Hi all,
Bit of background, I'm a student currently at the end of their 3rd year in a mechatronics degree, with little RC experiance, who has chosen to do UAV indoor / confined space navigation / object detection as their thesis topic. I'm essentially looking at getting a quad rotor copter and trying to add sensors / control systems onto the vehicle to ideally navigate in enclosed environments autonomosly or semi-autonomosly.
I'm realising that this is going to be a pretty challenging topic, both because of the inherent nature of instabilities in the QR, along with the changing air flow dynamics in an enclosed environment.On the flip side of things, it's a year long project that I'm going to dedicate many hours towards, and at least it's a thesis topic I find interesting.
So some questions that spring to mind at the moment -
i) Are you laughing at how much suffering I'm putting myself in for?
ii) Is this something viable to make a decent amount of headway in as a one person final year student investigation?
iii) How suitable do you think the ArduCopter platform would be in regards providing a basic framework (providing stability, auto leveling etc) to work off for future improvements?
Thanks for what's hopefully the beginning of a fun challenge, and unfortunately 1000's of questions.
Replies
i) Laughing? nope. Smirk? maybe.
ii) Quite viable to do! People have done it before, and it looks like the micro scale QR's (less than 6 inches in diameter) are much more suited to indoors than their bigger cousins
iii) I'll put it this way, Starting off with the ArduCopter is a HUGE boost to getting a project like yours going.
I took the hard route and tried to do it from scratch, (mostly) by myself, and still don't have it flying around wonderously. I've managed to get it to teeter-totter like it should, but my program needs to be tossed into the trash can the instant I can get something better.
I gotta admit though, I NEVER thought I'd learn this much about C++ programming, machine design & fabrication, and general application of physics. QR's are one heck of way to learn about embedded control systems.
I think you have a pretty high chance of success actually and I think arduCopter is a good platform to start working from because the basics all work and there's a lot of information and expertise on this website to help you along.
I've seen two other threads on object avoidance recently here and here.
For the indoor navigation you need either an optical flow sensor which should be available some day soon and/or you could try building something off agmatthews's wii stuff which is in his blog here.