In an article named "Puckish robots pull together," Nature describes the work done at the Polymorphic Robotics Laboratory (PRL) of the University of Southern California on self-reconfigurable teams of robots. There, Wei-Min Shen and his colleagues simulate the absence of gravity by creating a 2D representation of space by using an 'air-hockey table.' With jets of air flow blowing on the surface, the 30 cm-wide robots, working in pairs, evolve in a frictionless environment, pick elements such as girders to assemble structures like if they were in space. NASA will use these teams of autonomous robots to build space systems like 10 km-long arrays of solar panels and other huge spatial structures.
||Here is an illustration showing the concept of self-assemblying structures in space by using free-flying intelligent rope robots (Credit: PRL).|
As these robots will work in teams, the Nature article looks at the concept of group intelligence.
Several research groups have demonstrated that teams of mobile, communicating robots can perform complex tasks: for example, they can collaborate to push objects over a surface. This is reminiscent of the way ants show group intelligence when carrying out collective tasks such as foraging.
In space, however, there is the added complication of a weightless, friction-free environment, which can make movements harder to control. Two robots carrying separate components for assembly might easily collide, or career past each other.
To explore such difficulties, Shen and colleagues have created a two-dimensional analogue of space -- better known as an air-hockey table. Jets of air blow through a mesh of tiny holes on the table's surface, so that puck-shaped robots float in a virtually friction-free environment.
||Here is an illustration showing a '2D flight-test' on an air hockey table, extensible to future 3D flight-test in micro-gravity environment (Credit: PRL).|
Now, let's look at how the future space workers are teaming in pairs.
A key attribute of the robots is that pairs are linked by a cord. This helps to stop individual pucks from flying out of control and brings them together to assemble the components that they carry. When the tether is extended, each puck can wander independently, looking for building components.
So far the experiments have involved long, rigid girders with connectors at each end, which can be joined into flexible chains. Once each robot in a pair has found itself a girder, they pick them up using mechanical docking units and the tether is reeled in to pull the two pucks and their cargoes together.
||Here is an illustration showing a FIMER robot with its two free-flying heads and the way it can conncet itself to another robot(Credit: PRL).|
The researchers are now teaching the robots to assemble triangles. These triangular elements will then be assembled in more complex structures such as trusses.
The PRL site contains tons of presentations on the subject. Here are the three that I recommend reading: "Space Assembly and Service via Self-Reconfiguration" (PDF format, 13 pages, 606 KB), "Self-Assembly in Space via Self-Reconfigurable Robots" (PDF format, 6 pages, 328 KB), and "Docking Among Independent and Autonomous CONRO Self-Reconfigurable Robots" (PDF format, 6 pages, 280 KB).
And if you're interested by other approaches to robotic team building, here are the links to two previous stories, "Robots Developing Team Building Skills" and "The EvBots: When Evolution Trains Robot Teams."
Sources: Philip Ball, Nature, May 28, 2004; and various websites