Differences

This shows you the differences between two versions of the page.

--- projects:canadawest:progress [2014/07/18 00:49] – [Wednesday, July 16th 2014] canadawest
+++ projects:canadawest:progress [2015/02/10 05:07] (current) – [Monday, February 9th 2015] canadawest
@@ Line 22: / Line 22: @@
 I awoke this morning to a phone call from John, about the SFOC.  He reiterated mostly what I read online.\\
-Before I submit an SFOC, I'll want to put together a really good application.  He seemed to also be under the impression that I need a SFOC to launch a launch a balloon, but I don't think so.
+Before I submit an SFOC, I'll want to put together a really good application.  He seemed to also be under the impression that I need a SFOC to launch a balloon, but I don't think so.
 ====== Tuesday, Oct 30th 2012 ======
@@ Line 535: / Line 535: @@
 I also just got a package from Hobby King today.  I should be able to make my Pixhawk drone tonight and fly it soon.  Perhaps tomorrow.
+====== Wednesday, July 23rd 2014 ======
+Today was rained out, unlike the last few days.\\
+Yesterday I was able to do full autonomous flight with the APM 2.5 Bixler setup.  It took off and landed as it was supposed to.
+The first flight needed some help landing.  I had it land while it was still at 30m and it just glided the whole length of the field.  It would have glided into the bushes if I didn't take control at the end.\\
+The second was pretty much perfect.  So, I decided to screw with it, of course.  The third did a HUGE circle outside of where I wanted it to, but in the end it returned.  I had to make it avoid a tree on landing approach, though.
+I was hoping to do some tests with the large foamie, but it had a twitchy aileron servo, so I fixed that today.
+Hopefully tomorrow will be better weather.  I've got some new autopilot functions to try out, and it would be a bummer if I was rained out two days in a row.
+Otherwise, things are progressing quite well.
+====== Thursday, September 11th 2014 ======
+It has been a fun and productive summer.
+I went to Calgary to get my friend Carl into the hobby more. He is still on the simulator, but having fun.
+I volunteered to organize the Christmas dinner for my RC club. Then I won the door prize at the meeting. Nice...
+I have flown and crashed the large foamie a few times, and I believe that I am near getting it near to flying properly.  This should be exciting.
+I have been wrestling with the secret to getting FPV w/ OSD and Tablet controlled groundstation.  The key was in not being too dedicated to certain hardware.  I have 2 different 433Mhz transmitter modules.  Flytron and HobbyKing brand.  The Flytron is an Open Source solution, which I like in principle.  The Hobby King brand needs me to mod the cable in order to use it with my OSD.  Since I'm not the biggest fan of slashing factory provided cables apart, I was trying to to everything with the the Flytron chip since..  reasons.\\
+The Flytron looked like it was working, except it couldn't do everything from the tablet that I could do from the computer.  This was odd, since I had the impression that other people could edit waypoints, etc from their groundstations.  Then I just tried the HK module and it worked.  I couldn't use the OSD, but it gave me unparalleled grounstation control of my APM.
+That made me think that it had to be the Flytron chip.  After some digging into the code, and doing some tests, I was able to figure out unit didn't have any hardware collision detection or anything.  No retransmissions or anything.  I think it does CRC though.  Packets that are received are all in-tact.
+So I'm now writing my own TCP for the thing, with ACK, ACK-REQ, DATA and FIN states.  It's not so involved so that it's without some design flaws, but it should work well enough.
+If I succeed, I'll post my code here.
+====== Saturday, September 13th 2014 ======
+Technically it's Saturday.
+It's almost 4am, and I'm pretty sure I've got it.  As promised, {{:projects:canadawest:openbee.zip|Here is the code.}}\\
+It retransmits if needs be, and I even made it so that it does that less often too.
+Unfortunately, after a test with the system, it doesn't seem to be as flawless as I hoped.\\
+More debugging needed.  I'll update the ZIP file when I get something more bulletproof.
+In the mean time, I rewired the HK transmitters to work with an OSD cable.  I think this should work well too. I will tansition to this setup for the near future.
+====== Monday, October 6th 2014 ======
+I've just been doing a lot of flying, a lot of crashing and repairing. The drone on the big Bixler was working quite well. Autonomous landing seem to be the tricky part mostly with the landing though.\\
+It had an unfortunate incident with a tree and another mis-repair with a wooden spar versus a now metal spar.\\
+It should be ready to go now.
+The big news is that the large foamie is now ready to be tested again. Hopefully I can get out in the next couple of days and see if it flies.
+I also managed to get the Hobby King wireless communication up and running with the telemetry. It works like a dream; much better and the Flytron.
+====== Tuesday, February 3rd 2015 ======
+I have been flying the bixler with fair frequency this year, as weather allows.
+I have also been emboldened to broaden the scope of the drone part of this project to be controlled by an Air Traffic Controller, of my own making.
+It occurs to me that the immediate future of drones lies not in flying everywhere, but from designated site-to-site paths.  A controller interface that would track all aircraft along a given corridor and route them appropriately is an important piece of infrastructure.
+I believe most of the technology to accomplish this already exists.  There are many Mavlink compatible programs for controlling an aircraft. They are all capable of connecting to the drone via a network connection as well as directly to the USB transmitter.
+Mavproxy is a neccesary in-between that can share one transmitter across many devices, all using Mavlink over tcp/ip.
+A pilot can connect to it locally, as well as to an ATC. The drone can be programmed with waypoints that the ATC can read and manipulate.
+Development can also be facilitated by Software In The Loop simulations of drones.
+More to come.
+====== Tuesday, February 3rd 2015 ======
+This is turning out quite well. There is only one hurtle that eludes me, and that is Guided mode.  For the time being though, I don't think I need to worry about that funtion to make something functional.
+SITL is turning out to be extremely helpful. Hopefully by the end of the weekend, I will have something to show.
+====== Monday, February 9th 2015 ======
+[[http://droneatc.ca|droneatc.ca]] and [[http://droneatc.org|droneatc.org]] are now online.
+If you have MAVProxy set up, you should be able to point them there soon by entering:\\
+output add droneatc.ca:14550
+For now I just have 2 test drones flying around in circles.
+If you do connect, please note that at this time the site is in Alpha, and I take no responsibility should a drone crash due to interactions with this service. Yeah, disclaimers.
+I have tested connection to the domain from within my house, and it seems to work.\\
+I will do a test with my Pi connected to my phone's WiFi, but not at this time.
+Thhe algorythm I use is pretty basic, but it should stop collisions for the sake of demoing it as a proof-of-concept at the very least.\\
+For now, I intervene if the drones are within 20 metres of each other, or within 5 seconds of possibly colliding.
+It works great if they're chasing each other, doing a head-on collision, or most other angles.\\
+The only scenario that's causing a near-miss is when one is doing a tight turn towards the other and ends up turning into their broadside.\\
+This can easily be fixed by making the conditions where it intervenes more conservative. 20m and 5 seconds is pretty close to be manuvreing together when ther're planes traveling at 20m/s.
+I have some to-do items.  For one, I have to better track the state of the drones.  I am assuming the drone wants to be set to Auto when it is done.  That's not always the case.  I would like one to just loiter while the other flies at it from different directions.
+I also assume the minimum turning radius is 10m.
+All these things can be made better by profiling the droens better and assessing their capabilities automatically.
+I would like to try sending messages to the pilot's ground station, and perhaps finding a way to interact with them.
+It's worth mentioning that the current strategy the back-end takes is to prevent collisions by enforcing distance between two drones.  It will intervene even when two drones will fly by eack other within a certain distance of each other.
+I'm going to see if I can drum up some interest on other websites.