|
|
|
Unlocking Animals' Divine Dance
BNH speaks with A. Stephen Morse, an electrical engineering professor at Yale who was recently awarded a $2.6 million National Science Foundation (NSF) grant
|
Business New Haven
10/18/1999
By: BNH
|
Tell us about the research for which you received the NSF grant, and who is on your team.
My overall research deals with systems theory and automatic control, and the NSF-funded project is within that area. The National Science Foundation set up a program two or three years ago to encourage research in what they call intelligence systems, which is a sub-area of what they call analysis distributed information. We formed a team consisting of myself, Professor Peter Belhumeur, who is also at Yale and is an expert at computer vision. Computer vision deals with having a computer understand what is in a photograph. Then there four other scientists in the group: Roger Brockett from Harvard University and Naomi Leonard from Princeton, who has experience with underwater robotics - underwater robots. In addition to the four of us, there are two other principle investigators from the zoology department at the University of Washington/Seattle: Julia Parrish, who is an expert in the study of fish schooling, and Daniel Grunbaugh, who is also an expert at studying natural occurring groups, such as schools of fish, groups of deer or flocks of birds. He's our theoretically inclined member.
What is the objective of the project itself?
The project is to try, on the one hand, to understand the mechanisms or protocols or algorithms of how animals in groups are able to coordinate their motion. Part of the project is to try to understand the motion of fish schools, with the fish having been tagged or marked so that it is possible to track their motion in the water. If you're going to understand how fish move around in schools, then you're going to have to photograph the fish in schools and watch the motion of individual fish. This turns out to be a un-trivial problem if one is interested in the three dimensional motion of fish. In two dimensions there is a database; in three dimensions, there is no such database anywhere in the world. One of the goals of this project is to obtain such data and make a database. To do that, Peter Belhumeur, who is an expert in all phases of computer vision, will assist in the process of acquiring that information and doing the image-processing necessary to make the data useful. Once that data is acquired it will be put on the Internet for anybody to use. So that's one goal of the project.
What is the next step in your research?
The other main goal is to build a school of robotic fish, except that the fish that we are going to build will not swim like fish. Fish, of course, move their tails and do all sorts of things, but the fish we're talking about will move like an underwater boat or a submarine. We're more interested in the grouping behavior then we are in the individual behavior which would emulate a fish. We hope to build a school of about ten of these - let's call them 'submarines' for the moment, or 'autonomous underwater vehicles' - AUVs. We would put them into either a swimming pool that we've built here at Yale and/or the Princeton University Olympic swimming pool where we would experiment with various algorithms or protocols to implement on the various submerged vehicles in order to have them coordinate their motion. We'd like to have them put themselves through various kinds of maneuvers without human intervention - that's where the word 'autonomous' comes up. They're going to be thinking for themselves, we hope, and maintaining their own same spacing, using their own rules, which we will program into the submarines. We hope through this process to understand a little bit of how one might build a grouping of coordinated anything - underwater vehicles or a flock of helicopters or a team of ground-based robots.
What are some possible practical applications of this research?
As you know, the [John F.] Kennedy [Jr.] plane went down en route to Martha's Vineyard last month, and there was an underseas search for the aircraft. For a search covering such a large area as that one could imagine putting into the water a large number of autonomous vehicles and having them perform a coordinated search. So search underwater could be an application. One could imagine a flock of small autonomous flying aircraft to seek out and rescue hikers or other people lost. Certainly for exploration, for rescue, [though] not for actual rescue by small vehicles; they couldn't pick up a human being. Certainly they would be useful for delivering medical supplies and locating stranded people.
|
Go FirstGo PreviousGo
NextGo LastGo
to Index
|
|
