Reader, [Michael Rubenstein], sent in a project he’s been working on. Kilobot, as stated in the paper(pdf), overcomes the big problems with real world swarm robotics simulations; cost, experiment setup time, and maintenance. The robot can be communicated with wirelessly, charged in bulk, and mass programmed in under a minute. Typically, robots used for swarm research cost over a $100, so large scale experiments are left to software simulation. These, however, rarely include the real world physics, sensor error, and other modifying factors that only arise in a physical robot. Impressively enough, the kilobot comes in far under a hundred and still has many of the features of its costlier brothers. It can sense other robots, report its status, and has full differential steer (achieved, surprisingly, through bristle locomotion). There are a few cool videos of the robot in operation on the project site that are definitely worth a look.
