Last Thursday, Senator Jeff Flake of Arizona introduced the following amendment to the U.S. Department of Defense appropriations bill currently in Congress:
None of the amounts appropriated or otherwise made available by this Act may be obligated or expended for the development of a beerbot or other robot bartender.
This sounds like a joke, but it’s not: Legislation prohibiting Department of Defense funding of robot bartenders is on its way to becoming law. The reason why Senator Flake wants this to become law is based, at best, on a misunderstanding of how basic robotics research works. At worst, it's a deliberate decision to misrepresent the research for political gain.
In 2015, MIT researchers presented a paper at the Robotics: Science and Systems (RSS) conference on “Policy Search for Multi-Robot Coordination under Uncertainty” [PDF]. We wrote about it at the time, mostly because it’s interesting research, but also because, as a way of demonstrating the improvements, the MIT students set up a beer delivery task to show how multiple robots could work together efficiently, even if they weren’t able to communicate with each other all the time. It was a cool demo, and did a good job of showing how effective MIT’s new algorithms were.
The MIT work was funded in part by both the Office of Naval Research and the Air Force. This is not unusual: The Department of Defense (DoD) funds a lot of robotics research, and most of the time it’s not too much of a stretch to see how there could be potential defense applications for it. In this case, as we pointed out in our 2015 article: “in a warehouse or disaster environment, you probably have a bunch of robots all trying to coordinate to complete well defined tasks, probably without nearly as much information as they would like, and this kind of planner allows them to be consistent and efficient in their actions.” This also extends to other logistics tasks, the researchers said, like delivering medical supplies in hospitals.
Because of the demonstration involving beer delivery, Senator Jeff Flake is now using this research as an example of wasteful spending by the Department of Defense. This is not the first time a U.S. Senator has used robotics research for this purpose, and Senator Flake in particular is known for his efforts to reduce federal spending that he sees as wasteful. Here is a quote from Senator Flake’s presentation to the U.S. Senate introducing his amendment:
Mr. President, did you hear the one about three robots that walk into a bar? No, you haven't. It's not a joke, but rather a project paid for in part by the Department of Defense. These robots, called "beerbots" (you can see a depiction here), were programmed to serve cold beers to graduate students. Researchers say that programming methods used for beerbots can be applied to other multi-robot systems in restaurants and bars, but as you can see the private sector has already developed robot bartenders, or robartenders. They've been mixing drinks in bars and even on cruise ships for years now. And with our national debt exceeding $21 trillion, taxpayers should not have to pick up the Pentagon's tab for beerbots and many other unnecessary spending items which are in the bill we're considering right now.
This amendment was adopted by the Senate by unanimous consent.
The issue here is that Senator Flake is conflating a student demonstration of how research could potentially be applied (delivering beer) with the fundamental objective of the research (in this case, multi-robot planning under uncertainty). Senator Flake also does not make any reference to what the objective of the research actually was: solving real-world logistics challenges. Instead, he seems to think that MIT was using government funding to develop a commercial robotic bartender, and his solution is to amend the DoD budget to prevent something that wasn’t even happening in the first place, from happening again.
It's difficult to understand how Senator Flake could have missed all of the detailed and easily available information about this research. For example, you can see that he uses a screenshot from MIT’s video in his presentation on the Senate floor, but he appears to have ignored the explanation of the general importance of the research in that same video. And Senator Flake’s reference to the researchers saying that the robots could be used in restaurants and bars is taken directly from this MIT News story, which also makes clear (in the headline and elsewhere) that the system could be used to deliver medical supplies.
Robotics can be a complicated topic, but there are plenty of sources that Senator Flake could have asked for help in understanding this research, including the authors of the paper. We spoke with lead author Chris Amato, who is now an assistant professor at Northeastern University, to get his perspective on how his research is being interpreted (emphasis ours):
Our research was on very general algorithms for multi-robot coordination. Our target application domains were logistics problems such as delivering medical supplies. As a fun substitute, we used beer in the demo. The research really doesn’t have anything to do with beer and wasn’t about beer delivery.
According to Dave Levitan, journalist and author of the book Not A Scientist: How politicians mistake, misrepresent, and utterly mangle science, "this is unlikely to actually be a 'misunderstanding'." Instead, Levitan says, Senator Flake is most likely using this research as an opportunity to take on supposedly wasteful spending, by finding perfectly reasonable government funded research projects that can, says Levitan, "be made to sound ridiculous when put a certain way. He almost certainly does it on purpose, ignoring the true value of the research and counting on the public to not be well enough informed to push back. And no fact-checking happened because that would have ruined the bit."
The good news, I guess, is that this likely won’t have a tangible impact on robotics research. Instead of delivering beer, robots can demonstrate their capabilities by delivering any number of other things, as an MIT spokesperson pointed out to us in an email over the weekend:
To demonstrate the system, the students thought it would be fun to use cans of beer, but obviously any item could have been used to illustrate the sophisticated algorithms that were developed to enable the robots to complete the task.
Really, the only thing that this legislation is going to do is take away a little bit of freedom that grad students have to try to make their work fun sometimes, while reminding us how desperately we need more representatives in our government who have a better understanding of science and engineering.
We asked Senator Flake's office for comment, and a member of his communications staff pointed out that the Senator was clear in his statement that the programming methods used for the beerbots could also be applied to other multi-robot systems. Senator Flake doesn't have an issue with supporting robotics research, we were told, but he doesn't want federal funding to be used for the development of beerbots specifically.