Almost exactly a year ago, I wrote about the remarkable abilities of Scincus scincus, otherwise known as the sandfish, sand skink or sand swimmer. This little Saharan lizard can swim through sand like a fish through water, and the way in which it does it remained a mystery until researchers at Georgia Tech set out to tackle the problem. It turns out that the sandfish undulates its body in a sinusoidal wave motion, tucking in its limbs and modifying the frequency of undulation in order to control speed and manage its movement in sands of greater or lesser compaction.
At the “CRAB lab” at Georgia Tech, Daniel Goldman and Ryan Maladen's team have continued their research, moving further into mathematical modelling, iterated and integrated with lab experiments; they have also worked in collaboration with Paul Umbanowar at Northwestern University in Illinois who is one of the gurus of the bizarre world of granular materials, and who kindly helped with several illustrations in my book. The results not only shed light on one of nature’s wonders, but generate important advances in robotics. Recently, the team has reported on a newly developed prototype robot, constructed from easily available components, that directly simulates the skills of Scincus scincus, and could lead, for example, to vital developments in post-earthquake search and rescue efforts. The complete paper by the group is available online (the illustrations at the head of this post are taken from there), and the Canadian Discovery Channel has a great video report (although I do rather resent the presenters referring to sand as “dirt”). The work was described briefly in the New Scientist, but since the full version, by James Urquart, is only available by subscription, I’ll take the liberty of reproducing it here:
Lizard-like robot can 'swim' through sand
TO ADD to the robots that can crawl over land, fly through air and swim underwater comes one that can swim through sand. Such robots could help find people trapped in the loose debris resulting from an earthquake.
Navigating through sand is harder than moving through water or air because sand can behave as both a solid and a fluid. We have no equations to describe how such substances flow, let alone to predict how an object can "swim" efficiently through them.
But the sandfish lizard, Scincus scincus, can travel through sand effortlessly, so Daniel Goldman and Ryan Maladen's team at the Georgia Institute of Technology in Atlanta decided to find out how they do it. They found that once the sandfish is submerged, it tucks its limbs into its sides and propels itself forward by wiggling from side to side.
Working with Paul Umbanhowar of Northwestern University in Evanston, Illinois, the team plugged their results into a computer model, which they used to show that a snake-like robot with just seven body segments could travel through a granular medium like sand.
Encouraged, the team built a 35-centimetre-long version of the robot, made from seven aluminium segments linked by six motors, all clothed in spandex to prevent the motors from becoming jammed.
The team then tested their robot by burying it in a container filled with plastic spheres 6 millimetres across. When the robot undulated its body at a frequency similar to the lizard, they found it could move forward at speeds of up to 0.3 body lengths per wave cycle - just below the 0.4 body lengths per cycle that a submerged lizard can achieve.
The robot could eventually match the lizard for speed, says Goldman, if more jointed segments are added to make its movements smoother.
Howie Choset, a roboticist at Carnegie Mellon University in Pittsburgh, Pennsylvania, thinks that the physics and the biology-inspired approaches to robot movement will one day meet and "at the intersection will be a deeper understanding of how biology works and how to make robots better".
Amongst the many things that I find fascinating about this work is the fact that it is a carefully executed project that uses mathematical modelling in intimate combination with laboratory experiment – the “fieldwork” informs the maths which, in turn, informs the next phase of laboratory investigation. In a world where, in my view, we place undue faith in the validity of mathematical models alone, particularly when evaluating complex natural processes, this kind of “continuous feedback” approach is to be admired and emulated.
[ The Georgia Tech press release page has links to further sources]
Interesting too that traditional design so closely approximated technical findings, as in this example of a fond early memory:
http://www.letterbox.co.uk/20670-04382-LBPOCKETMONEYTOYS3-2/pocket-money-toys/up-to-3-pounds/jointed-snake
Greetings, Michael. I have been keeping up with your postings, and greatly enjoyed the set on writing a successful work of popular science.
Best wishes,
Richard
Posted by: Richard Bready | July 23, 2010 at 12:01 AM
Greetings Richard - and thanks for the analogy. I hadn't even thought of the design correlation with those toy snakes - but I remember them well myself!
Posted by: Sandglass | July 23, 2010 at 05:31 PM
I thought you had another entry related to this aside from the one at your link:
http://throughthesandglass.typepad.com/through_the_sandglass/2009/06/of-reptiles-and-robots---locomotion-in-sand.html
Apparently, my memory is still intact. :)
The tight coupling between modeling and experimentation is a really fantastic situation when it comes along. I'm sure it would be difficult to accomplish in some studies, but certainly more researchers could bring these together more often. Perhaps I have always been naive, but I have always thought that this was the core of the scientific method.
Posted by: F | July 23, 2010 at 07:47 PM
Hmmm! That's a creative ideas to use robot for right purpose.
Posted by: pcb prototype | December 07, 2011 at 05:54 AM