Litchi odd behavior when using multiple POIs

[ctaranto]

I saw a similar thread but the answer didn't address what I am seeing and the thread died.

I updated my M2Z firmware last week and wanted to try Litchi for the first time today with waypoints. For the most part, it worked awesome, with smooth transitions and accurate repeatable missions.

Until my last mission. 4 WPs and 2 POIs. The first two WPs pointing at POI 1 as "Focus POI". The third WP, still pointing at POI 1, set as Interpolate. WP four set to POI 2 and Interpolate. The WPs were pretty much in a straight line, about 1/4 mile in total distance (signal to controller was 5 bars the entire way).

The problem happens between WPs 1 and 2. The M2Z doesn't keep POI 1 at center while traveling between WP1 and WP2 (drifts a lot), but when it hits WP 2, POI 1 at center and is centered through WP 3. The interpolate between 3 and 4 is fine.

After confirming the right settings, I created a new mission to isolate the issue. Just WPs 1 and 2, same locations, height, speed, etc. and one POI. Worked perfectly (POI 1 at center the entire way).

I added a 2nd POI without any WPs pointing at it. The center point drifted the same way as before. Remove POI 2, perfect. Added POI 2 in a different location (this time way off to the other side of the mission). It drifted the same exact way.

Through all of these, I set WP 1 to pause for 5 seconds. When there were 2 POIs, the M2Z would point towards POI 1, but too low. After 5 seconds, it corrected the pointing, but then drifted as it moved. When there was 1 POI, the M2Z would point at 90-ish degrees for 5 seconds, and then when it started to move, immediately turned at POI 1 and kept it in center the entire way. This behavior was repeatable by removing and adding POI 2. Odd.

Is there a bug in Litchi that causes POIs to not be centered between points 1 and 2 when multiple POIs are used? I ran out of time to experiment with adding two WPs right next to each other to see if it was isolated to just 1 and 2.

 

[Aerial-Pixel]

I saw a similar thread but the answer didn't address what I am seeing and the thread died.

I updated my M2Z firmware last week and wanted to try Litchi for the first time today with waypoints. For the most part, it worked awesome, with smooth transitions and accurate repeatable missions.

Until my last mission. 4 WPs and 2 POIs. The first two WPs pointing at POI 1 as "Focus POI". The third WP, still pointing at POI 1, set as Interpolate. WP four set to POI 2 and Interpolate. The WPs were pretty much in a straight line, about 1/4 mile in total distance (signal to controller was 5 bars the entire way).

The problem happens between WPs 1 and 2. The M2Z doesn't keep POI 1 at center while traveling between WP1 and WP2 (drifts a lot), but when it hits WP 2, POI 1 at center and is centered through WP 3. The interpolate between 3 and 4 is fine.

After confirming the right settings, I created a new mission to isolate the issue. Just WPs 1 and 2, same locations, height, speed, etc. and one POI. Worked perfectly (POI 1 at center the entire way).

I added a 2nd POI without any WPs pointing at it. The center point drifted the same way as before. Remove POI 2, perfect. Added POI 2 in a different location (this time way off to the other side of the mission). It drifted the same exact way.

Through all of these, I set WP 1 to pause for 5 seconds. When there were 2 POIs, the M2Z would point towards POI 1, but too low. After 5 seconds, it corrected the pointing, but then drifted as it moved. When there was 1 POI, the M2Z would point at 90-ish degrees for 5 seconds, and then when it started to move, immediately turned at POI 1 and kept it in center the entire way. This behavior was repeatable by removing and adding POI 2. Odd.

Is there a bug in Litchi that causes POIs to not be centered between points 1 and 2 when multiple POIs are used? I ran out of time to experiment with adding two WPs right next to each other to see if it was isolated to just 1 and 2.

Have you tested it with VLM (Virtual Litchi Mission)? Here is a quick tutorial I've done on VLM, it might help you better pinpoint your issue.

 

[parkgt]

When people need help with a mission it always helps to post the mission so that it can be examined.

 

[ctaranto]

Thanks for the replies. I'll have to launch my Windows VM to try out VLM (I'm on Linux). I didn't want to post the mission because it's showing my house and privacy and all.

I was curious if there was a known issue with Litchi and having 2 POIs (even if one of them isn't being used).

I will also try to reproduce this not at my house and post that mission if need be.

Thanks...

 

[ctaranto]

VLM is pretty slick. Thanks for the video.

If the VLM is accurate, it looks like there may be a bug somewhere in either Litchi or the M2Z firmware.

The distance between WP 1 and 2 is about 210ft. The angle at WP 1 is 130 degrees; at WP 2 it's 244 degrees. The elevation at WP 1 is 12ft; at WP 2 it's 111ft. The POI has an elevation of 3ft and is much closer to WP 1 than WP 2.

If I add a few WPs along the way (say ~50 ft apart) and all point to POI 1, VLM shows the tracking to POI 1 perfectly. If I don't add those intermediate points, the tracking drifts like the real flight does.

I removed the height difference (kept the flight at 12ft). Didn't make a difference (in VLM).

I'll have to try this out for real (probably next weekend).

Any thoughts on long distances between WPs? Though 200ft doesn't seem very long to me...

 

[parkgt]

Any thoughts on long distances between WPs? Though 200ft doesn't seem very long to me...

Post the mission!

 

[ctaranto]

These are .csv exports of the mission set to a random house somewhere in MA.

Both have a single POI. The 3 point mission drifts way of the front of the house, and then back to the house, between points 1 and 2.

The more point mission has 3 additional points between points 1 and 2, and there is no drifting from the POI.

Hope this is what you were looking for...

 

[parkgt]

These are .csv exports of the mission set to a random house somewhere in MA.


Did you look at the file extensions? What I get on download are two text files; not two .cvs files.

Go to Mission Hub, load mission and then export as cvs and save. You should end up with a .csv not a text file(s) as above.

 

[ctaranto]

When I tried to upload .csv files, MavicPilots said invalid file format. So I renamed them to .txt. Just download and rename to .csv. They are .csv exports.

 

[parkgt]

When I tried to upload .csv files, MavicPilots said invalid file format. So I renamed them to .txt. Just download and rename to .csv. They are .csv exports.

Tried doing it for you but I can't change the file extension!

 

[ctaranto]

Odd. Probably Windows hiding the extension so you can't change it (there's a Windows setting to change that).

I renamed them back to .csv and zipped them. Let me know if that works. Thanks.

 

[parkgt]

Since neither of the missions you provided are anything like what you described in the first post it is impossible to guess what your problem was. Neither has 2 POIs or 4 WPs.

 

[ctaranto]

It's a simplified mission. In the virtual environment, the 2nd POI doesn't matter it seems. And I did not want to post a mission of where I live.

Did you notice that the in the mission with 3 WPs, the POI doesn't stay centered, but the mission with many more waypoints, it does center the POI?

 

[parkgt]

First off who would have known that is where you live if you didn't tell them? I set up missions on homes for RE jobs doesn't mean I live there. Do you think someone is out to find you?

The difference is not how many waypoints you have; it is how close the second is to the first. A second WP close to the first seems to more tightly focus the transition between them.

 

[ctaranto]

I'll not address the privacy concerns.

I agree that the tracking accuracy depends on how many waypoints are defined. And that doesn't seem right.

If I have 2 WPs and 1 POI, it shouldn't matter how far apart they are from the POI if they're both configured to point at the POI. Adding more WPs along the same line to "force" it to keep looking at a POI seems like a workaround to an issue.

If this is the way Litchi works, that's fine. I'll just keep inserting WPs when the distance between WPs exceeds 50 or so feet. But it seems like that should be unnecessary.

 

[parkgt]

I agree that the tracking accuracy depends on how many waypoints are defined. And that doesn't seem right.

That is not what I said.

 

[ctaranto]

You said "A second WP close to the first seems to more tightly focus the transition between them." By adding a WP closer to the first, I'm adding a WP, otherwise, if I physically moved the 2nd WP away from where I want to go to be closer to the 1st WP, I"m changing the mission and the drone wouldn't get to the intended destination.

Sure, I can create a mission that tracks the POI perfectly by placing WPs close to a POI. But that isn't the goal. Given two WPs and 1 POI, with both WPs pointing to the 1 POI, the focus during travel bettwen WPs should always be on the POI regardless of the distance between WPs.

Additionally, the mission isn't physically impossible. Adding a handful of points on the same line as the two distant WPs solves the POI problem (at least in software).

 

[DJ Wes]

Given two WPs and 1 POI, with both WPs pointing to the 1 POI, the focus during travel bettwen WPs should always be on the POI regardless of the distance between WPs.

This is false.

There is also a clear misunderstanding of how waypoints work and the difference between “Focus POI” and “Interpolate”. In fact, I’m not sure where to start so I’ll try to hit the “high points”.

• The use of POIs for setting the heading at waypoints ONLY defines the drone’s heading at those waypoints. The drone’s heading between waypoints is determined by smoothly transitioning from the heading defined at one waypoint to the heading defined at the next waypoint. There is nothing that makes the drone point at a POI while transitioning from one waypoint to the next other than an intelligently designed mission.
• “Focus POI” and “Interpolate” refer to the gimbal pitch, not the heading and therefore, should not be part of this discussion.
• The center point is not drifting as you mentioned. Instead, the heading of the drone is smoothly transitioning from what you defined at one waypoint to what you defined at the next waypoint.
• The word “tracking” should not be used in reference to waypoints and POIs. The drone is not tracking anything. Instead it is simply transitioning from one heading to another as it moves from one waypoint to another.

Giving specific answers to the questions about your mission would require the ability to see your mission.

 

[ctaranto]

Thanks for the response and info, DJ Wes.

Agreed that Focus POI and Interpolate are gimbal pitches and I conflated that with the POI for heading.

My understanding was if two WPs are defined to point to a single POI, there no heading from one WP to the next. With WPs defined as pointing at that the same thing, the entire trip from WP1 to WP2 should have had the drone facing POI 1 the entire time. My understanding is apparently not correct.

The explanation you provide about heading and transition from WP to WP does make sense, but it was clearly not what I expected. Now that I know how WPs with Litchi work, I will make sure I add more WPs to keep the heading towards the POI.

 

[ctaranto]

For those reading through this thread, Dronelink behaves pretty much as I had originally expected Litchi to. Setting up a Path, with two WPs about 300 ft apart with one POI. Configure two markers, one at each WP, to point to the POI. The drone (at least in simulation), points at the POI the entire way.

I imagine I'll use both Litchi and Dronelink depending on the type of mission I want to run. They both have their place.