Monday, December 14, 2009

Autopilot Ground Test

Today we tested the thermopile sensors of our autopilot. They stabilize the plane in yaw, pitch, and roll by analyzing infrared heat discrepancies between the sky and land. The videos below show our test.

Over January we will mount our digital camera beneath the Big Easy's wing and start making maps of Colby's Campus.

Sunday, December 13, 2009

Autopilot installation

Over the past few days we have been configuring and installing the brains of our UAV, a.k.a our autopilot. We decided to use an "off the shelf" autopilot called Attopilot V 1.8 which navigates by using a GPS sensor and pre-loaded 3 dimensional waypoints (longitude, latitude, and altitude). The Attopilot manipulates 4 control surfaces (rudder, elevator, ailerons, and throttle) to both stabilize and control the plane along its set course.

This is the cluster of wires coming out of the Attopilot. These wires connect to the remote control radio receiver (Rx), control surfaces, and sensors.

The XYZ thermopiles detect long wave infared light anywhere from 5.5 to 15 microns in length and log this data at 5 hz (5 times per second). The thermopiles are used to detect a heat gradient between the land and sky in order to stabilize the aircraft in yaw, pitch, and roll.

Picture of the XY thermopile.

Pitot Tube
The pitot tube determines the plane's relative airspeed by calculating the pressure differential between an externally and internally mounted tube. As the plane's speed changes the pressure from the external pitot tube should change while the internally placed tube's pressure should remain constant.

1/8'' diameter brass pitot tube. Tucker used a mini pipe cutter to trim the length of the Pitot tube.

Power Sensor
The power sensor falls in line between the electronic speed controller (ESC) and the Lithium Polymer batter (Lipo). The sensor measures and records the voltage and amount of miliamp hours consumed by the Lipo. The milliamp hours can then be divided by kilometers traveled to create a "mileage" of sorts in order to assess flight efficiency.

Global Positioning System (GPS)
The GPS use a number of satellites to triangulate the planes position.

GPS is the square sensor mounted on the nose of the fuselage.

Easy Star fuselage modified with Easy Glider wing. In our final setup we will attach our Pentax camera beneath the wing.

Attopilot Case Design/Build

Given the high cost and delicacy of the Attopilot, we designed a case to protect the device from impact in the event of a crash. We wanted to strike a balance between accessibility and protection. This meant being able to remove the Attopilot from the case with ease while also ensuring a tight and secure fit.

We started with some basic pencil and paper sketches before moving on to a 3D Google Sketchup model and then the real deal. We made a box out of plexiglass (cut by an exacto knife and fixed with super glue) and hot glued packing foam to the walls. For vertical stability we glued some EPP foam to the underside of the top of the case. We also built a removable door with holes for the two 1/4 inch silicon pressure sensor tubes. It is secured in place with an elastic band for easy access.

Sketchup model of the case without the foam or the removable door (download the file here)

Notice the opening in the top to run the wires out of the case

Monday, December 7, 2009

Pano image stitch

I used a free trial version of Pano stitching software to stitch together two of our aerial images from last friday. The software detects visual similarities between pictures and overlaps them based on those control points it creates. Here the software used the white soccer field lines to mash the two pictures together.

For our map making we will be using photo stitching software called Pic't Earth. Pic't Earth will use the GPS meta data associated with each of our images to stitch them together and georeference them on a digital map.

Sunday, December 6, 2009

Google Earth Vs. Colby: I-95

Here's a comparison of the image we captured roughly 250 feet above I-95 to the same stretch of highway in Google earth. Our picture below has no post-processing digital zoom. We used our Pentax Optio A40 12 megapixel camera fixed below the wing of our EPP Hornet to grab the shot. This image represents the photo quality which we'll use to construct aerial maps.

The Inevitable: Plane Crashes

The question is not if you will crash an RC plane but when. Over the last two months, we have spent hours climbing in trees, running anxiously through fields, and climbing over hills to find our crashed planes. All of our planes have crashed at least once and many have been retired to garbage cans and recycling bins throughout the Maine area. Anticipating these imminent crashes, we opted to go with foam airframes from early on. Not only are foam airframes very resilient but they are very easy to fix. Quick fixes maximize the amount of time we can keep are planes in the air and greatly decreases the amount of time it takes to learn how to fly.
Here is a brief tutorial we made that shows how to fix EPP with a glue gun. We were flying again 60 seconds after the end of the video.

A crash on the Maine Coast near Fort Williams in Cape Elizabeth.

First images with our Pentax Optio A40

Last Friday Foster and I went out to the Colby soccer field with our Pentax Optio A40 12 megapixel camera to test our remote image firing system. We attached an Infared Prism trigger to our camera and Rx reciever on board of our EPP v-tail hornet. Flipping a switch on our remote control radio triggers the Prism to send a infared signal to the Pentax camera to capture a digital image.

Foster and I testing the Prism trigger with the camera mounted under the wing of the hornet.

From Build Log

We sent up the plane and took some pictures of the soccer field and I-95. The high winds (20 mph) made it difficult to stabilize the plane for clear pictures, we did manage to get a few good ones though. In the upcoming weeks when we install our autopilot sensor suite onto the plane it should make for much more stable flights (i.e more consistently sharp pictures). The autopilot uses infared thermopile sensors which detect the heat gradient between the sky and land. The autopilot software uses this gradient to stabilize itself in yaw, pitch, and roll attitude directions.

Here's some of the sharp pictures we captured.

From Build Log

From Build Log

From Build Log

Here's some of the blurry images. We think this was caused by a sudden change in yaw, pitch, or roll of the plane's attitude mid photo capture.

From Build Log

From Build Log

From Build Log

Wednesday, December 2, 2009

Flight Tests on December 2

Today, Dan tucker and I were at the field testing airframes. We tested the standard Easy Star with an extended rudder with both 2 cell and 3 cell batteries, the Easy Star with a modified Easy Glider wing with the 2 cell and 3 cell, and our new V-Trainer with both the 2 cell and 3 cell batteries.

Discussion of a 3 cell lipo.

The EPP V-Trainer. We are thinking of using this for mapping in places without fields to land such as beaches and mountains.

Our Easy Star UAv airframe.
A look from above over Frat Row and Miller with an Easy Star with a 3 cell lipo.

Monday, November 30, 2009

Blackberry Video

Foster and I wanted a cheap way to grab some First Person View (FPV) footage to show to our professors, friends, and family as a proof of our concept. We decided to cut a hole in our Easy Star air plane fuselage to mount my Blackberry Curve. Mounted in our airframe my Blacberry could capture low quality video. I bought a 2 Gigabyte SD card from Rite Aid for $19.99 to increase video storage space and the phone was ready to film. Although the video quality is rather poor we think the smartphone serves as a cool platform to broadcast FPV in real time through its data channel. The mobile app Ustream could help us and others broadcast live video streams to the internet. Now I just have to convince my friend to mount his iphone 3GS onboard and we'll get some better quality!

This flight below was filmed with the Blackberry. We were flying RC but will be integrating attopilot into our easy star as soon as it arrives.

UPDATE: The new Motorola Droid phone has the capability of broadcasting 720 X 480 video live through Qik. See an example video here.