Hi there, I am trying to do some high g soaring and find that gps can’t take this (>6g) and apparently this is fundamental limitation. I think I can get the glider to pass repeatedly past a point and using some beam break measurements or similar know its position each time. Is it possible to push this position to the glider each time such that IMU drift doesn’t become a problem? I’m expecting 5-20seconds between returns. Could this work? Any ideas of how to update the location easily? Any other great ideas for flying precision loops (circa +/- 5inch) at high g’s. Thanks, Craig
What GPS unit do you have?
Apparently the F9P units have much better performance in turns than the regular M8 and similar units.
Check Tridges scripted aerobatics posts - this is exactly the sort of thing he’s been working on - well maybe not that high G forces, but certainly repeated turns and inverted flight.
I looked up the Zed F9P specs 
Operational limits2
Dynamics ≤ 4 g
Altitude 80,000 m
Velocity 500 m/s
but maybe it’s still your best hope??
Hi Shawn, thanks for the suggestion but I’ve been pretty deep down the gps vs g’s rabbit hole. It appears to but fundamentally difficult to produce a gps receiver that can handle high accelerations. The whole point of my post is that for this reason gps is not the solution. I’m keen to understand alternatives as described in my original question.
Thanks anyway
Craig.
The problem is probably related to flying inverted and high g maneuvering UAV being indistinguishable from a maneuvering missile 
.
1 Like
Maybe 
I only want my “missile” to fly in circles going not more that about 20seconds from the starting position - not much use for missile based naughtyness.
I have heard the high g = missile comment before, but I don’t think that is it.
Surface to ground missiles don’t do high g manoeuvres typically (except maybe as evasive manoeuvres and that could be done periodically allowing a new GPS lock after). Air to air and surface to air do high g manoeuvres but they use radar or IR to track the fasty movy thingys.
From what I can find the high g limitation is to do with high acceleration making the Karman filtering difficult - maybe making the comparison to timing a challenge as changing doppler and signal edge timing between pulses from different satellites is the issue. We’ve spoken to some GPS suppliers and gone down this rabbit hole a fair way.
IMHO your best bet would be to use camera based tracking with markers. You could try pretending to be vision based tracker and send updates only in a certain place though I would expect EKF to strongly dislike the idea.IIRC recently support was added for soft setting EKF position estimate for when dead reckoning starts reckoning wrong 
.
Higher allowed acceleration increases likely solution space making the problem harder. If you were running Integrated Inertial GNSS unit you could probably maintain high quality lock well above 4g by knowing where you should be and using that in acquiring updated GNSS fixes. Unfortunately that would require either getting your hands on a dedicated GNSS Inertial Unit or designing GNSS receiver and writing a complex code for a decent CPU to churn out the math fast enough.
I bet UBlox or their partners have such solutions but they are restricted to military government contractors and approved sales to allies for obvious reasons.
1 Like
Check with @tridge about how the EKF copes during these periods of GPS unreliability.
There could be options for a flight controller with better gyros and accels, maybe not needed.
Where exactly are we getting the idea that GPS is ill-suited to an accelerated environment? And why are we making bizarre and wildly inaccurate comparisons to missiles?
I’ve flown manned and unmanned aircraft that use ancient GPS technology by comparison in the same (or higher G) regimes without issue. In fact, recently I flew a flight test event where we spiked the G to 10.8 and sustained 9 through 180 degrees of heading change with no loss of positional SA using what amounts to a well filtered Neo-6M (more details are not possible due to strict customer agreement).
While examining the EKF as Shawn suggests is an excellent first step, I’d also suggest examining your GPS antenna quality and placement, along with a better receiver (like the F9P or inertially integrated models suggested above). I don’t think the accelerated turns are doing much to affect the GPS signal, rather, antenna masking is much more likely the issue, coupled with inertial drift.
Your goal of +/-5 inches in all planes is extremely lofty. Without reliable RTK precision or a marker/beacon based system as suggested, it’s likely impossible.
In all cases, I think you probably want to use the highest quality IMU that will fit in your craft. Have a look at the external systems here (at least one of which is integrated GPS/INS):
https://ardupilot.org/plane/docs/common-external-ahrs.html
1 Like