In [153]:

    

from treehopper.api import *
from treehopper.libraries.sensors.inertial import Bno055, Mpu9250
import numpy as np
from time import sleep
%matplotlib inline
import matplotlib.pyplot as plt
from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
from matplotlib.figure import Figure
from keras.models import Sequential
from keras.layers import Dense, LSTM, SimpleRNN


    

In [2]:

    

board = find_boards()[0]
board.connect()


    

In [3]:

    

imu = Bno055.probe(board.i2c)[0]


    

In [7]:

    

print(imu.accelerometer)


    

    




[-0.3125, 58.8125, -2.1875]

In [8]:

    

imu2 = Mpu9250.probe(board.i2c)[0]


    

In [10]:

    

imu2.accelerometer


    

    
Out[10]:





[1.03271484375, 0.025634765625, 0.0660400390625]

In [11]:

    

imu.auto_update_when_property_read = False
imu2.auto_update_when_property_read = False


    

In [141]:

    

imu_data = np.zeros((1000, 22))  # accel + gyro + mag xyz (9) + eular (3) + quaternion (4) + gravity (3) + linear_accel (3) = 22
imu2_data = np.zeros((1000, 9))  # accel + gyro + mag xyz (9)


    

In [142]:

    

for i in range(1000):
    imu.update()
    imu2.update()
    
    imu_data[i, 0] = imu.accelerometer[0]
    imu_data[i, 1] = imu.accelerometer[1]
    imu_data[i, 2] = imu.accelerometer[2]
    
    imu_data[i, 3] = imu.gyroscope[0]
    imu_data[i, 4] = imu.gyroscope[1]
    imu_data[i, 5] = imu.gyroscope[2]
       
    imu_data[i, 6] = imu.magnetometer[0]
    imu_data[i, 7] = imu.magnetometer[1]
    imu_data[i, 8] = imu.magnetometer[2]
    
    imu_data[i, 9] = imu.eular_angles[0]
    imu_data[i, 10] = imu.eular_angles[1]
    imu_data[i, 11] = imu.eular_angles[2]
    
    imu_data[i, 12] = imu.quaternion[0]
    imu_data[i, 13] = imu.quaternion[1]
    imu_data[i, 14] = imu.quaternion[2]
    imu_data[i, 15] = imu.quaternion[3]
    
    imu_data[i, 16] = imu.gravity[0]
    imu_data[i, 17] = imu.gravity[1]
    imu_data[i, 18] = imu.gravity[2]
    
    imu_data[i, 19] = imu.linear_acceleration[0]
    imu_data[i, 20] = imu.linear_acceleration[1]
    imu_data[i, 21] = imu.linear_acceleration[2]
    
    imu2_data[i, 0] = imu2.accelerometer[0]
    imu2_data[i, 1] = imu2.accelerometer[1]
    imu2_data[i, 2] = imu2.accelerometer[2]
    
    imu2_data[i, 3] = imu2.gyroscope[0]
    imu2_data[i, 4] = imu2.gyroscope[1]
    imu2_data[i, 5] = imu2.gyroscope[2]
    
    imu2_data[i, 6] = imu2.gyroscope[0]
    imu2_data[i, 7] = imu2.gyroscope[1]
    imu2_data[i, 8] = imu2.gyroscope[2]
    sleep(0.01)


    

In [143]:

    

plot(imu_data[:, 1])


    

    
Out[143]:





[<matplotlib.lines.Line2D at 0x19d65ddd438>]






    

In [146]:

    

#scale data
max_accel = np.max(np.abs(imu_data[:, 0:3]))
max_gyro = np.max(np.abs(imu_data[:, 3:6]))
max_mag = np.max(np.abs(imu_data[:, 6:9]))
max_eular = np.max(np.abs(imu_data[:, 9:12]))
max_quaternion = np.max(np.abs(imu_data[:, 12:16]))
max_gravity = np.max(np.abs(imu_data[:, 16:19]))
max_linear = np.max(np.abs(imu_data[:, 19:]))

max_accel2 = np.max(np.abs(imu2_data[:, 0:3]))
max_gyro2 = np.max(np.abs(imu2_data[:, 3:6]))
max_mag2 = np.max(np.abs(imu2_data[:, 6:9]))

print(max_accel)
print(max_gyro)
print(max_mag)

print(max_eular)
print(max_quaternion)
print(max_gravity)
print(max_linear)

print(max_accel2)
print(max_gyro2)
print(max_mag2)


    

    




93.625
1306.625
64.0625
57.59
0.998901367188
9.8
16.66
1.79260253906
32768.0
32768.0

In [150]:

    

scaled_imu_data = np.copy(imu_data)
scaled_imu2_data = np.copy(imu2_data)

scaled_imu_data[:, 0:3] = scaled_imu_data[:, 0:3] / max_accel
scaled_imu_data[:, 3:6] = scaled_imu_data[:, 3:6] / max_gyro
scaled_imu_data[:, 6:9] = scaled_imu_data[:, 6:9] / max_mag

scaled_imu_data[:, 9:12] = scaled_imu_data[:, 9:12] / max_eular
scaled_imu_data[:, 12:16] = scaled_imu_data[:, 12:16] / max_quaternion
scaled_imu_data[:, 16:19] = scaled_imu_data[:, 16:19] / max_gravity
scaled_imu_data[:, 19:] = scaled_imu_data[:, 19:] / max_linear

scaled_imu2_data[:, 0:3] = scaled_imu2_data[:, 0:3] / max_accel2
scaled_imu2_data[:, 3:6] = scaled_imu2_data[:, 3:6] / max_gyro2
scaled_imu2_data[:, 6:9] = scaled_imu2_data[:, 6:9] / max_mag2


    

In [181]:

    

in_out_neurons = 9  
hidden_neurons = 50

model = Sequential()
# model.add(Dense(9, input_dim=9, activation='relu'))
model.add(SimpleRNN(input_shape=(9, ???), units=32, return_sequences=False))  
# model.add(Dense(hidden_neurons, in_out_neurons))  
model.add(Dense(512, activation='relu'))
# model.add(Dense(512, activation='relu'))
# model.add(Dense(512, activation='relu'))
model.add(Dense(22))


    

In [182]:

    

model.compile(loss='mean_squared_error', optimizer='adam', metrics=['accuracy'])


    

In [183]:

    

model.fit(scaled_imu2_data, scaled_imu_data, epochs=100, batch_size=100)


    

    




---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-183-653a9aa287fa> in <module>()
----> 1 model.fit(scaled_imu2_data, scaled_imu_data, epochs=100, batch_size=100)

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\keras\models.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, **kwargs)
    963                               initial_epoch=initial_epoch,
    964                               steps_per_epoch=steps_per_epoch,
--> 965                               validation_steps=validation_steps)
    966 
    967     def evaluate(self, x=None, y=None,

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\keras\engine\training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, **kwargs)
   1591             class_weight=class_weight,
   1592             check_batch_axis=False,
-> 1593             batch_size=batch_size)
   1594         # Prepare validation data.
   1595         do_validation = False

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\keras\engine\training.py in _standardize_user_data(self, x, y, sample_weight, class_weight, check_batch_axis, batch_size)
   1424                                     self._feed_input_shapes,
   1425                                     check_batch_axis=False,
-> 1426                                     exception_prefix='input')
   1427         y = _standardize_input_data(y, self._feed_output_names,
   1428                                     output_shapes,

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\keras\engine\training.py in _standardize_input_data(data, names, shapes, check_batch_axis, exception_prefix)
    108                         ': expected ' + names[i] + ' to have ' +
    109                         str(len(shape)) + ' dimensions, but got array '
--> 110                         'with shape ' + str(data_shape))
    111                 if not check_batch_axis:
    112                     data_shape = data_shape[1:]

ValueError: Error when checking input: expected simple_rnn_8_input to have 3 dimensions, but got array with shape (1000, 9)

In [173]:

    

predicted = model.predict(scaled_imu2_data)


    

In [174]:

    

print(predicted.shape)


    

    




(1000, 22)

In [175]:

    

plot(scaled_imu_data[:, 0:3])


    

    
Out[175]:





[<matplotlib.lines.Line2D at 0x19d614ec048>,
 <matplotlib.lines.Line2D at 0x19d614ec208>,
 <matplotlib.lines.Line2D at 0x19d614ec400>]






    

In [178]:

    

plot(predicted[:, 0:3])


    

    
Out[178]:





[<matplotlib.lines.Line2D at 0x19d67ac17b8>,
 <matplotlib.lines.Line2D at 0x19d67ac1978>,
 <matplotlib.lines.Line2D at 0x19d67ac1b70>]






    

In [122]:

    

%matplotlib inline
fig = Figure()
FigureCanvas(fig)
original = fig.add_subplot(111)
original.plot(imu_data[:, 0:2])


    

    
Out[122]:





[<matplotlib.lines.Line2D at 0x19d658ff048>,
 <matplotlib.lines.Line2D at 0x19d658ff208>]

In [ ]: