DJI Mavic, Air and Mini Drones
Friendly, Helpful & Knowledgeable Community
Join Us Now

Lost GPS = No RTH, why? Not an actual incident, just wondering.

OA works without GPS. It doesn't need GPS. I can fly in my house and around the yard without GPS. It uses the obstacles in such a way, that it knows where it is relative to the obstacles. so it stays put or can move a little bit, if nothing is in its way. It will just hover if blocked by an obstcle.
Have you tested this without a VPS lock, meaning in full ATTI?
 
  • Like
Reactions: RodPad
Have you tested this without a VPS lock, meaning in full ATTI?
Yes. If i fly around in full atti, before I get my GPS locked, the drone flies fine, but will drift if i let go of the controls, but if it drifts towards an obstacle, it stops and hovers. I can let it drift all over the place when low to the ground. It just drifts and stops when near obstacles. If in atti and I fly straight toward an obstacle, it will go around it. GPS only keeps it from drifting when there are no obstacles seen by the VPS. When surrounded by obstacles, the VPS makes the drone stop moving completely and it hovers without drifting. It doesn't move, because the VPS doesn't see any direction that an obstacle wouldn't be avoided.
 
That's not actually quite true. The drone has acceleration and rotation sensors. There's no fundamental reason why it couldn't attempt to navigate back to the home point using dead reckoning based on its sensor inputs, in much the same way that INS systems used to guide aircraft across the Atlantic way before the days of GPS.

There's a very good reason: wind.
 
It wouldn't be nearly as accurate as GPS, of course, but it's better than just landing where ever it happens to be.

Boy, that's a very debatable assertion, that I don't have the patience to have. We'll just have to agree to disagree.
 
OA works without GPS. It doesn't need GPS. I can fly in my house and around the yard without GPS. It uses the obstacles in such a way, that it knows where it is relative to the obstacles. so it stays put or can move a little bit, if nothing is in its way. It will just hover if blocked by an obstcle.

OA relies of detecting convergence/divergence of objects in the view of two cameras. All that it requires is sufficient lighting.

VPS only requires one camera to implement optical flow algorithms.
 
  • Like
Reactions: StevenBrodsky
Dead Reckoning can't be compared with GPS positioning... it doesn't even come close.
I will point out that for decades commercial aviation used dead reckoning in the form of Inertial Naviation Systems to maintain aircraft separation over the oceans where there was no radar or radio communication with Air Traffic Control. The gyro and accelerometer chips in modern smart phones and drones are the modern-day equivalent of this technology. It works even in windy conditions, because the accelerometers detect wind-induced movements and the gyros inform the drone of which direction it's facing.

No, relying on those sensors is not as accurate as GPS over longer distances, but the drone is dependent on them for basic attitude and position-keeping because they're far more accurate over short times and distances than GPS is.

The Wikipedia article on inertial navigation indicates that GPS can be relied upon for an accuracy of about 9 metres, while inertial navigation can fix a position to within around 650 metres after an hour of flight. That implies an accuracy of around 65 metres (around 200 feet) after 6 minutes of flight.

I don't know about you, but getting the drone back within 200 feet of the launch point with the possibility of reestablishing communications and control sure seems safer to me than just having it land at some random, unknown location. The worst case is that it dead-reckons somewhere and ends up landing in a random spot anyway, which is really no worse than if it hadn't at least tried to return.

It just seems like a bit of a missed opportunity to me.
 
I don't know about you, but getting the drone back within 200 feet of the launch point with the possibility of reestablishing communications and control sure seems safer to me than just having it land at some random, unknown location. The worst case is that it dead-reckons somewhere and ends up landing in a random spot anyway, which is really no worse than if it hadn't at least tried to return.
How often does a drone lose GPS reception when out flying in the open?
Not often enough that it's worth worrying about.
And if it ever happened, you can still fly the drone home.
Not having RTH isn't an insurmountable problem.

 
How often does a drone lose GPS reception when out flying in the open?
Not often enough that it's worth worrying about.
And if it ever happened, you can still fly the drone home.
For sure, if you've still in communication with the drone. But where things get interesting is when whatever knocked out communication also knocked out the GPS reception.

It's a niche scenario, no doubt.
 
In addition, I am pretty sure I have seen Sar104 say the drone knows its speed, from the angle of tilt, although, in the absence of GPS, I think that would be airspeed and not ground speed.
It knows its heading and airspeed from the compass and tilt angle, but it can do better than that in terms of navigating without GNSS positioning. The accelerometers and rate gyros alone theoretically provide a full 3-D position and velocity solution, spoiled only by sensor drift, which is why it needs barometer, compass and GNSS positioning to correct for that drift via the sensor fusion (Kalman or similar) algorithms. MEMS devices generally don't perform as well as aviation-grade inertial units, exhibiting far more drift.

The inertial solution is relatively straightforward, and is described in section 2 of this paper:


As it happens, I originally derived that solution for precisely this purpose – to compare the purely inertial solution from DJI DAT data to the full sensor fusion solution, to see how long the inertial solution took to diverge significantly from reality. As you might expect, that depends somewhat on how well the inertial sensors are calibrated, as well as flight time. In some cases it would get the aircraft close to home, while in others, it wouldn't.

In any case, DJI control software doesn't attempt to make that calculation.
 
The Wikipedia article on inertial navigation indicates that GPS can be relied upon for an accuracy of about 9 metres, while inertial navigation can fix a position to within around 650 metres after an hour of flight. That implies an accuracy of around 65 metres (around 200 feet) after 6 minutes of flight.
That is with an INS that costs a couple hundred grand at least. Stuff we get in consumer devices is nowhere near accurate and low drift enough, GPS being available as a 2nd source to include in the sensor fusion and keep the rest in check is what's made the current stuff even possible.

Also an airplane can work from the assumption that it is always flying forward and is inherently stable, a multicopter is neither.
 
Last edited:
I don't know about you, but getting the drone back within 200 feet of the launch point with the possibility of reestablishing communications and control sure seems safer to me than just having it land at some random, unknown location. The worst case is that it dead-reckons somewhere and ends up landing in a random spot anyway, which is really no worse than if it hadn't at least tried to return.

It just seems like a bit of a missed opportunity to me.
This was what i was thinking. Instead of the drone landing, have it attempt to proceed in the direction of it's last known home point. Also, RTH should be set to go to just under 400ft on the way back, the GPS satellites may be picked up again on the way back, reestablishing the GPS system's location. So, during the time the drone is trying to fly back and increasing it's altitude GPS could be reestablished and all would be good in the world.
 
Last edited:
That is with an INS that costs a couple hundred grand at least.
Price alone is not necessarily indicative of performance, especially when you're comparing products across decades. In the days of aviation INS units, computers used to cost a lot too. Advances in silicon fabrication have changed all of that.

While the sensors onboard a drone aren't the ultimate in terms of accuracy, they're enough to get it at least heading in the right direction. And I still think that improves the odds and is better than nothing. It's easy enough to limit the non-GPS RTH flight time in order to prevent the drone from going too far afield.

I'm speaking in terms of principles here. But it would be really interesting to see an example of non-GPS dead-reckoning navigation to get an idea of just what it would or wouldn't be capable of. You'd think that a sufficiently savvy person would be able to cook something like a phone app up as a demonstration.
 
Price alone is not necessarily indicative of performance, especially when you're comparing products across decades. In the days of aviation INS units, computers used to cost a lot too. Advances in silicon fabrication have changed all of that.

While the sensors onboard a drone aren't the ultimate in terms of accuracy, they're enough to get it at least heading in the right direction. And I still think that improves the odds and is better than nothing. It's easy enough to limit the non-GPS RTH flight time in order to prevent the drone from going too far afield.

I'm speaking in terms of principles here. But it would be really interesting to see an example of non-GPS dead-reckoning navigation to get an idea of just what it would or wouldn't be capable of. You'd think that a sufficiently savvy person would be able to cook something like a phone app up as a demonstration.
I don't know about that. I just don't think someone will take the time and effort. And, i think they would have to modify firmware, which would be difficult. they would have to disable the RTH landing with GPS loss protocol. and change it to a RTH without landing protocol with no GPS. A daunting programming effort.
 
I don't know about that. I just don't think someone will take the time and effort. And, i think they would have to modify firmware, which would be difficult.
To be clear, when I mused about an app to demonstrate non-GPS dead reckoning I was talking about something that would run on a phone, not on the drone itself. Just a "proof of concept" to show what kind of accuracy one might expect from it, based on the idea that both phones and drones contain similar sensors.

Just more idle daydreaming on my part.
 
  • Like
Reactions: StevenBrodsky
To be clear, when I mused about an app to demonstrate non-GPS dead reckoning I was talking about something that would run on a phone, not on the drone itself. Just a "proof of concept" to show what kind of accuracy one might expect from it, based on the idea that both phones and drones contain similar sensors.

Just more idle daydreaming on my part.

There is no need to do that. You can take actual flight data and easily compute the inertial navigation solution to compare with the actual position and velocity of the aircraft. That directly shows the accuracy that you could expect from the sensors in the aircraft.
 
I will point out that for decades commercial aviation used dead reckoning in the form of Inertial Naviation Systems to maintain aircraft separation over the oceans where there was no radar or radio communication with Air Traffic Control. The gyro and accelerometer chips in modern smart phones and drones are the modern-day equivalent of this technology.

And that's the problem... they're not. Not even close.
 
This was what i was thinking. Instead of the drone landing, have it attempt to proceed in the direction of it's last known home point. Also, RTH should be set to go to just under 400ft on the way back,

Impossible. There's no way to sense AGL.

This points out the risk of setting RTH altitude, relative to takeoff altitude, close to 400ft. Depending on the terrain traversed on the way back, it's easy to exceed 400ft AGL, violate the rules, and even intrude into manned aircraft space.

I set my RTH between 150-200' depending on the flight.
 
I don't know about you, but getting the drone back within 200 feet of the launch point with the possibility of reestablishing communications and control sure seems safer to me than just having it land at some random, unknown location.

Unless I see it, 200ft radius most of the time would not be helpful. It includes homes, rugged wilderness, and all sorts of areas that are not easily searched, in some cases impossible.

I've spent well over an hour searching for my Avata in a grassy field, and only found it, about 80' from where it took off, because you can make it beep!
 
Unless I see it, 200ft radius most of the time would not be helpful. It includes homes, rugged wilderness, and all sorts of areas that are not easily searched, in some cases impossible.
The point was to get it back within radio communication range so that you could regain control. You don't need pinpoint accuracy to attempt that.
 
Last edited:
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
130,891
Messages
1,557,574
Members
159,902
Latest member
Modoc