Friday, July 20, 2007

Biomechanics – Cool stuff!

The physics posts that have shown up until now on this blog have been on boring things (albeit, close to my heart boring things) such as the second law and equilibrium self assembly. So, I have been thinking for a while now that I should write something on a cooler topic, even though I might know a lot less about it. The topic kind of chose itself this last week when my idle reading on collective intelligence (more on that in a subsequent post) took an interesting tangent.

The tangent is Biomechanics. As a discipline, in the broadest terms, this is a quest to understand the locomotion of living things. Of course, this quest can be addressed at many levels. But for the purposes of this post, we are going to restrict ourselves to the following question. Clearly, living things have lots of stuff on their mind (read neural control circuit) apart from locomotion. So, they have evolved in such a way that they must have a minimal feedback-limited control circuit that governs locomotion. If we could figure out what this minimal model is, we could apply it to robotics and hence be able to design robots that walk easily and hence have room left in them to build in other functions. Why would we want to do that? A standing example could be what happened to the Mars Rover Opportunity. The video below is a time lapsed footage of the rover extricating itself from some loose sand. In real time, it was stuck for a month [1].

You see, the Mars rover belonged to the old robotics paradigm of control that was not feedback-limited and certainly not minimal. This idea of minimality is something we learnt from watching nature solve complicated problems with ease [2]. So, people started asking, how do living things walk/run? A successful minimal model that came out of this study is what I call the “foot-forward strategy”. This model essentially said that the organism put half its feet down and kept the other half in the air (3 feet if you are a hexapod and one if you are human). It measured the resistance that the surface gave each of its feet and decided how far ahead to land the feet that are in the air and then keep repeating the process [3,4]. This model, that has just two ingredients could successfully explain the locomotion of a wide range of living things in a wide range (not exhaustive as you will see below) of terrains.

This strategy was implemented with great success in a robot called the Rhex. See below a video of Rhex zipping through all kinds of terrain, in a direction given to it by a guy with a remote control [5].

This is all well and good and a success to the method of scientific enquiry. But this model works only when the terrain is solid. Consider for example the following video of a spider moving on some uneven substrate. This substrate has holes that are larger than the foot of the spider and deeper than the length of its leg, but it still manages to zip across it (the video is slowed down 20 times).

So the question now is how do we understand this kind of motion. In attempt to study this question systematically, researchers decided to take this motion into the lab. Find below a video of a cockroach running across a wide mesh. Again the video is slowed down 50 times. It is actually moving very fast [6].

Now, there are several possible explanations for how the cockroach manages this. One possible explanation could be that this is an emergent (euphemism for “pleasantly unforeseen”?) consequence of the foot-forward strategy itself. But this the researchers can test readily, for they had the Rhex and we know for a fact that Rhex does not know anything other than the foot-forward strategy. So they did that and let Rhex run on the mesh to see what happens.

Oops! Rhex does not like the mesh! This tells us that there are two possibilities. The messier of the two is the possibility that the Cockroach has more than the simple foot forward strategy built into its neural circuit and we need to figure out what that is. But there is a simpler possibility. May be there is a physiological feature of the leg of the insect that we are missing. And this latter turns out to be the answer. Look at the cartoon along side of a cockroach. Its legs have spines or hairs or whatever you want to call them. These hairs have the property that they give easily in one direction (when pushed towards the leg) and are very stiff in the other direction, requiring loads greater than the weight of the insect in question to make them give. What the insect does when its foot lands ina hole is to use one of these pikes for leverage. You can go back to the video of the cockroach to see that this is indeed the case. So the claim now is the foot-forward strategy together with spikes or hairs are sufficient to negotiate terrain with gaps. The researchers tested this as well. They took Rhex and put spikes on his legs with the same properties as those on the legs of insects and put him back on the mesh. See the outcome in the video below.

It works! Rhex manages to get across, even though less elegantly than his real insect counterparts. The researchers of course performed other tests to verify the hypohesis. They took a cockroach and removed the hair from its legs and watched it stumble on the mesh. They took one of the fastest running creature on earth, the ghost crab (Ocypode quadrata) and let it run on the mesh. It struggled of course, because it runs on sand and hence has no spikes on its legs. And then they put spikes on its legs and watched it make it across the mesh successfully [7]. And so the researchers have successfully demonstrated that the foot forward strategy is enough even with holes in the terrain! This whole thread is a cool illustration of scientific methodology in general and that is one of the reasons I decided to write about it. The other of course are the cool videos. Are you as impressed by the coolness of it all as I am?

Asides, References and Disclaimers:


[1] I had the unique opportunity of watching the rover stuck on Mars live! I was at a NASA meeting at the Kennedy Space Center at the time and they were supposed to show us the rover in action live. But as it turned out, the Rover was stuck for the duration of the meeting.

[2] I looked for a reference on the control strategy for the rover, but could not find one. So, what I am saying here is hearsay. It must be classified or something. And the hearsay comes from a friend of mine that is Robotics researcher at UPenn, subject to my understanding of what he said.

[3] Notice that individual sensing gives an advantage to the multilegged creature. If the resistance on the front most leg is smaller than the hind ones, you know you are probably going onto softer terrain and step accordingly and keep your eyes, if you have them, on your food or predator or ipod or whatever.

[4] This of course is my minimal interpretation of the model. Find more details on this and other models in this Science review paper and the references there in.

[5] All the references on the development and implementation of the Rhex can be found in the website liked above. As an aside note that this project is funded by DARPA. So, unless we have secretly discovered an alien inhabited planet that the U.S is planning to invade, Rhex is more likely to be used in Iraq or whatever the next location for the war on terror is!

[6] This and all of the following is work done by Daniel Goldman and his collaborators. You can find all relevant references at his website. The videos are all stolen from there as well.

[7] The crab videos are on youtube as well. If you want to watch them, they are here and here.

PS: Apologies if this post showed up multiple times in your feed reader. Blogger screwed me over.

8 comments:

Born a Libran said...

Just as you had said, this was some pretty cool stuff... Shows how we keep learning from nature...

CuriousCat said...

Yes BaL, yes. Glad you see it too! And given all this people discounting evolution is beyond me!

Born a Libran said...

Why, you dont believe in Intelligent Design? Cant you see the robot which has been designed... uh.. intelligently for ...uh... our purposes? Seriously, I know where you are coming from.. but the amount of evidence for evolution is overwhelming but has not been well received by the public... Gotta blame the scientists for that to some extent..

CuriousCat said...

You are right of course, BaL. I was ranting a while back about scientists taking their role in outreach seriously, it is a post filed under junk I think...+ the evolution thought was perhaps too tangential...did you read Sean Carrol's argument (on Cosmic variance) about the fact that if animals had wheels then may be ID is reasonable, but not otherwise?

Born a Libran said...

Havent read too much of Sean Carrol... But I understand what he was saying... I wonder evolution never came up with round wheels and brakes though ;)

RTA Express said...

Dude nice summary of my (and Goldman, Lin, Koditschek and Full's) paper. I feel I've crossed a net-fame threshold, because somebody finally violated copyright to post our videos on YouTube. I won't snitch though...

One thing to add is: consider how much circuitry you save in a robot if you can design it to essentially throw its legs out in a stereotyped pattern, absent most electronic (equivalent to neural in biological systems) feeback, and it will cope on challenging substrates, due to it's leg shape, orientation, and compliance of various extensions. You just need a pattern generator, no need to load rHEX up with haptic or light sensors and all the associated electronics, because his mechanical system is tuned to deal with the environment.

Glad you enjoyed the paper-- btw, the reference is:

Spagna, J.C., D.I. Goldman, P-C. Lin, D.E. Koditschek, & R.J. Full. 2007. Distributed mechanical feedback in arthropods and robots simplifies control of rapid running on challenging terrain. Bioinspiration and Biomimetics 2: 9-18.

Cheers,
J. Spagna

CuriousCat said...

Thank you for stopping by Dr. Spagna. Cool stuff you have done. And I did read the biomechanics paper.
+do not snitch to youtube for a while, for this post has not been read widely yet!:) And please tell me if I should correct my write up because of lack of understanding of something and I shall do that

Anonymous said...

That's a cool stuff!