Bandwagon's Drones Thread

L Lenardo - I've mentioned a few times in here that the Phantoms don't log absolute elevation....it's elevation above take off. Don't change your image altitudes. Mark 3 GCPs, 3 images each, using the basic editor. Then reoptimize and do the rest (much quicker) in the raycloud editor.

Also, the P3 camera is garbage for this stuff. P4 is much better, still not quite as good as the SODA.

Here's a site I flew last week. About 2,000 acres. 5,100 photos (45GB).


Here's the prelim surface (something new I've been working on)
Broadmoor NED by PBSusa - 3D model

Sludig said:

Wonder how big that kind of thing is in Colorado. If it was steady work at all might make an interesting career change.

Click to expand...

Colorado has a lot of UAS activities. Not sure how much demand there is in the area, but a lot of uas-focused software companies there.

I really don't think the market for contractors/service providers that ONLY fly is as big as everyone thinks it is. 90% of the companies that benefit from this stuff just want to do it themselves, but they have no idea how to do it right now. I don't see much longevity in being *just* a service provider.

That's a terrible idea.

There's a site calibration for local coordinates. I cant bring it up on my phone, but look at the pix4d documentation for local site calibration.

Lenardo said:

except for a few minor wtf points it came out decent in the densified point cloud-textured not so much due to..trees and shadows-i turn off manmade points and the worst wtf goes away..

too many trees in one section fubar'd one side due to the camera not being able to see thru the canopy. the mosaic- overall- came out good, which is what i am going to use - at real side to adjust the walls and decks to asbuilt

Click to expand...

I'd really recommend checking your ortho against 10 other points, especially far away from the points you used for control.

I've learned a lot of additional steps to check my stuff since working with the surveyors, especially after finding out hoe unreliable pix4d's quality report is. Found a mistake in a landfill dataset this way, which i THINK was caused by flying too low. I flew it at about 180ft from launch point ("0"), and the terrain varied from -60' to +80'. Too much variance in GSD across the set and caused some crazy shit in the point cloud.

You should check out global mapper with the lidar module. Very affordable and makes it easy to check existing shots against the point cloud without having to manually identify them all.

Anyways, flew a 1/2 mile stretch of a busy road last week and we're hoping the drone data is good enough to use for a design surface and keep the crew out of the street.
I generated some points and the crew staked to them to verify yesterday. Averaging 0.06' vertical difference in areas covered by shadows during flight and 0.02' throughout the rest.

Lenardo said:

6 hundredths vertical for shadows and 2 for non shadows,,,that is Very good. for a topo that is phenominal 2 hundredths is basically did the point of the rod stick into the ground some or not... 2 hundredths is about a quarter of an inch (.0833333333'-surveyor's foot) to an inch so .02 is just under a quarter of an inch... boss wants to know if we should upgrade the p3 to the p4pro,, and i went, sure if we are going to use it more than once a year like we are doing now for the p3 (new tech takes a while to catch on in my area, the boss is 72, most of us are over 50...(i am 51) unlike in your area, there are not many wide open areas (lots and LOTS of trees covering lots, i got one lot i WANT to fly, but right now, cannot due to the fact that if i flew it, except for one spot where the cleared a path for digging septic test holes,, the entire lot is...tree. have to wait another month or two before i can fly it (leaves fall off of said trees)

like i said, for what i did. good result, good enough for me to go,, this wall that was supposed to be HERE, is actually HERE... (about 2' away), and go, hey they moved the addition 1.5', and go,,, deck being built here,,,and here,,,and here etc...

Click to expand...

We only checked about 15 points yesterday (out of 5,000 DSM points total) and I wanted to check a few more before working on my part of the project today (and the project is like 2 blocks from the office). Checked 10 points this morning and the average for 9 of them was even better at 0.018', but I found one that was -0.24', located under a powerline in this shadow-


I was going to have the guys check a couple more points in the lanes(been trying to avoid the main travel lanes) in that shadowy area, but I built a quick surface and decided we're just going to do this portion in the field either way. I was getting a variance between points of about 0.6', not matching the profile.

The rest of it looked really good, though. After all the checks and making the instructions for the field and taking it back out to them, took me about 2 hours to clean up the noisy points and send a .txt off to the surveyor.

We're doing 11 sites around Walla Walla for this project. The project above is on 2nd Ave, you can zoom in on the dot to load the ortho. MAP LINK
The surveyor is marking all the locates from the ortho right now while the crew gets my fill-in and finishes the curb work. The workflow is really starting to get streamlined.

And yea, I would definitely pick up a P4 Pro (Do NOT get the Pro+). A P4 Advanced would actually work, too. Double check my memory here, but I think the Advanced has the same exact camera, but no vision sensors....which is fine for us. We disable the vision sensors anyways because direct sunlight will fuck them up.

You guys should seriously replace the P3. The camera on it is absolute garbage for this stuff. The P4Pro camera is pretty damn good, and it blows anything else out of the water in the $1,500 range.

Some draftsmen and surveyors from the other offices are coming over next month for a little workshop on working with my raw outputs so they can start doing their own work eith it for each office.

Today was probably the most fun I've had at work since I started a year ago. Not to say I don't love my job, but it was just fun to spend all day going back and making examples of every "problem" I've found or mistake I've made. A few of the examples I made were of a vineyard and ended up coming in handy 3 hours later when the engineers were working on a new vineyard project with a different surveying company that just started using drones. I just realized today that they never knew what the fuck I was talking about when I said that giving them a surface could take 30 minutes to 20 hours, depending on what they needed it for. Took all of 13 seconds for it to make sense to them with some visuals. I'm a moron, should have done this 6 months ago.

Halfway through the ASPRS symposium at the Commercial UAV expo.

What a difference a year makes. Everything is a lot more realistic this year, in terms of capabilities, claims and expectations. There's also A LOT more people with real-world experience and conversations about workflow and findings.

Glad I came this year. It's a pretty exciting time.

Dji is actually (finally) making a big push into the commercial market. All of the LIDAR manufacturers are mounting om the dji m200 and m600

Saw a few examples a few hours ago.