Neural mechanisms underlying visual free-flight control in Drosophila melanogaster
Most flying insects rely heavily on vision for flight stabilization and course control. Understanding of how animals process visual information to generate appropriate motor output may on one hand lead to robust control strategies for autonomous flying robots, and on the other hand reveal fundamental neural principles of visual processing. Drosophila is an excellent model to approach this important problem empirically, due to its high specialization, relatively small brain size, and the genetic tools available for experimental manipulations. Our approach is to present dynamic visual stimuli during free-flight and measure the fly's responses to explore the underlying neural processing. This is done in a custom-designed wind tunnel with lateral projection screens, and two pan-tilt cameras. The 3D coordinates of the fruit fly's position and body posture are reconstructed in real-time from the information coming from the two cameras. First, we will determine the relevant physical quantities for motion processing (e.g., brightness, contrast, spatial and temporal frequency components). Then, we explore how fruit flies make use of this information to generate appropriate and robust flight control. A detailed analysis of Drosophila's visually elicited flight behaviour will allow an integrated understanding of its flight control system:
  1. 1. Visual parameters
  2. 2. Neural processing
  3. 3. Motor output
  4. 4. Reafferent visual feedback
Out of these four major stages, the first can be controlled, the last two can be observed, while the second -relatively hidden- can be experimentally explored and modified using genetic tools. For example, specific cell types with known or supposed function in a motion processing pathway can be specifically blocked and the effects of the experimental treatment assessed in the wind tunnel. Finally, computational modelling techniques will give the possibility to simulate complete control loops, which provides an important basis for possible future hardware realizations, including autonomous visually guided micro-robots.
Right hemisphere - Left hihsemepre