Mapping Accuracy with IP

[DigitalDrone]

If horizontal accuracy is set to 1cm then they are at mapping grade; one step below survey grade. Matching 5mm of accuracy is going to be more challenging and is probably only going to be possible via a surveyor with a total station. GPS is excellent for measuring northings and eastings but has some limitations when it comes to heights.

We are using a UAS to save time gathering horizontal data but we are also backstopping it with conventional survey methods and equipment. Both to spot check existing orthomosaic data and provide accurate height/elevation data. I do not think that traditional survey methods will be replaced by UAS flights in the near future. You still, after all have to set GCPs by the old tried and true method. I do think that where extreme accuracy is not an issue, or not a big issue, that UASs can provide a cheaper alternative vs conventional boots on the ground. (The preceding was my opinion. I am legally entitled to have them. My opinions hold no sway, influence any government, or amount to a hill of beans. But they are mine and mine alone.)

In what instance would extreme accuracy not be important? I've trying to get this answer for a while just to know what jobs I can do with out GCP's?

Thanks!

 

[bluelight.support]

In what instance would extreme accuracy not be important? I've trying to get this answer for a while just to know what jobs I can do with out GCP's?

That is wholly down to what your client wants - if they are looking for a pretty google map overlay which is current, then you can get away with just the gps from the drone, as 3-5m is fine for that.

Bearing in mind that accuracy under 1m is very very hard, as that falls under the professional realm, due to the differences in tectonic plates round the world, and different co-ordinate systems and epochs. As an example i heard recently, Hawaii moves around 8mm per year - so over a period of years a survey becomes wholly inadequate - so you need the most accurate data to start with and for that you need professional kit which can provide the audit able data to do this. Im afraid your options are really google map style data or spend money - ESRI for example uses very accurate data for GIS otherwise all your models are no good!

 

[RMartin]

In what instance would extreme accuracy not be important? I've trying to get this answer for a while just to know what jobs I can do with out GCP's?

Thanks!

I would think for real estate projects, photography, general mapping...jobs along those lines. When I say above construction, we are using the data collected to digitize sub-surface utility information in the trench. The accuracy on that would be measured in half the width of an excavator bucket (we get it a lot closer than that). Another example would be pier locations, which, in the future a utility would be required to be bored under a slab the operator could steer the sond and avoid hitting a pier (a poor construction practice but it happens).

General mapping needs to be performed to some reasonable degree of accuracy in which both you and the client agree upon. Examples would be park benches and/or trees for the local city parks department or street signage. Mapping items that are not dimensionally critical but are still good to know where they generally are.

Accuracy in construction matters. Accuracy in survey (where the surveyor has applied his stamp and becomes legally liable for its accuracy and content) is a huge deal.

And again, this is just my opinion. Take it for what it is worth.

 

[bluelight.support]

When I say above construction, we are using the data collected to digitize sub-surface utility information in the trench. The accuracy on that would be measured in half the width of an excavator bucket (we get it a lot closer than that). Another example would be pier locations, which, in the future a utility would be required to be bored under a slab the operator could steer the sond and avoid hitting a pier

I personally wouldn't offer that level of mapping without ground control points. UAV gnss is around 1-5m which is nowhere near accurate enough for what you are suggesting. Im not even sure liability insurance would cover you either.

 

[RMartin]

I personally wouldn't offer that level of mapping without ground control points. UAV gnss is around 1-5m which is nowhere near accurate enough for what you are suggesting. Im not even sure liability insurance would cover you either.

 

[RMartin]

Exactly. We use 2 Geo 7X Trimble GPS units with mapping grade (1-2cm) packages installed and a Nivo total station to set GCPs. We are noticing accuracy around 0.059 ft corrected via an RTK network on our deliverables. That is close enough for what we are doing. Without GCPs, I was getting around 30cm - 2m of accuracy of the onboard GPS unit on the Inspire.

 

[DigitalDrone]

How is Volume accuracy without GCP's? I was talking with a local operator and he was telling me stockpile surveying can do without GCP's and you can get fairly accurate results in the vertical volume measurements only.

This sounds like critical stuff and I can see why.

 

[bluelight.support]

How is Volume accuracy without GCP's? I was talking with a local operator and he was telling me stockpile surveying can do without GCP's and you can get fairly accurate results in the vertical volume measurements only.

I haven't done much in the way of volumes, but if vertical accuracy is important, then i would suggest gcp's would be as well, as the vertical data from the flight controller is the most inaccurate, therefore would have thought the gcp would compensate for this - maybe someone with better experience than me can guide you better

 

[RMartin]

How is Volume accuracy without GCP's? I was talking with a local operator and he was telling me stockpile surveying can do without GCP's and you can get fairly accurate results in the vertical volume measurements only.

GPS does really well with X and Y horizontal coordinates but on the vertical plane it has problems. You can have greater inaccuracies in the vertical by a foot or more so you are right, it is critical to provide checkpoints (GCPs that have been accurately measured, or as accurate as possible) to backcheck the data that the UAS is gathering.

Again, I am no expert on the topic but a little homework on the limitations of GPS and common sense should get you there.

 

[RMartin]

How is Volume accuracy without GCP's? I was talking with a local operator and he was telling me stockpile surveying can do without GCP's and you can get fairly accurate results in the vertical volume measurements only.

This sounds like critical stuff and I can see why.

I don't know if this will help or confuse the issue but I found this while trying to address and accuracy issue we are discussing with a contractor:

Scientific White Paper: How accurate are UAV surveying methods?

 

[Mogmen]

Exactly. We use 2 Geo 7X Trimble GPS units with mapping grade (1-2cm) packages installed and a Nivo total station to set GCPs. We are noticing accuracy around 0.059 ft corrected via an RTK network on our deliverables. That is close enough for what we are doing. Without GCPs, I was getting around 30cm - 2m of accuracy of the onboard GPS unit on the Inspire.

Excellent information here. Thank you all!

I too am mapping with an inspire 1 x5 15mm. And processing in house with pix4d. The frustration i am having is dealing with the extreme vertical shift in pix4d when using GCPs due to the incorrect altitude data in the exif. Are you guys experiencing this? If so what are doing to overcome this? This has put me on mission to find a mapping system which can write the correct altitude data in the exif.

I have been looking at the microdrones md41000. Does anyone out there have any experience with Microdrones?

 

[RMartin]

I too am mapping with an inspire 1 x5 15mm. And processing in house with pix4d. The frustration i am having is dealing with the extreme vertical shift in pix4d when using GCPs due to the incorrect altitude data in the exif. Are you guys experiencing this? If so what are doing to overcome this? This has put me on mission to find a mapping system which can write the correct altitude data in the exif.

Have you specified a spheroid when you insert your GCPs? GRS_1980 is the one we use. And yes, the vertical data you collect is
abysmal. Haven't figured that out yet. I've noticed it is sometimes off by more than half of the altitude I'm flying at, even when the start altitude is reading correctly when I take off.
We currently are only using data collected via UAS to digitize hardscape in GIS (saves tons of time of field work). The X-Y data is accurate enough...as accurate as the control points we set.

 

[Mogmen]

Stop looking at the DJI RTK system to work in the way you would expect it to within the mapping / survey industry. I have written loads of posts of this in different sections if you do a search. DJI in their wisdom do not transmit the RTK data through the flight control system into the camera - therefore the geo location data on the EXIF info on the JPG is not cm accurate. Apparently if you put in exact co-ordinates into ground station pro the system will fly to that co-ordinate, however, no accurate details are added to geo-location which makes the system useless for our purposes. Im sure DJI will come up with a survey solution and add many $$$$$ to it!

Ground control point, or cross reference with previous ground survey points for accuracy is the only way to go!

I am curious if you experience problems when processing with pix4d using GCPs

Have you specified a spheroid when you insert your GCPs? GRS_1980 is the one we use. And yes, the vertical data you collect is
abysmal. Haven't figured that out yet. I've noticed it is sometimes off by more than half of the altitude I'm flying at, even when the start altitude is reading correctly when I take off.
We currently are only using data collected via UAS to digitize hardscape in GIS (saves tons of time of field work). The X-Y data is accurate enough...as accurate as the control points we set.

I don't recall if i have specified a spheroid when inserting GCPs. I will try as you suggest. Thank you. I have read on this thread where bluelight uses an exif editor to change the altitude data en mass. I looked at those programs and i have no experience working from the command prompt. I guess i can learn. I would rather find a drone that does not have these issues.

Do you know if this issue is the case with all DJI flight controllers?

 

[bluelight.support]

I am curious if you experience problems when processing with pix4d using GCPs

this is down to the data going from the flight controller to the jpg so makes no difference on the final processing system. Without gap's using a survey grade gnosis system the data is unusable anyway for anything serious.

 

[RMartin]

this is down to the data going from the flight controller to the jpg so makes no difference on the final processing system. Without gap's using a survey grade gnosis system the data is unusable anyway for anything serious.

Does the EXIF editor work well for you? At the moment, the Z value is not that big a deal to us. In the future, it is going to become more important.

 

[bluelight.support]

to resolve the DJI negative figure (below sea level) - we use the exif editor in a batch process to make all levels above sea level. You still need ground control points and the external gnss system to get accurate z values, otherwise you could be unto many metres wrong!

 

[RMartin]

to resolve the DJI negative figure (below sea level) - we use the exif editor in a batch process to make all levels above sea level. You still need ground control points and the external gnss system to get accurate z values, otherwise you could be unto many metres wrong!

I've never encountered (to date) negative values but I have flown a few flights that recorded EXIF data that was half the altitude I was actually at. We've always used GCPs but Pix4D sometimes gets angry if it has to work hard on altitude figures. I think I am going to start editing the values to make life a little easier.

 

[r62ewa]

I am in the exact same boat as everyone here. Figuring out how to use a UAV for mapping, volumes, etc and what projects I can support without GCPs as I don't have the funds for survey grade GPS. Also, I have run into the negative elevation values in the EXIF as DJI records the altitude as Below Sea Level. Couple this with as RMartin has said in the elevations recorded are also erroneous most of the time and resulting processed point clouds and other outputs are not accurate.

I have been working with a University Remote Sensing organization and they were able to write a batch process and bit of Python code whereby I can quickly edit all of the altitudes of my DJI collected images based on ellipsoid height I set manually. The result is updated altitudes for all images written to a text file that is imported into Pix4D for processing to use in place of the EXIF values from the pictures. In this method, my 3D point clouds are much more accurate to real world elevations. One project I processed using the EXIF altitudes at near sea level had the elevation in the point cloud as -11m and when I fixed the altitudes using my script and processed the dataset again the same point was 5m, which is much more realistic to where I had flown.

What I do is run a batch script on all DJI collected images for a project and the batch process writes a text file for each image with relevant EXIF data for that image. I then run the Python script which asks me for an ellipsoid height, I enter it and it then iterates through all image text files from the batch process and combines all the results into a new single text file with updated elevations referenced to the ellipsoid height I set. I then substitute this text file in Pix4D to use instead of the EXIF of the images.

So far my projects are all near sea level, my challenge is how can I accurately determine an ellipsoid height at higher elevations without an accurate GPS.

 

[RMartin]

So far my projects are all near sea level, my challenge is how can I accurately determine an ellipsoid height at higher elevations without an accurate GPS.

So now we are playing "Stump the Chump?" Without GPS, or some survey experience, the only option I can think of would be to hire a surveyor to set control points around the area that you operate in. Those points could then be incorporated into your work for an accurate elevation. We have a network of control points around campus that we have had installed and expanded upon that we use for that purpose (among others).

 

[r62ewa]

So now we are playing "Stump the Chump?" Without GPS, or some survey experience, the only option I can think of would be to hire a surveyor to set control points around the area that you operate in. Those points could then be incorporated into your work for an accurate elevation. We have a network of control points around campus that we have had installed and expanded upon that we use for that purpose (among others).

Thanks. I live in a very remote area and control points are sparse or non-existent where I would be working. I don't have any surveying background but would really like to have enough knowledge to set accurate GCPs if I could afford the equipment, that of course is another limiting factor.

I guess I shall content myself with accepting non GCP related work until I can get my surveying up to snuff with some equipment.

Great thread.