Biological robots and xenobots

Kamm and colleagues, in a 2018 perspective on multi-cellular engineered living systems, set out the case for cell-based robots. They noted that conventional robots cannot self-emerge, self-assemble or self-heal¹. They added that the rules of interaction between living components at different scales remain elusive¹. Gumuskaya and colleagues later sorted biobots into three types: noncellular, hybrid and fully cellular. Noncellular biobots are entirely robotic but inspired by biological forms or behaviors⁷. Hybrid biobots combine biological and inanimate components, for example muscle tissue on inert scaffolds⁷. Fully cellular biobots form the third type, and xenobots belong there⁷.

Blackiston, Kriegman, Bongard and Levin (2023) reported a dispute over vocabulary. One author noted that the name xenobot rebrands the established animal cap nomenclature and avoided it as nontechnical. Team members disagree on the correct term, and the candidates include biobots, biological robots, engineered living systems and biomachines². The same paper warns that biological robots may never become a viable technology. It states that xenobots are not yet useful and cannot solve any problems³. Proposed uses such as cleaning the environment or delivering medicine remain visions.

Levin (2023) used xenobots to argue that cells hold behaviours the body normally suppresses. In the animal, frog skin cells form a passive protective layer. Liberated from the instructive influences of other cells, they form a self-motile Xenobot⁴. Levin also described kinematic self-replication. Given loose cells, the biobots corral them into piles that compact and form the next generation⁵. Pai and colleagues (2025) studied basal Xenobots made from Xenopus ectodermal explants. They asked how the transcriptome changes when a cell collective takes on a non-evolved body, without genomic editing or drug treatment⁸.

Gumuskaya and colleagues (2024) introduced anthrobots. These are ciliated spheroids of 30 to 500 microns that each begin as a single cell from the adult human lung. They self-construct over two weeks in extracellular matrix, then move in a low-viscosity habitat. Their movement ranges from tight loops to straight lines. The authors found that this behavioural diversity correlates significantly with morphological diversity⁶. Fotowat and colleagues (2026) characterised a self-assembling biobot to study wild-type neural cells in a non-standard body. They aim to inform the design of future innervated biological robots⁹.