Welcome Inspire Pilots!
Join our free DJI Inspire community today!
Sign up

questions about the cut off point

Joined
Aug 3, 2016
Messages
13
Reaction score
1
Age
49
Location
Halifax, Nova Scotia, Canada
New to the industry of using UAVs for Commercial use. I've got a trial going of DroneDeploy, pretty sweet. I can export a KML file or a .las cloud point file (still learning about what those are). I can see this being delivered to the customer via Dropbox, downloadable file from my site or simple USB stick. They can then load the file into their own software to conduct their business as usual.

I'm also looking at the extremely expensive Pix4D suite of products. My question comes down to this:
Is it up to me to have this software in my toolbox or is this something my client should already have?
Once I capture images, then what? Should I purchase/rent Pix4D calculate distance, volume, elevation etc and export them to file formats that they can use?

Just wondering where my work ends and where the client begins on this one -expectation. Anyone out there with experience on this and perhaps what they practice or their workflow would be a huge help.

Thanks
 
You can do a lot of he measurements and volumetric analysis that you mention in DroneDeploy, but that starts in their $99/month tier. Also if you want to do the real deal shapefile exports, you need to go to a higher price tier.

Pix4D is indeed extremely powerful, but you will also need an equally powerful computer to do the orthorectification in a timely manner. I currently run Agisoft Photoscan on my specced out 2015 15" MacBook Pro; it can only handle creating textured models from photo sets of around 100-125 photos and absolutely chokes on tiled models.

I'm looking hard at an RTK-capable fixed wing UAS for surveying, but like you I'm not sure if I'll go with the $250/month tier for DroneDeploy or get a new Mac and process the images myself.
 
  • Like
Reactions: Parker
New to the industry of using UAVs for Commercial use. I've got a trial going of DroneDeploy, pretty sweet. I can export a KML file or a .las cloud point file (still learning about what those are). I can see this being delivered to the customer via Dropbox, downloadable file from my site or simple USB stick. They can then load the file into their own software to conduct their business as usual.

I'm also looking at the extremely expensive Pix4D suite of products. My question comes down to this:
Is it up to me to have this software in my toolbox or is this something my client should already have?
Once I capture images, then what? Should I purchase/rent Pix4D calculate distance, volume, elevation etc and export them to file formats that they can use?

Just wondering where my work ends and where the client begins on this one -expectation. Anyone out there with experience on this and perhaps what they practice or their workflow would be a huge help.

Thanks
Unless you are a chartered/certified surveyor you cannot stand over any measurement or volumetric assessment unless you make it clear to the client that you are providing low accuracy estimates only.

Pix4d or Photoscan only produce as good as what is fed into them. Without GCP'S your accuracy is going to be way out anyway. This is only consumer grade GPS. I always state the parameters of accuracy, no liability resulting from the dataset being used outside of those parameters, and supplied as-is.

Depending on the nature of the datasets you are providing, the client should be capable of processing them and interpreting them for their own purposes, just don't leave yourself open to any lawsuit for example when a client sets-out a housing development and they find the data you supplied is 5m, or much more, out. Just make it clear to them. Keeps you right.

PFAW Holder
BNUC-S Qualified
 
  • Like
Reactions: Parker
Unless you are a chartered/certified surveyor you cannot stand over any measurement or volumetric assessment unless you make it clear to the client that you are providing low accuracy estimates only.

Pix4d or Photoscan only produce as good as what is fed into them. Without GCP'S your accuracy is going to be way out anyway. This is only consumer grade GPS. I always state the parameters of accuracy, no liability resulting from the dataset being used outside of those parameters, and supplied as-is.

Depending on the nature of the datasets you are providing, the client should be capable of processing them and interpreting them for their own purposes, just don't leave yourself open to any lawsuit for example when a client sets-out a housing development and they find the data you supplied is 5m, or much more, out. Just make it clear to them. Keeps you right.

PFAW Holder
BNUC-S Qualified

Thanks for this reply and experienced insight. I will indeed be adding a disclaimer regarding no liability and level of accuracy in the statement of work, which will be signed by myself and the client. The output would be a shapefile, KML etc.

Is there a way to determine the margin of error on accuracy?

Thanks again
 
Thanks for this reply and experienced insight. I will indeed be adding a disclaimer regarding no liability and level of accuracy in the statement of work, which will be signed by myself and the client. The output would be a shapefile, KML etc.

Is there a way to determine the margin of error on accuracy?

Thanks again

It depends on the level accuracy measurement you want. I know, sounds like a bum politicans answer.

The only way you can accurately work out the level of accuracy accurately is to compare it to an RTK control dataset. Even that will vary depending on the number of satellites you have and their azimuth on any given day. On a recent one I did the drone accuracy worked out at 0.8m. That is pretty good but not typical. It's normally much bigger than that. Thats not worked out by me by the way but by a clients GIS surveyor who set out the GCP's.

You can only state what DJI's hover hold tolerances are. 2.5m horizontal and 0.5m vertical. And that's not really related to an aggregated orthomossaic dataset out of Photoscan. Bearing in mind too that the reported GPS elevation figures can be many metres out in my experience. Using RTK and GCPs is the only way you can get accurate results I'm afraid.

The gear is expensive but you can hire it in and build it into your cost.

PFAW Holder
BNUC-S Qualified
 
  • Like
Reactions: Parker
The only way you can accurately work out the level of accuracy accurately is to compare it to an RTK control dataset.
RTK data capture is ideal for quick on-site confirmation, as you can click the button and get it to 1cm accuracy very quickly. The issue with that is the £6k on equipment, plus the £2k per year data plan to get the RTK dataset. (Assuming a hand-held unit with on of the UK providers and data plan for RTK). Another option is using RTK on the drone itself with a base station set-up at your site, this is again RTK but using correctional data between the drone and the base station. This is really expensive (think eBee at 25k) but does not need ground control points to get cm accuracy.

We went a slightly different route and use a £6k handheld 1cm accuracy unit and plot 5-10 ground control points but post process the data back at the office with free OS Rinex data to get the accuracy and then add the data into Pix4D, Ajisoft, or others (still finding a good one). We can get an Inspire 1 down to 1cm accuracy that way.

You can pick up units off eBay sometimes quite cheaply, just make sure you ge the serial number checked to make sure they're not cloned or stolen, and ensure the cm upgrades (if Trimble) are added.

I spent 18 months researching all this, and its a very specialist and not easy subject. Yell if i can help (or someone else yell if i got this wrong!)
 

Members online

Forum statistics

Threads
22,277
Messages
210,655
Members
34,324
Latest member
LewisuhaAbips