Complex systems are characterized by many independent components whose low-level actions produce collective high-level results. Predicting high-level results given low-level rules is a key open challenge; the inverse problem, finding low-level rules that give specific outcomes, is in general still less understood.
We present a multi-agent construction system inspired by mound-building termites, solving such an inverse problem. A user specifies a desired structure, and the system automatically generates low-level rules for independent climbing robots that guarantee production of that structure.
Robots use only local sensing and coordinate their activity via the shared environment. We demonstrate the approach via a physical realization with three autonomous climbing robots limited to onboard sensing. This work advances the aim of engineering complex systems that achieve specific human-designed goals.
In the case of mound-building termites, colonies comprising thousands of independently behaving insects build intricate structures, orders of magnitude larger than themselves, using indirect communication methods. In this process, known as stigmergy, local cues in the structure itself help to direct the workers.Werfel et al. (p. 754; see the Perspective by Korb) wanted to construct complex predetermined structures using autonomous robots. A successful system was designed so that for a given final structure, the robots followed basic rules or “structpaths” in order to complete the task.
1Wyss Institute for Biologically Inspired Engineering, Harvard University, Cambridge, MA 02138, USA.
2School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, USA.
Science 14 February 2014:
Vol. 343 no. 6172 pp. 754-758
Inspired by Swarming Termites, Robots Can Build Without Supervision
Thinking about insect swarms can make your skin crawl. But for some scientists, those thoughts inspired an idea for creating small robots that as a group can build a simple structure, with each bot working independently and without a blueprint.
Modeling the behavior of termites, a Harvard research team has assembled small robots and developed software that enables them to build something without supervision. Working with no central control, the robots instead are guided by simple rules as they interact with their environment, tools, and each other.
Ultimately, the goal is to have the bots swarming in places where it’s dangerous or difficult for humans to work, like disaster zones or in outer space.
The Harvard team’s theory and formulas were inspired by millions of termites that work autonomously to build mounds—some taller than 10 feet (3 meters)—in African rain forests and savannas.
Principal investigator Radhika Nagpal was joined by lead author Justin Werfel and electrotechnical engineer Kirstin Petersen. They describe their work in a paper in issue of Science.
In videos that accompany their paper, the researchers show teams of robots using foam blocks to build a variety of structures, from a simple set of stairs to a low wall surrounding what could become a room.
It’s impressive that the robots can stack things, said James McLurkin, a computer scientist at Rice University who is also looking to mimic the group behavior of insects. “They’re putting blocks on top of each other to build something,” McLurkin said. “No one’s done this before.”
The Harvard team’s simpler demonstration has three robots assembling four stairs that reach four blocks in height. Just to do that, the software and robots have to accomplish about 100 steps in correct order.
The bots perceive only the bricks and other robots in their immediate vicinity, using simple rules to move, climb, and carry bricks to the right spot.
A human user need only design a structure and the software automatically generates the rules that guide the robots. New formulas developed by the researchers turn the high-level goal into low-level rules.
Swarms of small, inexpensive robots operating independently offer advantages over other robot systems.
Multiple robots controlled by a central source can be difficult to scale, as larger teams can overwhelm a central coordinator, whether a computer or human. The robot hordes also can continue working if a few, or even many, individuals fail.
Insects Are Laughing at Us
The work reflects growing scientific interest in reproducing animal and insect behavior through mechanics and electronics to accomplish human goals. The specific concept of insect swarms is crucial to the success of future robotics, McLurkin said.
“Multiple robots working together will conduct search and rescue, patrol large geographic areas, map the floor of the ocean, and explore under your kitchen refrigerator,” he said.
But challenges loom large. Scientists struggle to control even five or ten robots at a time.
“Insects look at our goals of five to ten thousand robots and just chuckle,” McLurkin said, contrasting that with what bugs themselves manage: “They’re dealing with five to ten million.”
Scientists say they also work to help the public get past the “ick factor” when they speak of robots that swarm like insects, through outreach to schools and other groups.
Petersen said she often uses the argument that small, insectlike machines should be less intimidating than large, complex, and humanoid versions: “Wouldn’t you rather have small, simple robots you could easily crush?”
The researchers don’t envision the robot teams otherwise replacing human construction crews.
That’s because the robots, operating on their own, work with a randomness akin to robotic vacuums that crawl indiscriminately around a room.
“Because they are working on their own,” Petersen said, “the robots can’t operate as efficiently as a human crew.”
Source: National Geographic
Source: National Geographic