Cheap RC Helicopter Controlled via Motion Capture

Autonomous Helicopter from M W on Vimeo.


Recently we've seen a couple of really interesting university projects where quadcopters are controlled in motion capture stages. The video above demonstrates my progress while working at the USC Institute for Creative Technologies this summer over the course of about 2-3 weeks, to show that even cheap models can be controlled in motion capture stages, and some progress toward innovative control mechanisms (following a gloved hand).

Technical:

The model is a Horizon Hobby: Blade CX2 RTF Electric Coaxial Micro Helicopter controlled via a Spektrum DX5e receiving servo signals through an Endurance RC PCTx from a PC computer. The software running on the computer consists of a VRPN (Virtual Reality Peripheral Network) middle wear program that takes data from a Phase Space Motion Capture Stage OWL Server and outputs clean model position data (x,y,z and heading) to a VR engine written to Dr. Evan Suma. The model is controlled by 4 PID loops, one each for the X,Y,Z, and heading variables and includes adjustable clamps, offsets, ramps, throws, trims, and other functionality to make it easier to tune and adjust the control loops. All of these values are adjusted easily by editing a text document that is read-in to the software every time it is executed, to prevent a recompilation whenever any values need to be adjusted (Often when manually tuning a PID loop!). Also, included in the control code is a trapezoidal function that smooths
deltas to the PID loops to prevent shocks to the control system (thanks to Marc Bolas for that suggestion!).

The PID outputs are then sent over a USB Human Interface Device to the PCTx, converted to PWM and PPM signals, sent to the TX, through the RX and eventually arrive at the servos and motor controller and make the helicopter move as desired.

The oscillations you will see in the video footage indicate that the PID loops are NOT critically damped (ie they need to be better tuned). But I believe that for two weeks of work that's pretty good performance for an inexpensive single rotor helicopter.

Hope you enjoy!

P.S. With luck code and a build log may be released soon, stay tuned!

Views: 2186


Developer
Comment by Ryan Beall on July 14, 2010 at 6:01pm

Developer
Comment by Mark Colwell on July 14, 2010 at 8:10pm
Navy's Blade CX doesn't require proficiency in keeping other hand in pocket.

Moderator
Comment by Gary Mortimer on July 14, 2010 at 11:59pm
Nice one, interesting control idea, the method for getting airframes around a point intuitively and without lots of training will probably take up lots of time in lots of minds over the next year or two.

Admin
Comment by Morli on July 15, 2010 at 6:02am
Yep interesting . Imagine trying this experiment with Trex600 or even 450 chasing your hand :)).
Here is one more control method that caught my eyes recently. It is using the Nunchuck to control helis and planess. the RC blog is here , the project is by guy "fmkit" and his website where he sells the diy boards/controller is here
But the most important reason which caught my attention is using second remote control with a spotter who can take control (if and when needed ) without any complication ( master controller leaves all control stick in center in his RC. Now I see a possibility.


Developer
Comment by Mark Colwell on July 15, 2010 at 7:42am
Wow! fmkit has very professional radio modifications that I will attempt, Off to Goodwill for some recycled` Midland radios and 2 sets of fmkit Wi Nunchuck chip kits will be ordered as well. Nice pan & tilt stabilizer may be released soon too.

Admin
Comment by Morli on July 15, 2010 at 10:50am
Mike , the development you have done in the two week project is very impressive. My apologies if I have sort of highjacked your blog ( not intentionaly but due to similar non standard control systems)
@Yes Mark, I am interested too in trying out the Nunchuck idea and have 2 yaesu handies, FRS radio pair.Watch his youtube and vimeo videos. Very impressive work. With respect to UAV and diydroning , one of our concerns is safety and limitations related to testing in better controled scenarios ( LOS) , I felt that may be having two full controls( geographycally sperated could help us all improve/test the design in better way( just my openion and thoughts even though I am from place where I don't have such limitations ,but I still want to follow the common sence and all the needed safety/rules and regulations. The RC RX side radio along with his modified uC controller acts as lost model finder becon!! with very simple concept ( yes I have done some ham fox hunting with our ham toys and home brewed RDF ant too) since it uses two radios as RF media on both ends.I think non of us want to ever want to loose our model( wantedly or unwantedly), and there were few post earlier for lost model tracking here. Also the fact that The second controller with spotter IMHO will help a lot on top of the fact being alternate /non standard RC design. This is the video with diversity RC controller!! I am talking about
Here are his video links.

' />

http://vimeo.com/ppp/videos
Also watch his flying wing with bottle dangling below!!
He has already few fans from here and I think he got two more.:)

Admin
Comment by Morli on July 15, 2010 at 10:54am
:(( . too many typos , sorry , (borrowed laptop and internet connection with no MS word spell checker)
Comment by Mike W on July 15, 2010 at 3:57pm
@Ryan,
Impressive performance with such few components! Makes me think I almost over-engineered this one..

@Gary, others,
Thanks for the accolades!

@Morli,
Not a hijack at all, thanks for the posts, some really interesting ideas and other work! A note about a safety feature in my design.. One advantage of my setup was that in order for computer control to pass through to the helicopter someone had to physically hold back the trainer switch on the Spektrum radio as the computer was inputing through the buddy interface. More times than I care to admit I had to release that switch and either kill the helicopter or take control back using the TX and try to save the model manually. I know you can simply modify the switch but in its present configuration it turned out to be a valuable safety feature!
Better to be safe than sorry.

Developer
Comment by Ryan Beall on July 15, 2010 at 4:28pm
I'm not sure what your setup is, but the more you can offload the lower level flight controls such as altitude hold and attitde hold onto the autopilot the better. Keep the upper level navigation to external visual comp. If you try to do everything off board you will get to enjoy the wonderful world of latency! Great project! If you havn't already you should tag team with the guys using kestrel. They would love a controls guy who understands PID etc. Plus you could probably learn a ton from how they do things. I know I did. Good luck on finishing up if you havn't already

USNA '08
LTjg Beall USN

Developer
Comment by Ryan Beall on July 15, 2010 at 4:30pm
"how they do things" meaning Kestrel autopilot GUI and control law layout etc. sorry I wasn't specific.

Oh and I would throw in Go Navy, but I'll just drop the "At Least We are not Airforce!"

Keep me posted!

Comment

You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service