Quadrotor Helicopters have recently gained popularity as experimental testbeds due to their low cost, relative mechanical simplicity, and VTOL capabilities. In this project, we designed and built a quadrotor aircraft with an inner attitude stabilization loop, to be combined with an outer loop providing yaw, pitch, and roll commands for position hold; this outer loop was to be implemented using an on-board camera. The end goal of the project was to demonstrate the feasibility of an entirely self-contained, autonomous quadrotor aircraft for indoor flight; to our knowledge, this hasn't been done before.

System Design and Overview

Background

The technological world is characterized by a motif of automation and mechanization, with the goal in mind of trying to get machines to do tasks instead of humans. Such tasks are usually ones that are often menial, repetitive, computationally burdensome, ultra-precise, dangerous, or physically impossible. The latter two characteristics have become especially important in the development of Unmanned Autonomous Vehicles (AUVs).

There has been much progress in the development of unmanned aircraft, able to perform maneuvers and reconnaissance, otherwise compromised by having a human on board. But the development of the airplane-based UAVs has shown the limited capabilities and functions of such a design. There is a technological niche for hovercraft UAVs, due to the need for low-level, stable movement, and even indoor air vehicles.

With such a niche to be filled, there have been many who have begun building such devices. They are currently on a very expensive and high-end market, designed for military purposes, but academia is producing its own, cheaper less-robust models. Academia has failed to produce a robust design with mounted vision as of the writing of this document. This is where team j! hopes to design a more powerful design for application within the less-expensive markets.

Product Description

Our project is to design an unmanned autonomous quadrotor that can meet the developmental challenges and applications afforded by academia. A quadrotor is a four-propeller mini-helicopter which invites simple design by comparatively inexpensively mounting the rotors for the propellers in stationary locations at axes of rotation to compensate for a robust control system. We plan on going above and beyond the off-source camera development models of Massachusetts Institute of Technology and the GPS-guided controls developed at Stanford for a model whose control and computation can be performed on-board by its own vision system.

We realize the feasible limits of our model, that in real markets, we do not stand a chance of selling this thing to the military, so we are developing it for a more reasonable customer: a technology research group. This research group, Made-Up Fake Company, Inc., is our main customer for this project. They specialize in research and development of UAV software and applications, and have built many UAV's for environmental and testing purposes. They make sensor equipment for airborne applications and software for analyzing the equipment.

Made-Up Fake Company, Inc. is interested in expanding their repertoire. They want to be able to provide low-level and even in-door application sensing. They have had a few suggestions from mining engineering companies and rescue groups about the necessity of having a hovercraft or small helicopters. Remote-control helicopters are too weak for their applications, and are too unstable for others. They could not afford military-grade quadrotors or helicopters for a developing project. So they turned to academia.

The designs of MIT and Standford have reliance on stationary cameras or GPS sensing, but for the applications that Made-Up Fake Company, Inc. wants, these will not do. Enter: Brigham YoungUniversity's team j!.

j! specializes in on-board quadrotor development. Made-Up Fake Company, Inc. contacted j! about the possibility of developing a quadrotor unit. They wanted a simple design, mainly so that they would have a functioning prototype for their software and application development. They knew that making on-board camera processing was not an easy task and so they gave j! a semester to try to get it done.

Because of their needs as developmental researchers, we needed to discover what they were looking for, in a unique position in UAV markets.

Archive:

Our Quadrotor in its Pomp and Panache

One Quadrotor in the hand is better than two in the...

Our Accomplishments of the week

Significant Advances for a week:

• We have full communication between the gumstix and the autopilot! This means the functions and algorithms that we compute in the gumstix processor can send meaningful values to the autopilot (which controls the quadrotor motors).

• Altitude Sensor and Hold: we now have a functioning Altitude sensor that is sending values to the autopilot, tuned enough to fly and maintain altitude.
• POSIT: Our POSIT command appears to be working in prototype, and we're excited to put it into action on the gumstix
• On-board Vision Segmentation: We are now processing our LEDs on the gumstix. This required transferring vision code and also tuning camera gains in the driver.

Significant Advances for the previous week:

• We have tuned our on-board camera and have made a prototypical camera cable for easy testing on the on board gumstix.

• We have connected the gumstix and it is now fully functional: We are now processing vision on-board, thanks to our team members--here are two pictures taken from the xscale embedded system on-board:

j! Rocks! Pitch! Yaw! Roll!