Testing the limits of the YTH in "follow me" mode (and nearly crashing)

[BoostedPerspective]

So I would have to say 40mph is pretty safe. Maybe 45mph if the wind is right. It did really well, I was really impressed. And bear in mind I still have my CGO3 camera on. I want to do more testing with other modes as well, but I'm waiting until I get my CGO3+ back from Yuneec.

I do have questions about some odd behavior in the video. I'll let you guys watch it first, and then we can discuss. This was ALL done with "follow me" mode. The slow part (you'll know it) was RTH. (And you might be able to guess why that was happening) The very end was "follow me" with some stick input.

Enjoy.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[Steve Carr]

The follow me and watch me modes are intended for smooth video performance. Rapid acceleration, sudden changes in speed or direction will affect that quality. Also running in watch me mode with rapid acceleration will result in loss of altitude because the bird is pitching forward. The altimeter cannot track the change in altitude that quickly. If you do this at low altitude you are flirting with a crash. It's not a racing copter. It's designed as a video platform for stable filming.

 

[CAPTAINDRONE]

I left a comment on your vid. I'm surprised the Typhoon kept losing altitude when it appears the ground was rather flat. A close call at one point. Nice vid.

 

[BoostedPerspective]

The follow me and watch me modes are intended for smooth video performance. Rapid acceleration, sudden changes in speed or direction will affect that quality. Also running in watch me mode with rapid acceleration will result in loss of altitude because the bird is pitching forward. The altimeter cannot track the change in altitude that quickly. If you do this at low altitude you are flirting with a crash. It's not a racing copter. It's designed as a video platform for stable filming.

And yet it did absolutely fantastic. I'm impressed. Good nuff.

 

[mlpdx]

Wow, the H did an excellent job tracking at that speed, much, much faster than my P4 can track! That said, I'd prefer that they tweaked the algorithm to ensure a certain safety margin above ground level even if it meant failing to keep up (or at least allow that as an option). I was clinching my teeth in fear at a couple of points there, even though I knew it wasn't going to crash based on your video description. Thankfully, you chose an excellent place for this test -- nice and flat. Nice work, and I hope you continue to enjoy your hex!

 

[BoostedPerspective]

Wow, the H did an excellent job tracking at that speed, much, much faster than my P4 can track! That said, I'd prefer that they tweaked the algorithm to ensure a certain safety margin above ground level even if it meant failing to keep up (or at least allow that as an option). I was clinching my teeth in fear at a couple of points there, even though I knew it wasn't going to crash based on your video description. Thankfully, you chose an excellent place for this test -- nice and flat. Nice work, and I hope you continue to enjoy your hex!

Thank you sir! And yes, Now I've learned that near the top speed limit it seems to lose altitude. On the *really* close call we were gone. I had NO idea what it was doing back there, and didn't know until I got home and saw the footage. I actually yelled at my monitor. SO close. Now, as for the starting and stopping. Thoughts? Either it was losing signal and stopping, then finding signal again, or maybe it did see the altitude and put the brakes on?

 

[glider]

Did you check the altitude after it landed back at where it took off? Just curious.

 

[Steve Carr]

I left a comment on your vid. I'm surprised the Typhoon kept losing altitude when it appears the ground was rather flat. A close call at one point. Nice vid.

Hey Captain. I would have been surprised it didn't lose altitude. When a copter accelerates to max speed, in this case 45mph, it is unable to maintain altitude. If you are flying over a 100' you won't notice it, but if you're at 25' you will.

 

[Steve Carr]

Thank you sir! And yes, Now I've learned that near the top speed limit it seems to lose altitude. On the *really* close call we were gone. I had NO idea what it was doing back there, and didn't know until I got home and saw the footage. I actually yelled at my monitor. SO close. Now, as for the starting and stopping. Thoughts? Either it was losing signal and stopping, then finding signal again, or maybe it did see the altitude and put the brakes on?

The tiny chip in the Flight Controller has a lot of calculations to perform when flying in autonomous mode. The whole process depends on communication of GPS coordinates between the H and the ST10. The telemetry antenna is located inside the ST10 and is susceptible to interference and signal loss especially inside a vehicle. I would suspect the telemetry was dropping out. If you try the process again, make slower changes in speed and keep the ST10 pointed at the H. That will likely provide a better result. You can still reach max speed, just do it more gradually.

 

[BoostedPerspective]

The tiny chip in the Flight Controller has a lot of calculations to perform when flying in autonomous mode. The whole process depends on communication of GPS coordinates between the H and the ST10. The telemetry antenna is located inside the ST10 and is susceptible to interference and signal loss especially inside a vehicle. I would suspect the telemetry was dropping out. If you try the process again, make slower changes in speed and keep the ST10 pointed at the H. That will likely provide a better result. You can still reach max speed, just do it more gradually.

Yes agreed, I was losing telemetry, I could see it in the CSV file from the ST16 when I was playing with Dashware. It performed PERFECTLY in the beginning when the speed change was gradual. What came after was "lets see what happens when" And it did exactly what you would expect from it. My main concern was the starting/stopping in that last bit. Which was probably signal drop.

 

[glider]

Yes agreed, I was losing telemetry, I could see it in the CSV file from the ST16 when I was playing with Dashware. It performed PERFECTLY in the beginning when the speed change was gradual. What came after was "lets see what happens when" And it did exactly what you would expect from it. My main concern was the starting/stopping in that last bit. Which was probably signal drop.

What is your estimate of altitude drop?

 

[BoostedPerspective]

What is your estimate of altitude drop?

My altitude was 20ft at the start. Once it started going down, it was pretty quick at that speed. I didn't look to see what the minimum was, have to go check. We can probably guess... 3-5ft?

 

[BobW55]

Wow,
1) Your camera is so much sharper than mine. ( I know it is off your other bird)
2) Algorithm needs to be rewritten to hold alt.
3) maybe use the wizard on the top of the car or hold the St-16 out the window.
4) Would be interesting to see how she does on some of the mountain roads in your area, but at a slower speed. Can she hold altitude then?
5) Would it do better in follow me mode?
Nice video, Thank You for trying it out.

 

[PatR]

Trading altitude for airspeed is understandable. If the H is programmed to prioritize maintaining a set distance in Follow mode, and the subject was moving faster than the propulsion system could provide for, the H would start a descent to generate more airspeed. It appears it limits out between 40 and 45mph?

 

[BobW55]

Would be interesting if it would hold you in camera frame even if it could not keep up.
I know.......more firmware revisions.

 

[BoostedPerspective]

Wow,
1) Your camera is so much sharper than mine. ( I know it is off your other bird)
2) Algorithm needs to be rewritten to hold alt.
3) maybe use the wizard on the top of the car or hold the St-16 out the window.
4) Would be interesting to see how she does on some of the mountain roads in your area, but at a slower speed. Can she hold altitude then?
5) Would it do better in follow me mode?
Nice video, Thank You for trying it out.

1. Agree, I'm desperately hoping what they send back looks like this
2. I would agree here? Why go 45mph if you descend? For short bursts maybe?
3. I had the idea to put the wizard on the roof next time
4. I plan to try this, but with the YTH *behind* me. That way the remote/wizard hits the hill FIRST so the Typhoon can read and make the change.
5. That was follow me mode. I want to try watch me, but only when I get the panning camera back.

And no problem, always in the name of science. (and fun)

 

[BoostedPerspective]

Trading altitude for airspeed is understandable. If the H is programmed to prioritize maintaining a set distance in Follow mode, and the subject was moving faster than the propulsion system could provide for, the H would start a descent to generate more airspeed. It appears it limits out between 40 and 45mph?

According to the speedo in my car, 40 was fine. Wasn't until I did 45-50mph that it had issues.
Looking at the telemetry data, it held 40 just fine, but when pushed further it bounced around 43-46mph.

Easy take away from all this, when using follow mode, keep your speed under 40.

 

[sdharris]

Glad you remembered to press record, I shot a brilliant follow me last week and forgot to have it in record! My second attempt was less impressive so will try again this week.

I noticed you said it was shot in raw and graded in premiere. Can you post an example of the raw file as I am strugglng to grade Typhoon H footage shot in raw and it looks nothing like S-Log shot on the P3 which I assumed is what it'll look like.

 

[BoostedPerspective]

Glad you remembered to press record, I shot a brilliant follow me last week and forgot to have it in record! My second attempt was less impressive so will try again this week.

I noticed you said it was shot in raw and graded in premiere. Can you post an example of the raw file as I am strugglng to grade Typhoon H footage shot in raw and it looks nothing like S-Log shot on the P3 which I assumed is what it'll look like.

What kind of example would you like? Right from the camera?

 

[sdharris]

What kind of example would you like? Right from the camera?

Yeah right from camera shot with raw as I want to double check mine is not off. When I shot with raw the colour tint was so far off it took too long to grade in FCPX so I was considering trying Premier Pro. I was hoping that raw would look like s-log but it doesn't. If I have not binned the original files I will post an example later.