Saturday, March 23, 2013

More on robotics

TED has a feature of 8 videos on robotics, the first of which caught my eye because it's a flying utility robot that is highly agile and very small -- only a few inches in radius. (Agility and radius are inversely related exponentially, so radius (size) has a big effect on agility.. more on this later)

One by itself, or a fleet of them, might be quite helpful on any number of projects, especially construction, and especially construction in a high threat environment. One might think immediately of pipelines and buildings and bridges, but also buzzing about large antennas and denied areas like biohazard project facilities or demolition sites (after a disaster).

In this video, entitled "Robots that fly ... and cooperate", hosted by Vijay Kumar of the University of Pennsylvania, we learn a few things that seem like really unique ideas, and might even be useful in non-robotic humans in project situations:

For instance: implicit coordination; the ability to work in a team by the simply expedient of sensing co-located neighbors and sensing an object that the whole team is working on. These robots can do this; there's no explicit communication or leadership among them (self directed work and work flow)

To to accomplish such coordination, the robots operating system assumes:
  • decentralized control
  • local information only
  • agnostic to neighbors
  • adaptive on the fly
(I wonder if they took a page from the Agile Methods book?)

Since each robot is individually small so as to maximize agility, to do big tasks it necessary to scale up and work as a team. Now here's an interesting sets of physics: as you scale up (use robots in a cooperating team), the apparent or synthetic radius increases. (Not unlike a synthetic aperature used in radars and other optical devices.) However, agility goes down exponentially, in part because of increased inertia.

Does that sound like a real agile team of humans? Indeed, as you add team members, inertia increases and agility is inhibited. I don't think I want to go so far as to put mathematics to the human situation, but it sure predicts the robotic situation.

Now here's something cool: using the Kinect sensor from the XBOX 360, a variant of the robot can do an empirical coordinate system -- no GPS required. (Who said empirical process control -- advocated in agile methods -- was impractical and only the defined process control paradigm known to all the Six Sigma crowd would work?)

The application is obvious: for denied areas, and especially for denied areas with physical unknowns, self mapping is possible from sensing the environment.

(When I was a field operative in the intelligence community many years ago, where was this thing?)

Vijay Kumar, U Pennsylvania.

Check out these books in the library at Square Peg Consulting