Description du projet
De nouveaux algorithmes imitant des poissons affamés dans des eaux turbulentes facilitent la navigation humaine
La nature est une source d’inspiration pour les scientifiques et les ingénieurs qui cherchent à développer des systèmes artificiels dont les applications répondent aux besoins de l’homme. La capacité des organismes à percevoir leur environnement et à y réagir constitue un riche répertoire de phénomènes qui valent la peine d’être imités. Parmi ceux-ci figure la capacité des poissons à attraper leurs proies dans des environnements turbulents, applicable à des domaines tels que la recherche et le sauvetage ou encore la détection de mines explosives. Le projet RIDING, financé par l’UE, étudie la manière dont les poissons intègrent de multiples entrées sensorielles et combinent détection et navigation pour s’adapter aux eaux agitées et saisir leurs proies. Les connaissances acquises permettront d’élaborer des algorithmes fondés sur la physique, en s’appuyant la mécanique des fluides numérique et l’apprentissage automatique pour exécuter ces calculs.
Objectif
Living systems developed dramatically efficient strategies to sense and navigate turbulent environments. Understanding these strategies is key to many real world applications required to function in the presence of turbulence: from search and rescue to demining and patrolling. While much is known on navigation in smooth environments, these approaches fail in the presence of turbulence. RIDING aims at elucidating the computations organisms use to extract useful information from turbulent stimuli and navigate to a target. A key observation is that organisms rely on multiple sensory cues, despite the distortions due to turbulence. Explaining this puzzle requires blending fluid dynamics with biological behavior. I will achieve this goal by developing physics-based algorithms elucidating the computations that support three fundamental pillars of biological behavior: 1) combine navigation with sensing, 2) balance multiple senses, 3) adapt to different environments. The result will be a comprehensive theory integrating biological behavior in a computational framework based on fluid dynamics. Predictions will be tested via experiments on fishes, known to routinely perform turbulent navigation combining multiple senses across distinct sensory environments. This multidisciplinary project leverages methods from physics, computer science and biology. In summary, the objectives of RIDING are to:
O1. Assemble a massive dataset of chemical and mechanical signals emitted by a target using computational fluid mechanics and asymptotic methods.
O2. Develop algorithmic approaches for sensing and navigation using tools from machine learning trained on multiple sensory signals from O1.
O3. Examine how sensory signals from O1 and algorithms from O2 vary in different environments.
O4. Test predictions by recording prey capture in the laboratory using three species of fish. Analysis includes a fascinating species which evolved unique sensory “legs” to catch prey in different environment
Champ scientifique
Programme(s)
Régime de financement
ERC-COG - Consolidator GrantInstitution d’accueil
16126 Genova
Italie