requirements completion

Author: giodegas

SLAMRobot.py

@@ -4,7 +4,7 @@
 import numpy as np
 from collections import deque
 import random
-import math
+
 
 class SLAMAgent:
     def __init__(self, state_size, action_size):
@@ -20,6 +20,7 @@ def __init__(self, state_size, action_size):
         self.learning_rate_decay = 0.01
         self.randomActions = [0,1,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2]
         self.model = self._build_model()
+
     def _build_model(self):
         # Neural Net for Deep-Q learning Model
         model = Sequential()
@@ -33,11 +34,13 @@ def _build_model(self):
         model.compile(loss='mse',
                       optimizer=Adam(lr=self.learning_rate,decay=self.learning_rate_decay))
         return model
+
     def remember(self, state, action, reward, next_state, done):
             if reward == 0:
                 self.memory.append((state, action, reward, next_state, done))
             else:
                 self.tempMemory.append((state, action, reward, next_state, done))
+
     def act(self, state):
         currentMinDistance = 1
         for i in range(len(state[0])):
@@ -51,6 +54,7 @@ def act(self, state):
             return self.randomActions[random.randrange(len(self.randomActions)-1)], True
         act_values = self.model.predict(state)
         return np.argmax(act_values[0]), False  # returns action
+
     def replay(self, batch_size):
         minibatch = []
 
@@ -77,9 +81,11 @@ def replay(self, batch_size):
         if self.epsilon > self.epsilon_min:
             self.epsilon *= self.epsilon_decay
         print("epsilon attuale: " + str(self.epsilon))
+
     def save(self, fn):
         self.model.save(fn) #fn è il file name del file dei pesi dei neuroni alla fine del training
         print("epsilon attuale: " + str(self.epsilon))
+
     def load(self, name, lastRandomValue):
         self.model.load_weights(name)
         self.epsilon = lastRandomValue