Logging of mavlinks error - LPE GPS fault

David_Kraus · November 1, 2016, 9:30am

Hi
is possible to access error messages somewhere in log?
In Qgroundcontrol, during the flight, error mesage apper ( [lpe] gps fault 0 0 219 0 0 0, or another numbers) but in log I did not see any problems with GPS. Is possible to see error mesages in logs, to be able precisely found moment when error happens, and determine cause of problems.

log is here
http://logs.uaventure.com/view/usRNkkv5SxxFkSM7gvTcQc

Flight with pixracer and pihawk GPS module, stable PX4 FW, small 250 race copter

thanks for help
David

David_Kraus · November 1, 2016, 9:39am

Now I found in QGC history LPE GPS error came up in 9:36:10 and 9:36:54
What can cause this type of fault? log shows that all the time GPS have 9-10 sat, what shut be enough

Nicolas · November 1, 2016, 10:38am

The message you’re seeing is generated here:

github.com

PX4/Firmware/blob/7c2798269cc360ab4a39532fd2cc351fdf141faa/src/modules/local_position_estimator/sensors/gps.cpp#L192


	}


	Matrix<float, n_y_gps, n_y_gps> S_I = inv<float, 6>(C * _P * C.transpose() + R);


	// fault detection
	float beta = (r.transpose() * (S_I * r))(0, 0);


	if (beta > BETA_TABLE[n_y_gps]) {
		if (_gpsFault < FAULT_MINOR) {
			if (beta > 3.0f * BETA_TABLE[n_y_gps]) {
				mavlink_log_critical(&mavlink_log_pub, "[lpe] gps fault %3g %3g %3g %3g %3g %3g",
						     double(r(0)*r(0) / S_I(0, 0)),  double(r(1)*r(1) / S_I(1, 1)), double(r(2)*r(2) / S_I(2, 2)),
						     double(r(3)*r(3) / S_I(3, 3)),  double(r(4)*r(4) / S_I(4, 4)), double(r(5)*r(5) / S_I(5, 5)));
			}


			_gpsFault = FAULT_MINOR;
		}


	} else if (_gpsFault) {
		_gpsFault = FAULT_NONE;
		//mavlink_and_console_log_info(&mavlink_log_pub, "[lpe] GPS OK");

A GPS fault happens when the GPS measurements jump more than the position uncertainy/GPS measurement uncertainty allow. This usually happens in poor GPS conditions, such as when a siginificant part of the sky is occluded or in scenarios with severe multipath effects.

The health of the sensors is published to the estimator_status topic here:

github.com

PX4/Firmware/blob/43d8e5710c063269f9e53d3f8bf4f4cf4704a6cf/src/modules/local_position_estimator/BlockLocalPositionEstimator.cpp#L769




	for (int i = 0; i < n_x; i++) {
		_pub_est_status.get().states[i] = _x(i);
		_pub_est_status.get().covariances[i] = _P(i, i);
	}


	_pub_est_status.get().n_states = n_x;
	_pub_est_status.get().nan_flags = 0;
	_pub_est_status.get().health_flags =
		((_baroFault > FAULT_NONE) << SENSOR_BARO)
		+ ((_gpsFault > FAULT_NONE) << SENSOR_GPS)
		+ ((_lidarFault > FAULT_NONE) << SENSOR_LIDAR)
		+ ((_flowFault > FAULT_NONE) << SENSOR_FLOW)
		+ ((_sonarFault > FAULT_NONE) << SENSOR_SONAR)
		+ ((_visionFault > FAULT_NONE) << SENSOR_VISION)
		+ ((_mocapFault > FAULT_NONE) << SENSOR_MOCAP);
	_pub_est_status.get().timeout_flags =
		(_baroInitialized << SENSOR_BARO)
		+ (_gpsInitialized << SENSOR_GPS)
		+ (_flowInitialized << SENSOR_FLOW)
		+ (_lidarInitialized << SENSOR_LIDAR)

You will see this in the log as EST0.fault_flags.

David_Kraus · November 1, 2016, 12:07pm

thanks for answer
I flight on the free space, so I will try to change GPS position on copter, if it helps.
Will be nice to have some graph in logmuncher, where will be shown if some sensor or application does not failing.
Because it will be more difficult to find problem without connected QGroundControl
David

David_Kraus · November 2, 2016, 10:33am

I try some more fligts, I move GPS to by above batery.
Grounstation continue with warning of GPS nosy and LPE GPS error. But from logs are not visible any problems.

There are logs, I try two GPS modules
http://logs.uaventure.com/view/uB3Crpof8arLdwDXcKzJFb
http://logs.uaventure.com/view/WfFE7go4VQ2fLh9MXHXmnB

The are clerly visible, when is system switch to the position mode, copter start diverge from desired position.

I try to switch to EKF2 here is log
http://logs.uaventure.com/view/HWKKrTyuyAfNiJaySyX3nh
jumps in the position visible on the map are not real(copter stay only close to start point), estimator have very big error when position mode is on.

System is reporting that is something wrong with GPS, but Im at open field, GPS is on the hieghes position on copter and I try two diferent modules, so Im out of ideas what I can try.

I see magnetometer measurements are very nosy, because I using internal mag on pixracer, only few mm above power board (ASCP4). I can try connect external mag from GPS module. Is it possible that mags make kalman gains messy?

Have someone any idea what can be wrong? Or what can I try?
Thaks
David

David_Kraus · November 2, 2016, 10:35am

here is picture of position gps

David_Kraus · November 2, 2016, 6:53pm

I try another flights with external MAG, and warning from WGC keep coming, but flight performance were much better

http://logs.uaventure.com/view/A2oBLRoXP2Pf7ht8hFoYmE

I will perform more flights tomorrow, if is all OK