class Agent():
def __init__(self, brain_spec=p.BRAIN_SPEC):
#State and Physics init
self.agent_reset()
#Think_center init
self.brain_spec = brain_spec
self.head = Head(brain_spec)
def agent_reset(self):
#Physics
self.pos = np.array([p.INIT_X, p.INIT_Y])
self.velocity = np.array([0.0, 0.0])
self.acceleration = np.array([0, 0])
#State
self.score = 0
self.alive = True
self.memory = []
#Characteristics
self.fov = p.FOV
def moove(self, action):
self.acceleration = action
self.velocity += self.acceleration
self.pos += self.velocity
self.velocity *= p.FRICTION_COEF
def update(self, world_input, reward):
#add more maybe later
action = np.argmax(self.think(world_input))
self.moove(p.ACTION_DIC[action])
self.update_score(reward)
self.learn(world_input)
def update_score(self, reward):
self.score += reward
def think(self, world_input):
return (self.head.think(world_input))
def mutate(self):
self.head.mutate()
def die(self):
self.alive = False
def remember(self, world_input):
if self.memory:
memory[-1][1] = score - memory[-1][
1] #If the score is higher, this is good
memory.append([world_input, self.score])
