Combat: Tank¶

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

Import	`from pettingzoo.atari import combat_tank_v3`
Actions	Discrete
Parallel API	Yes
Manual Control	No
Agents	`agents= ['first_0', 'second_0']`
Agents	2
Action Shape	(1,)
Action Values	[0,5]
Observation Shape	(210, 160, 3)
Observation Values	(0,255)

Combat’s classic tank mode is an adversarial game where prediction, and positioning are key.

The players move around the map. When your opponent is hit by your bullet, you score a point. Note that your opponent gets blasted through obstacles when it is hit, potentially putting it in a good position to hit you back.

Whenever you score a point, you are rewarded +1 and your opponent is penalized -1.

Official Combat manual

Environment parameters¶

Some environment parameters are common to all Atari environments and are described in the base Atari documentation.

Parameters specific to combat-tank are

combat_tank_v2.env(has_maze=True, is_invisible=False, billiard_hit=True)

has_maze: Set to true to have the map be a maze instead of an open field

is_invisible: If true, tanks are invisible unless they are firing or are running into a wall.

billiard_hit: If true, bullets bounce off walls, in fact, like billiards, they only count if they hit the opponent’s tank after bouncing off a wall.

Action Space¶

In any given turn, an agent can choose from one of 18 actions.

Action	Behavior
0	No operation
1	Fire
2	Move up
3	Move right
4	Move left
5	Move down
6	Move upright
7	Move upleft
8	Move downright
9	Move downleft
10	Fire up
11	Fire right
12	Fire left
13	Fire down
14	Fire upright
15	Fire upleft
16	Fire downright
17	Fire downleft

Version History¶

v2: Minimal Action Space (1.18.0)
v1: Breaking changes to entire API (1.4.0)
v0: Initial versions release (1.0.0)

Usage¶

AEC¶

from pettingzoo.atari import combat_tank_v2

env = combat_tank_v2.env(render_mode="human")
env.reset(seed=42)

for agent in env.agent_iter():
    observation, reward, termination, truncation, info = env.last()

    if termination or truncation:
        action = None
    else:
        # this is where you would insert your policy
        action = env.action_space(agent).sample()

    env.step(action)
env.close()

Parallel¶

from pettingzoo.atari import combat_tank_v2

env = combat_tank_v2.parallel_env(render_mode="human")
observations, infos = env.reset()

while env.agents:
    # this is where you would insert your policy
    actions = {agent: env.action_space(agent).sample() for agent in env.agents}

    observations, rewards, terminations, truncations, infos = env.step(actions)
env.close()

API¶

class pettingzoo.atari.combat_tank.combat_tank.raw_env(has_maze=True, is_invisible=False, billiard_hit=True, **kwargs)[source]¶