This topic provides an overview of the available analyzers. These are all run by default
but you can specify to only run a subset using the -a argument (see
Selecting analyzers).
Each section lists a description of the analyzer, along with possible fails/warnings.
Note
This section is provided for reference only. The log itself will include all this information along with the “severity” of the issue, “series” (where the error occurred) and “evidence” (what caused a fail/warning to be raised).
Autopilots store configuration settings known as ‘parameters’. For proper analysis, logs must contain this parameter information. This test will FAIL if the input does not contain parameter information.
Fail: No parameters present
Pass: Parameters present
Note
No user action can be taken in response to this error.
A UAV often has several estimates of its altitude. This test will FAIL or WARN if the various vehicle’s altitude estimates diverge.
Fail: This altitude estimate differs from the canonical craft altitude
Warn: This altitude estimate differs from the canonical craft altitude
An autopilot checks many aspects of the aircraft’s state before allowing it to be armed - for example, that it has a good GPS fix. This test will FAIL if the craft ever arms when some arming checks are disabled.
Fail: Some of the arming checks were disabled when the craft was armed
The autopilot reports both the craft’s attitude and the attitude the craft believes it should be at. This test will FAIL or WARN if the vehicle’s desired attitudes and achieved attitudes are not within threshold delta values for more than a threshold time.
Fail: Desired attitude not achieved
Warn: Desired attitude not achieved
Warn: Vehicle attitude never set
A UAV often has several estimates of its attitude. This test will FAIL or WARN if the various vehicle’s attitude estimates diverge.
Fail: This attitude estimate differs from the canonical craft attitude.
Warn: This attitude estimate differs from the canonical craft attitude.
Many autopilots are capable of monitoring their own performance. This test will FAIL if problems are detected with the autopilot.
Fail: Severe scheduler overruns
Many autopilots are capable of monitoring their flight batteries.
This test will FAIL if the battery level falls below the battery failsafe threshold level, or if a battery failsafe event is received.
Fail: Battery fell below failsafe threshold
Fail: Battery failsafe event received
Pass: Battery never below failsafe
Compass calibration produces a set of parameters that specify expected compass discrepancies. This test will WARN or FAIL depending on the degree that these compass offset parameters exceed specified thresholds.
Fail: Compass offsets in parameters are out of bounds
Fail: Compass offset parameter set seen/set
Warn: Compass offsets in parameters are out of bounds
Warn: Compass offsets in parameters are zero
Pass: Compass offsets in parameters look reasonable
The strength and direction of the Earth’s magnetic field should be relatively constant and lie within certain thresholds. This test will FAIL or WARN if the compass vector length exceeds the respective threshold. Possible causes include flying near large metal objects.
Fail: Compass Vector Length above threshold
Fail: Compass Vector Length below threshold
Fail: Compass Vector Length delta exceeds threshold
Warn: Compass Vector Length below threshold
Warn: Compass Vector Length above threshold
Crashes are detected both heuristically and by explicit log messages. This test will FAIL if the vehicle appears to crash
Fail: Vehicle evaluated itself as crashed
Fail: Vehicle is past maximum allowed angle and running its motors
Warn: Vehicle’s attitude never updated
Pass: Never crashed
Vehicles typically need to progress through a sequence of arming steps before they can move. This test will FAIL if the craft did not arm.
Fail: The vehicle never armed
Pass: The vehicle armed
Determining whether a vehicle has ever flown in a log is done heuristically based on things like motor speeds. This test will FAIL if the craft did not ever seem to fly.
As evidence the test provides information about the whether the vehicle armed and whether it reaches the servo threshold required to take off.
Fail: The vehicle never seemed to take off
Pass: The vehicle appeared to fly
The Extended Kalman Filter (EKF) has many built-in checks to ensure that it is functioning correctly. This test will FAIL or WARN if EKF variances exceed the respective thresholds, or FAIL if the EKF status flags indicate errors.
For EKF status flag fails, the evidence field provides information about the specific estimates that are incorrect.
Fail: The EKF status report indicates a problem with the EKF
Fail: [variance] exceeds fail threshold
Warn: [variance] exceeds warn threshold
Warn: [variance] was never updated
Warn: EKF flags were never updated
Note
In the list above [variance] is one of: velocity, pos_horiz_variance, pos_vert_variance, compass_variance, terrain_alt_variance.
The accuracy and precision of GPS messages can vary depending on many factors including weather, ionospheric disturbances and number of satellites visible. This test will FAIL if the quality of the GPS information is poor.
The test compares the recorded number of satellites and HDOP (horizontal degree of precision) to threshold values and reports both values as “evidence”.
Fail: No 3D fix was ever acquired
Pass: First 3D GPS Fix Acquired
Gyroscopes sometimes start to register movement where there is none. This test will FAIL or WARN if the any gyroscope’s average acceleration on any axis begins to drift.
Fail: Gyroscope readings differ from first gyroscope
Warn: Gyroscope readings differ from first gyroscope
A UAV often has several estimates of its position. This test will FAIL or WARN if the various vehicle’s position estimates diverge.
Fail: This position estimate differs from the canonical craft position
Warn: This position estimate differs from the canonical craft position
A UAV can self-assess its sensors’ health. This test will FAIL if any sensor is detected as failed.
Fail: The craft’s assessment of its sensors indicate a problem
A log should not end while the vehicle appears to be moving under its own power. This test will FAIL if the vehicle still appears to be moving when the log ends.
Fail: Log ended while craft still flying
Pass: Log truncation not detected
Note
There are several possible causes for a truncated log (including power failure due to brownout, running out of memory for the log file, or failure of the logging sub-system).
The vehicle type is normally automatically detected by dronekit-la from the log. Sometimes the log does not contain sufficient information to make this determination. This test will FAIL if the craft type is never defined.
Fail: No information provided defined what type of vehicle was being analysed
Pass: Vehicle was appropriately defined
Note
Information about the vehicle type/frame allows a much deeper log analysis. This information is typically present in logs, but may be omitted.
If you get this error you should provide the information to the tool using the -m and -f flags as
shown:
./dronekit-la <files> -m copter -f quad
Tip
Solo tlogs do not include the frame and model information!
Description: A UAV often has several estimates of its velocity. This test will FAIL or WARN if the various vehicle’s velocity estimates diverge.
Fail: This velocity estimate differs from the canonical craft velocity
Warn: This velocity estimate differs from the canonical craft velocity