A Need Breed of Biologically-Inspired, Intelligent Robots
Nature knows best and scientists are taking cues from biological cells to produce new, radically different types of robotics.
Robotics that behave like biological cells
An all-star tag team of MIT, Columbia, Cornell, and Harvard are developing “particle robotics,” a system of small, simple, disk-shaped robots that can work together to navigate around obstacles and transport objects. Synergistically, they accomplish what no individual could on its own.
The individual units, known as cells or particles, are about 6-inches-wide and can form loose connections with one another via magnets in their outer panels. And while most other robots are complex, these cells are strikingly simple — in fact, they can’t even move on their own. But they can expand and contract, pushing and pulling on each other to move as a single, oscillating entity, guided by sensors that detect light sources.
The sensors and algorithms within each bot measure light intensity and broadcast that information to coordinate the movements of the cells. Those closer to the light expand first, initiating a “mechanical expansion-contraction wave” that drives the entire group toward its goal.
The little robots’ simplicity is their greatest advantage. Their disc-shaped bodies allow them to adopt different shapes to best suit the task at hand, be it navigating through tight gaps, or pushing objects, tasks that previous, square-shaped modular robots couldn’t do as well.
Simplicity also means scalability. In simulations, up to 100,000 particle robots worked together to maneuver around obstacles. The individual robots do not rely on, or even directly communicate with one another, so researchers can add or subtract cells at will, making the project almost infinitely scalable. And failsafe. With no centralized control or single point of failure, squads of up to 10,000 cells could maintain their movement (albeit at a slower rate) even when 20% of the units had failed.
The researchers’ goals are to simultaneously expand and minimize the project, creating a swarm of millions or billions of microscopic cell robots.
Robotics that are biological cells
A foray into experimental robotics, conducted by Tufts, Harvard, and University of Vermont, challenges the definition of the word robot. The project combines AI and microengineering to create “novel living machines” that are neither robots nor animals but “ living, programmable organisms.”
These small, millimeter-wide “Xenobots” are named after Xenopus laevis, the African clawed frog whose embryonic cells were used in their construction. A construction that began virtually on Vermont’s Deep Green supercomputer cluster. It spent months of processing time, analyzing designs for new life-forms, keeping the ones that produced the desired behaviors and discarding the rest – evolution in virtual form.
Researchers then physically recreated the most promising designs in the lab. They harvested skin and cardiac cells from frog embryos, and spliced them together using tiny forceps and a cauterizing electrode. The skin cells formed the “creature’s” body and the beating cardiac cells endowed it with movement. The cells self-organized into something completely foreign to the tree of life, something both living and dead.
In the petri dish, the xenobots moved on their own, either walking, swimming, or crawling, depending on their physiological design. In another test, scientists introduced a number of pellets, which the Xenobots encircled and pushed to the center of that circle, collectively and spontaneously.
Cyborgs like these boast numerous advantages. They don’t require chemically complex, expensive materials, as the Xenobots were made from 100% biological tissue. And they don’t create pollution when they reach the end of their 7-day life-cycle. They simply dissolve, producing the same ecological impact as a frog shedding its skin.
Minuscule biological machines boast a number of useful future applications, most notably in the realms of environmental and personal health. Microscopic automotons could search out and gather radioactive waste. Or they could help clear tiny plastics from the ocean by piling them up (as they did the pellets during lab tests).
Or they could be used in medical settings to perform janitorial services throughout the body. Bio-bots made from a patient's cells (to avoid immunological rejection) could scrape plaque from capillaries, or deliver medications directly to specific tissues.
An interesting few decades ahead
These projects highlight the inevitable fate of robotics engineering: the integration of living and nonliving components. Over the coming decades we’ll see not only robots that act as if they’re alive, but some that truly are.