Robot dinosaurs are the future of paleontology

Illustration for article titled Robot dinosaurs are the future of paleontology

Since dinosaurs died out 65 million years ago, there's only so much we can really know about how these creatures moved and lived. But as two scientists explain to us, building robot dinosaurs could unlock the secrets hidden in fossils.


This almost unspeakably awesome idea comes out of Drexel University. Paleontologist Dr. Kenneth Lacovara and mechanical engineer Dr. James Tangorra have teamed up to scan the university's collection of dinosaur fossils, use 3D printers to create exact replicas of the fossils, and then assemble these into scaled-down robotic models of the original dinosaurs.

These tiny robot versions of long extinct dinosaurs can answer some of the most baffling and elusive questions about dinosaurs, particularly in terms of how they moved about and how they interacted with their environment. Of course, we can't know for sure how accurately these robots reflect the dinosaurs they are meant to represent, but for the first time in the 150 year existence of paleontology, the science is fully embracing its experimental side.

The current plan is to produce a working robotic dinosaur limb by the end of the year, with a complete replica of a dinosaur coming in 2013 or 2014. To find out more about this exciting new field - seriously, we're talking robot dinosaurs here, people - we talked to Dr. Lacovara and Dr. Tangorra to find out more about their research.

How did this project get started? How did you come up with the idea of creating these robotic dinosaurs?

Dr. Kenneth Lacovara: I came to James, who was introduced through a mutual friend, and I wanted a way that I could replicate the material I was working with in the laboratory. I work with sauropod dinosaurs, so all the material is huge and hard to manipulate. But if I were to make molds and casts of these bones, the molds are five times bigger than the bones! It's a cost problem, it's a space problem, when you make a mold you only get so many pulls from it before it degrades.

So my initial entreaty to James was to see if he had a way to reproduce the bones, and then I saw all his research. Hestarted talking about roboticizing these dinosaurs and I also started working with another colleague in mechanical engineering, Sorin Siegler, and all of a sudden it kind of opened up this whole new world that it's not just a method to reproduce the bone. It's a way to let the bones tell us how they worked.


Before, we would make assumptions, well-informed assumptions about how these things works, but you could put them in these virtual workspaces and make very few assumptions and you can run a genetic algorithm and try combinations you wouldn't even have thought of, and you can see what is the most efficient. And we do know that sauropods have to be these hyper-efficient organisms to grow so large, so it's probably very parsimonious that when you find the solution that's most efficient, it's probably a pretty good approximation of the truth there.

Is this the first example of what we might call experimental paleontology? I know archaeology has a fairly well-developed experimental track, but I haven't heard of this sort of thing in paleontology.


Lacovara: Certainly, there are other paleontologists doing what you might call experimental paleontology. But it's still a limited but growing area in the field. I've seen computer models of how sauropods walk, I've seen some computer simulations of how bones are put together, but it's still an area that's just starting. For the most part, paleontology is still the same way now that it was 150 years ago.

How do you ensure that the robots you're creating are accurate reflections of the original dinosaur's movement and biomechanics?


Dr. James Tangorra: You can't. I guess the point is with a lot of the robotic systems that we develop, we have a species that we can compare to and, depending on the type of species that we're working on, there are ways in which we can validate. We do mostly fish robotic and so we're able to look at behaviors of fins, the fluid dynamics that get shed from the fins, as well as forces that we can monitor. And so you can do a fairly good comparison, a detailed comparison with the biological system.

But that's sort of the point with the dinosaur. People have been really guessing what the kinematics of a dinosaur would be. Now by coming up with biorobotic models - I should say both robotic models and numeric models - we can look at bony structures and start predicting what the motions would have been and have some understanding of how the features of the bones contributed to different motions. So rather than just looking at it and saying, "This is what must have occurred," we can start looking at different features and see how those features affect different behaviors.


Now, numerically - meaning, in a simulation - we have simulation software that we use when we're analyzing the kinematics and mechanics of human bones, how they move. And we can affect different properties and see the end result. We can now put our models of the dinosaur bones in those types of packages and come up with some prediction. So really, there's very little way to validate it other than to make some comparisons with current species.

What I think it will do is give us some insight into what likely occurred using a much more detailed analysis of features, like sizes, inertias, and we can manipulate elements of a bone. When you get a bone that's been fossilized, it doesn't necessarily mean that the bone looked like the fossil. It can be deformed over time. And so we can start looking at how small structural changes would have affected the original animal motions and at least come up with what we think are more detailed, more complete predictions and speculations about what occurred.


Will these robots look like "real" dinosaurs, with skin and everything, or are you just focused on getting the internal structure right?

Tangorra: Well, I don't know what a museum would do with this, but we're really focusing on the muscular-skeletal system. There is some use of skin and that could potentially be a source of compliance that would affect behavior, but at least now in the first pass we're more worried about bone and the bone properties. I'm just learning about dinosaurs myself so I'm very surprised about certain features that dinosaur bones have.


I envisioned dinosaurs to be much more akin to a reptilian version of an elephant than a bird, but you start seeing structural aspects of their bones which show they're much more like bird. Not just in the way they breathe but also in how the bones are much lighter - they call them pneumatic bones - they have essentially air pockets in them so that they become very light. So structurally they're very different from mammals. The way the muscles attach and the type of muscles they have are probably going to be very different with what we see with mammals.

But to get back to what's important, what we're looking at is the muscular-skeletal system. If we're trying to identify a muscle, some of the muscle properties might come from compliant or restrictive sheathes that serve on the muscles, so we may have to actually be manipulating properties that do appear visual, but we're not really too concerned about the skin structure until at least we're doing environmental tests of the robotic systems.


One use of the numerical and the robotic might be to actually look at how slight changes across the years may have come from environmental changes. You know, why is one pelvis much larger than another? It might be because the dinosaur is in a swamp that actually has a flow through it, so it gives them greater stability in a fluid environment. You never know with nature what was important, but the whole thing exists for a reason, and we hopefully will be able to start associating these features, both morphologically and physiologically, with behaviors.

Illustration for article titled Robot dinosaurs are the future of paleontology

Are you also trying to capture the internal structure of the original bone?

Tangorra: For the bones that you're using for kinematic studies, it's not important to have the inside, the inside structure. Because also you have a fossilized bone, so the inside structure that you see on a fossil has been mineralized, so you don't actually have the right properties. You have the appropriate structure, but you don't have the right properties. When you're doing the mechanical testing of a bone, then you have to introduce the internal structure.


That means you may have to be using a fossil that's broken or a bone that was broken so that you can see the internal structure. The internal structure doesn't matter so much for the overall kinematics. It will matter more when we're doing loading tests to see how bones may have deformed, but that's a combination of geometry and material properties. So then, in the experimental phase, we would have to wonder if the material properties were slightly different, we would have to see how those affect the individual bones' loading performance. But those are kind of two separate things.

Right now, the plan is to make much smaller, scaled down versions of the dinosaurs so that they're easier to work with. But would you ultimately like to print out massive, full-sized robotic versions of the original dinosaurs?


Tangorra: Well, we're a research laboratory, we want to build anything we can, so of course I've pictured a full-size dinosaur running around. Is it experimentally reasonable? Probably not. If you're going to build something for a museum that's life-size, then you're probably going to build something that just looks like it. Even if we were going to build a life-size leg, one of the problems of having a life-size leg bone is that it weighs a couple hundred pounds and you can't do very much with it, so that's partly why we want to have these scale models that are much easier to manipulate and slightly less dangerous to the experimenters.

But sure, once you nail down the kinematics, when you look at the actual behavior of the muscular-skeletal structure, the knee doesn't work like a hinge joint. There's a reason why you have the joint and the heads of the bones are arranged the way you do, they give you more degrees of freedom. Once we start seeing that on a dinosaur, you might in a museum piece want to demonstrate why the particular type of joint gave its behavior. But from a research perspective, you generally try to do scale models instead of things at full size.


Lacovara: Sure, that would be wonderful. The full scale is probably not necessary for the research, but it would be fabulous for education, for museum displays. So ultimately I would like to print full-scale models of the things that we're working on. And Drexel now has a museum, we have the Academy of Natural Sciences in Philadelphia, and I would love to see the dinosaurs one day displayed there in full size. It would be very impressive.

How did 3D printing help make this project possible?

Tangorra: It really simplifies it. That's what we're bringing to the table. 3D printing is such a wonderful new engineering tool that really allows us to go from a digital representation to a 3D model with the level of resolution that we desire. When you look at paleontologists, they've had model makers making casts out of the bones, and then you actually, to shrink something, you have to get a person to make a replication at a smaller scale. And there's error inherent in that, and once you make a small cast and you want to change any features, then you really have to make a new model and recast it, so it becomes very difficult, or at least time-consuming and expensive, to change the different features.


So with the 3D printing, once the scan was made and we put the numerical scan into a level of resolution that we wanted to work with - because computationally you have to carry a lot of information - but once we got it to the level that we wanted it, printing was a matter of a few hours. And we can print many, many copies, and everyone gets to have one, and we manipulate them and you break them and you redo them.

Some of the engineering challenges become imparting the right inertial characteristics so that when you get a 3D material it's made from a different material than bone and the distribution on the inside isn't the same. So when we're working with these and want them to have the right inertial characteristics then we have to build pockets into them and introduce different materials, but it's such a simple procedure to print them that that's not the challenge. I wouldn't say it's effortless, but it's very easy.


Lacovara: The technology is becoming much more accessible, it's coming down in price, the area that you can print is becoming much larger. I know now that car companies and aerospace companies have 3D printers that can print very large blocks now. So to make a big skeleton you can glue pieces together that are printed.

But one of the nice things about the printing technology is that you can go into the CAD program and you can draw in connections between pieces, you can actually draw in the armature that you want to use to meld the skeletons, and just print that along with the bones. You can scan in color, some printers can print in color. It will have greater fidelity than molding or casting, which tends to produce some defamation. This technology will produce the best replicas possible.


Our thanks to Dr. Lacovara and Dr. Tangorra for taking the time to talk with us. For more, check out Drexel University. Image by Patalakha Serg, via Shutterstock. Image of fossil being scanned via Drexel University.



This is a very interesting idea that I never would have thought of. I wonder how much this will feed back into robot design for other applications — paleontological solutions to locomotion problems in robots would be pretty cool.

Also, the Autobots thought of it first :P