Over 1,000 Robots Programmed to Swarm

Just over 1,000 robots were programmed to swarm and work together by scientists at Harvard University. This was called a case of artificial self-assembly. In nature, self-assembly occurs when bees swarm or when birds flock together and act with a common goal such as heading back toward the hive or heading south for the winter. The scientists who created the robots were with the School of Engineering and Applied Sciences and Wyss Institute for Biologically Inspired Engineering at Harvard University. Michael Rubenstein was the lead author and the study report was published in the journal Science.

The task of self-assembly requires the interactions of a vast number of individuals that otherwise act independently in limited and unreliable ways. In nature, the mechanisms for getting a large number of individuals to behave in a group with a common purpose are largely unknown but are of great interest to scientists. The research group at Harvard was able to create both physical systems (robots) and programmable algorithms that were capable of this complicated activity.

The program was able to get autonomous robots to act as a group based on the individual, autonomous robots cooperating through local interactions. The collective algorithm managed a design for shape formation that was highly robust to variability and errors. In simpler terms, the individual robots communicated with their neighbors. With the ability to program and control 1,024 robots, this was the largest number of robots to exhibit artificial self-assembly.

The robots used in the experiment were called Kilobots and only cost about $20 each. In tests of self-assembly, the robots were initially arranged on a flat surface in an arbitrary shape that was tightly packed. Then “seed” robots were placed next to the shape. The robots on the far side of the shape then began to crawl around the edge of the shape toward the seed robots to self-assemble. The robots were driven by motors that made them vibrate much like cell phones do when set to vibrate. The robots were only able to communicate with the robots nearest to them and infrared light was used for communication. An individual robot was not able to determine what the whole collective of robots was doing. The seed robots acted as the point of origin in a coordinate system and information about their positions was propagated outward throughout the swarm. This is how each individual robot oriented itself and knew how to remain in the shape. In the experiment, it took about 12 hours for the robots to take on the shape that was directed by the program.

While scientists that work on artificial self-assembly systems may view natural cases of self-assembly for inspiration and lessons, successful examples of creating self-assembly by artificial means can also offer suggestions to biologists for identifying and understanding self-assembly in nature. There is mutual benefit when either natural or artificial cases are explored. This case of artificial self-assembly was said to be analogous to how ants form bridges with their own bodies to allow other ants to cross a chasm. Ideas about future uses for this new technology have included robots that are tiny, like a grain of sand, being able to swarm and snake through very small areas. Both very small and very large robots could be made to swarm with this new technology.

By Margaret Lutze

Wall Street Journal

Leave a Reply

Your email address will not be published.