COOKIES: By using this website you agree that we can place Google Analytics Cookies on your device for performance monitoring. |
University of Cambridge > Talks.cam > Fluid Mechanics (DAMTP) > Plankton Blinders: navigating turbulence with limited information
Plankton Blinders: navigating turbulence with limited informationAdd to your list(s) Download to your calendar using vCal
If you have a question about this talk, please contact Professor Grae Worster. Plankton face the challenging task of navigating oceanic turbulence using only hydrodynamic and chemical cues to escape predators or catch prey. In this seminar, I will explore the canonical planktonic navigation problem of vertical migration, where a planktonic organism aims to move upward or downward as quickly as possible through turbulence, relying solely on local flow velocity gradients. I will begin with a theoretical model that assumes flow gradients remain approximately uniform and steady over a certain time. Under this assumption, the optimal swimming direction can be derived analytically. Numerical simulations show that agents implementing this strategy can ‘surf’ on turbulence, exhibiting average vertical speeds of up to twice their swimming speed. Experimental results from larval snails provide further support for this surfing strategy, as these organisms actively respond to velocity gradients, counteracting local vorticity. Next, I will explore the use of deep reinforcement learning techniques to learn similar navigation strategies in partially observable environments. Despite the current limitations in reinforcement learning within this context, it presents unique potential for solving more complex navigation problems, such as agents with memory, multiple sensors, or cooperative behaviour. Finally, I will present an approach to design passive flexible agents capable of mimicking these surfing behaviours, offering novel insights into both planktonic organisms navigating turbulent flows. This talk is part of the Fluid Mechanics (DAMTP) series. This talk is included in these lists:
Note that ex-directory lists are not shown. |
Other listsCambridge Disaster Research Network [POPULAR Science] Life Sciences and Society SeminarsOther talksTBA Engineering large, multipurpose microprocessor specifications (using the x86-64 architecture as a case study) Topological Invariants for G-kernels and Group Actions Operator Algebras Q&A Session Generalised Lagrangian Mean: formulation and computation |