Status of PathPlanner and FixedWingPathFollower?

[Felkk]

I just got my F3 shield soldered up and tested it out with HITL and now I want to see if I can get fixed wing waypoint navigation working.

Before I tried it out, I did a quick search for PathPlanner and PathFollower on the forums to see what their status is.

I found https://groups.google.com/forum/?fromgroups#!searchin/phoenixpilot/pathplanner/phoenixpilot/47Qw0FEltH4/_v-ymJ4Q5yQJ and https://groups.google.com/forum/?fromgroups=#!searchin/phoenixpilot/pathplanner/phoenixpilot/N-RMw7tonO8/04OWTu_MZigJ and was wondering if anyone has any comments on how to proceed on working on navigation.

[peabody124]

You should find that the existing code works for sequential navigation.  At least I was playing around with simulations not that long ago and fixed the issues I came across.  That is not to say it is "done" in any sense.  The second post you link to is talking about how to extend it in the future - I think a better short term goal is to make sure what we have in place is a solid foundation to build upon.

What we have currently:

- flight/Modules/FixedWingFollower  - This starts when you enable the fixedwingfolloewr and is what I tested in simulation, and seems to follow all the commands the waypoints support (flying arcs, circling a point, and flying a straight line).

- flight/Modules//PathFollower/fixedwingpathfollower.c - this is the code Kenn has put more hours on testing.  however it isn't linked into the current object representations for navigation.  More specifically there is a lot of math about following the path in this module which should be abstracted into the paths.c library file but the code is a bit beyond me to do that.

I think making sure to keep the representation of the path separated from the code to follow that path is a really good goal to strive for.

So I'd start by enabling those modules and testing it in HITL - it should work.  What might bite you quickly is memory which we have that new issue about.

[Kenn Sebesta]

I need to get back around to testing nav for aircraft. In principle, it should work. In reality it's not been tested. Give me a little bit and I will try to draw up some instructions on getting this working in simulation. Then we can proceed from there to getting it working on a specific plane, and finally to getting it working on general planes.

[Felkk]

Thanks Kenn,

A guide for setting up and configuring pathplanner would be great. I messed around with it today in HITL and whenever I switch my flight mode to RTH, the F3 crashes.

[Felkk]

Pics didn't insert correctly in my previous post.

I tried enabling the modules with the UAVO browser but I'm having some trouble. I'll detail my method below and I would appreciate it if you could tell me that I am doing it correctly.

Plug in board via USB.
Go to System Tab.
Expand Settings.
Expand Module Settings.
Expand State.
Enable PathPlanner and FixedWingPathFollower.

 
Click the green up arrow to send the settings to the board.
Changed states go to yellow.

 
Unplug the USB cable.
Plug USB back in.
View Module states and they still say they are disabled.

[peabody124]

So clicking the green arrow doesn't save settings.  You need to click the upward red arrow with a harddrive behind.  Perhaps we should revert to using words instead of icons (or make the icon small and add the text back).

However... I suspect (from my experience) what you will find is that is that once you do get it to save and power cycle, the board will reboot and then go into boot fault mode because it is out of memory.  This brings up 

https://github.com/TauLabs/TauLabs/issues/269 and

https://github.com/TauLabs/TauLabs/issues/276

which will be priorities for this.  I'm not terribly concerned about being able to get it going but ti will take some tweaking.  Possibly we'll have to right a specialized sensor/attitude module to optimize the memory but I hope not.  It _would_ be extremely useful to try and break things down though.  I did see some large chunks of ram sitting around though.  With GPS running there was 2200 bytes free.  The sensor module had around 1200 bytes free (but I had no GPS at the time so that's not with the full EKF running) and the event dispatcher another 1200 bytes.  Currently the fixed wing follower takes 1500 bytes and the planner takes 1000 so that should be enough to shuffle it around (and I bet they don't _need_ that much ram.

I'm sure there's also something eating up a chunk of ram too that we haven't accounted for.  It would be _extremely_ useful if you could make a custom build with as much stuff disabled (basically just keep Telemetry) and then see how much ram is free.  Repeat with all the drivers disabled in pios_board.c.  This will at least give us some baselines.  It might also be that the memory map doesn't even use the whole main chunk of ram, but I'm not sure.

[Felkk]

The red button works much better. Adding text boxes below the buttons to indicate what they do would be good.

Yep, the board went into boot fault mode. I'll try out stripped down builds. How are you able to calculate the used RAM and how much is free?

[peabody124]

To see it per module check out the TaskInfo object (removing that diagnostic information is another 200 bytes).  Check SystemStats for the total free heap.

[Felkk]

With the below makefile and clean builds  for the FlyingF3 I had the following values in the GCS. Should I post these results in https://github.com/TauLabs/TauLabs/issues/269 as well?

[Felkk]

Would have some time to write up a little explanation of using RTH? Now that the F3 has enough free memory, I tried it in HITL but was getting some weird results. The home waypoint seemed to be not working properly.

[msev]

Felkk would you mind sharing your shield design files, if you have drawn it on the computer or something?

[Felkk]

I just used perf board, pin headers and the pinout.txt to wire it up. There must be something wrong with my shield because the F3 no longer boots up while in the shield. I have had to resort to jumper wires and a breadboard https://dl.dropbox.com/u/12489931/IMG_20130218_192938_699.jpg until the real shield being discussed https://groups.google.com/forum/?fromgroups=#!topic/phoenixpilot/VeObyJb6rek is completed.

[Kenn Sebesta]

So I just set up HiTL and am starting to poke around again. A lot of things have changed, so I need to refamiliarize myself. I'll let you know how progress goes.

[Felkk]

I have been busy with class the past week, but I started looking into RTH and fixedwingpathfollower.c. After I enable FixedWingPathFollower in the GCS, reboot the board, and then try to set RTH as a flight mode, I receive a system configuration error. Are there any other modules or settings that need to be set for RTH to not set off the config alarm?

[Gary Mortimer]

Did someone say fixed wing Nav... Sorry not been paying enough attention. Which board will fly this?

[peabody124]

Ok, so it's working for me.  At least in SITL.

I threw down some waypoints with the PathPlanner gadget.  Then took off in stabilized mode, flipped to path planner and it did the right hting.  Then flipped into RTH and it started flying home.  I'm not sure of the status of HITL, it's not something I've ever tested.  If you are using linux it might be worth trying to enable the simulation module though and running in pure simulation mode.

The steps to get it running are

1. Enable the FixedWingFollower module

2. Enable the PathPlanner module

3. On a real system you would also need to enable GPS and set the estimation algorithm to INSOutdoor.

As a small aside the current holding behavior isn't very good.  There is a separate mode circle position at fixed radius, but for some reason that's kickign the path follower into a critical state.

I found the default settings way too aggressive (hence the snaking in the beginning of the path in teh first screenshot).  Here are the settings I adjusted too (this is on next).

https://dl.dropbox.com/u/6645063/fixedwing-sitl-20130310.uav

[peabody124]

Ok that bug is fixed

https://github.com/TauLabs/TauLabs/pull/392

I'm not sure that things are working sensible WRT the airspeed