Breaking large mission up and combining into orthomosaic
I have ready a couple other posts on here, one titled "Manage large areas missions" and I saw a suggestion in there on doing just that- but, it was about a year ago so I'm checking to see if it's the only option and maybe things have changed since then. Additionally, I've read a bit about KML and I'm still not sure I understand what I'm supposed to do with that in order to break up into smaller mission.
I have a site that is about 47 acres, but the bigger issue is that it's in the mountains and I know there's no way to maintain signal for the entire site from one single spot- so I'm definitely going to need to launch from several spots. Plus, I did read that 1000 images is about where you want to max out on mapping.
So, given the size and the terrain, I feel confident I'm going to need to break this into more than one mission.
Here's my concern- What happens along the edges during stitching? Is there going to be a problem with the exact line along where the multiple missions meet each other with the overlap being off?
When plotting out the map, I'm still not sure (and I did read a post that went into detail about copying and pasting from an original "large mission" into smaller missions and just sizing them down from there) about how to know the exact line where it's being broken up? Maybe I'm missing something, and maybe I'm worrying about something I shouldn't worry about, but I'm just trying to avoid having to re-do this more than once because of finding an error in the stitching process.
I hope all of that makes sense.
I'm using a Mavic A2 and a Samsung S8 and a mapping account.
Comments
20 comments
Hi Charles. The questions are widespread so I will try to help.
If you take Google Earth and draw out boundaries of each flight of the total mission, you can export them as KML files which can then be imported into DroneLink. But that is not necessary as you could simply create your missions right in DroneLink. One of the better uses for KML's is when your client tells you they need a specific area mapped (or just photographed). I ask them to submit a KML of the property boundary. If they do not know that, then I can use Regrid.com to obtain that. But not thinking you really need a KML for 47 acres.
Total maximum Image count allowed is greatly dependent on your computer system. If you have a power house computer then you can handled many more images than someone on an average machine. Those numbers vary GREATLY. If you are doing cloud processing then whatever company is doing the processing will provide those guidelines.
If sounds like you will have multiple flights from different take off points. I would just make sure you have some overlap in those flights and make darn sure you have good elevation data if you are flying with terrain follow.
The size of your mission and individual flights is only dependent on 1) how many batteries you have and 2) line of sight. The altitude you fly at determines the Ground Sampling Distance (how many pixels per inch, or pixels per CM). It also determines the number of photos you end up with.
You could fly over 47 acres at 375' with an 80% Front and 75% side in one flight and end up with under 250 images. If the area is totally covered with trees, you can have stitching problems, especially on a windy day.
I would think you should be able to fly 47 acres in 3 flights (3 different sides of the mountain). But that depends on visibility, accessibility, and line of sight. Hope this helps.
charles folds
Breaking into more than one mission is a good idea for large sites. Under normal circumstances I wouldn't consider 47 acres particularly large - I have a site I run every two weeks that is 400+ acres. But, if you cannot maintain a signal then that is definitely a concern.
First to address your concern - as long as the two maps have sufficient overlap, you won't have any trouble stitching as long as you are using good software. I run a lot of missions where battery swaps occur (which is a similar concept) and never have a problem.
For a site that large the free version of Zephyr won't give you enough images, so I would recommend trying WebODM or Maps Made Easy for reasonably priced (or free if you know how to install yourself) options. Good software is critical if you have things with a lot of similarities such as trees or water.
Other things to consider:
Hope that helps!
Barry
Thank you so much for that!
Based on my experimenting with plotting out the missions, I think you are spot on with the 3 flights- BUT, those will likely have to be launched from 3 different places if I have signal issues, there about 5 different ridgelines in the area.
I mapped it out at 200' terrain follow and it's around 1,200 images (and I'm ball parking that altitude right now but I suspect I'm not going to fly much higher based on the details she's wanting)- the most I've processed was a little over 500 images and that took a while for my computer to choke down, but it did it. So, I'm thinking over 1k is going to be pushing it.
No problem on the battery issue, I'm good there. And I'm headed out there soon to check take off points to make sure I've got good line of sight from each point- but I'm pretty confident about that. The property is about 70% tree cover so yes that was one of the main reasons I'm doing some research on this- I'm worried about those lines along where the different smaller missions will be connecting. THAT is the root of my question. Yes, I can plot out several missions on the property no problem, I've been experimenting with that- but I can't figure out how to make the overlap along those lines to be consistent with the overlap within the missions. And again, maybe I'm worrying about something that I shouldn't worry about?
When you match up several missions on the same property, how wonky can that line be between the missions if you don't have those flight lines spaced/overlapped exactly like the spacing/overlapping that dronelink will automatically do within each individual mission?
I would LOVE to see some samples of other people who have done this. I've hunted on youtube and other places but I'm not finding anything yet.
The plans don't have to match up - they just need to overlap. Each photo is geotagged so good software will figure out where it goes.
The problem with sharing these types of photos is that they are HUGE!
BTW - if you use something like WebODM - it will save the full resolution file as a "GeoTIFF". Those can be imported into Google Earth Pro and overlaid automatically - which is great for showing progression.
Ah-
Barry, you answered my question right there about the overlap, thank you!
I'm currently using ESRI Drone2Map because it plays well with ArcGIS Pro, which is where I'll be adding vector elements...but I'm going to check some other options. I've had good luck with Drone2Map though.
Yes, my concern was having the map from mission #1, not match up well with the map from mission #2 or #3 etc etc. But, you're saying the overlap shouldn't be an issue if the stitching software is good, and I'm pretty confident with ESRI products.
I think I'm going to do a quick test run here in my yard first to see how that looks- but just letting the missions overlap sounds like a great way to go if that works out.
I THINK this is little different than just a battery sway in a way because I'm going to have to fly the low flay area, then hike up to another ridgeline and fly another area, then hike to another ridgeline and do it again- ultimately having 3 different launch/land locations. Yes, I think you're right that if it was just a flat 47 acres it would probably do it from one spot, one mission, and just do some battery swaps- but here I think signal will be an issue.
I'm definitely going to need to spend a day flying around this site to check on signal from the remote, satellites, terrain, etc- it's definitely the biggest spot I've done, and also in terrain and shape the most complex. But hey, always good to learn some new tricks :-)
You guys are awesome and thank you SO VERY MUCH for the great advice and your time!!!
charles folds
You raise a good point. If you are taking off from different locations, there is a good chance that your flight altitude will be different since everything is measured from the point of takoff. Sure, you can use AGL option but that will have limited accuracy, so that is definitely something you are going to have to experiment with to avoid problems.
See how it works in Pix4dCapture. At the same time, you can see several independent missions and you can precisely plan connecting new missions with already existing ones. Maybe someday Dronlink will also have it?
Dronlink's biggest problem is the terrible demand for processor power when planning the map area in a web browser. The slightest shift, rotation or change of flight direction requires a powerful processor, I don't know why. It is very tedious. In Pix4DCaptura it works much better.
Oh dang- I didn't think about that! Hmmm...and yeah, I'm not sure an easy obvious solution there.
I suppose this is one of those issues (the large site with multiple take off spots specifically) that it would be cool if there were something built into the software to accommodate for that- of course I'm not sure what that would be.
Now this is making my head hurt- because yes, if there is enough variation in the altitude between the missions then that's going to wind up with a wonky looking product in the end.
vg1- I'm not sure because I haven't checked in a while, but the reason I started using dronelink to start with is because it was the only one compatible with a Mavic A2.
vg1
There is nothing to stop you from putting multiple parts of a mission all into a single mission file, then you can see them all together, and just disconnect in between and RTH.
As for the processing power. Pix4D uploads the missions to the drone instead of using virtual sticks. This limits functionality but means the drone handles everything locally. DL offers that as an option - you just need to use the convert function which I referenced earlier, and it will do the same thing as Pix4D and not need a powerful processor.
https://www.youtube.com/watch?v=4F-W4HKJlmA
charles folds
DL has a feature that allows you to specify the altitude above ground instead of take off. It works, but it is not always 100% accurate because it relies on Google data
I think I wrote incorrectly, writing about the processing power of the processor, I was thinking about the situation where on the computer in the web browser you plan the mapping area. Each change is very CPU intensive when you plan your mapping mission.
Barry,
"There is nothing to stop you from putting multiple parts of a mission all into a single mission file, then you can see them all together, and just disconnect in between and RTH."
This is the part I'm curious about, and maybe I'm just not digging in deep enough in the mission planning tools when I'm plotting out the mission. Are you saying that instead of making 3 separate missions, that I can make one mission and have it take off- fly a section- return to home- then I move to the other location for launch/land and fly the next section and return to the new home- them move to the 3rd and last location to launch/land and then complete. Essentially having one big continuous map missions, but with 3 different launch/land spots?
As opposed to creating 3 different missions and just have them overlap?
Yes, I understand the AGL, I've used that some before and it does make sense that will be my only option here, I'll just have to deal with the possible variation in how the terrain is measured.
charles folds
Yes...that's what I am saying.
Here is a plan with three maps...and with the drone disengaging (meaning the mission stops) in between each one.
Or...you could just have it RTH. I typically use disengage because if I have enough battery I might want to just hit resume and let it run for a bit. But...anyway. you get the concept.
Hey look at that!
Ok- yeah this is one of those challenges of learning software...I didn't even know the "disengage" component was a thing!
Now I know and THAT right there is the answer to the original question, but I just didn't know how to ask it.
Whew, so sorry for all the work there Barry and I appreciate your patience!
Now my challenge will just be on how it handles the topography of this site- we'll see how well the AGL does on this one :-)
So- for each of those individual maps I need to set and put in a "RTH" and that will be 3 different "home" point?
That's what I need it to do because I know I'm going to not have one home point that will be able to keep a signal the whole time. I'll start messing with assembling a new map now that I know about the disengage.
RTH goes back to where to took off from.
But what if I need to relocate after each of the "disengage"?
Like imagine on that map you did there, that if I'm standing in map #1 that I can't get enough signal to be able to fly map #2 if I stay at the home point of map #1. So, after it disengages from map #1 flight and return to home...I then have to walk over to the center point of the map #2 and relaunch. Wouldn't I then need to establish somehow that it needs to return to where I'm actually standing at that point (the center of map #2) rather than attempting to land at the original home point (center of map #1). Again, this is because of terrain I'm going to be losing signal if it tries to do that.
Every time you take off it marks the home point as the new take-off location - there are exceptions to that which we don't have to go into.
Ok cool- so I can break this 1 missions into 3 small connected missions using "disengage"- fly map area 1, return to home, then go to the place I need to launch and land the 2nd map and do it all over again for map 2, and the do the same for map 3.
I think I understand now- thank you again for your time!!
Please sign in to leave a comment.