|
Biological systems, from embryos to social
insects, get tremendous mileage by using vast numbers of cheap and
unreliable components to achieve complex goals reliably. We are rapidly
building embedded systems with similar characteristics, from self-assembling
robots to sensor networks. How do we understand and engineer collective
behavior?
In this talk I will describe two ongoing projects in my group. The goal of
the first project is to understand how local cell behavior affects global
tissue-level properties in the developing fruitfly wing. I will show an
abstract model of cell division that led to an unexpected prediction - that
the system intrinsically regulates polygonal cell shape to a fixed global
distribution. This distribution appears to be conserved across a diverse set
of organisms.
The goal of the second project is to design local behaviors for robots so
that they collectively achieve a desired global goal, such as constructing a
user-specified 2D block structure. I will discuss a set of local algorithms
we have developed, and show how transferring "sophistication" from the
robots to the blocks can result in higher efficiency and robustness. I will
also describe a prototype we have built using simple robots and RFID
enhanced blocks.
Bio:
Radhika Nagpal joined Harvard University as an Assistant Professor of
Computer Science in Sept 2004. She received her PhD degree in Computer
Science from the Massachusetts Institute of Technology in 2001. In between
she was a postdoc lecturer at MIT and then a research fellow at the Systems
Biology department at Harvard Medical School. Her research interests are
biologically-inspired approaches to multi-agent systems and modeling
multi-cellular behavior in biology. She recently received the 2005 Microsoft
New Faculty Fellowship award.
|