3d mapping training for yuneec h520 & starting in 3d mapping

[arruntus]

The idea is that yes, even if Google Earth is not reliable with heights. No problem, there are free databases that have DSMs from all over the world with an accuracy of 30m. It is not necessary anymore, the GSD we see in the programs is never accurate, it is an average. If you manage to lower that average you gain in quality.

 

[claudius62]

Well, I summarize.
On my drawing, Departure to 323 meters, I want to hover 50 meters above the highest point of the area (361 meters).
So I have to put 361 - 323 + 50 so 88 (meters) in the Altitude box.
I'm good or I have to go back to bed!

Check the box

 

[Wingman]

I may be totally wrong but as far as I know a true altitude is a vertical position above mean sea level (MSL) but you are flying above an elevated area with an elevation reaching up to 361 m. If a drone altitude is true one then putting 88m will make it well below AGL.

 

[claudius62]

I do not think you're right, because the reference altitude at takeoff of the drone is 0, if you want to make a route 30 m above your take-off point, you put the value 30 and it will make the route to 30 m above takeoff. if at the second point you put 50 m, it will go up to 50m and will be 50 m above your takeoff point.
The altitude you put is always relative to the take-off point.

 

[Wingman]

I have actually googled and found this Question on how Altitude is measured which explains that a drone altitude is not a true one. When a drone takes off it is set to 0 tn a launch site ground level no matter how elevated it is above MSL. So it is not altitude in aviation terms it is AGL. That's what I thought from the beginning but an Altitude term put me of that thought. So in your case 88m seems to be the right number.

 

[Wingman]

Thanks for correction claudius. I was wrong. I was writing the same after some research and at the same time you posted your reply.

 

[OnTheRopes]

So if the H520 cannot currently handle mapping at a uniform altitude above ground level using DroneDeploy, can other drones manage this, I need to be able to produce accurate topographical surveys on challenging terrain

 

[arruntus]

Few systems are capable of tracking the terrain. Of the best known ones that allow this, one is Ardupilot and the other is UGCS.

Don't get crazy about heights. The height 0 AGL with which you normally work on a drone is based on the height at sea level. As the instruments, barometer, altimeter are not 100% accurate you always work with a safety margin. If someone wants to fly over a mountain, it is not advisable to leave a margin of only 10 m because that failure in the readings can end in disaster.

 

[OnTheRopes]

Few systems are capable of tracking the terrain. Of the best known ones that allow this, one is Ardupilot and the other is UGCS.

Don't get crazy about heights. The height 0 AGL with which you normally work on a drone is based on the height at sea level. As the instruments, barometer, altimeter are not 100% accurate you always work with a safety margin. If someone wants to fly over a mountain, it is not advisable to leave a margin of only 10 m because that failure in the readings can end in disaster.

Okay, but if it does not track the height, can the software still stitch together van accurate topo survey if the terrain is hilly?

 

[Tuna]

Stitching will be accurate, but the detail might vary. It's like looking at a normal photograph - things closer to the camera have more visible detail. A chosen GSD is nominal at best, but gives a target for a survey.

Unfortunately, ground elevation data is not reliably available around the world, and the density of information changes in different regions, so even across the US you can find areas which have better data and areas which have worse. It's difficult to measure (satellites have to compensate for vegetation, buildings, snow cover and so on) and relies on a consistent reference (the world is far from a uniform sphere and has local gravitational and geographical bulges so things like sea level are not 'constant').

So any drone service offering terrain following has to include a lot of caveats - small local features are likely to be missed, some areas may have poor overage and the reference data may not be useful. When the ground within a survey area might change by more than the altitude of the drone, it's a recipe for disaster.

It's also worth remembering that the stitching software relies on a relatively level flight path to measure the heights of objects the camera sees. If you had a sloping survey height, the camera will see a ground surface that appears to be flat (always at the same distance from the camera). The software should compensate for that, but I've not managed to discover how accurate the analysis is under such conditions.

 

[OnTheRopes]

Stitching will be accurate, but the detail might vary. It's like looking at a normal photograph - things closer to the camera have more visible detail. A chosen GSD is nominal at best, but gives a target for a survey.

Unfortunately, ground elevation data is not reliably available around the world, and the density of information changes in different regions, so even across the US you can find areas which have better data and areas which have worse. It's difficult to measure (satellites have to compensate for vegetation, buildings, snow cover and so on) and relies on a consistent reference (the world is far from a uniform sphere and has local gravitational and geographical bulges so things like sea level are not 'constant').

So any drone service offering terrain following has to include a lot of caveats - small local features are likely to be missed, some areas may have poor overage and the reference data may not be useful. When the ground within a survey area might change by more than the altitude of the drone, it's a recipe for disaster.

It's also worth remembering that the stitching software relies on a relatively level flight path to measure the heights of objects the camera sees. If you had a sloping survey height, the camera will see a ground surface that appears to be flat (always at the same distance from the camera). The software should compensate for that, but I've not managed to discover how accurate the analysis is under such conditions.

That makes a lot of sense, I need to experiment with this preferably using the Typhoon H, I realise it will be difficult to get a uniform coverage of overlapped photos as it will have to be done using timelapse perhaps. I wonder if you have any plans to introduce mission planning into uavtoolbox, I am just contemplating buying it for some mapping trials with the H before deciding on a 520 or the other manufacturer.

 

[10-8]

That makes a lot of sense, I need to experiment with this preferably using the Typhoon H, I realise it will be difficult to get a uniform coverage of overlapped photos as it will have to be done using timelapse perhaps. I wonder if you have any plans to introduce mission planning into uavtoolbox, I am just contemplating buying it for some mapping trials with the H before deciding on a 520 or the other manufacturer.

Don't discount the ability to discern altitude changes in mapping software. It may not have the negative impact you fear. do your own maps and determine if it will work for your needs. If not, then you you can introduce GCPs for even greater accuracy. When mapping a hilly or sloping terrain, determine the height of the highest point in the map and set your mission altitude to around 30m above that point. Here are a couple of DSM models from mapping missions I have flown to show that even in areas that look flat on the Ortho, the DSM model can determine even the slightest change and slope.

 

[EneffTech]

Hello Wingman,

There are a number of good tutorials on youtube, of which some have already been mentioned by Claudius62. But I do know from experience you will need Ground Control Points (GCP's). The X and Y axis are not too bad, but it is the Z axis that needs some tying down.

Regards,

 

[OnTheRopes]

Don't discount the ability to discern altitude changes in mapping software. It may not have the negative impact you fear. do your own maps and determine if it will work for your needs. If not, then you you can introduce GCPs for even greater accuracy. When mapping a hilly or sloping terrain, determine the height of the highest point in the map and set your mission altitude to around 30m above that point. Here are a couple of DSM models from mapping missions I have flown to show that even in areas that look flat on the Ortho, the DSM model can determine even the slightest change and slope.

Thanks for the input, I have found a couple of companies offering courses on the subject, not sure how good they are but if they seem okay it should give me a start at least.

 

[Eyeup]

Here's another thought. If you want to use the 3D capability of the mapping apps then you can combine your nadir shots (camera strap down) with orbits at a couple of levels for more vertical detail. If you are doing this then the recommendation is that you set your altitude to 4x the highest point in the area (may be the topography or a structure. Remember to ensure sufficient overlap too. Check out the instructions on something like MapsMadeEasy. Their app may not work with Yuneec but however you capture your images, as long as they meet their spec your good to go. Here's a cheeky example for you Wilkinson Drive, and here's the 3D version. 3D Wilkinson Drive

 

[arruntus]

Here's another thought. If you want to use the 3D capability of the mapping apps then you can combine your nadir shots (camera strap down) with orbits at a couple of levels for more vertical detail. If you are doing this then the recommendation is that you set your altitude to 4x the highest point in the area (may be the topography or a structure. Remember to ensure sufficient overlap too. Check out the instructions on something like MapsMadeEasy. Their app may not work with Yuneec but however you capture your images, as long as they meet their spec your good to go. Here's a cheeky example for you Wilkinson Drive, and here's the 3D version. 3D Wilkinson Drive

A good example of a combination of the 3 flight modes required for a photogrammetry project. Nadir, that the H520 does it, oblique (in this we are in doubt since the last update) and circular, at different heights (for now can not).

 

[Eyeup]

A good example of a combination of the 3 flight modes required for a photogrammetry project. Nadir, that the H520 does it, oblique (in this we are in doubt since the last update) and circular, at different heights (for now can not).

So can't the H520 carry out an orbit POI mission with the camera gimbal set at 45deg (for example) and taking a set of interval images?

 

[Graham]

The H520 does not have POI. DataPilot can be programmed to fly surveys (and waypoint to waypoint) in straight lines only. Until the recent update, the camera could be programmed to fly obliques in survey mode, but now does nadir only. Nobody has provided a reason why oblique was removed and I can't think of one.
If you want to see what DataPilot can do, stand-alone versions are available here;
Downloads - Yuneec

 

[OgreCorsair]

Hey guys I just watched this video put out by drone u. I was hoping to get some thoughts on it is it accurate is there something they're missing or is the Phantom Pro 4 that much better than the typhoon?

 

[Kev Waite]

Hey guys I just watched this video put out by drone u. I was hoping to get some thoughts on it is it accurate is there something they're missing or is the Phantom Pro 4 that much better than the typhoon?

The guy obviously has never seen a H520 or Typhoon H fly when it loses a prop at 2.30. It does not spin around and stays completely under control.