The setup consists of a camera equipped robot in a small arena with obstacles, an LED display that transforms the camera input into patterns of vertical and horizontal stripes, a fly tethered to a wire and a monitoring system measuring the fly's movement. As the tethered fly tries to move forward, the monitoring system records its motion and translates it into motor commands for the robot. Images transmitted from the robot camera trigger patterns of stripes on the LED display, allowing the fly to react to obstacles.
The result is a fly piloting a robot car as it whizzes through a forest of pillars in the arena. The research aims at constructing a complete computational model of a fly's neural network and may create technology to control future flying robots.