mirror of
https://github.com/illiumst/marl-factory-grid.git
synced 2025-05-23 15:26:43 +02:00
69 lines
2.6 KiB
Python
69 lines
2.6 KiB
Python
import numpy as np
|
|
from pathlib import Path
|
|
from environments import helpers as h
|
|
|
|
|
|
class BaseFactory(object):
|
|
LEVELS_DIR = 'levels'
|
|
|
|
def __init__(self, level='simple', n_agents=1, max_steps=1e3):
|
|
self.n_agents = n_agents
|
|
self.max_steps = max_steps
|
|
self.level = h.one_hot_level(
|
|
h.parse_level(Path(__file__).parent / self.LEVELS_DIR / f'{level}.txt')
|
|
)#[np.newaxis, ...]
|
|
self.slice_strings = {0: 'level', **{i: f'agent#{i}' for i in range(1, self.n_agents+1)}}
|
|
self.reset()
|
|
|
|
def reset(self):
|
|
self.done = False
|
|
self.steps = 0
|
|
self.agents = np.zeros((self.n_agents, *self.level.shape), dtype=np.int8)
|
|
free_cells = np.argwhere(self.level == 0)
|
|
np.random.shuffle(free_cells)
|
|
for i in range(self.n_agents):
|
|
r, c = free_cells[i]
|
|
self.agents[i, r, c] = 1
|
|
self.state = np.concatenate((self.level[np.newaxis, ...], self.agents), 0)
|
|
return self.state, 0, self.done, {}
|
|
|
|
def step(self, actions):
|
|
assert type(actions) in [int, list]
|
|
if type(actions) == int:
|
|
actions = [actions]
|
|
self.steps += 1
|
|
r = 0
|
|
# level, agent 1,..., agent n,
|
|
collision_vecs = np.zeros((self.n_agents, self.state.shape[0])) # n_agents x n_slices
|
|
for i, a in enumerate(actions):
|
|
old_pos, new_pos, valid = h.check_agent_move(state=self.state, dim=i+1, action=a)
|
|
if valid:
|
|
self.make_move(i, old_pos, new_pos)
|
|
else: # trying to leave the level
|
|
collision_vecs[i, 0] = 1
|
|
for i in range(self.n_agents): # might as well save the positions (redundant)
|
|
agent_slice = self.state[i+1]
|
|
x, y = np.argwhere(agent_slice == 1)[0]
|
|
collisions_vec = self.state[:, x, y].copy() # otherwise you overwrite the grid/state
|
|
collisions_vec[i+1] = 0 # no self-collisions
|
|
collision_vecs[i] += collisions_vec
|
|
reward, info = self.step_core(np.array(collision_vecs), actions, r)
|
|
r += reward
|
|
if self.steps >= self.max_steps:
|
|
self.done = True
|
|
return self.state, r, self.done, info
|
|
|
|
def make_move(self, agent_i, old_pos, new_pos):
|
|
(x, y), (x_new, y_new) = old_pos, new_pos
|
|
self.state[agent_i+1, x, y] = 0
|
|
self.state[agent_i+1, x_new, y_new] = 1
|
|
|
|
def free_cells(self):
|
|
free_cells = self.state.sum(0)
|
|
free_cells = np.argwhere(free_cells == 0)
|
|
np.random.shuffle(free_cells)
|
|
return free_cells
|
|
|
|
def step_core(self, collisions_vec, actions, r):
|
|
return 0, {}
|