In 2020, a new lifeform arrived on Earth. More specifically, it arrived in a lab — the Levin Laboratory at Tufts University in Massachusetts. As alien species go, these were no little green men or any other science-fiction cliché. They looked more like tiny black specks of fine sand moving slowly around in a Petri dish. And while they’re not alien in the extraterrestrial definition, they certainly are in the sense that they’re strange. These so-called “xenobots” are living, biological automatons that may just signal the future of robotics as we know it.
“These don’t fit the classical definition of an organism because they can’t reproduce — although from a safety standpoint this [is] a feature and not a defect,” Douglas Blackiston, a senior scientist in the Allen Discovery Center at Tufts University, told Digital Trends. “They could be classified as an ‘imperfect organism.’ I think they do qualify as robots, however. Even though they are living, they are built from the ground up for a specific purpose. These aren’t something that has ever, or could ever, exist in nature — it’s a human-made construction.”
Let’s back up. Last year, researchers at Tufts created the world’s first tiny, living, self-powered robots. These xenobots have been designed to function in a swarm: Walking, swimming, pushing pellets, carrying payloads, and working together to “aggregate debris scattered along the surface of their dish into neat piles.” They’re able to survive for weeks without food and to heal themselves after lacerations. Oh, and they’re made out of bits of frog, reconfigured by an A.I.
To create the xenobots, the Tufts researchers took skin cells from fresh frog embryos (the species of frog is called a Xenopus laevis) and encouraged them to “reboot their multicellularity” in a new environment. Freed from the rest of the embryo, these skin cells formed what Michael Levin, the scientist after whom the Levin Lab is named, calls a “proto-creature,” complete with its own unique structure and behavior.
As the Tufts scientists were creating the physical xenobot organisms, researchers working in parallel at the University of Vermont used a supercomputer to run simulations to try and find ways of assembling these living robots in order to perform useful tasks.
“We use A.I. to ‘evolve’ different robot designs in a virtual world,” said Blackiston. “The computer is given a task, like ‘make a robot that can walk in a straight line,’ and it assembles millions of different combinations of virtual cells until it solves the problem … The computer then gives me a blueprint, and I get to work connecting the cells to make the living version. So, in a way, I’m taking orders from the computer.”
An initial paper about the work, a proof of principle that living robots exist, and that A.I. can design them to do simple things, was published last year. A second paper, published recently in Science Robotics, shows that steps have been taken to make these into useful tools.
Traditional developmental biology has focused on standard model systems, such as the fruit fly, the mouse, and the frog, and how their genomes encode hardware that creates a certain kind of body. The xenobots Levin and his fellow researchers are working on what he told Digital Trends is the “complementary question.” This concerns the “reprogrammability of the software of life,” and whether genetically normal cells be enticed to build something that’s quite different from their biological default.
“I think this is the beginning of a new approach in which a myriad of novel life forms are added to the standard toolkit of biologists that enables them to ask where bodyplans come from, how cooperation among cells works, how cellular collective intelligence is implemented, and how we can stimulate cell groups to make whatever we want,” Levin said. “Not only does this shed light on the relationship between genome and anatomy — since our xenobots have a totally standard frog genome — but it also enables useful synthetic living machines, and gives us a new sandbox in which to understand the rules of morphogenesis.”
The idea of biological robots is not new. In fact, it arguably pre-dates the modern conception of robots as largely ruggedized, metallic entities. The robots imagined by the Czech playwright, Karel Čapek, who coined the term “robot,” in his 1920 science-fiction play Rossum’s Universal Robots are biological in nature. They are created in a factory using synthetic organic material, making them more similar to the modern idea of androids than machines.
Other real-life researchers have also sought to combine the natural and machine world in interesting ways. The European Union-funded Flora Robotica program aims to “develop and investigate closely linked symbiotic relationships between robots and natural plants and to explore the potentials of a plant-robot society able to produce architectural artifacts and living spaces.” A project funded by the Office of Naval Research, meanwhile, focuses on the construction of an insect army of backpack-wearing cyborg locusts for carrying out tasks like bomb detection. At Zhejiang University, in China, researchers have created a setup that allows humans to mind-control the movements of rats by way of a technology called a brain-brain interface. Last year, researchers at Stanford University embedded low-power microelectronics in live jellyfish with the goal of enhancing their natural propulsion. And so on.
The difference between these projects and the xenobots is that the latter doesn’t simply use tech components to augment the abilities of a biological organism; it creates an entirely new biological organism that can — or, at least, will — be controlled like a wholly artificial robot.
“A.I.-designed xenobots explode the definitions of both robot and organism because they embody traits of both,” Josh Bongard, a professor in the University of Vermont’s Department of Computer Science, told Digital Trends. “They are like robots because they are designed to perform some useful function for humans autonomously. But they are also organisms in the sense that they are genetically unmodified frogs, just pushed into very different forms and functions.”
Xenobots, their creators promise, are likely to have a number of different applications, both far- and near-term. Levin suggested that near-term possibilities might include environmental cleanup and sensing, since the use of amphibian cells used to living in outdoor temperature water, which are biodegradable in about a week, could make them perfectly suited for these scenarios. The bots can metabolize dangerous chemicals and sense tiny amounts of pollutants. They even have basic, currently primitive, ways of recording environmental experiences — by glowing red and changing shape when exposed to certain conditions.
“On the environmental side, these could be used for bio-detection and bioremediation,” Blackiston said. “We could program the living robots to sense pollutants, and hopefully to seek them out and destroy them. Once they finish their job, they [could then] break down harmlessly in the environment, [without leaving] any artificial waste behind.”
The longer-term vision is focused on regenerative medicine. “Almost all problems of biomedicine — traumatic injury, aging, cancer, birth defects — could be defeated if we knew how to motivate cell collectives to build whatever complex organs we want,” said Levin.
The researchers speculate that it will be possible to build bots out of various different cell types for different use-cases. “You could imagine using a similar system to deliver drugs in a human patient, or aid in the repair process following an injury,” said Blackiston. “If made from a patient’s own stem cells, it would allow us to make biocompatible robots that are cleared from the patient naturally after they finish their job.”
There’s still a lot more work to do before this stage is reached. One challenge involves how best to control the bots. “[This problem] remains, for now, a complete mystery,” said Bongard. “We are working on this, and we hope to have new surprises to report on in the not too distant future.”
Blackiston said that one concept involves programming the bots with innate biological behaviors, which could potentially evolve as they age. In other words, the xenobots could be “born” with one purpose, and then switch to another as they get older.
Another hurdle involves speeding up the production of the bots. Currently, xenobots have to be built by hand, a process that, Blackiston noted, “requires a lot of time under the microscope and a great deal of fine motor control.” The researchers are looking at ways of adapting 3D bioprinters to automate the entire process, resulting in a sort-of conveyor belt production line for living robots.
One thing that’s sure: We’re likely to hear a whole lot more about xenobots as time goes by. The “xeno” in their name might stick around, but these are likely to become a whole lot more familiar to the world in the years to come.
- Optical illusions could help us build the next generation of AI
- Dyson lifts lid on ‘top secret’ project
- Finishing touch: How scientists are giving robots humanlike tactile senses
- Scientists think the next big solar storm could create an ‘internet apocalypse’
- The future of manufacturing: A look ahead to the next era of making things