When poor little fish are confronted with AI, robot-fish, virtual reality, and drones
Clément Sire (LPT Toulouse)
After introducing the collective behaviors observed in fish schools, I will outline a methodology for quantitatively measuring social interactions (attraction/repulsion; alignment) within animal groups. The reconstructed interactions can then be directly implemented in analytical models which quantitatively reproduce the collective motion of fish. I will also briefly address an alternative machine learning approach designed to produce realistic fish trajectories. In the latter part of the presentation, I will explore various applications of such analytical or machine learning behavioral models in the context of robotic and drone platforms, and a virtual reality setup… for fish! In particular, our robot-fish and VR setups constitute original and powerful tools to study the social dynamics of fish and their response to controlled perturbations.