Hi,
I have ArduRover v2.49 with Mission Planner 1.3.28 running with a 3DR Pixhawk with 3DR telemetry in a model boat.
I have now successfully completed a few missions albeit over small distances. While on the face of it, everything looks perfect, the boat continually wobbles (twitches) and telemetry logs indicate high frequency rudder cycling.
The telemetry replay for the 3 minute mission is on :
It looks very successful apart from the Ch1out values. This is the rudder.
The attached 'Rudder and ESC.jpg' shows mission output for both rudder and ESC (motor).
I had change to manual mode just before WP3 at about 10:46:40 to avoid a collision and that explains the ESC anomoly at that time. I realize I have to adjust (reduce) the desired cruise speed and that's not a problem. But the rudder activity is a concern.
The attached 'Rudder.jpg' shows a second by by second breakdown of rudder activity over a representative interval. This shows a 20 degree to port then a 20 degree to starboard almost every second. Hence the wobbles and twitches and probably why speed isn't rising.
I know this isn't a software or firmware issue - it's down to my own incompetence / unfamiliarity.
The final 'MP config.jpg' shows what is currently set. I'm sorry to be asking stupid questions, but is this the right screen to start modifying the rudder behavior ? Does anything look obviously wrong ? Or do you have any other advice or suggestions ?.
Thanks in advance for any suggestions, help, advice or pointers.
Tony
Replies
You're a hero Grant - thankyou. I really appreciate the effort you've gone to and will certainly get back to you after the next outing (which will be on Sunday). Will try adjusting the TConst first as suggested.
Yesterday I downloaded all parameters to a file. Then did a CLI reset and downloaded all default parameters to a file. Then uploaded the original (first) file parameters. Then used Notepad++ to compare the 2 files and highlight any differences from defaults. I then read up about any changed values that I didn't understand plus identified potential candidates for alteration. TConst was in my list, but is now is now at the top.
The explanation and analysis you've provided is superb and very clear. I'll let you know shortly after the Sunday sailing.
Thankyou very, very much.
Tony
Tony,
I have an idea. I did some extensive tuning today on my catamaran. Auto missions repeated turn left and right 45 degrees, 90, 135, 180. I played around with a lot of parameters and I am not sure the TConst is going to solve the problem. It may just make the zig zag slower. Why not put the controller on a one axis gimbal. Not powered but something that can allow the boat to roll around it. platform with a weight hanging on it. Does the code need to know if the boat is rolling??? Just a thought.
Regards,
David R. Boulanger
Hi David and thanks for the reply. In my opinion you are absolutely right. Last week I quickly re-soldered the motor connections and tested them. I then spent a couple of days static testing the setup. During that time I test-changed most parameters incrementally to their min and max values and measured Kort activity.
Interestingly your original suggestion of changing NAVL1_PERIOD yielded the most favorable result - although not really fixing the issue. Also spent a lot of time trying to force PixHawk or MP to ignore Roll/Pitch and use Yaw to control steering, because this is really what's required on a boat. It wont surprise you to hear that I failed !!! I can adjust the Rudder load factor, increase damping, reduce response times but that's about it.
I'm neither a perfectionist nor an idealist, but think (as you say) the real solution rests in decoupling or removing z axis inputs or at least getting the code to ignore it. This would also make the PIxhawk's job a lot easier (performance, consumption, heat etc). Damping would then provide the final tuning component, rather than being used to mask a problem which rests elsewhere.
Yesterday I ordered a couple of SD cards and an SD card reader and these should arrive today. I plan to remove the PixHawk SD card and examine the real content. Then download the Rover source code, make a simple name change in the code on something unimportant. Recompile and upload the code to one of the new SD cards. Put the new SD card in the PixHawk and see if my change is evident. I don't underestimate the challenges involved in the paragraph !!!
If the above is eventually successful I could then (in theory) rummage around and make additional changes to (a) Decouple Z axis input (b) Change Rover symbol (c) Enable some of the Mavlink buttons which are enabled in Plane but disabled in Rover. I appreciate that I'd be doing a personal (unofficial) code branch with all the usual caveats and wouldn't be able to upgrade without making similar changes in any official new Rover code release.
Some people can leap tall buildings in a single bound, but I'm one of those unlucky individuals who usually trip over a pebble.It will probably end in tears but I'm determined to give it a try :)
Thanks again,
Tony
Note that the code is not stored on the microSD card. The microSD only holds the logs. The code is stored in flash on the Pixhawk, and the parameters are stored in FRAM.
Changing microSD does not change the code at all.
Sorry Grant, Tom and team - I've let you down.
Yesterday my shame circuits went into overload when I lost all motive power and had to request help from the rescue boat. On retrieval, I discovered my soldering skills leave much to be desired. Motor pic is below.
Repairs in hand and will supply info as promised, as soon as I get her back in the water.
Apologies .........
Tony
@Tony,
What do you use for a rescue boat?
I have a small tugboat that I built from a Zippkits
It is quite powerful for such a small hull.
Regards,
Tom C AVD
Hi Hi Tom - your rescue boat looks ideal - small, strong and maneuverable. Thanks for sharing. I'm lucky - several club members rotate to operate a rescue service on a non-official rota. Pics below. Many thanks.
Hi Tony. If you could attach your dataflash log I can have a look. I suspect the rudder is trying to correct the rocking motion of the boat but I'll be able to tell from the log.
Thanks, Grant
Hi Grant. Sorry for the delay in response.
I've downloaded the dataflash.log for the single mission in question but the file size on the PC is 112MB. When zipped or rar'd the size is 25MB. I don't think I can attach files greater than 7MB in size in this forum. I'd really like to get you the info. Can I manipulate the dataflash log on the PC, or create a subset of the data in the dataflash log so I can get the size down. I know accurately a 2 minute time frame within which the issue is occurring.
Sorry to be a pain and thanks for any advice.
Tony
My last reply got cut off. Not sure why.
What I was trying to say was most people upload the larger files to some sort of cloud storage like dropbox, skydrive or google drive and then post the public link. But Tommy's generous ftp offer will also work. I'll have a look.
Thanks, Grant.
-
1
-
2
-
3
-
4
of 4 Next