def analyse_ekf(
ulog: ULog, check_levels: Dict[str, float], red_thresh: float = 1.0,
amb_thresh: float = 0.5, min_flight_duration_seconds: float = 5.0,
in_air_margin_seconds: float = 5.0, pos_checks_when_sensors_not_fused: bool = False) -> \
Tuple[str, Dict[str, str], Dict[str, float], Dict[str, float]]:
"""
:param ulog:
:param check_levels:
:param red_thresh:
:param amb_thresh:
:param min_flight_duration_seconds:
:param in_air_margin_seconds:
:param pos_checks_when_sensors_not_fused:
:return:
"""
try:
estimator_status = ulog.get_dataset('estimator_status').data
print('found estimator_status data')
except:
raise PreconditionError('could not find estimator_status data')
try:
_ = ulog.get_dataset('ekf2_innovations').data
print('found ekf2_innovation data')
except:
raise PreconditionError('could not find ekf2_innovation data')
try:
in_air = InAirDetector(
ulog, min_flight_time_seconds=min_flight_duration_seconds, in_air_margin_seconds=0.0)
in_air_no_ground_effects = InAirDetector(
ulog, min_flight_time_seconds=min_flight_duration_seconds,
in_air_margin_seconds=in_air_margin_seconds)
except Exception as e:
raise PreconditionError(str(e))
if in_air_no_ground_effects.take_off is None:
raise PreconditionError('no airtime detected.')
airtime_info = {
'in_air_transition_time': round(in_air.take_off + in_air.log_start, 2),
'on_ground_transition_time': round(in_air.landing + in_air.log_start, 2)}
control_mode, innov_flags, gps_fail_flags = get_estimator_check_flags(estimator_status)
sensor_checks, innov_fail_checks = find_checks_that_apply(
control_mode, estimator_status,
pos_checks_when_sensors_not_fused=pos_checks_when_sensors_not_fused)
metrics = calculate_ecl_ekf_metrics(
ulog, innov_flags, innov_fail_checks, sensor_checks, in_air, in_air_no_ground_effects,
red_thresh=red_thresh, amb_thresh=amb_thresh)
check_status, master_status = perform_ecl_ekf_checks(
metrics, sensor_checks, innov_fail_checks, check_levels)
return master_status, check_status, metrics, airtime_info
def calculate_imu_metrics(ulog: ULog,
in_air_no_ground_effects: InAirDetector) -> dict:
ekf2_innovation_data = ulog.get_dataset('ekf2_innovations').data
estimator_status_data = ulog.get_dataset('estimator_status').data
imu_metrics = dict()
# calculates the median of the output tracking error ekf innovations
for signal, result in [
('output_tracking_error[0]', 'output_obs_ang_err_median'),
('output_tracking_error[1]', 'output_obs_vel_err_median'),
('output_tracking_error[2]', 'output_obs_pos_err_median')
]:
imu_metrics[result] = calculate_stat_from_signal(
ekf2_innovation_data, 'ekf2_innovations', signal,
in_air_no_ground_effects, np.median)
# calculates peak and mean for IMU vibration checks
for signal, result in [('vibe[0]', 'imu_coning'),
('vibe[1]', 'imu_hfdang'),
('vibe[2]', 'imu_hfdvel')]:
peak = calculate_stat_from_signal(estimator_status_data,
'estimator_status', signal,
in_air_no_ground_effects, np.amax)
if peak > 0.0:
imu_metrics['{:s}_peak'.format(result)] = peak
imu_metrics['{:s}_mean'.format(
result)] = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal,
in_air_no_ground_effects, np.mean)
# IMU bias checks
imu_metrics['imu_dang_bias_median'] = np.sqrt(
np.sum([
np.square(
calculate_stat_from_signal(estimator_status_data,
'estimator_status', signal,
in_air_no_ground_effects,
np.median))
for signal in ['states[10]', 'states[11]', 'states[12]']
]))
imu_metrics['imu_dvel_bias_median'] = np.sqrt(
np.sum([
np.square(
calculate_stat_from_signal(estimator_status_data,
'estimator_status', signal,
in_air_no_ground_effects,
np.median))
for signal in ['states[13]', 'states[14]', 'states[15]']
]))
return imu_metrics
def getBarometerData(ulog: ULog) -> pd.DataFrame:
vehicle_air_data = ulog.get_dataset("vehicle_air_data").data
baro = pd.DataFrame({'timestamp': vehicle_air_data['timestamp'],
'sensor' : 'baro',
'baro_alt_meter': vehicle_air_data["baro_alt_meter"]})
return baro
def __init__(
self, ulog: ULog, min_flight_time_seconds: float = 0.0,
in_air_margin_seconds: float = 0.0) -> None:
"""
initializes an InAirDetector instance.
:param ulog:
:param min_flight_time_seconds: set this value to return only airtimes that are at least
min_flight_time_seconds long
:param in_air_margin_seconds: removes a margin of in_air_margin_seconds from the airtime
to avoid ground effects.
"""
self._ulog = ulog
self._min_flight_time_seconds = min_flight_time_seconds
self._in_air_margin_seconds = in_air_margin_seconds
try:
self._vehicle_land_detected = ulog.get_dataset('vehicle_land_detected').data
self._landed = self._vehicle_land_detected['landed']
except:
self._in_air = []
raise PreconditionError(
'InAirDetector: Could not find vehicle land detected message and/or landed field'
' and thus not find any airtime.')
self._log_start = self._ulog.start_timestamp / 1.0e6
self._in_air = self._detect_airtime()
def __init__(self, ulog: ULog, min_flight_time_seconds: float = 0.0,
in_air_margin_seconds: float = 0.0) -> None:
"""
initializes an InAirDetector instance.
:param ulog:
:param min_flight_time_seconds: set this value to return only airtimes that are at least
min_flight_time_seconds long
:param in_air_margin_seconds: removes a margin of in_air_margin_seconds from the airtime
to avoid ground effects.
"""
self._ulog = ulog
self._min_flight_time_seconds = min_flight_time_seconds
self._in_air_margin_seconds = in_air_margin_seconds
try:
self._vehicle_land_detected = ulog.get_dataset('vehicle_land_detected').data
self._landed = self._vehicle_land_detected['landed']
except:
self._in_air = []
raise PreconditionError(
'InAirDetector: Could not find vehicle land detected message and/or landed field'
' and thus not find any airtime.')
self._log_start = self._ulog.start_timestamp / 1.0e6
self._in_air = self._detect_airtime()
def __init__(self, ulog: ULog):
"""
:param ulog:
"""
super(EclCheckRunner, self).__init__()
try:
estimator_status_data = ulog.get_dataset('estimator_status').data
print('found estimator_status data')
control_mode_flags, innov_flags, _ = get_estimator_check_flags(
estimator_status_data)
self.append(
MagnetometerCheck(ulog, innov_flags, control_mode_flags))
self.append(
MagneticHeadingCheck(ulog, innov_flags, control_mode_flags))
self.append(VelocityCheck(ulog, innov_flags, control_mode_flags))
self.append(PositionCheck(ulog, innov_flags, control_mode_flags))
self.append(HeightCheck(ulog, innov_flags, control_mode_flags))
self.append(
HeightAboveGroundCheck(ulog, innov_flags, control_mode_flags))
self.append(AirspeedCheck(ulog, innov_flags, control_mode_flags))
self.append(SideSlipCheck(ulog, innov_flags, control_mode_flags))
self.append(OpticalFlowCheck(ulog, innov_flags,
control_mode_flags))
self.append(IMU_Vibration_Check(ulog))
self.append(IMU_Bias_Check(ulog))
self.append(IMU_Output_Predictor_Check(ulog))
self.append(NumericalCheck(ulog))
except Exception as e:
capture_message(str(e))
self.error_message = str(e)
self.analysis_status = -3
def calculate_ecl_ekf_metrics(
ulog: ULog, innov_flags: Dict[str, float], innov_fail_checks: List[str],
sensor_checks: List[str], in_air: InAirDetector, in_air_no_ground_effects: InAirDetector,
red_thresh: float = 1.0, amb_thresh: float = 0.5) -> Tuple[dict, dict, dict, dict]:
sensor_metrics = calculate_sensor_metrics(
ulog, sensor_checks, in_air, in_air_no_ground_effects,
red_thresh=red_thresh, amb_thresh=amb_thresh)
innov_fail_metrics = calculate_innov_fail_metrics(
innov_flags, innov_fail_checks, in_air, in_air_no_ground_effects)
imu_metrics = calculate_imu_metrics(ulog, in_air_no_ground_effects)
estimator_status_data = ulog.get_dataset('estimator_status').data
# Check for internal filter nummerical faults
ekf_metrics = {'filter_faults_max': np.amax(estimator_status_data['filter_fault_flags'])}
# TODO - process these bitmask's when they have been properly documented in the uORB topic
# estimator_status['health_flags']
# estimator_status['timeout_flags']
# combine the metrics
combined_metrics = dict()
combined_metrics.update(imu_metrics)
combined_metrics.update(sensor_metrics)
combined_metrics.update(innov_fail_metrics)
combined_metrics.update(ekf_metrics)
return combined_metrics
def calculate_ecl_ekf_metrics(
ulog: ULog, innov_flags: Dict[str, float], innov_fail_checks: List[str],
sensor_checks: List[str], in_air: InAirDetector, in_air_no_ground_effects: InAirDetector,
multi_instance: int = 0, red_thresh: float = 1.0, amb_thresh: float = 0.5) -> Tuple[dict, dict, dict, dict]:
sensor_metrics = calculate_sensor_metrics(
ulog, sensor_checks, in_air, in_air_no_ground_effects,
red_thresh=red_thresh, amb_thresh=amb_thresh)
innov_fail_metrics = calculate_innov_fail_metrics(
innov_flags, innov_fail_checks, in_air, in_air_no_ground_effects)
imu_metrics = calculate_imu_metrics(ulog, multi_instance, in_air_no_ground_effects)
estimator_status_data = ulog.get_dataset('estimator_status', multi_instance).data
# Check for internal filter nummerical faults
ekf_metrics = {'filter_faults_max': np.amax(estimator_status_data['filter_fault_flags'])}
# TODO - process these bitmask's when they have been properly documented in the uORB topic
# estimator_status['health_flags']
# estimator_status['timeout_flags']
# combine the metrics
combined_metrics = dict()
combined_metrics.update(imu_metrics)
combined_metrics.update(sensor_metrics)
combined_metrics.update(innov_fail_metrics)
combined_metrics.update(ekf_metrics)
return combined_metrics
def calculate_imu_metrics(ulog: ULog, multi_instance, in_air_no_ground_effects: InAirDetector) -> dict:
estimator_status_data = ulog.get_dataset('estimator_status', multi_instance).data
imu_metrics = dict()
# calculates the median of the output tracking error ekf innovations
for signal, result in [('output_tracking_error[0]', 'output_obs_ang_err_median'),
('output_tracking_error[1]', 'output_obs_vel_err_median'),
('output_tracking_error[2]', 'output_obs_pos_err_median')]:
imu_metrics[result] = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal, in_air_no_ground_effects, np.median)
# calculates peak and mean for IMU vibration checks
for imu_status_instance in range(4):
try:
vehicle_imu_status_data = ulog.get_dataset('vehicle_imu_status', imu_status_instance).data
if vehicle_imu_status_data['accel_device_id'][0] == estimator_status_data['accel_device_id'][0]:
for signal, result in [('gyro_coning_vibration', 'imu_coning'),
('gyro_vibration_metric', 'imu_hfgyro'),
('accel_vibration_metric', 'imu_hfaccel')]:
peak = calculate_stat_from_signal(vehicle_imu_status_data, 'vehicle_imu_status', signal, in_air_no_ground_effects, np.amax)
if peak > 0.0:
imu_metrics['{:s}_peak'.format(result)] = peak
imu_metrics['{:s}_mean'.format(result)] = calculate_stat_from_signal(vehicle_imu_status_data, 'vehicle_imu_status', signal, in_air_no_ground_effects, np.mean)
except:
pass
# IMU bias checks
estimator_states_data = ulog.get_dataset('estimator_states', multi_instance).data
imu_metrics['imu_dang_bias_median'] = np.sqrt(np.sum([np.square(calculate_stat_from_signal(
estimator_states_data, 'estimator_states', signal, in_air_no_ground_effects, np.median))
for signal in ['states[10]', 'states[11]', 'states[12]']]))
imu_metrics['imu_dvel_bias_median'] = np.sqrt(np.sum([np.square(calculate_stat_from_signal(
estimator_states_data, 'estimator_states', signal, in_air_no_ground_effects, np.median))
for signal in ['states[13]', 'states[14]', 'states[15]']]))
return imu_metrics
def getAirspeedData(ulog: ULog) -> pd.DataFrame:
airspeed = ulog.get_dataset("airspeed").data
airspeed = pd.DataFrame({'timestamp': airspeed['timestamp'],
'sensor' : 'airspeed',
'true_as': airspeed["true_airspeed_m_s"],
'indicated_as': airspeed["indicated_airspeed_m_s"]
})
return airspeed
def getMagnetometerData(ulog: ULog) -> pd.DataFrame:
vehicle_magnetometer = ulog.get_dataset("vehicle_magnetometer").data
mag = pd.DataFrame({'timestamp': vehicle_magnetometer['timestamp'],
'sensor' : 'mag',
'magnetometer_ga[0]': vehicle_magnetometer["magnetometer_ga[0]"],
'magnetometer_ga[1]': vehicle_magnetometer["magnetometer_ga[1]"],
'magnetometer_ga[2]': vehicle_magnetometer["magnetometer_ga[2]"]})
return mag
def getImuData(ulog: ULog) -> pd.DataFrame:
sensor_combined = ulog.get_dataset("sensor_combined").data
imu = pd.DataFrame({'timestamp': sensor_combined['timestamp'],
'sensor' : 'imu',
'accel_m_s2[0]': sensor_combined["accelerometer_m_s2[0]"],
'accel_m_s2[1]': sensor_combined["accelerometer_m_s2[1]"],
'accel_m_s2[2]': sensor_combined["accelerometer_m_s2[2]"],
'gyro_rad[0]': sensor_combined["gyro_rad[0]"],
'gyro_rad[1]': sensor_combined["gyro_rad[1]"],
'gyro_rad[2]': sensor_combined["gyro_rad[2]"]})
return imu
def calculate_sensor_metrics(ulog: ULog,
sensor_checks: List[str],
in_air: InAirDetector,
in_air_no_ground_effects: InAirDetector,
multi_instance: int = 0,
red_thresh: float = 1.0,
amb_thresh: float = 0.5) -> Dict[str, float]:
estimator_status_data = ulog.get_dataset('estimator_status',
multi_instance).data
sensor_metrics = dict()
# calculates peak, mean, percentage above 0.5 std, and percentage above std metrics for
# estimator status variables
for signal, result_id in [('hgt_test_ratio', 'hgt'),
('mag_test_ratio', 'mag'),
('vel_test_ratio', 'vel'),
('pos_test_ratio', 'pos'),
('tas_test_ratio', 'tas'),
('hagl_test_ratio', 'hagl')]:
# only run sensor checks, if they apply.
if result_id in sensor_checks:
if result_id == 'mag' or result_id == 'hgt':
in_air_detector = in_air_no_ground_effects
else:
in_air_detector = in_air
# the percentage of samples above / below std dev
sensor_metrics['{:s}_percentage_red'.format(
result_id)] = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal,
in_air_detector, lambda x: 100.0 * np.mean(x > red_thresh))
sensor_metrics['{:s}_percentage_amber'.format(result_id)] = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal, in_air_detector,
lambda x: 100.0 * np.mean(x > amb_thresh)) - \
sensor_metrics['{:s}_percentage_red'.format(result_id)]
# the peak and mean ratio of samples above / below std dev
peak = calculate_stat_from_signal(estimator_status_data,
'estimator_status', signal,
in_air_detector, np.amax)
if peak > 0.0:
sensor_metrics['{:s}_test_max'.format(result_id)] = peak
sensor_metrics['{:s}_test_mean'.format(
result_id)] = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal,
in_air_detector, np.mean)
return sensor_metrics
def getGpsData(ulog: ULog) -> pd.DataFrame:
vehicle_gps_position = ulog.get_dataset("vehicle_gps_position").data
gps = pd.DataFrame({'timestamp': vehicle_gps_position['timestamp'],
'sensor' : 'gps',
'alt': vehicle_gps_position["alt"],
'lon': vehicle_gps_position["lon"],
'lat': vehicle_gps_position["lat"],
'vel_N': vehicle_gps_position["vel_n_m_s"],
'vel_E': vehicle_gps_position["vel_e_m_s"],
'vel_D': vehicle_gps_position["vel_d_m_s"],
})
return gps
def getOpticalFlowData(ulog: ULog) -> pd.DataFrame:
optical_flow = ulog.get_dataset("optical_flow").data
flow = pd.DataFrame({'timestamp': optical_flow['timestamp'],
'sensor' : 'flow',
'pixel_flow_x_integral': optical_flow["pixel_flow_x_integral"],
'pixel_flow_y_integral': optical_flow["pixel_flow_y_integral"],
'gyro_x_rate_integral': optical_flow["gyro_x_rate_integral"],
'gyro_y_rate_integral': optical_flow["gyro_y_rate_integral"],
'gyro_z_rate_integral': optical_flow["gyro_z_rate_integral"],
'quality': optical_flow["quality"]
})
return flow
def check_if_field_name_exists_in_message(ulog: ULog, message: str,
field_name: str) -> bool:
"""
Check if a field is part of a message in a certain log
"""
exists = True
if message not in [elem.name for elem in ulog.data_list]:
exists = False
elif field_name not in ulog.get_dataset(message).data.keys():
exists = False
return exists
def calculate_imu_metrics(
ulog: ULog, in_air_no_ground_effects: InAirDetector) -> dict:
ekf2_innovation_data = ulog.get_dataset('ekf2_innovations').data
estimator_status_data = ulog.get_dataset('estimator_status').data
imu_metrics = dict()
# calculates the median of the output tracking error ekf innovations
for signal, result in [('output_tracking_error[0]', 'output_obs_ang_err_median'),
('output_tracking_error[1]', 'output_obs_vel_err_median'),
('output_tracking_error[2]', 'output_obs_pos_err_median')]:
imu_metrics[result] = calculate_stat_from_signal(
ekf2_innovation_data, 'ekf2_innovations', signal, in_air_no_ground_effects, np.median)
# calculates peak and mean for IMU vibration checks
for signal, result in [('vibe[0]', 'imu_coning'),
('vibe[1]', 'imu_hfdang'),
('vibe[2]', 'imu_hfdvel')]:
peak = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal, in_air_no_ground_effects, np.amax)
if peak > 0.0:
imu_metrics['{:s}_peak'.format(result)] = peak
imu_metrics['{:s}_mean'.format(result)] = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal,
in_air_no_ground_effects, np.mean)
# IMU bias checks
imu_metrics['imu_dang_bias_median'] = np.sqrt(np.sum([np.square(calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal, in_air_no_ground_effects, np.median))
for signal in ['states[10]', 'states[11]', 'states[12]']]))
imu_metrics['imu_dvel_bias_median'] = np.sqrt(np.sum([np.square(calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal, in_air_no_ground_effects, np.median))
for signal in ['states[13]', 'states[14]', 'states[15]']]))
return imu_metrics
def getRangeFinderData(ulog: ULog) -> pd.DataFrame:
range = pd.DataFrame()
try:
range_0 = ulog.get_dataset("distance_sensor", 0).data
rng_0 = pd.DataFrame({'timestamp': range_0['timestamp'],
'sensor' : 'range',
'data': range_0["current_distance"],
'quality': range_0["signal_quality"]
})
range = pd.concat([range, rng_0], ignore_index=True, sort=False)
except:
pass
try:
range_1 = ulog.get_dataset("distance_sensor", 1).data
rng_1 = pd.DataFrame({'timestamp': range_1['timestamp'],
'sensor' : 'range',
'data': range_1["current_distance"],
'quality': range_1["signal_quality"]
})
range = pd.concat([range, rng_1], ignore_index=True, sort=False)
except:
pass
try:
range_2 = ulog.get_dataset("distance_sensor", 2).data
rng_2 = pd.DataFrame({'timestamp': range_2['timestamp'],
'sensor' : 'range',
'data': range_2["current_distance"],
'quality': range_2["signal_quality"]
})
range = pd.concat([range, rng_2], ignore_index=True, sort=False)
except:
pass
return range
def check_if_field_name_exists_in_message(ulog: ULog, message_name: str,
field_name: str) -> bool:
"""
Check if a field is part of a message in a certain log
"""
if field_name is None:
return False
msg_data = ulog.get_dataset(message_name).data
field_name_list = dict.keys(msg_data)
for elem in field_name_list:
if field_name == field_name_wo_brackets(elem):
return True
return False
def getVioData(ulog: ULog) -> pd.DataFrame:
vehicle_visual_odometry = ulog.get_dataset("vehicle_visual_odometry").data
vio = pd.DataFrame({'timestamp': vehicle_visual_odometry['timestamp'],
'sensor' : 'vio',
'x': vehicle_visual_odometry["x"],
'y': vehicle_visual_odometry["y"],
'z': vehicle_visual_odometry["z"],
'qw': vehicle_visual_odometry["q[0]"],
'qx': vehicle_visual_odometry["q[1]"],
'qy': vehicle_visual_odometry["q[2]"],
'qz': vehicle_visual_odometry["q[3]"],
'vx': vehicle_visual_odometry["vx"],
'vy': vehicle_visual_odometry["vy"],
'vz': vehicle_visual_odometry["vz"]
})
return vio
def calculate_sensor_metrics(
ulog: ULog, sensor_checks: List[str], in_air: InAirDetector,
in_air_no_ground_effects: InAirDetector, red_thresh: float = 1.0,
amb_thresh: float = 0.5) -> Dict[str, float]:
estimator_status_data = ulog.get_dataset('estimator_status').data
sensor_metrics = dict()
# calculates peak, mean, percentage above 0.5 std, and percentage above std metrics for
# estimator status variables
for signal, result_id in [('hgt_test_ratio', 'hgt'),
('mag_test_ratio', 'mag'),
('vel_test_ratio', 'vel'),
('pos_test_ratio', 'pos'),
('tas_test_ratio', 'tas'),
('hagl_test_ratio', 'hagl')]:
# only run sensor checks, if they apply.
if result_id in sensor_checks:
if result_id == 'mag' or result_id == 'hgt':
in_air_detector = in_air_no_ground_effects
else:
in_air_detector = in_air
# the percentage of samples above / below std dev
sensor_metrics['{:s}_percentage_red'.format(result_id)] = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal, in_air_detector,
lambda x: 100.0 * np.mean(x > red_thresh))
sensor_metrics['{:s}_percentage_amber'.format(result_id)] = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal, in_air_detector,
lambda x: 100.0 * np.mean(x > amb_thresh)) - \
sensor_metrics['{:s}_percentage_red'.format(result_id)]
# the peak and mean ratio of samples above / below std dev
peak = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal, in_air_detector, np.amax)
if peak > 0.0:
sensor_metrics['{:s}_test_max'.format(result_id)] = peak
sensor_metrics['{:s}_test_mean'.format(result_id)] = calculate_stat_from_signal(
estimator_status_data, 'estimator_status', signal,
in_air_detector, np.mean)
return sensor_metrics
def get_innovation_message_and_field_names(
ulog: ULog,
field_descriptor: str,
topic: str = 'innovation') -> Tuple[str, List[str]]:
"""
:param ulog:
:param field_descriptor:
:param topic: one of (innovation | innovation_variance | innovation_test_ratio)
:return:
"""
field_names = []
message = get_innovation_message(ulog, topic=topic)
field_name = get_field_name_from_message_and_descriptor(message,
field_descriptor,
topic=topic)
innov_data = ulog.get_dataset(message).data
if field_name in innov_data:
field_names.append(field_name)
else:
i = 0
while '{:s}[{:d}]'.format(field_name, i) in innov_data:
field_names.append('{:s}[{:d}]'.format(field_name, i))
i += 1
if field_name.endswith('_vel'):
field_names.append('{:s}_h{:s}[0]'.format(field_name[:3],
field_name[-3:]))
field_names.append('{:s}_h{:s}[1]'.format(field_name[:3],
field_name[-3:]))
field_names.append('{:s}_v{:s}'.format(field_name[:3],
field_name[-3:]))
return message, field_names
def get_flight_report(ulog_file):
print(ulog_file)
try:
ulog = ULog(
ulog_file,
['vehicle_status', 'battery_status', 'vehicle_gps_position'])
vehicle_status = ulog.get_dataset('vehicle_status')
battery_status = ulog.get_dataset('battery_status')
utc_avail = False
# Processing arming data.
armed_data = vehicle_status.data[
'arming_state'] # 1 is standby, 2 is armed.
armed_tstamp = vehicle_status.data['timestamp'] # in us.
was_armed = np.any(armed_data == 2)
if not was_armed:
del ulog
return False
try:
gps_position = ulog.get_dataset('vehicle_gps_position')
start_time = time.localtime(gps_position.data['time_utc_usec'][0] /
1E6) # in us since epoch
# boot_time = time.localtime(gps_position.data['timestamp'][0]/1E6) # in us since FC boot
utc_avail = True
except:
utc_avail = False
batt_v_start = battery_status.data['voltage_filtered_v'][0]
batt_v_end = battery_status.data['voltage_filtered_v'][-1]
# batt_v_tsamp = battery_status.data['timestamp'] # in us.
arm_disarm_idx = np.where(
armed_data[:-1] != armed_data[1:]
) # Find where arming state changes by comparing to neighbor.
arm_disarm_events = armed_data[arm_disarm_idx]
arm_disarm_events_tstamp = armed_tstamp[arm_disarm_idx]
arm_disarm_durations = (
arm_disarm_events_tstamp[1::2] - arm_disarm_events_tstamp[0::2]
) / 1E6 # Assume every second event is arming. Can't be armed on boot!
arm_disarm_total_time = arm_disarm_durations.sum()
arm_min, arm_sec = divmod(arm_disarm_total_time, 60)
arm_min = int(arm_min) # because it's an integer.
if utc_avail:
print('System powered on {0}'.format(time.asctime(start_time)))
else:
print('No GPS. No UTC time')
print('Log file name: {0}'.format(ulog_file))
print('Arm duration: {0} min {1} sec'.format(str(arm_min),
str(arm_sec)))
print('Start voltage: {0}V, End voltage: {1}V'.format(
str(batt_v_start), str(batt_v_end)))
print('\n\n')
del ulog
return True
except IndexError:
# print('Empty or invalid file')
return False
except Exception as e:
# print(str(e))
return False
def analyse_ekf(
ulog: ULog, check_levels: Dict[str, float], red_thresh: float = 1.0,
amb_thresh: float = 0.5, min_flight_duration_seconds: float = 5.0,
in_air_margin_seconds: float = 5.0, pos_checks_when_sensors_not_fused: bool = False) -> \
Tuple[str, Dict[str, str], Dict[str, float], Dict[str, float]]:
"""
:param ulog:
:param check_levels:
:param red_thresh:
:param amb_thresh:
:param min_flight_duration_seconds:
:param in_air_margin_seconds:
:param pos_checks_when_sensors_not_fused:
:return:
"""
try:
estimator_states = ulog.get_dataset('estimator_states').data
print('found estimator_states data')
except:
raise PreconditionError('could not find estimator_states data')
try:
estimator_status = ulog.get_dataset('estimator_status').data
print('found estimator_status data')
except:
raise PreconditionError('could not find estimator_status data')
try:
_ = ulog.get_dataset('estimator_innovations').data
print('found estimator_innovations data')
except:
raise PreconditionError('could not find estimator_innovations data')
try:
in_air = InAirDetector(
ulog, min_flight_time_seconds=min_flight_duration_seconds, in_air_margin_seconds=0.0)
in_air_no_ground_effects = InAirDetector(
ulog, min_flight_time_seconds=min_flight_duration_seconds,
in_air_margin_seconds=in_air_margin_seconds)
except Exception as e:
raise PreconditionError(str(e))
if in_air_no_ground_effects.take_off is None:
raise PreconditionError('no airtime detected.')
airtime_info = {
'in_air_transition_time': round(in_air.take_off + in_air.log_start, 2),
'on_ground_transition_time': round(in_air.landing + in_air.log_start, 2)}
control_mode, innov_flags, gps_fail_flags = get_estimator_check_flags(estimator_status)
sensor_checks, innov_fail_checks = find_checks_that_apply(
control_mode, estimator_status,
pos_checks_when_sensors_not_fused=pos_checks_when_sensors_not_fused)
metrics = calculate_ecl_ekf_metrics(
ulog, innov_flags, innov_fail_checks, sensor_checks, in_air, in_air_no_ground_effects,
red_thresh=red_thresh, amb_thresh=amb_thresh)
check_status, master_status = perform_ecl_ekf_checks(
metrics, sensor_checks, innov_fail_checks, check_levels)
return master_status, check_status, metrics, airtime_info
def create_pdf_report(ulog: ULog, output_plot_filename: str) -> None:
"""
creates a pdf report of the ekf analysis.
:param ulog:
:param output_plot_filename:
:return:
"""
# create summary plots
# save the plots to PDF
try:
estimator_status = ulog.get_dataset('estimator_status').data
print('found estimator_status data')
except:
raise PreconditionError('could not find estimator_status data')
try:
estimator_innovations = ulog.get_dataset('estimator_innovations').data
estimator_innovation_variances = ulog.get_dataset(
'estimator_innovation_variances').data
innovation_data = estimator_innovations
for key in estimator_innovation_variances:
# append 'var' to the field name such that we can distingush between innov and innov_var
innovation_data.update(
{str('var_' + key): estimator_innovation_variances[key]})
print('found innovation data')
except:
raise PreconditionError('could not find innovation data')
try:
sensor_preflight = ulog.get_dataset('sensor_preflight').data
print('found sensor_preflight data')
except:
raise PreconditionError('could not find sensor_preflight data')
control_mode, innov_flags, gps_fail_flags = get_estimator_check_flags(
estimator_status)
status_time = 1e-6 * estimator_status['timestamp']
b_finishes_in_air, b_starts_in_air, in_air_duration, in_air_transition_time, \
on_ground_transition_time = detect_airtime(control_mode, status_time)
with PdfPages(output_plot_filename) as pdf_pages:
# plot IMU consistency data
if ('accel_inconsistency_m_s_s'
in sensor_preflight.keys()) and ('gyro_inconsistency_rad_s'
in sensor_preflight.keys()):
data_plot = TimeSeriesPlot(
sensor_preflight,
[['accel_inconsistency_m_s_s'], ['gyro_inconsistency_rad_s']],
x_labels=['data index', 'data index'],
y_labels=['acceleration (m/s/s)', 'angular rate (rad/s)'],
plot_title='IMU Consistency Check Levels',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# vertical velocity and position innovations
data_plot = InnovationPlot(innovation_data,
[('gps_vpos', 'var_gps_vpos'),
('gps_vvel', 'var_gps_vvel')],
x_labels=['time (sec)', 'time (sec)'],
y_labels=['Down Vel (m/s)', 'Down Pos (m)'],
plot_title='Vertical Innovations',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# horizontal velocity innovations
data_plot = InnovationPlot(
innovation_data, [('gps_hvel[0]', 'var_gps_hvel[0]'),
('gps_hvel[1]', 'var_gps_hvel[1]')],
x_labels=['time (sec)', 'time (sec)'],
y_labels=['North Vel (m/s)', 'East Vel (m/s)'],
plot_title='Horizontal Velocity Innovations',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# horizontal position innovations
data_plot = InnovationPlot(
innovation_data, [('gps_hpos[0]', 'var_gps_hpos[0]'),
('gps_hpos[1]', 'var_gps_hpos[1]')],
x_labels=['time (sec)', 'time (sec)'],
y_labels=['North Pos (m)', 'East Pos (m)'],
plot_title='Horizontal Position Innovations',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# magnetometer innovations
data_plot = InnovationPlot(
innovation_data, [('mag_field[0]', 'var_mag_field[0]'),
('mag_field[1]', 'var_mag_field[1]'),
('mag_field[2]', 'var_mag_field[2]')],
x_labels=['time (sec)', 'time (sec)', 'time (sec)'],
y_labels=['X (Gauss)', 'Y (Gauss)', 'Z (Gauss)'],
plot_title='Magnetometer Innovations',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# magnetic heading innovations
data_plot = InnovationPlot(innovation_data,
[('heading', 'var_heading')],
x_labels=['time (sec)'],
y_labels=['Heading (rad)'],
plot_title='Magnetic Heading Innovations',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# air data innovations
data_plot = InnovationPlot(
innovation_data, [('airspeed', 'var_airspeed'),
('beta', 'var_beta')],
x_labels=['time (sec)', 'time (sec)'],
y_labels=['innovation (m/sec)', 'innovation (rad)'],
sub_titles=[
'True Airspeed Innovations', 'Synthetic Sideslip Innovations'
],
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# optical flow innovations
data_plot = InnovationPlot(innovation_data,
[('flow[0]', 'var_flow[0]'),
('flow[1]', 'var_flow[1]')],
x_labels=['time (sec)', 'time (sec)'],
y_labels=['X (rad/sec)', 'Y (rad/sec)'],
plot_title='Optical Flow Innovations',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# plot normalised innovation test levels
# define variables to plot
variables = [['mag_test_ratio'], ['vel_test_ratio', 'pos_test_ratio'],
['hgt_test_ratio']]
y_labels = ['mag', 'vel, pos', 'hgt']
legend = [['mag'], ['vel', 'pos'], ['hgt']]
if np.amax(estimator_status['hagl_test_ratio']
) > 0.0: # plot hagl test ratio, if applicable
variables[-1].append('hagl_test_ratio')
y_labels[-1] += ', hagl'
legend[-1].append('hagl')
if np.amax(estimator_status['tas_test_ratio']
) > 0.0: # plot airspeed sensor test ratio, if applicable
variables.append(['tas_test_ratio'])
y_labels.append('TAS')
legend.append(['airspeed'])
data_plot = CheckFlagsPlot(
status_time,
estimator_status,
variables,
x_label='time (sec)',
y_labels=y_labels,
plot_title='Normalised Innovation Test Levels',
pdf_handle=pdf_pages,
annotate=True,
legend=legend)
data_plot.save()
data_plot.close()
# plot control mode summary A
data_plot = ControlModeSummaryPlot(
status_time,
control_mode,
[['tilt_aligned', 'yaw_aligned'],
['using_gps', 'using_optflow', 'using_evpos'],
['using_barohgt', 'using_gpshgt', 'using_rnghgt', 'using_evhgt'],
['using_magyaw', 'using_mag3d', 'using_magdecl']],
x_label='time (sec)',
y_labels=['aligned', 'pos aiding', 'hgt aiding', 'mag aiding'],
annotation_text=[['tilt alignment', 'yaw alignment'],
[
'GPS aiding', 'optical flow aiding',
'external vision aiding'
],
[
'Baro aiding', 'GPS aiding',
'rangefinder aiding', 'external vision aiding'
],
[
'magnetic yaw aiding',
'3D magnetoemter aiding',
'magnetic declination aiding'
]],
plot_title='EKF Control Status - Figure A',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# plot control mode summary B
# construct additional annotations for the airborne plot
airborne_annotations = list()
if np.amin(np.diff(control_mode['airborne'])) > -0.5:
airborne_annotations.append(
(on_ground_transition_time,
'air to ground transition not detected'))
else:
airborne_annotations.append(
(on_ground_transition_time,
'on-ground at {:.1f} sec'.format(on_ground_transition_time)))
if in_air_duration > 0.0:
airborne_annotations.append(
((in_air_transition_time + on_ground_transition_time) / 2,
'duration = {:.1f} sec'.format(in_air_duration)))
if np.amax(np.diff(control_mode['airborne'])) < 0.5:
airborne_annotations.append(
(in_air_transition_time,
'ground to air transition not detected'))
else:
airborne_annotations.append(
(in_air_transition_time,
'in-air at {:.1f} sec'.format(in_air_transition_time)))
data_plot = ControlModeSummaryPlot(
status_time,
control_mode, [['airborne'], ['estimating_wind']],
x_label='time (sec)',
y_labels=['airborne', 'estimating wind'],
annotation_text=[[], []],
additional_annotation=[airborne_annotations, []],
plot_title='EKF Control Status - Figure B',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# plot innovation_check_flags summary
data_plot = CheckFlagsPlot(
status_time,
innov_flags, [['vel_innov_fail', 'posh_innov_fail'],
['posv_innov_fail', 'hagl_innov_fail'],
[
'magx_innov_fail', 'magy_innov_fail',
'magz_innov_fail', 'yaw_innov_fail'
], ['tas_innov_fail'], ['sli_innov_fail'],
['ofx_innov_fail', 'ofy_innov_fail']],
x_label='time (sec)',
y_labels=[
'failed', 'failed', 'failed', 'failed', 'failed', 'failed'
],
y_lim=(-0.1, 1.1),
legend=[['vel NED', 'pos NE'],
['hgt absolute', 'hgt above ground'],
['mag_x', 'mag_y', 'mag_z', 'yaw'], ['airspeed'],
['sideslip'], ['flow X', 'flow Y']],
plot_title='EKF Innovation Test Fails',
annotate=False,
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# gps_check_fail_flags summary
data_plot = CheckFlagsPlot(
status_time,
gps_fail_flags, [[
'nsat_fail', 'pdop_fail', 'herr_fail', 'verr_fail',
'gfix_fail', 'serr_fail'
], ['hdrift_fail', 'vdrift_fail', 'hspd_fail', 'veld_diff_fail']],
x_label='time (sec)',
y_lim=(-0.1, 1.1),
y_labels=['failed', 'failed'],
sub_titles=[
'GPS Direct Output Check Failures',
'GPS Derived Output Check Failures'
],
legend=[[
'N sats', 'PDOP', 'horiz pos error', 'vert pos error',
'fix type', 'speed error'
],
[
'horiz drift', 'vert drift', 'horiz speed',
'vert vel inconsistent'
]],
annotate=False,
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# filter reported accuracy
data_plot = CheckFlagsPlot(
status_time,
estimator_status, [['pos_horiz_accuracy', 'pos_vert_accuracy']],
x_label='time (sec)',
y_labels=['accuracy (m)'],
plot_title='Reported Accuracy',
legend=[['horizontal', 'vertical']],
annotate=False,
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# Plot the EKF IMU vibration metrics
scaled_estimator_status = {
'vibe[0]': 1000. * estimator_status['vibe[0]'],
'vibe[1]': 1000. * estimator_status['vibe[1]'],
'vibe[2]': estimator_status['vibe[2]']
}
data_plot = CheckFlagsPlot(status_time,
scaled_estimator_status,
[['vibe[0]'], ['vibe[1]'], ['vibe[2]']],
x_label='time (sec)',
y_labels=[
'Del Ang Coning (mrad)',
'HF Del Ang (mrad)', 'HF Del Vel (m/s)'
],
plot_title='IMU Vibration Metrics',
pdf_handle=pdf_pages,
annotate=True)
data_plot.save()
data_plot.close()
# Plot the EKF output observer tracking errors
scaled_innovations = {
'output_tracking_error[0]':
1000. * estimator_status['output_tracking_error[0]'],
'output_tracking_error[1]':
estimator_status['output_tracking_error[1]'],
'output_tracking_error[2]':
estimator_status['output_tracking_error[2]']
}
data_plot = CheckFlagsPlot(
1e-6 * estimator_status['timestamp'],
scaled_innovations,
[['output_tracking_error[0]'], ['output_tracking_error[1]'],
['output_tracking_error[2]']],
x_label='time (sec)',
y_labels=['angles (mrad)', 'velocity (m/s)', 'position (m)'],
plot_title='Output Observer Tracking Error Magnitudes',
pdf_handle=pdf_pages,
annotate=True)
data_plot.save()
data_plot.close()
# Plot the delta angle bias estimates
data_plot = CheckFlagsPlot(
1e-6 * estimator_status['timestamp'],
estimator_status, [['states[10]'], ['states[11]'], ['states[12]']],
x_label='time (sec)',
y_labels=['X (rad)', 'Y (rad)', 'Z (rad)'],
plot_title='Delta Angle Bias Estimates',
annotate=False,
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# Plot the delta velocity bias estimates
data_plot = CheckFlagsPlot(
1e-6 * estimator_status['timestamp'],
estimator_status, [['states[13]'], ['states[14]'], ['states[15]']],
x_label='time (sec)',
y_labels=['X (m/s)', 'Y (m/s)', 'Z (m/s)'],
plot_title='Delta Velocity Bias Estimates',
annotate=False,
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# Plot the earth frame magnetic field estimates
declination, field_strength, inclination = magnetic_field_estimates_from_status(
estimator_status)
data_plot = CheckFlagsPlot(1e-6 * estimator_status['timestamp'], {
'strength': field_strength,
'declination': declination,
'inclination': inclination
}, [['declination'], ['inclination'], ['strength']],
x_label='time (sec)',
y_labels=[
'declination (deg)',
'inclination (deg)', 'strength (Gauss)'
],
plot_title='Earth Magnetic Field Estimates',
annotate=False,
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# Plot the body frame magnetic field estimates
data_plot = CheckFlagsPlot(
1e-6 * estimator_status['timestamp'],
estimator_status, [['states[19]'], ['states[20]'], ['states[21]']],
x_label='time (sec)',
y_labels=['X (Gauss)', 'Y (Gauss)', 'Z (Gauss)'],
plot_title='Magnetometer Bias Estimates',
annotate=False,
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# Plot the EKF wind estimates
data_plot = CheckFlagsPlot(1e-6 * estimator_status['timestamp'],
estimator_status,
[['states[22]'], ['states[23]']],
x_label='time (sec)',
y_labels=['North (m/s)', 'East (m/s)'],
plot_title='Wind Velocity Estimates',
annotate=False,
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
def create_pdf_report(ulog: ULog, output_plot_filename: str) -> None:
"""
creates a pdf report of the ekf analysis.
:param ulog:
:param output_plot_filename:
:return:
"""
# create summary plots
# save the plots to PDF
try:
estimator_status = ulog.get_dataset('estimator_status').data
print('found estimator_status data')
except:
raise PreconditionError('could not find estimator_status data')
try:
ekf2_innovations = ulog.get_dataset('ekf2_innovations').data
print('found ekf2_innovation data')
except:
raise PreconditionError('could not find ekf2_innovation data')
try:
sensor_preflight = ulog.get_dataset('sensor_preflight').data
print('found sensor_preflight data')
except:
raise PreconditionError('could not find sensor_preflight data')
control_mode, innov_flags, gps_fail_flags = get_estimator_check_flags(estimator_status)
status_time = 1e-6 * estimator_status['timestamp']
b_finishes_in_air, b_starts_in_air, in_air_duration, in_air_transition_time, \
on_ground_transition_time = detect_airtime(control_mode, status_time)
with PdfPages(output_plot_filename) as pdf_pages:
# plot IMU consistency data
if ('accel_inconsistency_m_s_s' in sensor_preflight.keys()) and (
'gyro_inconsistency_rad_s' in sensor_preflight.keys()):
data_plot = TimeSeriesPlot(
sensor_preflight, [['accel_inconsistency_m_s_s'], ['gyro_inconsistency_rad_s']],
x_labels=['data index', 'data index'],
y_labels=['acceleration (m/s/s)', 'angular rate (rad/s)'],
plot_title='IMU Consistency Check Levels', pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# vertical velocity and position innovations
data_plot = InnovationPlot(
ekf2_innovations, [('vel_pos_innov[2]', 'vel_pos_innov_var[2]'),
('vel_pos_innov[5]', 'vel_pos_innov_var[5]')],
x_labels=['time (sec)', 'time (sec)'],
y_labels=['Down Vel (m/s)', 'Down Pos (m)'], plot_title='Vertical Innovations',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# horizontal velocity innovations
data_plot = InnovationPlot(
ekf2_innovations, [('vel_pos_innov[0]', 'vel_pos_innov_var[0]'),
('vel_pos_innov[1]','vel_pos_innov_var[1]')],
x_labels=['time (sec)', 'time (sec)'],
y_labels=['North Vel (m/s)', 'East Vel (m/s)'],
plot_title='Horizontal Velocity Innovations', pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# horizontal position innovations
data_plot = InnovationPlot(
ekf2_innovations, [('vel_pos_innov[3]', 'vel_pos_innov_var[3]'), ('vel_pos_innov[4]',
'vel_pos_innov_var[4]')],
x_labels=['time (sec)', 'time (sec)'],
y_labels=['North Pos (m)', 'East Pos (m)'], plot_title='Horizontal Position Innovations',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# magnetometer innovations
data_plot = InnovationPlot(
ekf2_innovations, [('mag_innov[0]', 'mag_innov_var[0]'),
('mag_innov[1]', 'mag_innov_var[1]'), ('mag_innov[2]', 'mag_innov_var[2]')],
x_labels=['time (sec)', 'time (sec)', 'time (sec)'],
y_labels=['X (Gauss)', 'Y (Gauss)', 'Z (Gauss)'], plot_title='Magnetometer Innovations',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# magnetic heading innovations
data_plot = InnovationPlot(
ekf2_innovations, [('heading_innov', 'heading_innov_var')],
x_labels=['time (sec)'], y_labels=['Heading (rad)'],
plot_title='Magnetic Heading Innovations', pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# air data innovations
data_plot = InnovationPlot(
ekf2_innovations,
[('airspeed_innov', 'airspeed_innov_var'), ('beta_innov', 'beta_innov_var')],
x_labels=['time (sec)', 'time (sec)'],
y_labels=['innovation (m/sec)', 'innovation (rad)'],
sub_titles=['True Airspeed Innovations', 'Synthetic Sideslip Innovations'],
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# optical flow innovations
data_plot = InnovationPlot(
ekf2_innovations, [('flow_innov[0]', 'flow_innov_var[0]'), ('flow_innov[1]',
'flow_innov_var[1]')],
x_labels=['time (sec)', 'time (sec)'],
y_labels=['X (rad/sec)', 'Y (rad/sec)'],
plot_title='Optical Flow Innovations', pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# plot normalised innovation test levels
# define variables to plot
variables = [['mag_test_ratio'], ['vel_test_ratio', 'pos_test_ratio'], ['hgt_test_ratio']]
y_labels = ['mag', 'vel, pos', 'hgt']
legend = [['mag'], ['vel', 'pos'], ['hgt']]
if np.amax(estimator_status['hagl_test_ratio']) > 0.0: # plot hagl test ratio, if applicable
variables[-1].append('hagl_test_ratio')
y_labels[-1] += ', hagl'
legend[-1].append('hagl')
if np.amax(estimator_status[
'tas_test_ratio']) > 0.0: # plot airspeed sensor test ratio, if applicable
variables.append(['tas_test_ratio'])
y_labels.append('TAS')
legend.append(['airspeed'])
data_plot = CheckFlagsPlot(
status_time, estimator_status, variables, x_label='time (sec)', y_labels=y_labels,
plot_title='Normalised Innovation Test Levels', pdf_handle=pdf_pages, annotate=True,
legend=legend
)
data_plot.save()
data_plot.close()
# plot control mode summary A
data_plot = ControlModeSummaryPlot(
status_time, control_mode, [['tilt_aligned', 'yaw_aligned'],
['using_gps', 'using_optflow', 'using_evpos'], ['using_barohgt', 'using_gpshgt',
'using_rnghgt', 'using_evhgt'], ['using_magyaw', 'using_mag3d', 'using_magdecl']],
x_label='time (sec)', y_labels=['aligned', 'pos aiding', 'hgt aiding', 'mag aiding'],
annotation_text=[['tilt alignment', 'yaw alignment'], ['GPS aiding', 'optical flow aiding',
'external vision aiding'], ['Baro aiding', 'GPS aiding', 'rangefinder aiding',
'external vision aiding'], ['magnetic yaw aiding', '3D magnetoemter aiding',
'magnetic declination aiding']], plot_title='EKF Control Status - Figure A',
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# plot control mode summary B
# construct additional annotations for the airborne plot
airborne_annotations = list()
if np.amin(np.diff(control_mode['airborne'])) > -0.5:
airborne_annotations.append(
(on_ground_transition_time, 'air to ground transition not detected'))
else:
airborne_annotations.append((on_ground_transition_time, 'on-ground at {:.1f} sec'.format(
on_ground_transition_time)))
if in_air_duration > 0.0:
airborne_annotations.append(((in_air_transition_time + on_ground_transition_time) / 2,
'duration = {:.1f} sec'.format(in_air_duration)))
if np.amax(np.diff(control_mode['airborne'])) < 0.5:
airborne_annotations.append(
(in_air_transition_time, 'ground to air transition not detected'))
else:
airborne_annotations.append(
(in_air_transition_time, 'in-air at {:.1f} sec'.format(in_air_transition_time)))
data_plot = ControlModeSummaryPlot(
status_time, control_mode, [['airborne'], ['estimating_wind']],
x_label='time (sec)', y_labels=['airborne', 'estimating wind'], annotation_text=[[], []],
additional_annotation=[airborne_annotations, []],
plot_title='EKF Control Status - Figure B', pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# plot innovation_check_flags summary
data_plot = CheckFlagsPlot(
status_time, innov_flags, [['vel_innov_fail', 'posh_innov_fail'], ['posv_innov_fail',
'hagl_innov_fail'],
['magx_innov_fail', 'magy_innov_fail', 'magz_innov_fail',
'yaw_innov_fail'], ['tas_innov_fail'], ['sli_innov_fail'],
['ofx_innov_fail',
'ofy_innov_fail']], x_label='time (sec)',
y_labels=['failed', 'failed', 'failed', 'failed', 'failed', 'failed'],
y_lim=(-0.1, 1.1),
legend=[['vel NED', 'pos NE'], ['hgt absolute', 'hgt above ground'],
['mag_x', 'mag_y', 'mag_z', 'yaw'], ['airspeed'], ['sideslip'],
['flow X', 'flow Y']],
plot_title='EKF Innovation Test Fails', annotate=False, pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# gps_check_fail_flags summary
data_plot = CheckFlagsPlot(
status_time, gps_fail_flags,
[['nsat_fail', 'gdop_fail', 'herr_fail', 'verr_fail', 'gfix_fail', 'serr_fail'],
['hdrift_fail', 'vdrift_fail', 'hspd_fail', 'veld_diff_fail']],
x_label='time (sec)', y_lim=(-0.1, 1.1), y_labels=['failed', 'failed'],
sub_titles=['GPS Direct Output Check Failures', 'GPS Derived Output Check Failures'],
legend=[['N sats', 'GDOP', 'horiz pos error', 'vert pos error', 'fix type',
'speed error'], ['horiz drift', 'vert drift', 'horiz speed',
'vert vel inconsistent']], annotate=False, pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# filter reported accuracy
data_plot = CheckFlagsPlot(
status_time, estimator_status, [['pos_horiz_accuracy', 'pos_vert_accuracy']],
x_label='time (sec)', y_labels=['accuracy (m)'], plot_title='Reported Accuracy',
legend=[['horizontal', 'vertical']], annotate=False, pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# Plot the EKF IMU vibration metrics
scaled_estimator_status = {'vibe[0]': 1000. * estimator_status['vibe[0]'],
'vibe[1]': 1000. * estimator_status['vibe[1]'],
'vibe[2]': estimator_status['vibe[2]']
}
data_plot = CheckFlagsPlot(
status_time, scaled_estimator_status, [['vibe[0]'], ['vibe[1]'], ['vibe[2]']],
x_label='time (sec)', y_labels=['Del Ang Coning (mrad)', 'HF Del Ang (mrad)',
'HF Del Vel (m/s)'], plot_title='IMU Vibration Metrics',
pdf_handle=pdf_pages, annotate=True)
data_plot.save()
data_plot.close()
# Plot the EKF output observer tracking errors
scaled_innovations = {
'output_tracking_error[0]': 1000. * ekf2_innovations['output_tracking_error[0]'],
'output_tracking_error[1]': ekf2_innovations['output_tracking_error[1]'],
'output_tracking_error[2]': ekf2_innovations['output_tracking_error[2]']
}
data_plot = CheckFlagsPlot(
1e-6 * ekf2_innovations['timestamp'], scaled_innovations,
[['output_tracking_error[0]'], ['output_tracking_error[1]'],
['output_tracking_error[2]']], x_label='time (sec)',
y_labels=['angles (mrad)', 'velocity (m/s)', 'position (m)'],
plot_title='Output Observer Tracking Error Magnitudes',
pdf_handle=pdf_pages, annotate=True)
data_plot.save()
data_plot.close()
# Plot the delta angle bias estimates
data_plot = CheckFlagsPlot(
1e-6 * estimator_status['timestamp'], estimator_status,
[['states[10]'], ['states[11]'], ['states[12]']],
x_label='time (sec)', y_labels=['X (rad)', 'Y (rad)', 'Z (rad)'],
plot_title='Delta Angle Bias Estimates', annotate=False, pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# Plot the delta velocity bias estimates
data_plot = CheckFlagsPlot(
1e-6 * estimator_status['timestamp'], estimator_status,
[['states[13]'], ['states[14]'], ['states[15]']],
x_label='time (sec)', y_labels=['X (m/s)', 'Y (m/s)', 'Z (m/s)'],
plot_title='Delta Velocity Bias Estimates', annotate=False, pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# Plot the earth frame magnetic field estimates
declination, field_strength, inclination = magnetic_field_estimates_from_status(
estimator_status)
data_plot = CheckFlagsPlot(
1e-6 * estimator_status['timestamp'],
{'strength': field_strength, 'declination': declination, 'inclination': inclination},
[['declination'], ['inclination'], ['strength']],
x_label='time (sec)', y_labels=['declination (deg)', 'inclination (deg)',
'strength (Gauss)'],
plot_title='Earth Magnetic Field Estimates', annotate=False,
pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# Plot the body frame magnetic field estimates
data_plot = CheckFlagsPlot(
1e-6 * estimator_status['timestamp'], estimator_status,
[['states[19]'], ['states[20]'], ['states[21]']],
x_label='time (sec)', y_labels=['X (Gauss)', 'Y (Gauss)', 'Z (Gauss)'],
plot_title='Magnetometer Bias Estimates', annotate=False, pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
# Plot the EKF wind estimates
data_plot = CheckFlagsPlot(
1e-6 * estimator_status['timestamp'], estimator_status,
[['states[22]'], ['states[23]']], x_label='time (sec)',
y_labels=['North (m/s)', 'East (m/s)'], plot_title='Wind Velocity Estimates',
annotate=False, pdf_handle=pdf_pages)
data_plot.save()
data_plot.close()
def process_logdata_ekf(filename: str,
check_level_dict_filename: str,
check_table_filename: str,
plot: bool = True,
sensor_safety_margins: bool = True):
## load the log and extract the necessary data for the analyses
try:
ulog = ULog(filename)
except:
raise PreconditionError('could not open {:s}'.format(filename))
ekf_instances = 1
try:
estimator_selector_status = ulog.get_dataset(
'estimator_selector_status', ).data
print('found estimator_selector_status (multi-ekf) data')
for instances_available in estimator_selector_status[
'instances_available']:
if instances_available > ekf_instances:
ekf_instances = instances_available
print(ekf_instances, 'ekf instances')
except:
print('could not find estimator_selector_status data')
try:
# get the dictionary of fail and warning test thresholds from a csv file
with open(check_level_dict_filename, 'r') as file:
reader = csv.DictReader(file)
check_levels = {
row['check_id']: float(row['threshold'])
for row in reader
}
print('Using test criteria loaded from {:s}'.format(
check_level_dict_filename))
except:
raise PreconditionError(
'could not find {:s}'.format(check_level_dict_filename))
in_air_margin = 5.0 if sensor_safety_margins else 0.0
for multi_instance in range(ekf_instances):
print('\nestimator instance:', multi_instance)
# perform the ekf analysis
master_status, check_status, metrics, airtime_info = analyse_ekf(
ulog,
check_levels,
multi_instance,
red_thresh=1.0,
amb_thresh=0.5,
min_flight_duration_seconds=5.0,
in_air_margin_seconds=in_air_margin)
test_results = create_results_table(check_table_filename,
master_status, check_status,
metrics, airtime_info)
# write metadata to a .csv file
with open('{:s}-{:d}.mdat.csv'.format(filename, multi_instance),
"w") as file:
file.write("name,value,description\n")
# loop through the test results dictionary and write each entry on a separate row, with data comma separated
# save data in alphabetical order
key_list = list(test_results.keys())
key_list.sort()
for key in key_list:
file.write(key + "," + str(test_results[key][0]) + "," +
test_results[key][1] + "\n")
print('Test results written to {:s}-{:d}.mdat.csv'.format(
filename, multi_instance))
if plot:
create_pdf_report(ulog, multi_instance,
'{:s}-{:d}.pdf'.format(filename, multi_instance))
print('Plots saved to {:s}-{:d}.pdf'.format(
filename, multi_instance))
return test_results
def getVehicleLandingStatus(ulog: ULog) -> pd.DataFrame:
vehicle_land_detected = ulog.get_dataset("vehicle_land_detected").data
land = pd.DataFrame({'timestamp': vehicle_land_detected['timestamp'],
'sensor' : 'landed',
'landed': vehicle_land_detected["landed"]})
return land