We are an interdisciplinary team of engineering and computer science students developing an autonomous quadcopter platform capable of indoor reconnaissance for entry in the AUVSI International Aerial Robotics Competition.
The AVTv4 uses the Hoverthings 550x frame with four off-the-shelf motors and ESCs. It is stable in a hover and can be flown with a joystick. We use a 9DOF sensor stick from Sparkfun to support sensor fusion with an onboard Cortex M3 processor on a custom control board. The board speaks serial over XBee and can be run with Ubuntu Linux.
We ideally would like to support university research and designed the platform with an eye toward graduate students. A Python plugin system makes it easily extendable. We wanted a quadcopter platform that, once bought and put together, could carry up to a kilogram of additional sensors to support research into machine learning and more sophisticated vision. It does not come with a GPS.