Zigzagging could help robots find land mines and leaky pipes.
A zigzag path is the best way to track smells in turbulent air or water, a new mathematical model suggests1. The finding could help researchers to design smell-seeking robots, and to understand how animals search for food and mates.
In calm, still air, odorant molecules spreading out from a source of smell simply get less concentrated further away from the source. All an insect or robot must do to locate the source is follow the trail in the direction of increasing odorant concentration. A gentle breeze creates a plume of odorant downwind. This time, pinpointing the smell source is a matter of finding the plume and figuring out the wind direction.
But wind is usually turbulent, full of swirling eddies. It can pick up a pocket of odorant and carry it a long way. So odour hangs in stirred-up air in more or less random bursts - a confusing smell signal.
A sniffing robot or creature sitting in such a plume far from the source will sense only occasional patches of odour that might be coming either from the direction of the source or from some other direction determined by the random flurries of air.
To process this random information quickly and find its way to the source, any search strategy must be statistical, say Eugene Balkovsky of the University of Chicago and Boris Shraiman of Bell Labs in New Jersey. The searcher must estimate the direction based on a sample of several odour bursts.
And it "should actively explore the space," the pair suggests. Waiting passively for whiffy bursts of air to arrive won't do.
On encountering an odour patch in a turbulent stream of air, Balkovsky and Shraiman propose, the searcher should assume that it comes from an ever-widening cone-shaped area upstream, and set out to explore all of this cone in a zigzagging trajectory.
At some point along this path the searcher is likely to encounter another patch, which then relocates the probable source within a new cone. It must set out exhaustively to explore this new cone. Each detection of an odour patch refines the search area and brings the searcher nearer to the source.
In this way, the researchers show, a searcher can track back efficiently from a single odour patch to find the source-even if it starts from outside the main odorant plume.
This is indeed what some pheromone-seeking moths appear to do. They move upstream against the wind, but with constant detours to left and right - motions called casting and zigzagging. Some sea creatures such as lobsters may adopt similar tactics to smell their way towards food and mates in turbulent water.
Robots that follow smell are being developed for a variety of applications. Some could sense hazards such as leaking gas pipes, buried explosives like land mines, or chemical warfare agents. Others might function in swarms, laying odour trails that prevent one another from duplicating their paths, for example in cleaning operations.
Balkovsky, E. & Shraiman, B. I. Olfactory search at high Reynolds number. Proceedings of the National Academy of Sciences USA, 9,(02). /vol>(2002).