Those wasteful robots!
In line with the unfortunately common American anti-intellectualism, Senator Tom Coburn from Oklahoma recently called NSF out on its “wasteful spending” by listing a bunch of projects the significance of which he doesn’t understand, including three projects that involve robots. Here they are:
1) The laundry-folding robot
Released to much fanfare, this UC Berkeley project demonstrates that a robot can autonomously fold towels. It’s kind of a neat demo — or it would be if it didn’t have to be sped up 50 times for the video — but the real significance of this engineering achievement is not in simply getting a robot to do this particular task, much as people hate doing it themselves. What’s really going on is an advance in machine vision and motion planning algorithms to a unprecedentedly refined levels. The machine vision part is needed to figure out a correct spatial model of a hanging piece of cloth from flat camera views: where are its sides and corners? And the motion planning is needed to grasp, hold, and fold the object. So it’s not that we’re so pleased that a robot can now fold laundry. It’s that folding laundry shows that the robots’ capabilities advanced much in the direction of autonomous perception, object modeling and motion. Capabilities, you know, that can be used for amazingly practical things like manufacturing, or warehouse work.
2) The bike-riding robot
This robot rides a bike. Another demo that maybe looks useless but isn’t. The real drive of the instrumented bicycle project is to test our understanding of the mechanical dynamics of bike-riding by engineering a system with controllable parameters that does the same thing. Why is that good science? Because it requires a rigorous description of a complex and dynamic (meaning: it’s stable in movement only, it will fall if stopped) mechanical system, and because it provides insight into an even more complex feedback system that is a human bike rider. And practical implications are certainly not that difficult to imagine, including sport performance analysis, safer bike design, and perhaps two-wheeled autonomous vehicles that would find use in the military, which I’m sure the Senator doesn’t find wasteful at all.
3) The robot rodeo
An educational event where teachers learned to program robots and use them in their classroom. It’s pretty silly to think that an event that used less than $7k of taxpayer money and provided amazing training for hundreds of teachers in a fun way, training that is likely to improve math and computer science education in their classrooms, is wasteful. There’s really nothing to comment on here.
But I have two more general thoughts on this. One is how the Senator understands the premise that there should be no sacred cows that are immune to questioning when it comes to spending. Sure, question away. But think before you question about something: that it’s precisely the job of the NSF to question researchers’ proposals and fund those that, by consensus of expert opinion, are most likely to be both successful and impactful. And that if you don’t take the time to at least try to understand what the underlying research contributes beyond the flashy demo, then you’ll end up looking like a fool. And that it’s only appropriate that some wacky and less than obviously applicable “blue sky” research projects should be funded: haven’t you read, Senator, Richard Feynman’s memoirs and the classic tale of one frivolous project leading to a major discovery and the Nobel prize in physics? Of course you haven’t.
And two is that perhaps roboticists (and maybe other scientists) could do a better job of explaining what the drive behind the flashy demo is. It’s harder to argue that autonomous perception and motion planning are frivolous than it is to argue that laundry folding is.