Senior Design: SUAS Drone Competition
For my senior design project, I worked on a team of 5 to create an autonomous airframe for the AUVSI’s Student Unmanned Aircraft System competition. For this project we used a microcontroller to take sensor measurements, communicate with a ground station, and survey the ground to identify targets using a camera.
We built a helicopter with an onboard autonomous system for piloting, navigating, and managing a camera payload. The airframe is being built to meet the specifications of the 2014 SUAS competition held by The Association of Unmanned Vehicle Systems International, AUVSI. Our helicopter was a nitro-methane fuel-based helicopter, controlled by the Pixhawk autopilot board. The software platform we used was the open-source Arducopter system, which we tailored our single-rotor airframe.
In addition to helping with the design, assembly, and mechanical maintenance, I was responsible for the electronics design, sourcing, and testing. I was also part of the programming team, which ensured that the ground station and on-board processing units efficiently handled its data lines, such as the many sensor inputs as well as the image processing.
One of my responsibilities on the team was to design and build our electronic subsystem by identifying our needs, mapping data and power flow, researching components, and assembling/testing the system. This required managing multiple power requirements, data frequencies, and processors (onboard as well as at a ground station). Our final drone stabalized itself and flew to GPS waypoints, streamed 1080p video to a ground station for analysis, and dynamically adjusted its flightplans based on data or instructions from base.
By using quantitative methods for making design choices we were able to efficiently address the competition requirements and put our team at an advantage. One chose was to build a single rotor copter over a plane or quadcopter because it was both suited for fuel-based flight yet wasn’t bound to flying in a straight path. A copter’s greater degrees of motion in the air over a fixed-wing plane meant we could easily change our flight plan and act in a tighter time/positional window.
While there is a lot of documented work on electronically powered drones, in this competition we had to fly over a large area and address a large task list, all while being scored heavily for time. Using a LiPo-powered airframe would have meant returning to base to swap batteries, which would cost a lot of points. We instead developed a fuel- based system, which allowed us a full flight on a single tank and more effectively addressed our competition requirements.
Because of the unique shape of the targets at the competition (geometric signs with alphanumerics on them), we chose to use blob analysis to find the center of gravity of blobs scattered in the view, and focus on blobs that have overlapping centers of gravity, indicating a character centered on a sign.
Our main competition frame was equipped with over a thousand dollars worth of hardware, so when we tested new features, we did so on a smaller single rotor copter. The one we chose was a smaller model from the product line from which we bought our larger frame, which kept our test rig as close to the original as possible. We made sure to choose an adequate size for handling the electronics payload, so we were able to simply move the onboard electronics from one to the other. Using this system we were able to conduct test flights more frequently, and without worry of damaging any expensive components.