Xenobots: Researchers Build Next-Gen Living Robots


In 2020, a group of computer scientists and biologists from the University of Vermont (UVM) and Tufts University created Xenobots — unique, small self-healing biological bots from frog cells. They could push a payload, walk around, and still show cooperative behavior in the presence of a pack of other Xenobots.

The next generation: Xenobots 2.0

The same group has now developed life forms that self-assemble a body from single cells, do not need muscle cells to stimulate, and also show the capability of recordable memory. The latest Xenobots travel faster, have longer lifespans, and navigate distinct environments than the first version, and yet, they can work collectively in teams and heal themselves if damaged. The consequences of the new analysis were published in Science Robotics.

Corresponded to Xenobots 1.0, the millimeter-sized automatons were created in a “top-down” method by manually placing surgical shaping and tissue of frog skin and cardiac cells to generate movement, the latest version of Xenobots comes with a “bottom-up” process. The biologists at Tufts exerted stem cells from eggs of the African frog Xenopus laevis. Hence these robots are called “Xenobots.”  

These cells allow them to self-assemble and develop into spheroids, where few cells after some days adapted to generate cilia — tiny hair-like projections that move back and forth or revolve in a particular direction. 

Rather than employing manually sculpted cardiac cells whose spontaneous rhythmic contractions allow the Xenobots 1.0 to scuttle around, cilia give the new spheroidal bots “legs” to drive them quickly across a surface. In a frog or human for that matter, cilia would commonly be located on mucous surfaces, like in the lungs, to eliminate the pathogens and other foreign material. On the Xenobots 2.0, they are re-purposed to provide quick movement.

“We are observing the extraordinary mobility of cellular collectives, which create a rudimentary unique ‘body’ that is completely different from their default — in this situation, a frog — despite having a perfectly normal genome,” said Michael Levin, Director of the Allen Discovery Center and Biology Professor at Tufts University, and author of the research.

 “In a frog embryo, cells unite to produce a tadpole. Here, eliminated from that context, we notice that cells can re-purpose their genetically encrypted hardware, like cilia, for new roles such as movement. Amazingly, cells can immediately take on new roles and produce new body systems and functions without lengthy periods of evolutionary selection for those characteristics.”

“In a way, the Xenobots are much similar to a traditional robot. Only we use tissues and cells rather than artificial elements to form the shape and produce the predictable response.” said senior researcher Doug Blackiston, who co-first authored the study with research technician Emma Lederer. “On the biology point, this method is assisting us to know how cells interact as they communicate with one another during development, and how we might adequately regulate those interactions.”

While the Tufts researchers built the mechanical organisms, researchers at UVM were occupied operating computer simulations that showed distinct forms of the Xenobots to understand if they might show diverse responses, both individually and in teams. 

Employing the Deep Green supercomputer group at UVM’s Vermont Advanced Computing Core, the team, directed by Josh Bongard, robotics experts, and computer scientists and under thousands of millions of casual environmental situations using an evolutionary algorithm. These simulations were utilized to recognize Xenobots most ready to work collectively in packs to accumulate large piles of debris in a field of particles.

“We know the job, but it’s not at all obvious — for people — what a successful design should look like. That’s where the supercomputer comes in and examines over the range of all feasible Xenobot packs to find the pack that does the task best,” says Bongard. “We want Xenobots to do beneficial work. Right now we’re providing them easy jobs, but eventually, our goal is for a new kind of living mechanism that could, for example, pick up microplastics in the ocean or contaminants in soil.”

It rolls out the latest Xenobots are very fast and more reliable at jobs such as trash collection than the previous model, working mutually in a pack to clean a petri dish and gather massive piles of iron oxide particles. They can also travel through narrow capillaries or cover large flat surfaces. These researches also recommend that the in silico simulations could in the future optimize new traits of biological bots for more complicated operations. One major highlight added in the Xenobot upgrade is the capability to record data.

Xenobots  Scientist
Doug Blackiston, PhD, co-first author on the study

Now with memory

A primary highlight of robotics is the potential to record information and utilize that information to transform the robot’s actions and performance. With that in mind, the Tufts researchers developed the Xenobots with a read/write skill to store one bit of data, using a fluorescent reporter protein called EosFP, which usually shows the green light. In the presence of light at 390nm wavelength, the protein shows red light instead of green.

The cells of the frog embryos were infused with carrier RNA coding for the EosFP protein before stem cells were extracted to build the Xenobots. The modern Xenobots now own a built-in fluorescent switch that can store exposure to blue light at approximately 390nm.

The Scientist examined the memory function by releasing 10 Xenobots to float around a surface on which one point is lighted with a beam of 390nm light. After two hours, they noticed that three bots show red light. The rest showing their original green light, efficiently storing the “travel experience” of the bots.

This proof of principle of molecular memory could be continued in the future to identify and store not only light but also the presence of chemical pollutants, radioactive contamination, drugs, or a disease condition. Further engineering of the memory function could facilitate the storing of various stimuli (more bits of data) or allow the bots to deliver compounds or modify action upon the sensation of stimuli.

Xenobots  lights
Using a fluorescent protein, Xenobots store exposure to blue light, by turning green, Credit: Doug Blackiston

Xenobots can heal themselves

“The organic stuff we are using has multiple characteristics we would like to implement in the bots — cells can act as storing devices to record information, sensors, computation networks and communication, motors for locomotion,” said Levin. “One thing the Xenobots and up-coming variants of biological bots can do that their plastic and metal counterparts have a problem is building their body design as the cells grow and mature, and then restoring and repairing themselves if they become damaged. Healing is a genetic characteristic of living organisms, and it is stored in Xenobot biology.”

The latest Xenobots were exceptionally proficient at healing and would close the majority of a severe full-length damage half their thickness within 5 minutes of the injury. All damaged bots were able to eventually heal the injury, restore their shape and resume their job as before.

According to Levin another benefit of a biological robot is metabolism. Unlike plastic and metal robots, the cells in a biological bot can consume and break down chemicals and work like small factories secreting and synthesizing proteins and chemicals. The entire field of synthetic biology — which has mainly concentrated on reprogramming single-celled organisms to create useful molecules — can now be employed in multicellular organisms.

Like the first Xenobots, the enhanced bots can endure up to ten days on their embryonic energy stores and manage their responsibilities without supplementary energy sources. However, they can also carry on at an adequate rate for several months if kept in a “soup” of nutrients.

What the experts are really after

Pleasant information about the biological bots and what we can acquire from them is manifested in a TED talk by Michael Levin.

During his TED Talk, professor Levin explains not only the extraordinary potential of biological robots to do useful tasks in the environment or possibly in healing purposes, but he also aims at what may be the most precious advantage of this study — using the bots to know how individual cells grow together, interact, and specialize to form a bigger organism, as they do in nature to produce a frog or human. It’s a unique way that can give a framework for regenerative medicine.

Realizing the enormous future for this technology, the University of Vermont and Tufts University have set up the Institute for Computer Designed Organisms (ICDO), to be formally started in the upcoming months, which will collectively pull resources from each university and outside sources to build living robots with frequently complex skills.

Journal Reference:

Douglas Blackiston, Sam Kriegman, Michael Levin, Emma Lederer, Simon Garnier, Joshua Bongard. A cellular platform for the development of synthetic living machinesScience Robotics, 2021; 6 (52): eabf1571 DOI: 10.1126/scirobotics.abf1571


Leave a Reply

Your email address will not be published. Required fields are marked *