Bandwagon's Drones Thread

I don't know any super secret squirrel places for parts....you've checked the places I would look.

Except maybe hobby king's US west or US east warehouse. It takes 1-2 days for me to get something from US West

Mapping Lesson of the day - don't sacrifice overlap for the sake of batteries!


The other two guys drove almost 3 hours yesterday to map 260 acres. I had the survey uploaded to the drone already, but I set it at 65% overlap instead of the usual 70%. I thought that the ground was almost completely flat, so I figured 65% was fine, and I was very worried about being able to cover it all with the batteries we have. Turns out there were a few moderate elevation changes, and the first two tries at processing either failed or had big gaps missing.

I thought I was going to have to drive back out there and re-fly all of it, so that would have been about an 8 hour mistake.

I ended up pulling the .log files off the FC and tagging all the photos, added some tie points and changed a few of the matching options and left it running overnight with my fingers crossed.
Woke up to find this guy chugging along, with about 1450 out of 1600 pictures enabled. Fuck yea!
I need to keep a better eye on altitude info when making survey flight paths off-site.



Edit - 2 quick examples of before and after-

I also made this for someone that is trying to extract IMU/Attitude info from the flight logs so that they can be attached to the pictures to calculate irradiance for NDVI/NDRE.

Apparently the data is easier to extract from the Tlogs (GCS logs) than from the .logs (onboard logs), so he keeps asking for the Tlog files.

This picture might seem a little out of context since it was after about a 2 hour email conversation, but basically - The city picture is a timeline, and the red lines from "SHUTTER" are supposed to be the GPS and IMU data you're pulling from the log to attach to the pictures. Depending on the rate, Tlogs can get an update anywhere from .5seconds to 5 seconds, and are not as complete as the on board logs. Since we can have 10 degree attitude changes in a second, there's a very real possibility (very likely, actually) of attaching incorrect information to the pictures when using the Tlogs. This is even worse when you lose connection on an autonomous mission, since now you're tagging every picture that was taken during that time with the last set of data that was received by the GCS. The BEST way to do it is with a serial connection between FC and camera. Syncing and tagging with the onboard log will have the same result, but it's a pain in the ass. Tlogs are garbage for this kind of task, since the entire goal is extreme accuracy.

I saw that at lunch today. Not exciting.

It's actually just for an elevation model. He's a software engineer that's trying to come up with a way to accurately assess volume using photogrammetry. The flight we did at 200ft, early in the year, shows promise. I feel like this one was too late, and it should have been done at 200ft. Oh well, I voiced my concern a few times.
I've been wanting to have some comparative examples (like what soil samples) for ever. I've probably done 7 or 8 jobs for free, because the person/company was supposed to give me the results of satellite imagery, soil samples and greenseeker values in exchange, but people are flaky as shit. I don't do anything for free anymore.

I just got a contract from OSU today, which I've been trying to get since last year. Just like one of the other schools we're flying for, they have permanent sensors throughout their ~25 acres of test plots, which they are going to share with us, along with mentioning us in the white paper that an under grad is doing on it. Pretty excited about this one - It'll be 9 flights over the summer, and the people that are analyzing the data actually have the credentials to do so (unlike me).

Picasso_sl said:

Search Web soil survey for online soil surveys. You can upload shapefiles to pull them for a perimeter as well.

Volumes on a complex surface can be impossible because you have to have a reference obv, you could digitize a usgs quad as a last case. On a flat surface you can drape a feature line around the perimeter and infill a surface to compare.

Click to expand...

I'll probably give this a go just out of curiosity, but I'm not going to be able to make any kind of educated comparison between soil surveys and the reflectance map results. Hopefully we'll get something good out of this OSU undergrad's thesis.

I finally made some progress with a survey company here. Had the 2nd meeting, they're going to use us on a 2 mile survey of a creek. I'm going to process it about 4 or 5 different ways, transitioning the interior GCPs into checkpoints, so that we can see how many are needed per acre to get the level of accuracy they want. If they're happy, they're going to start using us as a subcontractor on some jobs this summer, including a massive project for the Army. One of the co-owners literally slammed his hands down on the table and looked at me and said "Do you understand the implications of what I'm telling you right now? This is a 5 million dollar contract...this has the potential to be very lucrative for all of us".

I don't think I've ever wanted to impress anyone more than I do on this upcoming creek survey.

For the paper. I swear to god I thought I was smiling the whole time. It felt pretty damn awkward...and all of us were trying to hide a chew (including the photographer).

Thought I'd share.

Perfect example of one of the reasons I do processing locally-

"By using the Service, you grant DroneDeploy a non-exclusive, irrevocable, fully paid and royalty-free, transferable, sublicensable, worldwide license to use, copy, reproduce, process, adapt, modify, publish, transmit, display, and distribute your User Content."

Yeah, no thanks.

Going to be rehashing some of the stuff earlier in the thread . . . so pardon in advance.

How large was that test field you have been using? And how capable would your setup be for a 1000 acre plot? How many acres can you reliably do per battery with your current setup (2 scans at different alts etc)?

Nm, 260 acres with 2 batteries?

LachiusTZ_sl said:

Yeah, no thanks.

Going to be rehashing some of the stuff earlier in the thread . . . so pardon in advance.

How large was that test field you have been using? And how capable would your setup be for a 1000 acre plot? How many acres can you reliably do per battery with your current setup (2 scans at different alts etc)?

Nm, 260 acres with 2 batteries?

Click to expand...

The test field we usually fly is about 96 acres. At 400ft with the canon, we can fly this in two batteries if it's dead calm. With the RedEdge, it's 3 batteries, landing at about 50% left on the 3rd. With 15mph winds, it would be 3-4 batteries with the canon.

Depending on the camera and weather, you can get anywhere from 40-70 acres per battery with the Solo. It's definitely not the most efficient setup, but it's cheap, reliable, allows much more control for surveys, and you don't get the rough landings like with a fixed wing.

As a comparision, the 2nd fixed wing I built can just barely cover 400 acres in a single battery, which is anywhere from 58 to 68 minutes. It's much harder to setup and plan missions, and I guarantee you it would be destroyed in 1-2 landings if I tried to let one of the other guys use it.

We won't be using the Solo forever, but it's the only thing (IMO) that can do the job, and has a price tag to match the actual value. In perfect weather conditions, we might be able to do 300acres an hour (versus 400 with most fixed wings). It only takes us about 3 minutes to do a battery swap.

If 1,000 acres is expected to be your normal/most common request, look into an eBee. Just know that that amount of data is probably going to take somewhere in the realm of a 2-6 days to process locally, and you'll pay quite a bit to have someone else do it. Make sure you have the business first.

Lenardo_sl said:

boss is being leary about the drone thing, i might just have to buy one - myself - and "prove" that it can cut Days out of our work scheduling when we have to topo larger sites..


i have a perfect one i could fly too - a subdivision project that is almost complete and ready for a final asbuilt plan, was 64 acres- of which ~half was developed(rest was wetlands). would take me an hr to get there.

Click to expand...

You still have to place GCPs...but how many you can cut out by using a drone is what we're trying to figure out now. You can have something to show your boss by spending $1500.

edit: if you're not interested in Ag at all, getting a Phantom3 or Phantom4 and using dronedeploy might actually be easier for you. I still dont really like DJI and I would only be able to offer a limited amount of help, but the P3/P4+dronedeploy combo is pretty easy to use now. I'd offer to do a free flight for you if he paid to bring us out there, but you could probably buy two for yourself with the airfare.

I've never been able to make a 3D model of something sitting on a table or on the floor in my house. I've probably tried 10 or 15 different times. The software is pretty susceptible to lighting and shadows, and needs 80-90% overlap for 3d model kind of stuff. Being that close to something in weird lighting is a pain in the ass to try to make work. If you ever get something to work well, I'd love to hear specifics on how you did it. I want to make a 3d model of the interior and exterior of a building, but I gave up early after trying small objects. I wonder if a tripod would work wonders.

Anyways, are you using Agisoft for the Phantom?