teste_com_rede_neural.py

# executa os movimentos de acordo com a decisao da rede neural
import numpy as np
import cPickle as pickle

from math import sqrt
from pybrain.datasets.supervised import SupervisedDataSet as SDS


import vrep
import time

from manage_joints import *


model_file = 'model_4_3_hidden_50_epochs.pkl'
net = pickle.load(open(model_file,'rb'))



vrep.simxFinish(-1) # just in case, close all opened connections
clientID = vrep.simxStart('127.0.0.1',19997,True,True,5000,5) # Connecta com o VREP. Por padrao ele ja abre essa porta.
if clientID == -1:
    exit (10)


Head_Yaw=[];Head_Pitch=[];
L_Hip_Yaw_Pitch=[];L_Hip_Roll=[];L_Hip_Pitch=[];L_Knee_Pitch=[];L_Ankle_Pitch=[];L_Ankle_Roll=[];
R_Hip_Yaw_Pitch=[];R_Hip_Roll=[];R_Hip_Pitch=[];R_Knee_Pitch=[];R_Ankle_Pitch=[];R_Ankle_Roll=[];
L_Shoulder_Pitch=[];L_Shoulder_Roll=[];L_Elbow_Yaw=[];L_Elbow_Roll=[];L_Wrist_Yaw=[]
R_Shoulder_Pitch=[];R_Shoulder_Roll=[];R_Elbow_Yaw=[];R_Elbow_Roll=[];R_Wrist_Yaw=[]
R_H=[];L_H=[];R_Hand=[];L_Hand=[];
Body = [Head_Yaw,Head_Pitch,L_Hip_Yaw_Pitch,L_Hip_Roll,L_Hip_Pitch,L_Knee_Pitch,L_Ankle_Pitch,L_Ankle_Roll,R_Hip_Yaw_Pitch,R_Hip_Roll,R_Hip_Pitch,R_Knee_Pitch,R_Ankle_Pitch,R_Ankle_Roll,L_Shoulder_Pitch,L_Shoulder_Roll,L_Elbow_Yaw,L_Elbow_Roll,L_Wrist_Yaw,R_Shoulder_Pitch,R_Shoulder_Roll,R_Elbow_Yaw,R_Elbow_Roll,R_Wrist_Yaw,L_H,L_Hand,R_H,R_Hand]

get_all_handles(3, clientID,Body)

time.sleep(4)

joints_movements = []

for i in range(0, 100):
    sensors = get_joint_values(clientID,Body)
    for s in range(10):
        a = net.activate(sensors)
        joints_movements.append(a)
        sensors = [sum(x) for x in zip(sensors,a)]

    i = i+10

    for movement in joints_movements:
        JointControl(clientID, 0, Body, movement)
        time.sleep(0.001)

    joints_movements = []