Litchi: distortion between the GPS coordinates of a pre set task and the reality on the ground?

[Bucho]

Hello
I'm starting on Litchi software.
After having pre set a WayPoints mission on Litchi - Hub, I observe a GPS positioning problem once arrived on the perimeter of flyover.
Indeed, once arrived at the scene I download the mission then I ask Litchi to run it : to my surprise, the drone is more than 100 meters away to fetch the WayPoint 1 !
It is as if the GPS was distorted.
Here are my equipment and software versions:
- Phantom 3 Professional, DJI Go 2.7.2, Remote controller 1.6.0 - firmware 8.1.80;
- Litchi 3.3.0;
- smartphone Samsung S6 Edge+ acting as a hotspot for the tablet Samsung Tab S2 connected to the Remote Controller;
- on the field, the GPS position ans the localisation on the map, for Home, is perfectly accurate.
Can you help me to correct the offset between GPS coordinates in the pre set mission and the reality on the ground?
Thank you.

 

[JockC]

Hello
Indeed, once arrived at the scene I download the mission then I ask Litchi to run it : to my surprise, the drone is more than 100 meters away to fetch the WayPoint 1 !

I don't see that problem with my Litchi flights (iPad).
It may seem like a very basic question but do you have rounded corners turned on for your waypoints (curve size greater than 0)? If so, then the drone will never actually reach the individual waypoints. Instead, it will round off all corners which will result in the drone starting it's turn before it actually reaches the next waypoint.
On your planning map, the yellow lines will show the direct link between each successive waypoint whilst the light blue lines will show the actual path that the drone will fly. If curved corners are set then those light blue lines may be considerably short of the WP.

 

[Bucho]

I don't see that problem with my Litchi flights (iPad).
It may seem like a very basic question but do you have rounded corners turned on for your waypoints (curve size greater than 0)? If so, then the drone will never actually reach the individual waypoints. Instead, it will round off all corners which will result in the drone starting it's turn before it actually reaches the next waypoint.
On your planning map, the yellow lines will show the direct link between each successive waypoint whilst the light blue lines will show the actual path that the drone will fly. If curved corners are set then those light blue lines may be considerably short of the WP.

Thanks JockC.
In fact my question seems basic, I have no doubt that Litchi works perfectly well on such a subject.
What I seek are the proper settings for the missions.
Here are some details on what I asked:
- Guidance mode: Waypoint
- Fashion path: rounded corners
- Radius default curve: 75%
The orientation to my POI is set in my waypoints, for example one I see Litchi planned:
- Bend radius: 12 m,
- Orientation: 103 °
- POI: 1
Greetings.
The mission is at : Mission Hub - Litchi

 

[JockC]

Thanks JockC.
In fact my question seems basic, I have no doubt that Litchi works perfectly well on such a subject.
What I seek are the proper settings for the missions.
Here are some details on what I asked:
- Guidance mode: Waypoint
- Fashion path: rounded corners
- Radius default curve: 75%
The orientation to my POI is set in my waypoints, for example one I see Litchi planned:
- Bend radius: 12 m,
- Orientation: 103 °
- POI: 1
Greetings.
The mission is at : Mission Hub - Litchi

Thanks for the flight info. I'm not sure if my response addressed your problem but if you look at your waypoint 10, for example, you will see that the actual flight path (blue line) will be some distance from the actual waypoint.
The remainder of your settings are pretty much as I would set up my own. One thing you may wish to set is the action the drone is to take when it gets to the last waypoint. You can get it to go home, go back to waypoint no. 1 etc.

 

[Bucho]

Thanks for the flight info. I'm not sure if my response addressed your problem but if you look at your waypoint 10, for example, you will see that the actual flight path (blue line) will be some distance from the actual waypoint.
The remainder of your settings are pretty much as I would set up my own. One thing you may wish to set is the action the drone is to take when it gets to the last waypoint. You can get it to go home, go back to waypoint no. 1 etc.

I have not left the mission continue, seeing the way things were going!
Indeed, the search for WayPoint 1 by the drone had significantly distant from the left edge of the lake (the drone was already beyond the Waypoint 8): If I had left things going on, it would have been possible that the signal will eventually be lost !
Thank you.

 

[Mark The Droner]

I've noticed that GPS fixes high in the air are not directly above the same GPS fixes on the surface of the earth. That's probably because there is no such thing as GPS fixes high in the air. The angles of satellite reception are different so the fix high in the air is a little off. It shouldn't be 100 meters off though.

 

[Bucho]

I've noticed that GPS fixes high in the air are not directly above the same GPS fixes on the surface of the earth. That's probably because there is no such thing as GPS fixes high in the air. The angles of satellite reception are different so the fix high in the air is a little off. It shouldn't be 100 meters off though.

Thanks, Marknmd.
The explanation that you give should have a rather negligible influence on the ground, or the drone pretended to go towards the Waypoint 1, then he went straight to the Waypoint 8 and went to overtake!
Pre task set seemed perfectly ignored!

 

[JockC]

The mission is at : Mission Hub - Litchi

Just as an aside, I noticed that you have actions at each of the waypoints (eg take photo, stop for 1 second etc). None of those actions will work when you have curved turns at those waypoints because the drone will never actually arrive at that waypoint. This may not be relevant to your current issue but it's worth noting, in any case. It actually has a note re this just above the actions section in each waypoint setting.

 

[John Decker]

I have experienced a similar offset, but it is within the parameters of gps accuracy. That is, gps signals tend to be accurate to within 1-3 metres. On the ground, my missions usually start about 2m away from where I drew them on the map.
100 metres would indicate something else is wrong. Either you're downloading the wrong mission or the water is playing tricks, or..?
Can you run another, basic (say 2 or 3 waypoints) mission in a field or park just to check and compare?

 

[Bucho]

Just as an aside, I noticed that you have actions at each of the waypoints (eg take photo, stop for 1 second etc). None of those actions will work when you have curved turns at those waypoints because the drone will never actually arrive at that waypoint. This may not be relevant to your current issue but it's worth noting, in any case. It actually has a note re this just above the actions section in each waypoint setting.

Yes JocjC, I knew all this, I recorded these actions so that they are active as soon as I switch to straight lines instead of curved turns.
Thank you.

 

[Bucho]

I have experienced a similar offset, but it is within the parameters of gps accuracy. That is, gps signals tend to be accurate to within 1-3 metres. On the ground, my missions usually start about 2m away from where I drew them on the map.
100 metres would indicate something else is wrong. Either you're downloading the wrong mission or the water is playing tricks, or..?
Can you run another, basic (say 2 or 3 waypoints) mission in a field or park just to check and compare?

Good comments, John, thank you.
I do not have the opportunity to find myself back at Lake Cruseilles, but you are right: I have already experienced a course with 6 WayPoints, pre configured for the land of my model aircraft club ; again the drone went for the Waypoint 1 ... but (only) a dozen meters away.
For the Lake Cruseilles, there were 12 Waypoints, but this number does not seem significant: configuring from 6 to 12 waypoints can not imply that the potential distortion should be from 10 to 100 meters.
There is something I do not perhaps understand in the settings in Litchi?

 

[John Decker]

Simplify things. Maybe there is something in the misson or the location that is throwing things off.. Anything more than 3 metres is something other than a gps fault.

 

[alokbhargava]

Did you wait for the home point set before running the mission?

Why do you say your actual waypoint 1 was 100 m away from the set coordinates?

 

[Wolfiesden]

I've noticed that GPS fixes high in the air are not directly above the same GPS fixes on the surface of the earth. That's probably because there is no such thing as GPS fixes high in the air. The angles of satellite reception are different so the fix high in the air is a little off. It shouldn't be 100 meters off though.

Um...I don't think so.

GPS was originally implemented for military use aboard aircraft and missiles along with surface vehicles. In fact, one of the factors that lead to its being opened to civilian uses was the shooting down of flight 007 and the Regan administration began the process to open up GPS for the SPECIFIC use of aircraft, virtually ALL of which fly higher than we are permitted to fly our drones. Only later did civilian uses on the ground become popular.

Later Clinton/Gore pressed to have selective availability shut down, again, for the express purpose of aviation safety and accuracy.

If it was so fundamentally flawed for use in the air, I really doubt that multiple presidents would have staked their reputations on making it available to the aviation industry and highly doubtful that the military would have adopted its use in combat if it couldn't be relied upon for accurate positions.

 

[Wolfiesden]

Not really sure what is causing your issues. I am running Litchi 3.4.0 (recent release) using a similar setup as you with my Galaxy S5 as a hotspot for my nVidia K1 which runs Litchi. This past weekend, I ran this mission and it operated flawlessly.
Mission Hub - Litchi

My home point was on the shore near points 1 and 24. I lost signal between 12 and 18 being behind trees on the point.

 

[Mark The Droner]

Um...I don't think so.

GPS was originally implemented for military use aboard aircraft and missiles along with surface vehicles. In fact, one of the factors that lead to its being opened to civilian uses was the shooting down of flight 007 and the Regan administration began the process to open up GPS for the SPECIFIC use of aircraft, virtually ALL of which fly higher than we are permitted to fly our drones. Only later did civilian uses on the ground become popular.

Later Clinton/Gore pressed to have selective availability shut down, again, for the express purpose of aviation safety and accuracy.

If it was so fundamentally flawed for use in the air, I really doubt that multiple presidents would have staked their reputations on making it available to the aviation industry and highly doubtful that the military would have adopted its use in combat if it couldn't be relied upon for accurate positions.

A GPS fix is a point. It's not a line. You don't agree?

I wouldn't call it "fundamentally flawed" since it does what it's supposed to do.

I have noticed that my aircraft will often hover at a GPS point in the air and it will not be directly above the same GPS point on the ground (e.g. RTH descent). So this makes me think it's because of what I described. But I acknowledge it's possible the fix simply reaccessed and shifted.

The way you seem to describe GPS is that every fix is a vertical line, not a point, and that line is perfectly perpendicular to the earth's surface. Is that right?

 

[Meta4]

The way you seem to describe GPS is that every fix is a vertical line, not a point, and that line is perfectly perpendicular to the earth's surface. Is that right?

A GPS location is a 2-dimensional reference with no z co-ordinate.
Unless you specify an altitude above a datum, the point could be anywhere on a line through the point and the centre of the earth.

 

[sar104]

A GPS location is a 2-dimensional reference with no z co-ordinate.
Unless you specify an altitude above a datum, the point could be anywhere on a line through the point and the centre of the earth.

More correctly, a GPS location is represented as a 2-D set of plane coordinates on a mathematical repesentation of the earth's MSL surface (the projection) plus an optional, orthogonal (vertical) coordinate above MSL for that projection if the receiver has enough satellites locked for a 3-D position fix. The plane coordinates define a line normal to the projected surface of the earth so, since most projections are not based on a spherical earth, that line will not, in general, pass through the earth's center (except at the poles or on the equator).

In any case, it would be correct to say that there should be no horizontal offset with altitude.

 

[Wolfiesden]

Well said gentlemen.

 

[Mark The Droner]

So, in summary, it's a perfect vertical line, and perfectly perpendicular to the earth's surface. Is that right?