Improve GPS signal in flight

Hello, I would like to do a vertical mapping of a wall of a building using DroneDeploy’s vertical mapping function, however due to the site’s limitations there’s only limited GPS signal reception (say only 3 or 4 satellites’ signal captured) , which makes the mission tricky. Does anyone know if there are ways to improve the reception in GPS signals or is there any other ways to let the drone / software to locate itself?
Any kind of help would be appreciated, thanks in advance.

1 Like

What is around the area? Unless there is a huge electromagnet or power substation there shouldn’t be any reason why you wouldn’t get the minimum 7 to fly. Have you tried taking off in ATTI mode to let it hover over obstructions and then switch to P and start the mission?

I am working in an area full of skyscrapers, it literally feels like sitting at the bottom of a well surrounded by concrete walls, I would have to fly upwards mostly by hand to the top to capture a GPS signal, and it would very likely be lost again when it comes down, hence I am looking for suggestions. Anyway thanks a lot for the suggestion. :slight_smile:

I understand and have been in that situation myself many times. There’s no way to augment the GPS for guidance so getting it into a manual mode and flying it manually is going to be your only option. Or you just need to find a different place to fly from.

Perhaps try an RTK drone (if available) which can connect to a local, high precision network.

That is unlikely to make any difference. It has nothing to do with accuracies. If you can’t see the satellites you can’t see the satellites. No need for an $8k drone to tell you that.

1 Like

@MichaelL Interested to learn more.

RTK can certainly assist when GPS accuracies are weak when it comes to reconstruction as it improves image tags. More precise drone navigation may be a different matter though.

I guess my point was that this thread has nothing to do with accuracies. If you read the OP he can’t get enough satellites and RTK isn’t going to help with that.

1 Like

Actually I’m having multiple problems with weak GPS signal, so the first one that I’m facing is during the flight where controlling is tricky, the next step would be the photo captured may have a problem of not having GPS data with it, therefore solutions in both aspects are all welcome. I guess RTK would be handy in some way in my case?

With the problems as you have described no it will not help. Understand that RTK runs on the same satellites. There are no additional satellites nor do they work any differently in their communication with the drone. The difference is that a GNSS receiver acting as a fixed base station is introduced which by viewing the same satellites and being set upon a known coordinate allows for corrections to be sent to the drone so that the accuracy of the coordinate as related to the position of the drone goes from a +/-1m down to a 2-3cm accuracy. This requires a different GPS chipset that is equivalent to surveying equipment whereas the native GPS on the drone is more akin to your cell phone.


For the last part of your statement saying that you are not seeing GPS data in images and symptoms of low satellite count that you are seeing could be an indicator of something being wrong with the GPS module on your drone. You will need to pull the logs off the drone and submit them to a service like AirData to diagnose.

What drone?

Maybe I should be clearer on this, during our experiment with less satellite count during flights, there may be a chance of photos having inaccurate location datum with it (e.g. wrong height), thus with DroneDeploy it would create an ‘incorrect’ mapping with those. Pretty sure our drone is working fine, recently done some other regular mapping with it and was perfectly fine.

Assuming you are connecting to RTK over NTRIP with a service that offers local base stations, strong satellite connection may not be required and RTK may be a good solution. But no guarantees as Michael said.

More accurately, “Receiving RTK corrections by connecting to a CORS via NTRIP.” would be the way to describe it. This is a very weak link in this community and quite honestly a major reason why Surveyors do not like drone mappers. DroneDeploy needs to send their team to a class or two so that we can propagate the correct information to a group of users that have no idea what surveying entails or even what GNSS is. Saying something like “a strong satellite connection may not be required”, what does that even mean? Do you know what the factors are that define a “strong satellite connection”? I’m not trying to be ugly but this is a subject that can go bad quickly and this information needs to be represented very precisely or we’ll end up with emoji’s for terms.

I will keep reiterating until we understand that a certain quality of signal is required. That quality is typically defined by a mask elevation, the SnR and PDOP values. You can Google what all that means. The RTK connection is also dependent on the number of satellites that are common to both the rover (drone) and the base. If you do not have 5 satellites you do not get RTK. If you do not have 7 satellites you do not fly. If you have 7 satellites but only 4 of them meet the QoS then you can fly but you don’t get RTK and so on.

1 Like

Hi @MichaelL. My thought here is complementary to yours. RTK may augment image tag accuracies in the case of weaker (generally 5-10) satellite connections. As you said, there is a minimum of 5 satellites required for RTK initialization, which drops to 4 thereafter to maintain the GNSS aspect of RTK triangulation. Given 6 satellites are required to record the DJI home point, RTK can initialize (albeit not at full accuracy) at lower satellite counts. I imagine your RTK experience has shown insufficient survey accuracies at lower (5-10) satellite counts, but the original thought here is that it may be helpful for Victor if the goal is improved accuracy tags for photogrammetry, not necessarily survey-grade outcomes.

@VictorK I would recommend backing off the wall as much as possible and flying higher, adjusting the gimbal to -20 degrees or more (in Advanced settings) in order to still fit the facade into the capture area. Our teams are also happy to assist directly if you message into us. Good luck flying!



WILL ABSOLUTELY enhance the accuracy of the geotags but the number of satellites is dependent completely on the configuration of the GNSS modules. either way this would not explain their not being any geotag information present as the OP may be suggesting.

Yes, an RTK float may occur but I have never seen a GNSS solution that allows even that to happen with less than 5 satellites as the OP suggests. Users need to understand some terminology that a drone in a sense flies on what is called a single solution. It will almost never receive any better than 1m accuracy and may be as poor as 10m. An RTK float can be as good as decimeter and as poor as 5m but is much more repeatable. RTK is the desired solution and should will be repeatable down to 1cm of accuracy.

This is not possible in the correct configuration because the solution would go the float and the survey would halt. This is a major thing that needs to happen in the future with drone RTK. I have already proven with PPK analysis that the Phantom 4 RTK provides an RTK confidence of about 95-97% meaning that it does drop to float about 4% of the time but recovers. The main requirement being that the 4% does not happen all at once. If it were to occur in one timespan a 15 minute flight could be in float for 36 seconds which could be 18 images. When you spread those images out over the flight though and understand that a near-fix float is drastically better than the best native GNSS performance then it is typically a wash. We feel the need in certain situations though to verify the RTK capture and verify that it indeed is dispersed and not isolated.

We’ll keep going back to this point but if the OP can’t maintain 6-7 satellites consistently they will not be able to do anything to fix it other than change locations or not fly regardless of an alternate GNSS technology.

1 Like

As someone who has flown extensively with RTK, written software to proxy NTRIP corrections, and spent hundreds of hours testing/optimizing “virtual” station techniques for blending CORS, I can attest to the statements by @MichaelL – this is not a situation where the OP’s unique application will be aided by adding kinematics, either real-time or post-processed.

Although technically PPK could indeed help his second issue; it will not have any effect on his real-time concerns, of course.

We’ve done some terrestrial work in Chicago where we’ve had problems (strangely we haven’t had the same issues in Manhattan, yet) - IMHO, there are a couple things that could help; however, the problem is how they would affect your payload-capacity being that you would prefer to do this in the air.

To elaborate on that – the issue is not limited to GNSS signals being blocked when surrounded by tall buildings; rather, the equal/greater issues is that you are potentially receiving signals from satellites that are being rejected because they are erroneous as a result of bouncing off the buildings.

Simply put, you are “seeing” fewer satellites, sure. However, if you were to obtain quality results from the exposed satellites, you could theoretically overcome the issue because accuracy would be substantially improved.

In not sure if you posted the device that you’re using, but one option might be using an Ardupilot device were you could use a cutting-edge GNSS chipset of your own choosing, to pickup the maximum number of concurrent channels on multiple constellations/bands, including L2C, L5, etc. Off the top of my head, I’m guessing ublox’s F9P would be the easiest route. There are two others whose name escapes me, that are similar/better but I’m not sure if they’ve been vetted as well in the Ardu ecosystem. In short – This would increase the number of signals that you have access to. While that might not be a panacea, it’s at least something.

The simplest and most pragmatic option is probably flying manual to about 100’. In addition, if you can handle the payload, use a survey-grade antenna with the vertical blocks for superior multi-path rejection (in hopes of extracting what you can / cleaning up the good signals you might not be able to use, currently); those same antennas have nice wide ground-plane blocking, which is also very helpful in your situation. And as mentioned above, you would be using something that was dual-feed to help facilitate the larger number of concurrent constellations/locks.

To clarify about the 100’ comment… You don’t need to get to the top of hte buildings to be able to substantially increase your signal locks. As you get to a certain level, you’ll see reduced terrestrial issues (including both noise-pollution and the crosstalk of the signal itself); you’ll also gain more exposure to satellites as the angle of sky-view increases quite a bit, even rising to just a third of the surrounding building’s height.

To keep this brief/easy-to-understand, I’m abbreviating in such a way that favors ease over accuracy. Unfortunately this is a bit of complex topic and doing it justice, while being entirely accurate, would require a much more in-depth or possible more confusing explanation. However, for all intents, the information I’ve provided should give you some ideas of the options, the issues/overhead, and the reason that manual flying is probably going to be the easiest option.

One last note… They do make devices for your situation (layout in areas that cannot receive GPS signals, such as construction layout inside buildings). They’ve largely been replaced with highly automated/advanced total stations; however, you can still find them around. Essentially they would use RF triangulation and known-points (ala GPS, or static input) to calibrate your location. These devices would output NMEA strings (think of this as the GPS middleware/simplest form of standardized communication that any device which interfaces with GPS’ typically understands).

Obviously the problem with that system is not only the overhead of purchasing and the logistics of configuring on your busy/insecure site; the greater issue being that you need a devices that you know can take a serial NMEA string and will run with that. Again that would limit you to something like Ardu ecosystem. There are probably other options; however, I’m not aware of any.

I hope that helps :slight_smile:


Good stuff! One thing that I wish the rtk drone manufacturers would allow us to do is adjust the basic values for the satellite filtering. Simple adjustments to the Mask and SnR could make the difference in flying or not. We’ve had several locations where a 10° mask included too many dirty satellites and we could not get a fix but when we increase the mask to 15° it was seeing only the acceptable satellites above the structures. We typically run PPK if we are not getting at least 15 satellites but that’s not going to do us any good if we can’t even get enough to take off and or lose them when we get down to a certain level.