Issue
22 Volume
1-2 Number
2009 Year
Contact
Cornell University
222 Day Hall
Ithaca, NY 14853-2801
 
P (607) 255-7200
F (607) 255-9030
›  E-mail VP Research
›  Download vCard

At Cornell

This Is Not Your Ordinary Ivy League

Hod Lipson, Mechanical and Aerospace Engineering/Computing and Information Science Lipson
Download Full Article
Share | E-mail | Print

How Do We Design and Make Robots That Can Mimic Every Aspect of Nature?

Design and Make

I look for ways to use evolution to design things. Robotics is the visible part of what I do, but underlying the robotics are two basic questions: Can we design machines that can design other machines? Can we make a machine that can make other machines? These are fundamental engineering questions. Engineering is about designing and making things. The ultimate thing we can design and make is the thing that can design and make other things. If we can do this, then we will have understood the underlying engineering processes well enough to automate them. It is the ultimate engineering challenge.

As time moves forward, the kinds of things that the engineering community needs to design and make become increasingly complex. The only way to sustain this growth and complexity of products, the next generation of iPods for example, is by automating some of the processes. We must elevate the generic process from designing and making the products themselves to designing the processes that can in turn make new products.

The Biology Connection

Nature is my inspiration. Nature is always designing and making things through the evolutionary process—through development, growth, and learning. It is amazing. Its accomplishments dwarf those produced by the best teams of human engineers. If it is done in nature, however, I know it is possible, and there must be a way to automate some of it. Engineers can only do a small portion of this now, so we have a long way to go.

While these questions permeate many engineering fields, we focus on robotics, as these represent a notoriously difficult, visible, and potentially high-impact challenge.

Growing Robots

You have seen plenty of robots in factories and in Hollywood movies. They are made out of rigid components, and while you might be able to reprogram them to do different things, the actual physical body is fixed. It cannot be changed. If something breaks, it cannot be fixed—game over. Most of the products today are like that: their morphology cannot be changed. This is not the case in nature. Over a lifetime, bodies grow and change. If something breaks, it heals. If one muscle is used more, it expands. Over a developmental timescale, a baby develops and matures. Over evolutionary timescales, the body plans of a species change to match new environments and adapt to new tasks.

How can we make robots that can change their physical structure? Grow? Adapt? We are trying to make machines that literally do that. We look at mechanisms that will allow materials to develop in an adaptive way—to heal, respond, and adapt to the situation. I am very excited about this project.

To read the article in its entirety view below or download the full article.

Back to Top