def test_benchmark_time_series():
dataset = bdp.DataSet()
trial = bdp.Run('700', dataset)
timeSeries = cbi.benchmark_time_series(trial)
testing.assert_allclose(trial.taskSignals['RollAngle'], timeSeries['p'])
testing.assert_allclose(trial.taskSignals['RollRate'], timeSeries['pD'])
testing.assert_allclose(trial.taskSignals['RollRate'].time_derivative(),
timeSeries['pDD'])
testing.assert_allclose(trial.taskSignals['SteerAngle'], timeSeries['d'])
testing.assert_allclose(trial.taskSignals['SteerRate'], timeSeries['dD'])
testing.assert_allclose(trial.taskSignals['SteerRate'].time_derivative(),
timeSeries['dDD'])
testing.assert_allclose(trial.taskSignals['ForwardSpeed'], timeSeries['v'])
testing.assert_allclose(trial.bicycleRiderParameters['g'], timeSeries['g'])
testing.assert_allclose(trial.taskSignals['PullForce'], timeSeries['F'])
def select_runs(riders, maneuvers, environments):
"""Returns a list of runs given a set of conditions.
Parameters
----------
riders : list
All or a subset of ['Charlie', 'Jason', 'Luke'].
maneuvers : list
All or a subset of ['Balance', 'Balance With Disturbance', 'Track
Straight Line', 'Track Straight Line With Disturbance'].
environments : list
All or a subset of ['Horse Treadmill', 'Pavillion Floor'].
Returns
-------
runs : list
List of run numbers for the given conditions.
"""
dataset = bdp.DataSet()
dataset.open()
table = dataset.database.root.runTable
runs = []
for row in table.iterrows():
con = []
con.append(row['Rider'] in riders)
con.append(row['Maneuver'] in maneuvers)
con.append(row['Environment'] in environments)
con.append(row['corrupt'] is not True)
con.append(int(row['RunID']) > 100)
if False not in con:
runs.append(row['RunID'])
dataset.close()
return runs
def load_trials(runs, H):
"""Returns a dictionary of Run objects which were successfully computed and
list of runs which had errors.
Parameters
----------
runs : list
A list of run numbers to try to load.
H : dictionary
The H vector for the lateral force contributions to steer torque and
roll torque for each rider.
Returns
-------
trials : dictionary
A dictionary of bicycledataprocess.Run objects. All in `runNums` must
be present. These trials have an additional attribute `H` which holds
the value for the bicycle model.
errors : list
A list of run numbers that had errors (primarily in the bump finding.
"""
trials = {}
errors = []
dataset = bdp.DataSet()
for i, r in enumerate(runs):
try:
trial = bdp.Run(r, dataset, filterFreq=15.)
except bdp.bdpexceptions.TimeShiftError:
errors.append(r)
except IndexError:
errors.append(r)
else:
trial.H = H[trial.metadata['Rider']]
trials[r] = trial
return trials, errors
import sys
sys.path.append('..')
from load_paths import read
import cPickle
import numpy as np
import matplotlib.pyplot as plt
import bicycledataprocessor as bdp
from canonicalbicycleid import canonical_bicycle_id as cbi
with open(read('pathToGoodRuns')) as f:
goodRuns = cPickle.load(f)
with open(read('pathToIdMat')) as f:
idMat = cPickle.load(f)
dataset = bdp.DataSet()
roll = {k: [] for k in idMat.keys() + ['Whipple', 'Arm']}
steer = {k: [] for k in idMat.keys() + ['Whipple', 'Arm']}
allH = cbi.lateral_force_contribution(['Charlie', 'Jason', 'Luke'])
for runNum in goodRuns:
trial = bdp.Run(runNum, dataset, filterFreq=15.)
rider = trial.metadata['Rider']
timeseries = cbi.benchmark_time_series(trial, subtractMean=True)
for k, model in idMat.items():
rollrsq, steerrsq, fig = cbi.input_prediction(timeseries, model)
roll[k].append(rollrsq)
steer[k].append(steerrsq)
fig.savefig('plots/' + str(runNum) + '-' + k + '.png')
path_to_config = os.path.join(os.getcwd(), 'bicycle-data.cfg')
config = SafeConfigParser()
config.read(path_to_config)
PATH_TO_BICYCLE_PARAMETER_DATA = \
config.get('data', 'PATH_TO_BICYCLE_PARAMETER_DATA')
PATH_TO_INSTRUMENTED_BICYCLE_DATA = \
config.get('data', 'PATH_TO_INSTRUMENTED_BICYCLE_DATA')
# select some runs
riders = ['Charlie', 'Jason', 'Luke']
maneuvers = [
'Balance', 'Balance With Disturbance', 'Track Straight Line',
'Track Straight Line With Disturbance'
]
environments = ['Horse Treadmill', 'Pavillion Floor']
data_set = bdp.DataSet(pathToDatabase=PATH_TO_INSTRUMENTED_BICYCLE_DATA)
runs = data_set.select_runs(riders, maneuvers, environments)
# initialize the data structures
time_series = {}
stats = {
'Root Mean Square of the Error': [],
'Coefficient of Determination': [],
'Maximum Error': []
}
max_num_samples = 0
for run in runs:
ts = {}
ts['Steer Tube Torque'] = None
def test_benchmark_lstsq_matrices():
dataset = bdp.DataSet()
trial = bdp.Run('700', dataset)
A, B, F = cbi.whipple_state_space(trial.metadata['Rider'], 1.0)
H = np.dot(np.linalg.inv(B[2:]), F[2:])
timeSeries = cbi.benchmark_time_series(trial, subtractMean=False)
M, C1, K0, K2 = trial.bicycle.canonical(nominal=True)
fixedValues = cbi.benchmark_canon_to_dict(M, C1, K0, K2, H)
rollParams = [
'Mpp', 'Mpd', 'C1pp', 'C1pd', 'K0pp', 'K0pd', 'K2pp', 'K2pd', 'HpF'
]
A, B = cbi.benchmark_lstsq_matrices(rollParams, timeSeries, fixedValues)
testing.assert_allclose(trial.taskSignals['RollRate'].time_derivative(),
A[:, 0])
testing.assert_allclose(trial.taskSignals['SteerRate'].time_derivative(),
A[:, 1])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed'] * trial.taskSignals['RollRate'],
A[:, 2])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed'] * trial.taskSignals['SteerRate'],
A[:, 3])
testing.assert_allclose(9.81 * trial.taskSignals['RollAngle'], A[:, 4])
testing.assert_allclose(9.81 * trial.taskSignals['SteerAngle'], A[:, 5])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed']**2 * trial.taskSignals['RollAngle'],
A[:, 6])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed']**2 * trial.taskSignals['SteerAngle'],
A[:, 7])
testing.assert_allclose(-trial.taskSignals['PullForce'], A[:, 8])
testing.assert_allclose(np.zeros_like(trial.taskSignals['PullForce']), B)
rollParams = ['Mpp', 'Mpd', 'C1pp', 'C1pd', 'K0pp', 'K0pd', 'K2pp', 'K2pd']
A, B = cbi.benchmark_lstsq_matrices(rollParams, timeSeries, fixedValues)
testing.assert_allclose(trial.taskSignals['RollRate'].time_derivative(),
A[:, 0])
testing.assert_allclose(trial.taskSignals['SteerRate'].time_derivative(),
A[:, 1])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed'] * trial.taskSignals['RollRate'],
A[:, 2])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed'] * trial.taskSignals['SteerRate'],
A[:, 3])
testing.assert_allclose(9.81 * trial.taskSignals['RollAngle'], A[:, 4])
testing.assert_allclose(9.81 * trial.taskSignals['SteerAngle'], A[:, 5])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed']**2 * trial.taskSignals['RollAngle'],
A[:, 6])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed']**2 * trial.taskSignals['SteerAngle'],
A[:, 7])
testing.assert_allclose(H[0] * trial.taskSignals['PullForce'], B)
steerParams = [
'Mdp', 'Mdd', 'C1dp', 'C1dd', 'K0dp', 'K0dd', 'K2dp', 'K2dd', 'HdF'
]
A, B = cbi.benchmark_lstsq_matrices(steerParams, timeSeries, fixedValues)
testing.assert_allclose(trial.taskSignals['RollRate'].time_derivative(),
A[:, 0])
testing.assert_allclose(trial.taskSignals['SteerRate'].time_derivative(),
A[:, 1])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed'] * trial.taskSignals['RollRate'],
A[:, 2])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed'] * trial.taskSignals['SteerRate'],
A[:, 3])
testing.assert_allclose(9.81 * trial.taskSignals['RollAngle'], A[:, 4])
testing.assert_allclose(9.81 * trial.taskSignals['SteerAngle'], A[:, 5])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed']**2 * trial.taskSignals['RollAngle'],
A[:, 6])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed']**2 * trial.taskSignals['SteerAngle'],
A[:, 7])
testing.assert_allclose(-trial.taskSignals['PullForce'], A[:, 8])
testing.assert_allclose(trial.taskSignals['SteerTorque'], B)
steerParams = [
'Mdp', 'Mdd', 'C1dp', 'C1dd', 'K0dp', 'K0dd', 'K2dp', 'K2dd'
]
A, B = cbi.benchmark_lstsq_matrices(steerParams, timeSeries, fixedValues)
testing.assert_allclose(trial.taskSignals['RollRate'].time_derivative(),
A[:, 0])
testing.assert_allclose(trial.taskSignals['SteerRate'].time_derivative(),
A[:, 1])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed'] * trial.taskSignals['RollRate'],
A[:, 2])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed'] * trial.taskSignals['SteerRate'],
A[:, 3])
testing.assert_allclose(9.81 * trial.taskSignals['RollAngle'], A[:, 4])
testing.assert_allclose(9.81 * trial.taskSignals['SteerAngle'], A[:, 5])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed']**2 * trial.taskSignals['RollAngle'],
A[:, 6])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed']**2 * trial.taskSignals['SteerAngle'],
A[:, 7])
testing.assert_allclose(
trial.taskSignals['SteerTorque'] +
H[1] * trial.taskSignals['PullForce'], B)
steerParams = ['Mdp', 'C1dd', 'K0dp', 'K2dp', 'K2dd']
A, B = cbi.benchmark_lstsq_matrices(steerParams, timeSeries, fixedValues)
testing.assert_allclose(trial.taskSignals['RollRate'].time_derivative(),
A[:, 0])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed'] * trial.taskSignals['SteerRate'],
A[:, 1])
testing.assert_allclose(9.81 * trial.taskSignals['RollAngle'], A[:, 2])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed']**2 * trial.taskSignals['RollAngle'],
A[:, 3])
testing.assert_allclose(
trial.taskSignals['ForwardSpeed']**2 * trial.taskSignals['SteerAngle'],
A[:, 4])
testing.assert_allclose(
H[1] * trial.taskSignals['PullForce'] +
trial.taskSignals['SteerTorque'] -
fixedValues['Mdd'] * trial.taskSignals['SteerRate'].time_derivative() -
fixedValues['C1dp'] * trial.taskSignals['ForwardSpeed'] *
trial.taskSignals['RollRate'] -
fixedValues['K0dd'] * 9.81 * trial.taskSignals['SteerAngle'], B)
#!/usr/bin/env python
import bicycledataprocessor as bdp
import os
import numpy as np
import matplotlib.pyplot as plt
pathToBicycleMechanics = '/media/Data/Documents/School/UC Davis/Bicycle Mechanics'
pathToDatabase = os.path.join(
pathToBicycleMechanics, 'BicycleDataProcessor/InstrumentedBicycleData.h5')
pathToParameterData = os.path.join(pathToBicycleMechanics,
'BicycleParameters/data')
dataset = bdp.DataSet(fileName=pathToDatabase)
dataset.open()
trial = bdp.Run('00699',
dataset,
pathToParameterData=pathToParameterData,
filterFreq=40.)
dataset.close()
golden_mean = (np.sqrt(5) - 1.0) / 2.0
fig_width = 6.0
fig_height = fig_width * golden_mean
params = {
'backend': 'ps',
'axes.labelsize': 8,
'axes.titlesize': 10,
'text.fontsize': 8,
try:
filename = '../data/steer-torque-data.h5'
with open(filename, 'r') as f:
time_series = pandas.read_hdf(filename, 'time_series')
stats = pandas.read_hdf(filename, 'stats')
except IOError:
# select some runs
riders = ['Charlie', 'Jason', 'Luke']
maneuvers = [
'Balance', 'Balance With Disturbance', 'Track Straight Line',
'Track Straight Line With Disturbance'
]
environments = ['Horse Treadmill', 'Pavillion Floor']
data_set = bdp.DataSet()
runs = data_set.select_runs(riders, maneuvers, environments)
# initialize the data structures
time_series = {}
stats = {
'Root Mean Square of the Error': [],
'Coefficient of Determination': [],
'Maximum Error': []
}
max_num_samples = 0
for run in runs:
ts = {}
ts['Steer Tube Torque'] = None
