Such a poser.

There has been a slight lapse in updates here, since we are now about to finish the project (or rather finish off what we're working on right now for someone else to continue) and we are currently writing our parts for the final report. Since we have run in to quite a few speed bumps along the way, we didn't manage to complete the whole project and produce a complete prototype, however a lot of the grunt work has been done. Hopefully we'll find the right place on the internet to upload all the code so that it'll be of use to someone else. We'll also post the final report here, so if someone's interested they can read it when we're done.

We were actually about to fly it again yesterday, however one of the motor axles broke and we have nowhere enough time to get another one, since the final presentation is monday which marks the official end of this project.

However, Albin has taken our little lovechild to a studio and taken some very flattering pictures, which combined with a few images from in-flight and a 3D-reconstruction made by Alexander makes for some nice presentation material.

Flying against a clear-blue sky.

About to land.

Studio pic 1, simply the copter.

The app started on a phone with the telemetry module next to it.

The copter with it's best friend the controller.

A fused 2D-mapping of an area outside the school.

The same image, but in 3D.

ArchCopter apping!

The Android app has gotten nowhere enough attention here on the blog, so it's about time we correct that! This is what greets you when starting the application:

ArchCopter app splash screen!

Progress has been steady and according to plan and Albin has produced a selling poster view of the app, seen below:

Promotional poster of the application.

What it does is that it takes asks the user to specify a polygon (the corners of the area in question), then it divides it into segments of appropriate width, then sets waypoints such that the whole area is swept.

Apparently the app should be ready for testing soon, and hopefully the copter will be up to the task. 

Testflight again!

So, after Jens cut down the print-time of the hexacopter from about 7 days to 3 with some smart redesigns we where ready to fly again! Since we only have one Raspberry Pi camera and don't want to crash it (yet),  we used a old digital camera and used some duckttape to attach it to our machine. It was a bit bulky and the video isn't that exiting, but it's still a video from mid air!

And here is the flight we where the camera was attached.

The flight after we noticed midair that we had forgot the antenna to the RC-controller. The magic here was that Jens quickly set the hexa to land (it can do that automatically) and somehow it got that command and did the best landing so far.

Me (Anton) and Alexander are working on the case for the RasPi and all the electronics, and hopefully the whole design including camera, gimbal and all the sensors will be easily attached and done in the end of the week.

Stay tuned!

Fatal flaw

Another video of a stable flight. 

Gonna try a backflip next time so that you won't get bored!

So, this is starting to become routine now: We assemble the copter, head out and make one very successful flight. Then we land and try to take off again only to crash. However, this time we had a fairly calm landing, but it broke nonetheless. After further analyzis, we realized that we have a very crucial design flaw regarding the current design of the copter. The small contact area of the crossbar in the lattice work is not enough to sustain the amount of stress we put on it, causing the layers to separate.

Layer separation.

Since it takes about 2½ hours to print one arm, I was not very keen on just reinforcing the bars, so this called for a completely new design, the fourth iteration of the arm.

Arm v4

It is printed in two parts, then rotated around the x and y axis creating a completely symmetric piece. It is also screwed together keeping the layers together, so layer separation should not be an issue with this iteration. Screwed together it forms an I-beam (http://en.wikipedia.org/wiki/I-beam) which is extremely tolerant when it comes to shear and bending loads. On the downside it does not handle torsion very well, but we havn't seen much of that to be honest. One major upside of the new design is that it takes a lot less time to print, one of the reasons being that you can print 4 parts (2 arms) at a time.. These 4 were printed after lunch yesterday:

Four of the new arms!

Since it is very long (18cm) it tends to warp while printing, but another benefit of the design is that since we bolt it together, the individual pieces align and become straight as an arrow!

It took some redesigning of the centerpiece aswell (not something we really wished for since the old one took 5½ hours to print, but since the socket has changed it is a necessity). I decided to make it modular (so that if it breaks, you can just change the parts needing changing) according to the same concept as the arms with two pieces per direction, as follows:

New modular centerpiece arms v3.

And it all comes together as below:

Centerpiece v3 assembled.

That is all from construction for now!

Dummy copter on a stick

The work on determining which speed we should fly at has begun. 

A number of factors must be considered; camera frame-rate, image blurriness. The more images of the same area the better the 3d reconstruction but at the cost of memory and time to compute 3d point cloud. 

Since the copter is waiting replacement parts and some code for the camera we conducted a simple test with the new dummy copter. (Its a smartphone on a stick...)

Plenty of duct tape was used.

By shooting a low res film and walking with the stick raised high to simulate copter flight at approx 3 m over the ground and then checking how the number of feature matches in the images decreased as distance between images increased. 

Since walking with a cameraphone on a stick 3 meters up did not produce as properly oriented and equally spaced pictures as as one would have hoped we should not focus so much attention on the result.

//Alexander.O

Good morning!

I'm glad to report that we are almost back in business, we have now printed all of the parts damaged in the fall, apart from the centerpiece. This has however gone through some changes that hopefully will not impact performance, but has reduced printing time from 8 to 5 hours. We have installed Teamviewer (A remote monitoring software) and a webcam on the computer connected to the 3D-printer, so now we can monitor it from our office. Or lab. Or whatever this should be called.

Screenshot showing the 3D-printing software along with the webcam stream from the computer. Here we recently started printing the centerpiece.

Today we have quite rainy weather here in the north of Sweden, but hopefully it will clear up after lunch when the centerpiece is ready, so that we can get it in the air again.


Emma also tested the reflashed GPS yesterday (We have started using the russian equivalent of GPS, the GLONASS) and is happy to report a mean error of 6m in urban setting. This is to be compared with the 15m that GPS gave. This is of course depending on how many satellites you are able to lock on.

She has also designed and printed a water-resistant case for the GPS module. I must say it is looking rather dashing with its plexi glass roofing!

GPS casing.

Nothing lasts forever

And some things last for a very short time indeed.
Yesterday we had the maiden voyage of the ArchCopter v2 and we had a very good first flight, as illustrated below.

First stable flight. First flight ever actually.

It is very sensitive, the smallest amount of throttle will send it towards the heavens like from out of a cannon, making the landings a bit hard, but the new and improved landing gears (as seen below) seemed to handle it very well!

New, jointed landing gears.

But, all things eventually come to an end, and during the last flight, we lost one of the motors in mid-air (one of the ESC plugs vibrated loose), causing it to spin out of control. This caused the hexa to start spinning around it's own axis, which in itself is not a huge problem, but before we were able to bring it down safely a strong gust of wind caused it to go belly up 10m above the ground. After that, needless to say, all we could do was pray for the electronics to survive. Which it did. Phuh!

We also caught it on tape for your amusement, enjoy:

And then there was silence.

Today has been spent salvaging the parts that survived, which turned out to be surprisingly many! Erik has also mounted the magnetometers on a couple of PCB-boards and soldered on plugs, as such:

Very fancy electronics.

Albin has not been idle either, as we can see below:

Screen dump of the main menu of the ArchCopter app. In swedish. Translated it says:

Connect

Trace flight

Scan area

Settings

About ArchCopter

The glorious return

Hello everybody, sorry to have kept you waiting for further updates but the lot of us (apart from Anton) has been on vacation. Now, however, the lot of us are back and production is in full swing!

So here's our updates, category by category:

Construction:

Chassi version 2 is assembled, including motors and APM!

Jens holding the assembled ArchQuad 2.0. Note that this picture is taken with the Raspberry Pi Cam!

We have had some issues with the APM (I think it's due to the new firmware released), but after clearing the EEPROM and wiping the APM and setting it up from scratch everything worked again.

We've also installed a custom battery in the 9XR (the controller used) and are now running it on a 7.4V battery from DFrobot (namely: http://www.dfrobot.com/index.php?route=product/product&filter_name=battery&product_id=437#.UgJDgpL0Eb0) and it seems to like it! Just remember to set the voltage warning according to the battery, otherwise it will beep like crazy.

Other than that, I (Jens) have kept busy with updating the document governing the project and helping the other member while Anton has been working on the cameras.

Plan is that we finish the landing gears today and go out flying tomorrow! If everything goes as planned, we will post a video here tomorrow!

Camera:

Anton has been struggling against the RaspPi cam requiring a bit larger current than the wall-socket->USB converter could provide, but once the issue was identified it was solved fairly easy. The RaspPi cam is now up and running, now it is just an issue of what framerate to use and how to save it to either USB memory or SD-card.

Demonstration image taken with the RaspPi cam. This image has not been Photoshop'ed. Promise.

Anton has also received an IR camera CMOS module as seen in the picture below:

IR cam module.

However it does not seem trivial at all to get it to work, so that's something Anton (probably in collaboration with Alex) will be working on in the near future.

Interface:

Fancy splash screen of our app!

Albin is working hard on the android app aswell and he's vaguely optimistic which is a good sign. Since a program is never really done, it's hard to define "done", but hopefully the program will do everything needed of it.

Sonar/GPS:

Most of this week has been spent dealing with the sonars and trying to get a feel for them. A case for them has also been printed, as you can see.

Sonar casing which attaches directly below the battery pack.

Emma has done some research regarding the accuracy of the sonars, and they are actually pretty good and can detect even narrow objects at a fair distance. Her next project is wallowing through the APM code and finding a good way of implementing five sonars.

Magnetometers:

Since Erik (see picture under Camera) has basically everything set up regarding the magnetometers, he has spent the week researching and defining the interfaces that we will use, i.e how everything will communicate and be supplied with power. Not an easy task, but has to be done nevertheless.