Zurich university's computer lab has been busy demonstrating autonomous drones at this week's CeBIT technology fair in Hanover, Germany.
The quadrotors are part of the Robotics and Perception Group's mission to develop machines – ground robots and drones – that can navigate by themselves.
Researchers at the Swiss university have developed the technology that allows the drone shown here to fly without relying on external infrastructure, such as GPS or motion-capture systems.
This quadrotor drone can navigate autonomously and reconstruct its environment in three dimensions.
To capture its flight path, the robot uses a single camera, seen here mounted on its underside.
The robotics lab at Zurich University is led by Professor Davide Scaramuzza. It was founded in February 2012 and is part of the Department of Informatics.
On the top of the quadrotor, you can see the drone's computer system. Information from the camera and an inertial measurement unit are fed into the onboard smartphone processor to compute the craft's motion.
Because the system used by the drone allows it to fly autonomously without any external infrastructure, it could be useful for search-and-rescue operations as well as for inspecting remote or inaccessible locations.
This picture shows one of the Zurich robot lab's quadrotors flying over a simulated landscape.
In the foreground, the screen shows a three-dimensional view of the terrain over which the drone is navigating.
As well as enabling the craft to navigate using waypoints, images captured by the under-belly camera allow the system to compute a 3D reconstruction of the quadrotor's environment in real time.
Research work carried at Zurich university's Robotics and Perception Group covers both ground and micro flying robots, as well as multi-robot heterogeneous systems that combine the two.
This picture gives you an idea of the small scale of the Swiss robot lab's quadrotor craft, seen here being worked on by one of the Robotics and Perception Group's researchers.
Because the aim of the research work is to make the Zurich robots active in their environment and capable of gaining knowledge from it, they are designed to cope with sudden movements in the landscape.
The system's robust visual odometry - the use of data from motion sensors to calculate changes in position - and probabilistic dense reconstruction algorithms mean the drone is not disturbed by movements beneath it.
The drone still captures an accurate 3D scan of objects under its flightpath.
The visual odometry, planning, and control algorithms run entirely onboard the drone on its smartphone processor.