the "Last Person to blah blah blah" thread is now old enough to vote (in the US)

That confirms that they are likely doing ( digital ) inertial integration of angle increments, and thus need a rate calibration ( angle per second ) on differential accelerometer outputs.

That means that prior to the first ( blade-breaking ) crash the rotational aspect of the inertial platform wasn't properly rate adjusted with respect to that all-too-important local gravity field. The error accumulated to the threshold when it became a grossly incorrect input to the ( homeostatic ) control loop ie. gravity never sleeps and always points to actual down. Your description indicates this was an issue on several axes.

At a guess it is possible/probable that the craft has the bare/minimal set of accelerometers. This rotation rate business can be solved by having more on board but suitably placed & analysed.

{ There is a related issue with 'spinning top' type gyros where there is a just sufficient number of gimbal axes ( if you go to four then it doesn't 'lock' and drift effects may be suppressed ). There is a ( semi- ) humorous story* upon gimbal lock with Apollo 11 to the effect that it is difficult to have a fourth one delivered when in space. }

The essential plan is to first solve for very accurately for 'down' ( a vector ) by over-specified inputs ( more accelerometers ), then use that down vector in a further lot of spin manoeuvres ( not about the vertical axis ) to count full rotations against/versus the rate output from the accelerometers. To be purer you can group accelerometers in pairs and use a 'difference of differences' to get higher order corrections if you like. But I have no doubt just doubled/tripled/quadrupled the cost of the device .... and, to my knowledge, you are intending future use as a strike platform ! :-)))))

Who would probably say no anyway ..... not because of you personally or your particular plans, but simply that if they say yes to you that might set a precedent. I would admit to some sympathy with that view having been a light plane pilot. It would likely be a danger to the human carrying craft but not so for the obvious reason of actual collision, but unwise responses from distracted pilots. For me I would let the drone 'eat propeller' as it were. However that's a bit cavalier and it is very expensive to have one replaced and/or re-machined. Bird strikes - even squashier biological matter - have been famously known to bring down a relative Goliath. Now I'm not at all saying you are intending to be a nuisance etc, but you could well be viewed as one regardless. Aviators are very propriety towards 'their' airspace ! :-0

Cheers. Mike.

* Nothing actually 'locks'. A series of manouevres can lead to a chance alignment of two or more gimbal axes, in which case from then on you lose a degree of freedom ie. a 3D gyro functions as a 2D one. You have to reset the device and then take some physical bearings etc to realign to the real universe. A fourth gimbal is actively tracked and monitored, keeping it away from chance alignment with the other three. If any two of the inner gimbals align, then the fourth can recover their non-degenerate alignment without reset etc. While quite unlikely : if the three inner ring axes coincide simultaneously then you're stuffed without reset. I think the idea here is that other sequelae from whatever generated such a scenario indicate platform loss being the least of your problems ie. way out of control spinning.

Well of course we all knew that didn't we folks? And written in Basic Engly Twenty Fido, oh yes, with deep joy.

Believe I have completed the coding logic for the Pi2 tank/crawler. Trying to combine forward/reverse/left/right into one joystick on a PS3 game controller did not work due to stick sensitivity, i.e., it is hard to move the stick forward without deflecting the stick left or right causing unwanted turning motion. I decided that the left stick would be forward/backward and the right stick left/right turn and the buttons on the front to control the pan/tilt logic on the camera mount. Seems to have worked out well. At least I do not need 10 opposable fingers to reach this button configuration unlike my first attempt. Here is a video I put together using Linux's kdenlive video editing package (I need to spend more time with this package). I attempted a split screen mode where two videos were cobined in sync with each other, i.e., the onboard camera and a 2nd camera. There is no sound because the motors are distracting. The drive motors have metallic gear transmissions built in and the servos also feature metal gears. The servos are a bit "too much" for the weight of the camera, but I used what I had. The video can be seen here - a short tour of two crunchers.

Now when the printer arrives I will be able to print a suitable body for the crawler.

At a minimum you need two reference directions, preferably close to orthogonal eg. down, plus one in the horizontal plane ( magnetic north here ). You still need some rate knowledge/calibration ( something per angle ) to get intermediate vectors.

Cool.

Yep. Civilian GPS gives you rough ( enough ? ) location but really no orientation information. You could hover for half an hour to improve the fix but why would you ?

The Gemini/Apollo craft joysticks had several modes. The question is whether that control input selects either position or orientation, or the rate change of those. All had spring return to neutral position plus detente ( meaning lock/inactivate ). Do you know what the first two executed commands were upon landing the Eagle on the Moon ? Shutting off the descent engine and controller to detente. Having succeeded they didn't want to inadvertently (a) jump up off the surface nor (b) flip the lander over because they left the attitude adjustment active.

I was looking at those lines ( aplenty ! ) and those would be an expensive encounter. Even flying between two phase wires without touching could result in circuit completion, by the presentation of a lower resistance pathway. :-)

It would almost certainly be the standard airspace definition for such a facility ie. cylindrical shape with ground datum as the lower boundary within the 5 mile radius. Large airports may have an inverted cone definition ie. the lower boundary increases linearly with distance from the ( ground level ) cone tip/centre. But there are variants/combo galore and typically many special local rulings. Beware*.

{ Tullamarine or Melbourne Airport has an airspace which is the union of a central cylinder and inverted cone, both having a common vertical axis of symmetry. But the tip of the cone is underground. It's a gradual cone eg. some 25km away you still have to stay under 2000ft, going to 8000ft in another 20km. But that 8000ft is then a lower bound for the regional airspace ie. large body commercial IFR flights only. Since those extents covered one side of my training area to the other then it was drilled into my brain. Alas the transponder tells all, where the central radar facility 'tweets' my plane once per sweep and it responds by saying Who I Am. Darn. The flight school never got a 'please explain' letter on my account, in any event excuses are neither tolerated nor are useful ! In theory you could pop up and duck down a few metres into the cone between sweeps and remain not in breach but I think that would be labelled as a sporting move. :-)) }

It was Michael Collins in 11 that politely brought the topic up.

Not having a four gimbal gyro was a definite design decision made. Adding a fourth one in certainly greatly increases the complexity ( read : usability and reliability ). That added outer ring has to be independently driven to keep it away from co-linearity with the axes of any of the inner three ( while tracking & recording said adjustments ). It's not actually used moment to moment for orientation, but available to avoid full reset and calibration to external markers ie. a lock state on the inner three.

Cheers, Mike.

* As Gary points out this is the pilot's ( your ) responsibility, ignorance is no defence etc. From the right source eg. the local pilot's shop there will be maps to define in detail the height vs position information that you need. I don't know what happens in the USA but DownUnda there is a potential penalty of $10K applicable ( plus loss of licensing etc ). The official sensitivity to incursions upon controlled airspace has gone exponential in recent years, for obvious reasons. The benefit of any doubt is no longer on offer. We don't yet have an F-18, or three, on CAP.