In 2020, scientists  developed organic, biocompatible robots that independently moved and interacted with other objects. This breakthrough design revolutionized approaches to environmental care, drug delivery, along with countless other applications.

Think of a robot. Did you imagine R2-D2? Maybe even a Roomba? We often think of robots as chunks of metal and grinding gears with the occasional flickering bulb atop an antenna. We don’t imagine something organic or biological. Robots thrive off synthetic fuels and electricity, not cells and nutrients. However, innovations in artificial intelligence (AI) change this.

AI has come a long way since the times of Ferranti Mark 1 – a machine of the 1950s that bested masters of checkers. As we step into a new decade of the 21st century, AI has helped transform  robots into our personal assistant Alexa and social-learning companions like Moxie. While we’re distracted by Alexa’s ability to know our favorite song or Moxie’s cute animated face, we forget their environmental impact. Stripped down, Alexa and Moxie are chemicals, metals, and plastics that release toxins into the environment as they degrade.

This begs the question: Can AI be incorporated into a biodegradable robot? Xenobots say yes.

Robots made of cells

Xenobots are programmed from stem cells harvested from their namesake – the African clawed frog (Xenopus laevis) [1]. Stem cells are unspecialized cells; meaning, unlike oxygen-transporting blood cells or sensory neurons, they don’t have specific functions.

Instead, their ability is to differentiate or transform into virtually any other cell type. Remember how Ditto can become any Pokémon? Same concept. This makes stem cells, and Ditto, pluripotent.

Pluripotent stem cells self-organize [2]. They innately assemble to grow and morph into any tissues in the human body.  Scientists utilize this pivotal ability to design xenobots that can virtually take on any shape and form.

The algorithm

Created from frog-derived pluripotent stem cells, xenobots are programmed by an evolutionary algorithm. This algorithm was developed by a collaboration between the University of Vermont, Tufts University, and the Wyss Institute for Biologically Inspired Engineering at Harvard University.

Evolutionary algorithms are AI-based calculations that use mechanisms of biological processes – think reproduction and Darwinian evolution – to find solutions for a given problem or goal [3]. In this case, the goal is to engineer a functional robot, defined by its ability to independently move.

 Once optimal designs for the xenobots are calculated, they’re put to the test on a petri dish. The bots’ behaviors are observed and compared to the goal. If the goal is not met, the designs are plugged back into the algorithm with new constraints. These constraints help the algorithm reconfigure the robot to yield better designs. 

Compare evolutionary algorithms to playing Wordle. Your objective is to guess the secret word. Each guess is your design of what the word might be. If it’s the wrong design, you try again, but now you know your constraints – the letters that don’t belong in the word or letters that are in the wrong space. Scientists carry out these iterative processes until they achieve the optimal design for a functional xenobot.

Coming to life

A xenobot’s design is brought to life through two basic components: passive blocks (green) and contractile blocks (red). The passive blocks do not move. These blocks made of stem cells act as a body, while contractile blocks are the legs. The contractile motions are achieved through a layer of frog heart muscle cells attached to the underside of the body. The biological function of heart muscle cells is to form cardiac tissues that pulse as heartbeats. These cells lend their contractile abilities to xenobots, enabling them to slide across surfaces. Micro-tweezers were used to manipulate the blob of cells into a replica of the simulated design.

Comparison of simulated and observed xenobot movements. Adapted with permission from Kriegman et al. [1] under the Creative Commons License 4.0.

The xenobots’ movements were tested in three applications: swarming, manipulating their surroundings, and transporting objects. With these foundational movements, xenobots can gather plastic pollutants from oceans, precisely administer medical drugs, and travel throughout the human body to remove toxins or buildup [1].

Xenobots successfully display movements and behaviors predicted through AI simulations. Adapted with permission from Kriegman et al. [1] under the Creative Commons License 4.0.

Self-sustainability

Unlike our phones after a shocking drop onto the concrete, xenobots are self-repairable. When torn apart by micro-tweezers, these bots can completely heal themselves. Wound healing is something that is unique to complex cellular processes – a space where metal, plastics, and glass can’t compete. This maximizes the xenobots’ lifetime use while minimizing resource consumption and e-waste.

Xenobot repairs itself after injury. Adapted with permission from Kriegman et al. [1] under the Creative Commons License 4.0.

Just one year after these bots debuted, the same research group discovered that xenobots can also reproduce, just not in the traditional sense. We typically associate reproduction with two parents producing offspring. Some of us may remember more self-sufficient methods from biology class like budding in yeast or binary fission where bacteria asexually reproduce by splitting in half.

Xenobots reproduce by means of kinematic self-replication. When placed in a petri dish filled with their own base composition (single stem cells), the bots spontaneously swarm. The circular motions aggregate the single cells into spheres that form more xenobots within a matter of days.

This type of propagation has never been seen on the scale of whole organisms until now [4].  

Not all xenobots are created equal. For kinematic self-replication to work, the bots must be reconfigured into a specific shape and given just the right amount of building materials and nutrients. In this case, the T. rex-looking xenobot with big legs and tiny arms developed a year ago would not work. We need a shape that can corral the dispersed stem cells. With this goal, AI tells us a xenobot taking on the form of Pac-Man is our best bet [4].

An ethical dilemma?

Just the phrase “living robots” instills fear of humanity being surpassed and ruled by our own technological creations. These fears are not irrational. In an interview with Forbes magazine, Micheal Levin, co-author and director of Tufts University’s Center for Regenerative and Developmental Biology, reveals, “When we start to mess around with complex systems that we don’t understand, we’re going to get unintended consequences” [5].

Scientific discovery is a double-edged sword. Before we jump to conclusions and become apprehensive of starring in the next episode of Netflix’s Black Mirror, remember, the most that  xenobots can do right now is spin and scuttle around. These rudimentary actions do not pose threats to us. In fact, such simple actions have already inspired solutions to many of our problems.

Unlike Alexa and Moxie, xenobots are made of biodegradable materials. They’re easier to break down and will not release toxins, in contrast to  their plastic and metal counterparts.

Xenobots are biocompatible because of their cellular composition. Since stem cells may also be sourced from human tissues, xenobots can advance patient-specific therapeutics. Their form and function rely on the native pluripotent and self-organizing capabilities of stem cells. No genetic modifications required. This means xenobots can’t independently take off and evolve, as co-author Douglas Blackiston reports to NPR [6]. 

 At 1 millimeter, xenobots are smaller than the thickness of your credit card, making them ideal for solving problems in tight spaces which may include removing plaque buildup along your arteries [1]. This is a big step in preventing cardiovascular diseases – one of the global leading causes of death [7]. When hooked up to the right sensors, they can pinpoint hard-to-reach areas of radiation and remove toxic waste [1].  

With limitless resources, their autonomous abilities to self-propagate and heal make them robust and scalable for regenerative medicine, tissue repair, and more [1]. However, the caveat  is these limitless resources. Without a long-term supply of cells and nutrients, these bots would cease to function. Another checkpoint is the AI algorithm. It lets us control and predict the bots’ behaviors, ensuring they do not go beyond the scientist’s intentions.  

Xenobots are malleable and can take on virtually infinite forms. Each form is a potential solution and the list of applications goes on. As robotics expert, computer scientist, and corresponding author Dr. Bongard shared with CNN: “Most people think of robots as made of metals and ceramics but it’s not so much what a robot is made from but what it does, which is act on its own on behalf of people” [8].

As it stands now, exploring the benefits of xenobots helps us more so than harms us. From curing diseases to protecting the environment, they can be good for us. After all, they are organic.

References

1. Cover image: Kriegman, S., Blackiston, D., Levin, M., & Bongard, J. (2021, December 1). Kinematic self-replication in reconfigurable organisms (Movie S1)[Video]. YouTube. https://youtu.be/-tKlIZXHiOo 
2. Kriegman, S., Blackiston, D., Levin, M., & Bongard, J. (2020). A scalable pipeline for designing reconfigurable organisms. Proc Natl Acad Sci U S A, 117(4), 1853-1859. doi:10.1073/pnas.1910837117. https://creativecommons.org/licenses/by/4.0/legalcode
3. Libby, A. R. G., Briers, D., Haghighi, I., Joy, D. A., Conklin, B. R., Belta, C., & McDevitt, T. C. (2019). Automated Design of Pluripotent Stem Cell Self-Organization. Cell Syst, 9(5), 483-495 e410. doi:10.1016/j.cels.2019.10.008
4. Simoncini, D., Schiex, T., & Zhang, K. Y. (2017). Balancing exploration and exploitation in population-based sampling improves fragment-based de novo protein structure prediction. Proteins, 85(5), 852-858. doi:10.1002/prot.25244
5. Kriegman, S., Blackiston, D., Levin, M., & Bongard, J. (2021). Kinematic self-replication in reconfigurable organisms. Proc Natl Acad Sci U S A, 118(49). doi:10.1073/pnas.2112672118
6. Chandler, S. (2020). World’s First ‘Living Robot’ Invites New Opportunities and Risks. Retrieved from https://www.forbes.com/sites/simonchandler/2020/01/14/worlds-first-living-robot-invites-new-opportunities-and-risks/?sh=2fa59cfa3caf.
7. Neuman, S. (2021, December 1). Living robots made in a lab have found a new way to self-replicate, researchers say. Retrieved from https://www.npr.org/2021/12/01/1060027395/robots-xenobots-living-self-replicating-copy
8. World Health Organization. (2022). Cardiovascular diseases (CVDs). Retrieved from https://www.who.int/en/news-room/fact-sheets/detail/cardiovascular-diseases-(cvds)
9. Hunt, K. (2021, November 29). World’s first living robots can now reproduce, scientists say. Retrieved from https://www.cnn.com/2021/11/29/americas/xenobots-self-replicating-robots-scn/index.html