188 lines
6.9 KiB
Python
188 lines
6.9 KiB
Python
import joblib
|
|
from sklearn.calibration import LabelEncoder
|
|
from agentActionType import AgentActionType
|
|
import time
|
|
from garbage import Garbage, GarbageType, RecognizedGarbage
|
|
from garbageCan import GarbageCan
|
|
from turnCar import turn_left_orientation, turn_right_orientation
|
|
from garbageTruck import GarbageTruck
|
|
from typing import Tuple, Dict
|
|
from gridCellType import GridCellType
|
|
from gameContext import GameContext
|
|
from agentOrientation import AgentOrientation
|
|
import pygame
|
|
from bfs import find_path_to_nearest_can
|
|
from agentState import AgentState
|
|
|
|
import tensorflow as tf
|
|
from keras.models import load_model
|
|
import keras.utils as image
|
|
from keras.optimizers import Adam
|
|
import numpy as np
|
|
|
|
|
|
def collect_garbage(game_context: GameContext) -> None:
|
|
while True:
|
|
start_agent_state = AgentState(game_context.dust_car.position, game_context.dust_car.orientation)
|
|
path = find_path_to_nearest_can(start_agent_state, game_context.grid, game_context.city)
|
|
if path == None or len(path) == 0:
|
|
break
|
|
move_dust_car(path, game_context)
|
|
next_position = calculate_next_position(game_context.dust_car)
|
|
game_context.grid[next_position] = GridCellType.VISITED_GARBAGE_CAN
|
|
can = game_context.city.cans_dict[next_position]
|
|
can.is_visited = True
|
|
_recognize_garbage(game_context.dust_car, can)
|
|
pass
|
|
|
|
def _recognize_garbage(dust_car: GarbageTruck, can: GarbageCan) -> None:
|
|
tree_model = joblib.load('machine_learning/model.pkl')
|
|
optimizer = Adam(learning_rate=0.001)
|
|
neural_model = load_model('machine_learning/neuralModel.h5', compile=False)
|
|
neural_model.compile(optimizer=optimizer)
|
|
for garbage in can.garbage:
|
|
predicted_class = predict_class(garbage, tree_model, neural_model)
|
|
garbage_type: GarbageType = None
|
|
if predicted_class == 'PAPER':
|
|
garbage_type = GarbageType.PAPER
|
|
elif predicted_class == 'PLASTIC_AND_METAL':
|
|
garbage_type = GarbageType.PLASTIC_AND_METAL
|
|
elif garbage_type == 'GLASS':
|
|
garbage_type = GarbageType.GLASS
|
|
elif predicted_class == 'BIO' :
|
|
garbage_type = GarbageType.BIO
|
|
elif predicted_class == 'MIXED':
|
|
garbage_type = GarbageType.MIXED
|
|
print(predicted_class)
|
|
recognized_garbage = RecognizedGarbage(garbage, garbage_type)
|
|
dust_car.sort_garbage(recognized_garbage)
|
|
|
|
def predict_class(garbage: Garbage, tree_model, neural_model) -> str:
|
|
if garbage.img is None:
|
|
return predict_class_from_tree(garbage, tree_model)
|
|
return predict_class_from_neural_model(garbage, neural_model)
|
|
|
|
def predict_class_from_tree(garbage: Garbage, tree_model) -> str:
|
|
attributes = [garbage.shape, garbage.flexibility, garbage.does_smell, garbage.weight, garbage.size, garbage.color, garbage.softness, garbage.does_din]
|
|
encoded = attributes_to_floats(attributes)
|
|
return tree_model.predict([encoded])[0]
|
|
|
|
def predict_class_from_neural_model(garbage: Garbage, neural_model) -> str:
|
|
img = image.load_img(garbage.img, target_size=(150, 150))
|
|
img_array = image.img_to_array(img)
|
|
img_array = np.expand_dims(img_array, axis=0)
|
|
img_array /= 255.
|
|
|
|
predictions = neural_model.predict(img_array)
|
|
prediction = np.argmax(predictions[0])
|
|
if prediction == 0:
|
|
return "BIO"
|
|
if prediction == 1:
|
|
return "GLASS"
|
|
if prediction == 2:
|
|
return "MIXED"
|
|
if prediction == 3:
|
|
return "PAPER"
|
|
if prediction == 4:
|
|
return "PLASTIC_AND_METAL"
|
|
|
|
def attributes_to_floats(attributes: list[str]) -> list[float]:
|
|
output: list[float] = []
|
|
if attributes[0] == 'Longitiudonal':
|
|
output.append(0)
|
|
elif attributes[0] == 'Round':
|
|
output.append(1)
|
|
elif attributes[0] == 'Flat':
|
|
output.append(2)
|
|
elif attributes[0] == 'Irregular':
|
|
output.append(3)
|
|
|
|
if attributes[1] == 'Low':
|
|
output.append(0)
|
|
elif attributes[1] == 'Medium':
|
|
output.append(1)
|
|
elif attributes[1] == 'High':
|
|
output.append(2)
|
|
|
|
|
|
if attributes[2] == "Yes":
|
|
output.append(0)
|
|
else:
|
|
output.append(1)
|
|
|
|
if attributes[3] == 'Low':
|
|
output.append(0)
|
|
elif attributes[3] == 'Medium':
|
|
output.append(1)
|
|
elif attributes[3] == 'High':
|
|
output.append(2)
|
|
|
|
if attributes[4] == 'Low':
|
|
output.append(0)
|
|
elif attributes[4] == 'Medium':
|
|
output.append(1)
|
|
elif attributes[4] == 'High':
|
|
output.append(2)
|
|
|
|
if attributes[5] == 'Transparent':
|
|
output.append(0)
|
|
elif attributes[5] == 'Light':
|
|
output.append(1)
|
|
elif attributes[5] == 'Dark':
|
|
output.append(2)
|
|
elif attributes[5] == "Colorful":
|
|
output.append(3)
|
|
|
|
if attributes[6] == 'Low':
|
|
output.append(0)
|
|
elif attributes[6] == 'Medium':
|
|
output.append(1)
|
|
elif attributes[6] == 'High':
|
|
output.append(2)
|
|
|
|
if attributes[7] == "Yes":
|
|
output.append(0)
|
|
else:
|
|
output.append(1)
|
|
return output
|
|
|
|
|
|
def move_dust_car(actions: list[AgentActionType], game_context: GameContext) -> None:
|
|
for action in actions:
|
|
street_position = game_context.dust_car.position
|
|
has_to_render_street = False
|
|
if action == AgentActionType.TURN_LEFT:
|
|
game_context.dust_car.orientation = turn_left_orientation(game_context.dust_car.orientation)
|
|
elif action == AgentActionType.TURN_RIGHT:
|
|
game_context.dust_car.orientation = turn_right_orientation(game_context.dust_car.orientation)
|
|
elif action == AgentActionType.MOVE_FORWARD:
|
|
game_context.dust_car.position = calculate_next_position(game_context.dust_car)
|
|
has_to_render_street = True
|
|
game_context.dust_car.render(game_context)
|
|
if has_to_render_street:
|
|
if game_context.grid[street_position] == GridCellType.STREET_HORIZONTAL:
|
|
game_context.render_in_cell(street_position, "imgs/street_horizontal.png")
|
|
elif game_context.grid[street_position] == GridCellType.STREET_VERTICAL:
|
|
game_context.render_in_cell(street_position, "imgs/street_vertical.png")
|
|
elif game_context.grid[street_position] == GridCellType.SPEED_BUMP:
|
|
game_context.render_in_cell(street_position, "imgs/speed_bump.png")
|
|
pygame.display.update()
|
|
time.sleep(0.15)
|
|
|
|
|
|
def calculate_next_position(car: GarbageTruck) -> Tuple[int, int]:
|
|
if car.orientation == AgentOrientation.UP:
|
|
if car.position[1] - 1 < 1:
|
|
return None
|
|
return (car.position[0], car.position[1] - 1)
|
|
if car.orientation == AgentOrientation.DOWN:
|
|
if car.position[1] + 1 > 27:
|
|
return None
|
|
return (car.position[0], car.position[1] + 1)
|
|
if car.orientation == AgentOrientation.LEFT:
|
|
if car.position[0] - 1 < 1:
|
|
return None
|
|
return (car.position[0] - 1, car.position[1])
|
|
if car.position[0] + 1 > 27:
|
|
return None
|
|
return (car.position[0] + 1, car.position[1]) |