Learn Posture

use machine learning to recognize robot's posture (following the example in scikit-learn-intro.ipynb )

1. Data collection

We have colleceted data before, you need to add new data if you want to add new posture.

the dateset are in robot_pose_data folder
each file contains the data belongs to this posture, e.g. the data in Back file are collected when robot was in "Back" posture
the data file can be load by pickle, e.g. pickle.load(open('Back')), the data is a list of feature data
the features (e.g. each row of the data) are ['LHipYawPitch', 'LHipRoll', 'LHipPitch', 'LKneePitch', 'RHipYawPitch', 'RHipRoll', 'RHipPitch', 'RKneePitch', 'AngleX', 'AngleY'], where 'AngleX' and 'AngleY' are body angle (e.g. Perception.imu) and others are joint angles.

2. Data preprocessing



In [51]:

    
%pylab inline
import pickle
from os import listdir, path
import numpy as np
from sklearn import svm, metrics

ROBOT_POSE_DATA_DIR = 'robot_pose_data'









    



Populating the interactive namespace from numpy and matplotlib






    



WARNING: pylab import has clobbered these variables: ['clf', 'permutation']
`%matplotlib` prevents importing * from pylab and numpy



In [52]:

    
classes = listdir(ROBOT_POSE_DATA_DIR)
print classes









    



['Frog', 'HeadBack', 'Left', 'Knee', 'Crouch', 'Back', 'Belly', 'Right', 'Sit', 'Stand', 'StandInit']



In [84]:

    
def load_pose_data(i):
    '''load pose data from file'''
    data = []
    target = []
    # YOUR CODE HERE
    
    filename = path.join(ROBOT_POSE_DATA_DIR, classes[i])
    data = pickle.load(open(filename))
    target = [i] * len(data)
    return data, target



In [61]:

    
# load all the data
all_data = []
all_target = []
# YOUR CODE HERE

print 'total number of data', len(all_data)









    



total number of data 200



In [86]:

    
# shuffule data
permutation = np.random.permutation(len(all_data))
n_training_data = int(len(all_data) * 0.7)
training_data = permutation[:n_training_data]

3. Learn on training data

In scikit-learn, an estimator for classification is a Python object that implements the methods fit(X, y) and predict(T). An example of an estimator is the class sklearn.svm.SVC that implements support vector classification.



In [56]:

    
clf = svm.SVC(gamma=0.001, C=100.)

learning



In [57]:

    
# YOUR CODE HERE









    Out[57]:





SVC(C=100.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,
  gamma=0.001, kernel='rbf', max_iter=-1, probability=False,
  random_state=None, shrinking=True, tol=0.001, verbose=False)

predicting



In [58]:

    
clf.predict(all_data[-1]), all_target[-1]









    Out[58]:





(array([10]), 10)



In [81]:

    
def evaluate(expected, predicted):
    print("Classification report:\n%s\n" % metrics.classification_report(expected, predicted))

    print("Confusion matrix:\n%s" % metrics.confusion_matrix(expected, predicted))



In [82]:

    
expected = []
predicted = []
# YOUR CODE HERE

evaluate(expected, predicted)









    



Classification report:
             precision    recall  f1-score   support

          0       1.00      1.00      1.00         5
          1       1.00      1.00      1.00         7
          2       1.00      1.00      1.00        15
          3       1.00      1.00      1.00         5
          4       1.00      1.00      1.00        20
          5       1.00      1.00      1.00         6
          6       1.00      1.00      1.00         8
          7       1.00      1.00      1.00         7
          8       1.00      1.00      1.00        18
          9       1.00      1.00      1.00         8
         10       1.00      1.00      1.00        41

avg / total       1.00      1.00      1.00       140


Confusion matrix:
[[ 5  0  0  0  0  0  0  0  0  0  0]
 [ 0  7  0  0  0  0  0  0  0  0  0]
 [ 0  0 15  0  0  0  0  0  0  0  0]
 [ 0  0  0  5  0  0  0  0  0  0  0]
 [ 0  0  0  0 20  0  0  0  0  0  0]
 [ 0  0  0  0  0  6  0  0  0  0  0]
 [ 0  0  0  0  0  0  8  0  0  0  0]
 [ 0  0  0  0  0  0  0  7  0  0  0]
 [ 0  0  0  0  0  0  0  0 18  0  0]
 [ 0  0  0  0  0  0  0  0  0  8  0]
 [ 0  0  0  0  0  0  0  0  0  0 41]]

4. Evaluate on the test data



In [83]:

    
expected = []
predicted = []
# YOUR CODE HERE

evaluate(expected, predicted)









    



Classification report:
             precision    recall  f1-score   support

          0       1.00      1.00      1.00         5
          1       1.00      1.00      1.00         3
          2       1.00      1.00      1.00         5
          3       1.00      1.00      1.00         5
          4       0.91      1.00      0.95        10
          5       1.00      1.00      1.00         4
          6       1.00      1.00      1.00         2
          7       1.00      0.75      0.86         4
          8       1.00      1.00      1.00         8
          9       1.00      1.00      1.00         3
         10       1.00      1.00      1.00        11

avg / total       0.98      0.98      0.98        60


Confusion matrix:
[[ 5  0  0  0  0  0  0  0  0  0  0]
 [ 0  3  0  0  0  0  0  0  0  0  0]
 [ 0  0  5  0  0  0  0  0  0  0  0]
 [ 0  0  0  5  0  0  0  0  0  0  0]
 [ 0  0  0  0 10  0  0  0  0  0  0]
 [ 0  0  0  0  0  4  0  0  0  0  0]
 [ 0  0  0  0  0  0  2  0  0  0  0]
 [ 0  0  0  0  1  0  0  3  0  0  0]
 [ 0  0  0  0  0  0  0  0  8  0  0]
 [ 0  0  0  0  0  0  0  0  0  3  0]
 [ 0  0  0  0  0  0  0  0  0  0 11]]

5. Deploy to the real system

We can simple use pickle module to serialize the trained classifier.



In [68]:

    
import pickle
ROBOT_POSE_CLF = 'robot_pose.pkl'
pickle.dump(clf, open(ROBOT_POSE_CLF, 'w'))

Then, in the application we can load the trained classifier again.



In [71]:

    
clf2 = pickle.load(open(ROBOT_POSE_CLF))
clf2.predict(all_data[-1]), all_target[-1]









    Out[71]:





(array([10]), 10)



In [ ]: