Hello Fellow Yuneec Pilot!
Join our free Yuneec community and remove this annoying banner!
Sign up

RBJ Austin Redevelopment

Joined
Nov 28, 2019
Messages
477
Reaction score
269
Age
47
Location
Austin, Texas
Just a project share mapping and modeling with the H520. Maps use a PPK method and the modeling uses Carlson Precision 3D Topo. Image processing, storage and sharing through DroneDeploy.

 
  • Like
Reactions: arruntus

arruntus

Premium Pilot
Joined
Aug 19, 2017
Messages
1,851
Reaction score
729
What a great day you picked, the air looks clean.

The 360 panorama has pretty good resolution, it looks great.

Thanks for sharing :D
 
  • Like
Reactions: chascoadmin

slw74

Part 107 certificate
Premium Pilot
Joined
Sep 5, 2018
Messages
94
Reaction score
77
Age
69
Location
Matthews, North Carolina
Wow, a very nice display of your work. Amazing 360. I would be interested in how you setup the mission so I can learn more about how to use the H520 more effectively being as I am a newbie to its capabilities.
Thanks,
Stephen
 
Joined
Nov 28, 2019
Messages
477
Reaction score
269
Age
47
Location
Austin, Texas
Wow, a very nice display of your work. Amazing 360. I would be interested in how you setup the mission so I can learn more about how to use the H520 more effectively being as I am a newbie to its capabilities.
Thanks,
Stephen
Thank you. This was a pretty good cross-section of work as far as what we do from day to day. Ask away and I will answer what I can!
 
Joined
Jul 11, 2018
Messages
226
Reaction score
81
Location
Poland
Hi there,

@chascoadmin - these are very good jobs! Impressed as with the provious projects you shared.
I have a feeling that the present ones (2 d maps) are consistent in terms of GSD and flight paths. Am I right? Could you please for the sake of us learning from yours vast experience, share flights' details: alt., gsd, overlap, camera you used and its settings etc.? Was the camera set for unprocessed or any other color mode?

How about 3 d models - was the camera set for nadir or anything else?


Is there any specific reason you use DroneDeploy? Is it any better than the other solutions?

Great jobs.


Greetings from Poland,
Pawel
 
Joined
Nov 28, 2019
Messages
477
Reaction score
269
Age
47
Location
Austin, Texas
Hi there,

@chascoadmin - these are very good jobs! Impressed as with the provious projects you shared.
I have a feeling that the present ones (2 d maps) are consistent in terms of GSD and flight paths. Am I right? Could you please for the sake of us learning from yours vast experience, share flights' details: alt., gsd, overlap, camera you used and its settings etc.? Was the camera set for unprocessed or any other color mode?

How about 3 d models - was the camera set for nadir or anything else?


Is there any specific reason you use DroneDeploy? Is it any better than the other solutions?

Great jobs.


Greetings from Poland,
Pawel
Hi Pawel, thank you for your interest in my maps! Below is a description of a typical flight and more detail on the RBJ site.

Obviously the first thing we have to do is look for obstructions and the highest point or top of structures to make sure that we get clearance, but define the pattern so that we get the best GSD possible.

My default settings are 75% overlaps both side and front. Altitude being 200ft. The main adjustments that I will make our according to that high point in which I decrease my side overlaps and increase my front. In this project the tallest building was 160ft and I like to maintain 50ft above that structure, both for clearance and decent shots of the roof. This structure took me to 220ft while maintaining the 75% front and side overlaps. This provided a GSD of 0.69in/px. I always make sure I am at 1 in/px or lower. Because there was a very tall structure and more particularly several buildings in a non-uniform configuration I chose to do a manual flight oblique flight. This was again done at 220ft. Camera pitch was at 55 degrees. And exposure was zero.

The flight was performed with an H520 and an E90 camera. Camera settings were with no processing. I perform post-processing with ACDSee to minimize blowout, shadows and increase color saturation. I think after trying many different softwares that ACDC is the best option for non-destructive batch editing of imagery of this nature. I have one filter that I initially run every time and do fine tuning if necessary after that.

The H520 was fitted with an Emlid M+ module on a custom configuration that I created. You can search both the forum.dronedeploy.com and community.emlid.com forums for more details on that setup. This allowed me to set a base station and collect gnss data for PPK processing. While I do use PPK, I also set ground control points on every map also utilizing a Emlid RS+ base station and an additional Emlid RS+ receiver as a rover. The entire setup cost me about $2,000. This brought the total cost of the Drone solution to about $6,000.

You can see by the 6th picture that I flew around the perimeter of each building and was able to capture all sides and the deep enclosed areas. Those actually came out better than I expected because of the heavy shadows, but I am certain that the image post-processing, reducing those shadows and acquiring more detail led to a better model.

I choose to use DroneDeploy because of our corporate environment and my need to make sharing and utilization by the rest of our project management team as easy as possible. Their user interface is by far the easiest to navigate and analyze the main content that they are interested in. Those are the ortho mosaic with CAD overlay, the 360 pano and the materials volume metrics. drone deploy has been a very good partner and in my opinion has the best support and forum related to mapping though the forum is also attended by several pilots from different industries so we are able to learn from each other. the only downfall that I have found to drone deploy so far which is not necessarily a negative point in our situation is that because they are processing a very high number of maps that their settings on point cloud creation medium to medium-high settings. this produces a very good terrain model, but can be a little short on very intricate structures. I work closely with them in a beta roll to resolve these issues.

I do maintain a copy of Agisoft Metashape for special cases where a map on DroneDeploy is not quite detailed enough or is low on the points count for the point cloud, but that is only about 1 in 10 cases.

Overall we have chosen to move to the H520 as our default mapping platform because of the retirement of the DJI Phantom 4 Pro. Even if the P4P hadn't been retired I probably would have moved to the H520 because of its versatility. We are working with them on the survey application in their DataPilot software. We have retired all but one of the P4P's and still use it for some progress photo and video applications.
 
Joined
Jul 11, 2018
Messages
226
Reaction score
81
Location
Poland
Hi Pawel, thank you for your interest in my maps! Below is a description of a typical flight and more detail on the RBJ site.

Obviously the first thing we have to do is look for obstructions and the highest point or top of structures to make sure that we get clearance, but define the pattern so that we get the best GSD possible.

My default settings are 75% overlaps both side and front. Altitude being 200ft. The main adjustments that I will make our according to that high point in which I decrease my side overlaps and increase my front. In this project the tallest building was 160ft and I like to maintain 50ft above that structure, both for clearance and decent shots of the roof. This structure took me to 220ft while maintaining the 75% front and side overlaps. This provided a GSD of 0.69in/px. I always make sure I am at 1 in/px or lower. Because there was a very tall structure and more particularly several buildings in a non-uniform configuration I chose to do a manual flight oblique flight. This was again done at 220ft. Camera pitch was at 55 degrees. And exposure was zero.

The flight was performed with an H520 and an E90 camera. Camera settings were with no processing. I perform post-processing with ACDSee to minimize blowout, shadows and increase color saturation. I think after trying many different softwares that ACDC is the best option for non-destructive batch editing of imagery of this nature. I have one filter that I initially run every time and do fine tuning if necessary after that.

The H520 was fitted with an Emlid M+ module on a custom configuration that I created. You can search both the forum.dronedeploy.com and community.emlid.com forums for more details on that setup. This allowed me to set a base station and collect gnss data for PPK processing. While I do use PPK, I also set ground control points on every map also utilizing a Emlid RS+ base station and an additional Emlid RS+ receiver as a rover. The entire setup cost me about $2,000. This brought the total cost of the Drone solution to about $6,000.

You can see by the 6th picture that I flew around the perimeter of each building and was able to capture all sides and the deep enclosed areas. Those actually came out better than I expected because of the heavy shadows, but I am certain that the image post-processing, reducing those shadows and acquiring more detail led to a better model.

I choose to use DroneDeploy because of our corporate environment and my need to make sharing and utilization by the rest of our project management team as easy as possible. Their user interface is by far the easiest to navigate and analyze the main content that they are interested in. Those are the ortho mosaic with CAD overlay, the 360 pano and the materials volume metrics. drone deploy has been a very good partner and in my opinion has the best support and forum related to mapping though the forum is also attended by several pilots from different industries so we are able to learn from each other. the only downfall that I have found to drone deploy so far which is not necessarily a negative point in our situation is that because they are processing a very high number of maps that their settings on point cloud creation medium to medium-high settings. this produces a very good terrain model, but can be a little short on very intricate structures. I work closely with them in a beta roll to resolve these issues.

I do maintain a copy of Agisoft Metashape for special cases where a map on DroneDeploy is not quite detailed enough or is low on the points count for the point cloud, but that is only about 1 in 10 cases.

Overall we have chosen to move to the H520 as our default mapping platform because of the retirement of the DJI Phantom 4 Pro. Even if the P4P hadn't been retired I probably would have moved to the H520 because of its versatility. We are working with them on the survey application in their DataPilot software. We have retired all but one of the P4P's and still use it for some progress photo and video applications.
Hi,

many thanks for your prompt and very comprehensive explanation (as always).

Appreciate your tips. They are very valuable and even though I am more into ag mapping/surveying they are very helpful. Careful mission planning is really key for success. Ag mapping might not involve high building and/or structures yet there are normally still a lot of other obstacles like trees, and power lines and energy windmills etc. one needs to take precautions of. They are really a headache especially when flying a fixed wing.


I will try to find your posts at dronedeploy and emlid forums touching on ppk/rtk issue. Thanks. PPK/RTK are also used in ag mapping.

One final question about e90, if I may - what are the h520 speeds which are safe for stitching the camera's output? E90 is a rolling shutter camera and therefore my question is whether you have experienced any difficulties in that field. I know that it will depend among other things on a light condition as well as altitude but having this in mind I tend to err on a safe side and usually fly not faster than 5 m/s (this translates to 11,2 mile/hour).

Pawel
 
Joined
Nov 28, 2019
Messages
477
Reaction score
269
Age
47
Location
Austin, Texas
Hi,

many thanks for your prompt and very comprehensive explanation (as always).

Appreciate your tips. They are very valuable and even though I am more into ag mapping/surveying they are very helpful. Careful mission planning is really key for success. Ag mapping might not involve high building and/or structures yet there are normally still a lot of other obstacles like trees, and power lines and energy windmills etc. one needs to take precautions of. They are really a headache especially when flying a fixed wing.


I will try to find your posts at dronedeploy and emlid forums touching on ppk/rtk issue. Thanks. PPK/RTK are also used in ag mapping.

One final question about e90, if I may - what are the h520 speeds which are safe for stitching the camera's output? E90 is a rolling shutter camera and therefore my question is whether you have experienced any difficulties in that field. I know that it will depend among other things on a light condition as well as altitude but having this in mind I tend to err on a safe side and usually fly not faster than 5 m/s (this translates to 11,2 mile/hour).

Pawel
Ag isn't a whole lot different besides the hardware than what I typically do. Our scope on this project has been done for 6-months now so I am flying for the owner to completion. Only about 20% of our work is building construction so my typical focus is on civil excavation, utilities and concrete. Capturing good horizontal and terrain elevations are more important there.

Curious on the Ag as I have talked to a couple of farmers in our area about next grow season and was wondering what software you are using for analysis and collaboration? Also, do you find that flying parallel with the rows produces better imagery and point clouds?

On the E90, I typically setup the mission at 75% overlaps, observe the photo interval and the mission duration and then go back to the Mission Start pane and manipulate the speed. As you can see this plan is a 3.6s interval (super-slow) and 28m long.

1577989108279.png

Going back and forth between that and the bottom of the Survey pane I get the speed set to what keeps the mission duration the same to see what the program is actually using. There may be an easier way to see it, but mine always seems to default back to 11mph when I check the box.

1577989209410.png

I then up the speed until I get the Photo Interval on the Survey pane to about 2 seconds. I have run down to 1.6 seconds, but that is totally dependent on whether or not the card you use can write that fast and lighting conditions. In essence I am getting more front overlaps. From measuring my maps I am getting about 230ft on the Y-axis at a 220ft altitude. This tells me that I need a 55ft trigger distance to accomplish 75% overlap. I know DroneDeploy basis all of their flight plan parameters according to the specs on the P4P and then it adjusts depending on what drone you actually connect to at flight, but I am not sure of the algorithm on DataPilot. In my calculations it is not correct. On this mission I got a 53.6s photo interval at 20mph! I examined the photos in comparison to a still shot and of course the are slightly blurred, but in comparison to a shot taken by the DataPilot default there is no difference.

1577991423484.png
1577991512090.png

This leads into your question about the rolling shutter. Across many drones I have tested, if you leave the camera parameters on auto a rolling shutter is always more blurred when compared to a global shutter. This is one reason that really peeves me about DJI letting the P4P go. It is by far the best camera for mapping at a price point that is good for most people. Since you fly a fixed-wing I assume that your camera is as good or better, but for multi-copters with built-in cameras I haven't seen a competitor. I am awaiting the release of the Autel EVO 2. I don't want to go off on a tangent, but it is smaller and worries me that it will not be able to handle wind conditions in our area as well and it surely will not get the battery life as advertised. All of that said, most processing softwares have corrections for rolling shutters so I have seen only slightly diminished in accuracies. In the order of 0.2in on my GCP RMSEs.
 

slw74

Part 107 certificate
Premium Pilot
Joined
Sep 5, 2018
Messages
94
Reaction score
77
Age
69
Location
Matthews, North Carolina
chascoadmin,
Wow! You have provided an incredible wealth of information on how to advantage the H520 features and just how you lay that out for success. I am overwhelmed trying to understand all of the details, but it is exciting to try to sort it all out while seeing how you've done it. I sincerely appreciate your willingness to help others understand how to use this new technology in a very meaningful way!
All the best,
Stephen
 
  • Like
Reactions: chascoadmin
Joined
Nov 28, 2019
Messages
477
Reaction score
269
Age
47
Location
Austin, Texas
chascoadmin,
Wow! You have provided an incredible wealth of information on how to advantage the H520 features and just how you lay that out for success. I am overwhelmed trying to understand all of the details, but it is exciting to try to sort it all out while seeing how you've done it. I sincerely appreciate your willingness to help others understand how to use this new technology in a very meaningful way!
All the best,
Stephen
I appreciate that and do what I can and wish I could be more of an influence than what I have been able to be recently on the beta team. It seems that Yuneec is choosing to shutdown many platforms that we have been using and communication has been at a minimum. I sincerely believe that this could be a premier mapping platform, especially with the intro of the Leica Ion camera, if they would allow me to assist and actually run some testing. It is shown with the H3, but in my experience with their releases of new models they are not very different than the predecessors and I hope there is a way I can get that camera on one of my H520's. I am pretty confident that in combination with continued testing of the survey solution would surpass what we were doing with the P4P and even the P4RTK if we can do some work on the H520 RTK model that is supposed to be release in February.
 
Joined
Jul 11, 2018
Messages
226
Reaction score
81
Location
Poland
Curious on the Ag as I have talked to a couple of farmers in our area about next grow season and was wondering what software you are using for analysis and collaboration? Also, do you find that flying parallel with the rows produces better imagery and point clouds?
Hi chascoadmin,

Thanks again for your feedback.

As for ag mapping and our experience - we feel that flying perpendicular or so to the plants' rows gives somehow better results. But actually a lot depends on various factors such as e.g. wind speed and its direction. I always try to fly perpendicular to the wind to secure that no picture is mistaken/misrecorded due to wind pushing my drone to fly faster than a camera can trigger and write. Furthermore this approach let me fly slightly longer as the drone does not need to fight the wind flying against it. So if only the wind allows I try to avoid flying parallel to the rows. I know that a multirotor is not that susceptible to wind pushing it but at a cost of faster battery drainage. Anyway my fixed wing habits apply to h520.

I have also noticed that this approach helps flying over very homogeneous fileds with a complete canopy coverage.

Other thing is what you and your farmers want to get out of you lying their fields - is it about vegetation indices (biomass, vigor, variable rate applications etc.) or things like plant count, gap detection, flowering density etc.? These would employ different cameras and approach.


As for the software we use pix4d but to my knowledge the drone deploy you use offers similar capabilities as for vi calculation and sharing.

Pawel
 
Joined
Nov 28, 2019
Messages
477
Reaction score
269
Age
47
Location
Austin, Texas
Hi chascoadmin,

Thanks again for your feedback.

As for ag mapping and our experience - we feel that flying perpendicular or so to the plants' rows gives somehow better results. But actually a lot depends on various factors such as e.g. wind speed and its direction. I always try to fly perpendicular to the wind to secure that no picture is mistaken/misrecorded due to wind pushing my drone to fly faster than a camera can trigger and write. Furthermore this approach let me fly slightly longer as the drone does not need to fight the wind flying against it. So if only the wind allows I try to avoid flying parallel to the rows. I know that a multirotor is not that susceptible to wind pushing it but at a cost of faster battery drainage. Anyway my fixed wing habits apply to h520.

I have also noticed that this approach helps flying over very homogeneous fileds with a complete canopy coverage.

Other thing is what you and your farmers want to get out of you lying their fields - is it about vegetation indices (biomass, vigor, variable rate applications etc.) or things like plant count, gap detection, flowering density etc.? These would employ different cameras and approach.


As for the software we use pix4d but to my knowledge the drone deploy you use offers similar capabilities as for vi calculation and sharing.

Pawel

Great information Pawel thanks for sharing! I will definitely keep it in mind as I continue discussions. What fixed-wing do you fly? Do you use a multispectral camera as well?
 

New Threads

Forum statistics

Threads
20,048
Messages
231,821
Members
24,392
Latest member
TomF