Contact: Jakob Engel

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.

We are currently looking for students interested to work on this project, both as research assistant (HiWi), or in the coruse of a Master-Thesis. Please contact us if you are interested. Note that I expect you to be well familiar with C++ and ideally have some background (lectures / projects / seminars) in the area of computer vision or robotics. You should also read (and agree to) William Freeman's tips on How to do research.

This video shows a prototype of the Crazyflie Nanocopter equipped with a nano-camera and analogue video transmitter providing a live-stream of the video on the PC. It was developed in our lab by Oliver Dunkley.

This video shows an invited talk at GameDuell TechTalk Explore in Berlin this year by Jakob Engel.

This project is embedded in and supported by the Software Campus.