steffen/marl-factory-grid
1
0
Fork 0
mirror of https://github.com/illiumst/marl-factory-grid.git synced 2026-01-15 23:41:39 +01:00
Code Issues Packages Projects Releases Wiki Activity

merged RL agent with action map plotting and added it to end of agents train loop

Browse Source
This commit is contained in:
Chanumask
2024-05-12 11:48:05 +02:00
parent 5c5a09be3f
commit f8aa4e0c03
3 changed files with 52 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
marl_factory_grid/algorithms/marl/a2c_dirt.py
View File

@@ -17,6 +17,7 @@ from collections import deque
from marl_factory_grid.environment.actions import Noop from marl_factory_grid.environment.actions import Noop
from marl_factory_grid.modules import Clean, DoorUse from marl_factory_grid.modules import Clean, DoorUse
from marl_factory_grid.utils.plotting.plot_single_runs import plot_action_maps
class Names: class Names:
@@ -571,8 +572,7 @@ class A2C:
if self.cfg[nms.ENV]["save_and_log"]: if self.cfg[nms.ENV]["save_and_log"]:
self.create_info_maps(env, used_actions) self.create_info_maps(env, used_actions)
self.save_agent_models() self.save_agent_models()
plot_action_maps(env, [self])
@torch.inference_mode(True) @torch.inference_mode(True)
def eval_loop(self, n_episodes, render=False): def eval_loop(self, n_episodes, render=False):

2
marl_factory_grid/levels/quadrant.txt
View File

@@ -1,4 +1,4 @@
########### ##########D
#---####### #---#######
#-----##### #-----#####
#------#### #------####

87
marl_factory_grid/utils/plotting/plot_single_runs.py
View File

@@ -1,4 +1,5 @@
import json import json
import os
import pickle import pickle
from os import PathLike from os import PathLike
from pathlib import Path from pathlib import Path
@@ -121,50 +122,60 @@ def plot_routes(factory, agents):
def plot_action_maps(factory, agents): def plot_action_maps(factory, agents):
renderer = Renderer(factory.map.level_shape, cell_size=80, custom_assets_path={ base_dir = os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
'green_arrow': 'marl_factory_grid/utils/plotting/action_assets/green_arrow.png', assets_path = {
'yellow_arrow': 'marl_factory_grid/utils/plotting/action_assets/yellow_arrow.png', 'green_arrow': os.path.join(base_dir, 'utils', 'plotting', 'action_assets', 'green_arrow.png'),
'red_arrow': 'marl_factory_grid/utils/plotting/action_assets/red_arrow.png', 'yellow_arrow': os.path.join(base_dir, 'utils', 'plotting', 'action_assets', 'yellow_arrow.png'),
'grey_arrow': 'marl_factory_grid/utils/plotting/action_assets/grey_arrow.png', 'red_arrow': os.path.join(base_dir, 'utils', 'plotting', 'action_assets', 'red_arrow.png'),
'wall': 'marl_factory_grid/environment/assets/wall.png', 'grey_arrow': os.path.join(base_dir, 'utils', 'plotting', 'action_assets', 'grey_arrow.png'),
}) 'wall': os.path.join(base_dir, 'environment', 'assets', 'wall.png'),
}
renderer = Renderer(factory.map.level_shape, cell_size=80, custom_assets_path=assets_path)
directions = ['north', 'east', 'south', 'west'] directions = ['north', 'east', 'south', 'west']
wall_positions = swap_coordinates(factory.map.walls) wall_positions = swap_coordinates(factory.map.walls)
wall_entities = [RenderEntity(name='wall', probability=0, pos=np.array(pos)) for pos in wall_positions]
action_entities = list(wall_entities)
dummy_action_map = load_action_map("example_action_map.txt") for agent_index, agent in enumerate(agents):
for agent in agents: if hasattr(agent, 'action_probabilities'):
# if hasattr(agent, 'action_probability_map'): action_probabilities = unpack_action_probabilities(agent.action_probabilities)
# for y in range(len(agent.action_probability_map)): for action_map_index, probabilities_map in enumerate(action_probabilities[agent_index]):
for y in range(len(dummy_action_map)): wall_entities = [RenderEntity(name='wall', probability=0, pos=np.array(pos)) for pos in wall_positions]
# for x in range(len(agent.action_probability_map[y])): action_entities = list(wall_entities)
for x in range(len(dummy_action_map[y])): for position, probabilities in probabilities_map.items():
position = (x, y) if position not in wall_positions:
if position not in wall_positions: if np.any(probabilities) > 0: # Ensure it's not all zeros which would indicate a wall
# action_probabilities = agent.action_probability_map[y][x] sorted_indices = sorted(range(len(probabilities)), key=lambda i: -probabilities[i])
action_probabilities = dummy_action_map[y][x] colors = ['green_arrow', 'yellow_arrow', 'red_arrow', 'grey_arrow']
if sum(action_probabilities) > 0: # Ensure it's not all zeros which would indicate a wall
# Sort actions by probability and assign colors
sorted_indices = sorted(range(len(action_probabilities)),
key=lambda i: -action_probabilities[i])
colors = ['green_arrow', 'yellow_arrow', 'red_arrow', 'grey_arrow']
for rank, direction_index in enumerate(sorted_indices): for rank, direction_index in enumerate(sorted_indices):
action = directions[direction_index] action = directions[direction_index]
probability = action_probabilities[direction_index] probability = probabilities[direction_index]
arrow_color = colors[rank] arrow_color = colors[rank]
if probability > 0: if probability > 0:
action_entity = RenderEntity( action_entity = RenderEntity(
name=arrow_color, name=arrow_color,
pos=position, pos=position,
probability=probability, probability=probability,
rotation=direction_index * 90 rotation=direction_index * 90
) )
action_entities.append(action_entity) action_entities.append(action_entity)
renderer.render_multi_action_icons(action_entities) renderer.render_multi_action_icons(action_entities)
def unpack_action_probabilities(action_probabilities):
unpacked = {}
for agent_index, maps in action_probabilities.items():
unpacked[agent_index] = []
for map_index, probability_map in enumerate(maps):
single_map = {}
for y in range(len(probability_map)):
for x in range(len(probability_map[y])):
position = (x, y)
probabilities = probability_map[y][x]
single_map[position] = probabilities
unpacked[agent_index].append(single_map)
return unpacked
def load_action_map(file_path): def load_action_map(file_path):