Protein^®

Canadian roboticists are giving their amphibious robot a Jaws-like ability to home in on a human diver's beating flippers. Fortunately their six-flippered yellow machine is friendly - the idea is to help it learn routes from humans to patrol endangered marine ecosystems.

A video - also embedded below - shows AQUA swimming in Barbados last year. It can also walk by rotating its flippers like wheels.

The new tracking software looks for any rhythmic motion at a frequency of between 1 and 2 times a second - about the rate divers move their flippers at. To avoid confusion with other rhythmic patterns - say an evenly spaced shoal of fish passing by - it also checks for the specific colour of the diver's flippers.

Tests so far have used recorded footage from trials last year, the next step is to try it for real.

Right now the system only responds to the amplitude of the beating flippers. In future it could attend to more features of the rhythm to tell different people apart - so you wouldn't have to worry about your underwater companion swimming off with someone else. Previous studies have shown that walking gaits can be used to identify people, so swimming patterns may be the same.

Read more about AQUA at its website, where you can read a paper on fin-tracking(pdf).

Tom Simonite, online technology reporter