Teams of humans--from soldiers to athletes--routinely work together to accomplish socially oriented tasks such as search-and-rescue missions. Robots, on the other hand, have so far been best deployed alone or under remote control. Now, however, Harvard University claims to have created an autonomous robot small and cheap enough to try out algorithms that coordinate team-like behaviors among legions of autonomous robots.
Today, there is no shortage of algorithms designed to emulate the team-oriented behaviors of cooperating humans, but their execution can only be done in software simulations or, at best, among a small group of robots. Harvard University, on the other hand, envisions legions of coin-sized robots executing the test algorithms designed to demonstrate cooperative behaviors among automatons.
Legions of coin-sized Kilobots work together to solve problems that require cooperation among socially oriented automatons. (Photo courtesy of Michael Rubenstein)
Created by members of the Self-Organizing Systems Research Group, led by Harvard professor Radhika Nagpal, the Kilobot reduces the size and cost of robots to the point of enabling large-scale hardware emulations of cooperative behaviors rather than just software simulations.
Harvard has licensed its cost-reduced Kilobot design to the Swiss robot manufacturer K-Team Corp. (Yverdon-les-Bains, Switzerland), which markets the robots and their development system. According to Nagpal, the Kilobot has been cost-reduced using a variety of novel techniques that together make hardware emulation of cooperative behaviors feasible for the first time. For instance, instead of costly conventional motors for wheeled locomotion, the Kilobot uses inexpensive vibrational motors attached to its spindly legs. Turning on one motor causes off-center centripetal forces that rotate the robot, whereas turning on two adjacent motors simultaneously causes forward motion at a fraction of the price of conventional motors.
The Harvard researchers--comprised of Nagpal and Michael Rubenstein, a postdoctoral fellow, along with Christian Ahler, a Harvard fellow at its School of Engineering and Applied Sciences and the Wyss Institute--recently demonstrated 25 Kilobots working together to emulate a variety of cooperative behaviors, including lining up in formations like soldiers, foraging like animals and synchronizing fully autonomous behaviors like a team. Once an algorithm is perfected with just a few robots, the Kilobot architecture allows it to be expanded to include potentially thousands of autonomous robots working together.
The researchers expect Kilobot algorithms to be perfected for research into the swarming behaviors of real insects, such as to collectively moving large objects or building nests, as well as human-oriented cooperative behaviors, such as searching through rubble for survivors after an earthquake. The Kilobot could also test out algorithms that use thousands of robots to remove pollutants from hard-to-reach areas.
Funding was provided by the National Science Foundation and the Wyss Institute.