Return to land.....spectacular FAIL

I took my Iris out for its second flight today and had some interesting moments with RTL.

I had a few attempts which were sucessful, one by the Transmitter and another with an Android Tablet.

They worked fine so I had a third attempt.  This time when I used the Transmitter, rather than flying to the launch point my drone started to circle around the launch point.  The circling got wider and further away from the Launch Point until it crashed into a tree.

I also had a scary moment yesterday with RTL by using the Transmitter however I was problem free using the android tablet.

Any ideas how and why I had such an epic fail? 

You need to be a member of diydrones to add comments!

Join diydrones

Email me when people reply –

Replies

  • Interesting

    I had a similar experience this weekend. At our local sports field. Was empty sunday afternoon.

    I programmed a mission and it did it and came back..SUCCESS

    Replace batttery...launch it again...waypoints success. Flicked switch to RTL...and then it jsut took off.

    It was if it had decided home was elsewhere as the direction seemed reasonably intentional.

    I should have ditched it or manually took over but I tried to recover autopilot ..but it kept going further away. (Out of the park)

    After a bit of panic I managed to drop it get it over a clear ground and bring it down

    SO..my question. is there anywhere in the log that records where the pixhawk "thinks" its RTL location is.

    GPS seemed fine as it did the way points correctly. but on RTL it didnt want to go back to the power on location.

    Strange.

    -P

    PS: Noob experience. Next flights will be becoming fully comfortable flying in stabilise etc to manually recover, during a panic situation.

     

  • FWIW, in software simulation I see this when the sensor data provided to the APM gets out of date, etc causing a feedback loop which causes the simulation to fly in circles.  In particular, in simulation, if the data from the simulated frame is not kept up to date, the APM tries to correct by providing PWM data based on bad information.

    Hearing a real world flight where the frame goes into a what I call a "feedback loop flight circle" is scaring me now because I was worried what I was seeing in simulation may not be isolated only to simulation.  I've corrected the simulation (for the most part) in my code base but imo, I have a weird feeling that there is still a very sensitive timing in the APM feedback path that "laggy" simulations just aggravate and real world flights dodge this issue since most of the time, it just runs fast enough.

    I would get the logs to some APM devs on drones-discuss google groups and see if they can provide any insight since they maintain and develop the APM software.

    • Hello

      I am very new and green behind the ears to the world of Drones.  Apart from two initial flights with my Iris I have no experience of how to retrieve any potential information.

      Would it be possible for you to point me in the right direction of people that would find such information useful, and how I could extract this information from my drone?

      Cheers

    • So am I, mostly been spending my time pulling my hair out over the code and only got to fly the Iris once.  =(

      MavLink should be capturing your flight data (extension is tlog) which can be analyzed by a GCS(Ground Control Station).  Youtube has some videos on how to analyze flight data using Mission Planner where the log is and it will parse out the data.

      http://copter.ardupilot.com/wiki/downloading-and-analyzing-data-log...

      https://www.youtube.com/watch?v=oVuFGzFV--Q&list=PLo3xVJ5BJ5039...

      is one such link if you google:

      • "Analyzing MavLink log data"
      • "Analyzing MavLink Mission Planner
      • "Analyzing MavLink Mission Planner youtube"
  • If it started behaving oddly like that I'd put it into stablise immediately and land it. You could connect to Mission Planner and check the settings or reload the Iris parameters.

This reply was deleted.