Inspired by marine animals, a new study has successfully used measurements of local flow properties, such as velocity and pressure, to build up a picture of the previously unknown global flow field. The findings, published recently in JFM, will help to direct future research on the deployment of sensory arrays that could be used to guide underwater autonomous vehicles.

Across fluid dynamics in general, the approach is often to consider a flow field from the point of view of an observer i.e. someone looking down on the entire flow from the outside. Researchers are able to control the parameters of the flow and observe what effect the changes have on the overall flow properties, or as lead author Eva Kanso puts it “in lab experiments and simulations we often play God”.

The starting point for Eva was to think about the flow properties from the point of view of being in the flow field. “If we have access only to the local flow properties, how can we use this information to find out about the global flow field?”

Marine animals of course have been doing this for as long as they have been around on the Earth. They sense information locally at their own length scale and time scale and then use this to determine their swimming patterns. Eva describes a famous experiment in which harbour seals were blindfolded and put into a pool with small swimming robots. The seals use their whiskers to track moving objects via hydrodynamic cues and as such they could successfully find the robots despite their lack of sight.

The fundamental problem can be described mathematically as an inverse problem: given information for a specific subdomain how can this be extrapolated to find the solution across the full domain. This is much more difficult than the usual ‘forward’ problems as the solutions constructed to solve an inverse problem are often multi-valued or the problem itself can have multiple solutions.

Eva and her team began with local measurements of the velocity and then used these data to construct the global velocity vector field. The fact that velocity is a vector quantity – and so has both magnitude and direction – greatly increases the difficulty of the problem compared to the use of a simpler scalar quantity such as concentration.

Previous studies modelling the response of male moths to a chemical cue released by a female have been relatively successful, in part due to the fact that the chemical concentration is a scalar. Expanding these methods to vector quantities remains a challenge but the team were able to do so with relative success, demonstrating their techniques for the canonical problem of a circular cylinder in uniform flow. The velocity and vorticity fields from the model are shown in figure (a).

The next step for the team is to be able to use the constructed global flow field to then inform the movements of the sensing device, much in the same way that a marine animal is able to adjust its movements in response to local measurements. This could ultimately lead to improved development of autonomous robotic vehicles.

Currently underwater vehicles must surface periodically to obtain a GPS signal to track their location, whilst large and expensive camera rigs are required on the vehicle itself to be able to navigate at depth. Building autonomous vehicles that use the principle of sensory information in the same manner as marine animals would not only reduce the number of cameras required but also remove the need to surface entirely, thus allowing for longer and more expansive journeys into the deep sea realm.

From Eva’s point of view though, the real excitement in this work relates back to the original motivation: “How and why do animals do this? What computations are happening in their brains? It’s really a neurosensory problem and understanding what’s going on would be fascinating…”

This paper is freely available for two weeks*

Colvert, B., Chen, K., & Kanso, E. (2017). Local flow characterization using bioinspired sensory information. Journal of Fluid Mechanics, 818, 366-381. doi:10.1017/jfm.2017.137

*from the date of the post