Terrain Aware or Fixed altitude for Mapping (Ortho, 3D and cut/fill calculation)

[queeg730]

Hi folks,
I want to create maps / surveys as accurately as I can, but see a lot of conflicting messages re using terrain awareness as opposed to simple fixed altitude flying.

What are the thoughts on using mapping apps (Map Pilot / Drone Deploy etc), WITH as opposed to WITHOUT terrain awareness for this type of work?

Being something of a newbie at mapping, (although an experienced UAV operator) I would be greatly indebted for any ideas / suggestions as to the various pros & cons.

Thanks in advance for any assistance here!
Mike

 

[R Martin]

Hi folks,
I want to create maps / surveys as accurately as I can, but see a lot of conflicting messages re using terrain awareness as opposed to simple fixed altitude flying.

What are the thoughts on using mapping apps (Map Pilot / Drone Deploy etc), WITH as opposed to WITHOUT terrain awareness for this type of work?

Being something of a newbie at mapping, (although an experienced UAV operator) I would be greatly indebted for any ideas / suggestions as to the various pros & cons.

Thanks in advance for any assistance here!
Mike

Fixed altitude does not take terrain (or your distance from the focal point to the ground) into account and your GSD varies. Terrain flight "follows" the terrain and maintains a constant height over the ground and gives you a stable GSD. If accuracy is your goal, then terrain following is another tool to use to ensure you are getting the best data possible.

I got away from the pay as you go apps a long time ago and switched to propriety software to run my aircrafts flight planning so I really can't comment on their usefulness any more.

 

[queeg730]

Fixed altitude does not take terrain (or your distance from the focal point to the ground) into account and your GSD varies. Terrain flight "follows" the terrain and maintains a constant height over the ground and gives you a stable GSD. If accuracy is your goal, then terrain following is another tool to use to ensure you are getting the best data possible.

I got away from the pay as you go apps a long time ago and switched to propriety software to run my aircrafts flight planning so I really can't comment on their usefulness any more.

Thanks R Martin for your reply...
I am leaning towards terrain aware for all mapping flights, and appreciate that it would result in a more consistent GSD for the ortho maps.... I wasn't so sure about the 3D models though.... but nevertheless, I think I will go with terrain aware for all.

thanks again mate.
Mike

 

[R Martin]

Thanks R Martin for your reply...
I am leaning towards terrain aware for all mapping flights, and appreciate that it would result in a more consistent GSD for the ortho maps.... I wasn't so sure about the 3D models though.... but nevertheless, I think I will go with terrain aware for all.

thanks again mate.
Mike

Just remember that each aircraft has its own GPS unit and the quality varies. My Inspire was good to about 2 meters. My big aircraft is much better but we use RTK to correct in real-time. Even given that, things go south so we also supplement GPS data with survey-grade control/check points. Even if the aircraft is spot on, you can PROVE that it is accurate with good control when you process your deliverables.

 

[clolsonus]

I would love to hear more feedback about drone deploy's new terrain aware feature used in the wild. One comment (granted from a while back) was that it may not properly consider the transit (commute) altitude when flying multi-battery missions. Due to issues with the DD app on our ipad mini 4, I've switched all flying over to pix4d capture which does some things better, some things worse.

For now all my flying projects have been put on hold, but when we were flying up to a week ago, we were flying in areas with significant terrain.

Just to share my misery: I made a pilot/planning error in my most recent mavic flight. The result was 3 broken props and a slightly broke/bent right front landing skid (the little plastic piece that is maybe 1.5" tall.) Basically my mistake was two-fold: (1) looking back I screwed up in making my area outline and covered some area I did not intend to cover, (2) I misunderestimated the amount of actual terrain rise at the one corner of our area. I believe I ended up flying through one tree and then made contact maybe 50-75' up on the second tree. I will try to attach some pictures that maybe tell the story.

I was able to produce a map right to the crash scene for whatever that's worth ... in the orthomap you can see the images getting smaller and smaller as the terrain rises up underneath. Again, totally my screw up ... planning and estimation mistakes, also due to road conditions we ended up launching from an area I hadn't anticipated which was towards the bottom of the area, not the top. Anyway, if anyone needs to chuckle at my screw up, go ahead, I deserve it!

Edit. I can estimate the ground elevation from the final pictures and know the mavic flight altitude. From that, I estimate the mavic tumbled approximately 55' to the ground. :-(

 

[R Martin]

I would love to hear more feedback about drone deploy's new terrain aware feature used in the wild. One comment (granted from a while back) was that it may not properly consider the transit (commute) altitude when flying multi-battery missions. Due to issues with the DD app on our ipad mini 4, I've switched all flying over to pix4d capture which does some things better, some things worse.

The last time I used Capture it did not have terrain following built in but that was over a year ago. Has that been added recently? I had to replace my Inspire with an H520-E90 (because that was the only Yuneec that was rated to work with Capture). It would still be really nice to have that feature available for the smaller aircraft.
Sorry to hear about you augering in. I slammed a wall indoors on my one attempt at indoor flight and struck that service from my offerings immediately. I am not the legend that I believe myself to be.

 

[clolsonus]

The last time I used Capture it did not have terrain following built in but that was over a year ago. Has that been added recently? I had to replace my Inspire with an H520-E90 (because that was the only Yuneec that was rated to work with Capture). It would still be really nice to have that feature available for the smaller aircraft.
Sorry to hear about you augering in. I slammed a wall indoors on my one attempt at indoor flight and struck that service from my offerings immediately. I am not the legend that I believe myself to be.

No, as far as I know, pix4d capture doesn't have any terrain following features at this time. But I like that it expects and is setup for being offline when you are in the field doing the actual flying (automatically downloads offline maps, etc.)

One nice thing pix4d has is contour lines on it's non-satellite base maps ... I wish I would have studied those just a little bit more closely when planning my last flight. I can go back now and totally see my screw up.

Hope everyone is staying safe out there and finding ways to pivot and offset lost income from jobs (or spouses jobs.)

 

[adm_geomatics]

I ran a DroneDeploy multi-battery TA mission yesterday and it worked well. As with @R Martin I normally map with my H520 now and it does not have TA, but that only matters if you have more than about 50ft in elevation change in your terrain from where you take off. Otherwise, it does make quite a big difference in the relative accuracy of your map just from a stitching and aerial triangulation perspective.

 

[HawkView]

I have been flying for stockpile measurements at a 90-acre quarry for several years that has more than a 400-foot elevation change. I have always used MapPilot with terrain Awareness. I tried using DroneDeploy a long while back when it was first beta testing but had different issues during battery changes.

I have used DroneDeploy's Terrain Awareness on several sites with less dramatic elevation change now that it is out of beta, and not had any issues. However, I have not tried it at the large quarry site yet, since I know MapPilot works there.

 

[queeg730]

I have been flying for stockpile measurements at a 90-acre quarry for several years that has more than a 400-foot elevation change. I have always used MapPilot with terrain Awareness. I tried using DroneDeploy a long while back when it was first beta testing but had different issues during battery changes.

I have used DroneDeploy's Terrain Awareness on several sites with less dramatic elevation change now that it is out of beta, and not had any issues. However, I have not tried it at the large quarry site yet, since I know MapPilot works there.

Thanks Hawkview.....
My quarry is about the same size... can I ask what elevation / altitude you use to map your site??? (eg 200 ft / 60 m with TA??)
thanks,
Mike

 

[HawkView]

300 feet with a 75/70 overlap. There is a sheer face on one side of the site, and between the inaccuracies of the terrain data and the aircraft GPS, I had a close call on one of the early flights so I bumped it up to be sure there was lots of room for error.

Thanks Hawkview.....
My quarry is about the same size... can I ask what elevation / altitude you use to map your site??? (eg 200 ft / 60 m with TA??)
thanks,
Mike

 

[queeg730]

300 feet with a 75/70 overlap. There is a sheer face on one side of the site, and between the inaccuracies of the terrain data and the aircraft GPS, I had a close call on one of the early flights so I bumped it up to be sure there was lots of room for error.

Fabulous!!!!!
thanks a million mate.

Mike

 

[adm_geomatics]

300 feet with a 75/70 overlap. There is a sheer face on one side of the site, and between the inaccuracies of the terrain data and the aircraft GPS, I had a close call on one of the early flights so I bumped it up to be sure there was lots of room for error.

It may not be possible in your scenario, but I generally set my flight path to run parallel with the contours so that the change in elevation to follow the surface is very minimal and hard grade breaks like that become allot safer.

 

[HawkView]

Correct, I also made that adjustment after I learned how to do that with MapPilot.

It may not be possible in your scenario, but I generally set my flight path to run parallel with the contours so that the change in elevation to follow the surface is very minimal and hard grade breaks like that become allot safer.

 

[queeg730]

It may not be possible in your scenario, but I generally set my flight path to run parallel with the contours so that the change in elevation to follow the surface is very minimal and hard grade breaks like that become allot safer.

I am familiar with setting TA in Map Pilot, but I am not sure that I understand the "run parallel with the contours" bit.. don't know how to set that in Map Pilot... Because the contours weave all over the place (on the horizontal plane I mean), how can I set flight path to run parallel with the contours....
Maybe I am missing the obvious here! :

 

[HawkView]

You can rotate the direction of the grid with two fingers on the screen. You have to determine the best direction for the site overall, it is still the same "lawnmower" pattern.

I am familiar with setting TA in Map Pilot, but I am not sure that I understand the "run parallel with the contours" bit.. don't know how to set that in Map Pilot... Because the contours weave all over the place (on the horizontal plane I mean), how can I set flight path to run parallel with the contours....
Maybe I am missing the obvious here! :

 

[adm_geomatics]

I am familiar with setting TA in Map Pilot, but I am not sure that I understand the "run parallel with the contours" bit.. don't know how to set that in Map Pilot... Because the contours weave all over the place (on the horizontal plane I mean), how can I set flight path to run parallel with the contours....
Maybe I am missing the obvious here! :

You'd need to review the terrain in something like Google Earth. Most tracts have a major direction that they fall, but of course there will still be some with perpendicular slopes that you just try to use the most common slope.

Something like this would be tough.

 

[adm_geomatics]

Parallel has one slope that would have my attention, but crossing the valley in a perpendicular route is quite a bit more active.




...and DroneDeploy. Notice how many less data points there are on the somewhat parallel (first) mission. I am sure they could be tuned better and worse respectively.

 

[yarrr]

Can anyone point me to literature that would suggest 'terrain following' would result in a more accurate point cloud, particularly when using high accuracy geotags (PPK)?

I understand the argument that the orthophoto could be better due to a more consistent pixel size, but I don't believe this affects the actual 3d reconstruction.

The entire concept of structure from motion is that key points are different distances from the camera- thus creating relief.

Perhaps a constant altitude makes sense for sites with large relief (like a quarry); but I'm skeptical for flatter sites, or for projects such as stockpiles.

Again, is there actual data available on this subject?

 

[adm_geomatics]

Can anyone point me to literature that would suggest 'terrain following' would result in a more accurate point cloud, particularly when using high accuracy geotags (PPK)?

I understand the argument that the orthophoto could be better due to a more consistent pixel size, but I don't believe this affects the actual 3d reconstruction.

The entire concept of structure from motion is that key points are different distances from the camera- thus creating relief.

Perhaps a constant altitude makes sense for sites with large relief (like a quarry); but I'm skeptical for flatter sites, or for projects such as stockpiles.

Again, is there actual data available on this subject?

I will contact an engineer and see what I can get, but you have to think about it with images per pixel and how that pixel looks. When tie-points are found they are matched and if an object looks different to the algorithm then it may not match it. This is the whole point of computer vision. Secondly if those points are different distances and thus angles that the point may not be seen combined with the matching problem then you end up with fewer tie-points for the reconstruction which means that you are missing possibly important features in between the points in the cloud. You are correct that the more intense the slopes the more important this becomes, but then that is the point of terrain following.