

identical_aliens: If set to true, each alien will have the some randomly chosen sprite. This argument overrides the random_aliens argument.

random_aliens: If set to True, each alien’s sprite is randomly chosen from all possible sprites. If random_aliens and synchronized_aliens are both False, each alien’s sprite is chosen cyclically from all possible sprites.