The Value of Interdisciplinary and International Collaborative Research

June 9,  2015, 4 pm





The most interesting, rewarding, and productive work of my career has been highly collaborative, involving scientists in other disciplines than computer science and other nations than the United States.  My first serious research project occurred while I was an undergraduate student at Princeton University as part of a team of computer scientists and organic chemists.  The theme of interdisciplinary science continued at Stanford University as part of the Heuristic Programming Project. All of the research in that laboratory was done in partnership with scientists, engineers, and physicians.  The theme continued when I joined NASA and created what became a large intelligent systems laboratory of computer scientists working with space scientists, aeronautical engineers, and mission controllers.  It continues to this day as part of my work with the US Air Force Office of Scientific Research (AFOSR).

Why bother with what may seem as the additional burden of having to learn about scientific disciplines outside of computer science?  First, the lessons of the real world are often very different than those of the “toy” domains often common in computer science research.  Take automated planning systems as an example.  The blocks world domain commonly used in early robotic planning research emphasized the issue of dealing with combinatoric explosions—the operators were simple, but if you ran into a roadblock, you had to backtrack back to an early stage in planning.  But this is very different from a problem like planning experiments in molecular biology.  Here is you run into a roadblock—for example a step makes the temperature too hot to do the next step—there is lots of knowledge about how to add a corrective step to fix the problem and not have to backtrack.  Second, working with real data and problems makes the technology transition process enormously easier and faster.  And finally, it’s just a lot more fun and interesting, at least in my experience, to work with really smart people in domains other than your own.

While I have always had international collaborators, my recent work with AFOSR has reinforced the importance of international science.  A current major emphasis is on understanding how to build effective human-machine teams.  Here cultural differences really matter as can be seen, for example, by different attitudes toward humanoid aspects of robotics in Asia, Europe, and the United States.  We just began a program that involves 15 research groups in Korea, Japan, Malaysia, Australia, England, Slovakia, Argentina, and the United States to explore how human-like appearance and behavior affects trust and performance in such teams.