In this research project we will explore visual navigation methods for Nanocopters, extremely small Quadrocopter with a flight-weight of less than 50g. Possible future applications are e.g. exploration of a collapsed building after a natural desaster, surveillance and inspection of difficult-to-reach areas, or simply as personal flying camera, to take pictures from a fully new perspective. In contrast to regular quadrocopters with a flight-weight of above 500g, nanocopters can safely be used indoors and close to people, can fly through even smaller gaps or windows, and - acting fully atonomously - can easily be deployed as a swarm.
The extraoridnarily small payload of such platforms poses new challenges, in particular as it imposes heavy constraints on the on-board sensor capabilities. Monocular nano-cameras - being lightweight, cheap and small, while at the same time providing rich information about the environment - are an ideal choice. In this project we will focus on off-board computations, i.e., perform costly computations on a ground-based laptop.
The aim of this research project is two-fold:
- Develop suitable hardware prototypes (nanocopter + nano-camera), and corresponding software interfaces to facilitate visual navigation with off-board computation
- Adapt and develop monocular visual navigation / SLAM methods suited for the high agility and comparatively low sensor quality of such a system.
This project is embedded in and supported by the Software Campus.
We provide some open-source code for the Crazyflie; for more details see the corresponding paper (see below). The Code is on GitHub
|Conference and Workshop Papers|
|Visual-Inertial Navigation for a Camera-Equipped 25g Nano-Quadrotor , In IROS2014 Aerial Open Source Robotics Workshop, 2014. [bib] [pdf] [video]|
|Visual Inertial Control of a Nano-Quadrotor , Master's thesis, Technical University Munich, 2014. [bib] [pdf]|