Welcome to InspirePilots.com

Join the leading DJI Inspire community for free!

Understanding the Inspire's sensors and what they do....

Discussion in 'Inspire 1 Discussion' started by neoben, Jul 6, 2015.

  1. neoben

    Joined:
    Jul 23, 2014
    Messages:
    92
    Likes Received:
    18
    Howdy...

    I think there's a lot of misunderstanding about what the various sensors 'do' on the DJI Quads (maybe by me), which makes people say things like 'I had a flyaway... my compass was off'.... This makes no sense to me, but maybe I'm way off base myself... So, to spur the debate... Here's what I understand...

    There are basically 4 sensors (not including the vision system) in the Inspire... IMU, Compass, barometer, and GPS.

    The IMU keeps the bird level (which is why there is drift in Atti mode), and can also detect acceleration in various directions.

    The compass is directional, which will keep the bird point the right direction, and flying in a straight line.

    The GPS is positional, and is intended to keep the bird in the same spot, barring user input on the joystick.

    The barometric sensor is designed to (along with the GPS) maintain a constant altitude (barring user input).

    So, if you're seeing the Inspire spin slowly by itself, or not fly in a straight line... that's a compass issue...

    ... but I don't think a flyaway is going to be cause by compass problems, as it's not a positional system. More likely, it's either a faulty GPS reading (the bird is trying to get somewhere else) which can be solved by switching to atti mode, or it's an IMU problem, which is worse, because if the Inspire doesn't know what level is, it can fly off at any speed and in any direction, and you can't really fix this by switching to atti mode.

    You can tell if the fly away was caused by GPS or IMU if you have video recording, because with a GPS problem, the camera will stay level, but with an IMU issue, the camera will tilt to whatever the IMU thinks 'level is...

    So, to summarize my ridiculously long post..

    1) Yaw or spin issues while stationary or in straight flight - Compass problem

    2) Altitude holding problems (rare) - Barometric Sensor problem

    3) Fly away or loss of control with level camera - GPS issue

    4) Fly away with tilted camera - IMU issue.

    also, there could easily be combinations of the two...

    Lastly... the Inspire 'should' be able to fly in a very stable manner with a poor GPS signal and poor compass calibration... In fact, in previous quads I've owned, you can turn the GPS and compass off all together with no ill effects... However, the IMU is the 'critical' component and without a stable IMU sensor, the bird will be fighting you at best, and uncontrollable at worse...

    I welcome your thoughts and expansions on my lengthy dissertation...

    Best,

    Ben
     
    Allmay, STClassic01, AERO 1 and 2 others like this.
  2. Luis Morales

    Joined:
    Apr 15, 2015
    Messages:
    270
    Likes Received:
    120
    Location:
    Madrid
    the baro can cause "fly aways" too, as sudden temp changes may vary its reading.
     
  3. The Editor

    The Editor Moderator
    Staff Member

    Joined:
    Aug 7, 2013
    Messages:
    6,723
    Likes Received:
    3,879
    Basically correct except that GPS is not used for altitude/height at all as it is way too inaccurate for this (GPS can have +/- 400ft error when it comes to altitude). The barometric sensor has a resolution of 10cm so that is used exclusively for height data.

    As well as a three dimensional accelerometer, the IMU also contains a MEMS gyro unit that will measure rotational thrust so it is not just the magnetometer that is producing rotational information.

    A poorly calibrated/erroneous compass can most definitely cause a 'fly-away' and is the single most common cause of this phenomenon.
    When the compass data is out/skewed it causes a conflict with the GPS in calculating true/magnetic North. This conflicting information is constantly at odds with what the GPS is seeing as North and the flight controller will try and correct the position of the aircraft causing what is commonly known as The Toilet Bowl Effect or 'TBE'. This error systematically increases and the circles get larger and larger with an experienced pilot unable to correct for the orientation differences and what appears to be non sensical stick input.
     
    Soflms and Ascender like this.
  4. neoben

    Joined:
    Jul 23, 2014
    Messages:
    92
    Likes Received:
    18
    I could see the Baro causing an altitude change, but not a fly away... Maybe a 'fly up'? :)

    Best,

    Ben
     
  5. Luis Morales

    Joined:
    Apr 15, 2015
    Messages:
    270
    Likes Received:
    120
    Location:
    Madrid
    fly ups are also called fly aways, anything that is the machine flying without control is called fly away.

    @The Editor the GPS is used for altitude too, it actually uses both, the accuracy when you have 8/9 sats its pretty good probably 1/3 meters error it usually used in conjunction with the baro, i can imagine the accuracy of 16 sats will be even better.
     
  6. skylabimaging

    Joined:
    Mar 17, 2015
    Messages:
    561
    Likes Received:
    247
    Location:
    Seattle, Washington

    Eh, he is right. It technically does do a fly up or down....and this shouldn't be confused with fly away or loss of control, although it still falls under "brown pant moments." Back in 2011 when DJI released the wkm, it had barometer issues in the imu that caused it to fluctuate in altitude, it wouldn't fly away but it could float up or down based on it's readings...
     
  7. Luis Morales

    Joined:
    Apr 15, 2015
    Messages:
    270
    Likes Received:
    120
    Location:
    Madrid
    the first Ardupilot had the barometer bare with no cover, the result was the sometime depending on your setup direct sun light will hit it and it will just fly away (up full throttle) full throttle means you have cero control.

    back them that was called fly away ;), thought now days we mostly use it for GPS glitches
     
    STClassic01 likes this.
  8. skylabimaging

    Joined:
    Mar 17, 2015
    Messages:
    561
    Likes Received:
    247
    Location:
    Seattle, Washington
    Meh, that's reasonable though, that's ardupilot...
     
    The Editor and Luis Morales like this.
  9. The Editor

    The Editor Moderator
    Staff Member

    Joined:
    Aug 7, 2013
    Messages:
    6,723
    Likes Received:
    3,879
    I am always happy(?) To be proved wrong but my understanding has always been that it is extremely unwise to use GPS for height data for anything aviation. Don't forget that the Inspires height is always relative to zero from its take off point (that effectively resets the baro).
    GPS altitude measures the user (or in our case the Inspires) distance from the center of the satellites orbits. These measurements are referenced to geodetic altitude or ellipsoidal altitude in some GPS equipment. Most equipment manufacturers utilize a mathematical model in the GPS software which roughly approximates the geodetic model of the earth and reference altitude to this model. As with any model, there will be errors as the earth is not a simple mathematical shape to represent. What this means is that if you were flying your Inspire at a few feet above the seashore, and it relied on GPS for its height data you could easily see your altitude as -15 meters (or worse).
    The geodetic model of the earth can have much more than this amount of error at any specific point and you have the GPS error itself to add in. As a result of this combined error, the Inspire could easily see a +/-40 meter error in some spots if it relied on GPS height data. (There were a lot of "errors" used in that sentence - sorry).
    If you have a search around on the internet you will find that GPS altitude inaccuracy is pretty well documented. :(
     
  10. Luis Morales

    Joined:
    Apr 15, 2015
    Messages:
    270
    Likes Received:
    120
    Location:
    Madrid
    took me a while to find the reference in the RCG thread.

    this is a picture of the inspire 1 gps module chip

    [​IMG]
    [​IMG]

    http://www.u-blox.com/images/downloads/Product_Docs/NEO-M8_DataSheet_(UBX-13003366).pdf

    it says it has an accuracy for horizontal position of 2/2.5m that is not too bad for a gps, and as i said before its not that the GPS is used, is that its used with the barometer, both values are used.
    the same can be seen in the ardupilot code and i bet most autopilots do the same.

    you are right that gps altitude data is not used when alone but it is combined with other data to get better results. this can even be tested if your curios, i done it with a pixhawk and a naza, put the controller with no gps and hover, and them do it with the gps you will be surprised on how much it improves.
     
    Ascender likes this.
  11. The Editor

    The Editor Moderator
    Staff Member

    Joined:
    Aug 7, 2013
    Messages:
    6,723
    Likes Received:
    3,879
    Yup - Its a standard Ublox module and you are quoting horizontal accuracy which is not altitude, that would be vertical accuracy (or was that a typo) !;)
     
  12. Auspete

    Joined:
    Apr 1, 2015
    Messages:
    55
    Likes Received:
    25
    For a "layman" I think I learnt something from all this - thanks guys
     
    The Editor and Ascender like this.
  13. STClassic01

    Joined:
    May 20, 2015
    Messages:
    54
    Likes Received:
    8
    Location:
    Cleveland, OH USA
    Naza M was like that. to resolve the issue you would put black electrical tape on the sides to shield the sun.
     
    VFTECH and The Editor like this.
  14. Gregg Smith

    Joined:
    Jul 17, 2015
    Messages:
    94
    Likes Received:
    22
    Location:
    Austin, TX
    Does anyone have info on what good readings are for the sensors? When I open the screen I see all the different numbers but couldn't tell you if they are in tolerance or not. Which would be a good thing to know before flying.
     
  15. Scotflieger

    Joined:
    Mar 29, 2015
    Messages:
    1,244
    Likes Received:
    578
    Location:
    Buckie, Scotland, UK
    Your Gyro/Accelerometer values should be zero (0) or +/-1 and the compass Z and Mod should between 1300 and 1500 as shown below.

    [​IMG]
     
    Larry W. likes this.