For decades, the field of robotics has been defined by a faithful, if mechanical, imitation of nature. Engineers have labored to replicate the gait of a dog, the flight of an insect, or the logic of a human brain using silicon and steel. But a new frontier in bioengineering is abandoning imitation in favor of direct synthesis. Instead of building machines that act like biology, researchers are building machines out of it.
The latest milestone in this shift is the "neurobot," a tiny, free-swimming assemblage of living cells that possesses its own rudimentary nervous system. As reported in the journal Advanced Science, these biological machines are not merely clusters of tissue; they are self-directed systems featuring neurons that autonomously wire themselves into functional circuits. The development, led by Tufts University biologist Michael Levin and his colleagues, represents a significant leap from earlier "xenobots"—millimeter-scale living robots first unveiled in 2020—moving toward internal, biological control mechanisms rather than reliance on external stimulation or pre-programmed geometry.
From Xenobots to Neurobots: A Lineage of Living Machines
The intellectual lineage here matters. Xenobots, constructed from embryonic frog cells, demonstrated that biological tissue could be coaxed into novel configurations capable of locomotion, self-healing, and even rudimentary replication. Those early constructs, however, lacked anything resembling a nervous system. Their behavior emerged from the contractile properties of cardiac cells and the physical architecture imposed by researchers—closer to a wound-up toy than an autonomous agent.
Neurobots mark a qualitative departure. The inclusion of neurons that self-organize into functional circuits introduces a layer of internal coordination absent from prior generations. In conventional robotics, the analogy would be the difference between a mechanical automaton following a fixed cam pattern and a system with an onboard processor capable of adapting its output. The distinction is not merely technical; it raises foundational questions about where mechanism ends and agency begins.
This trajectory also reflects a broader shift in synthetic biology. The field has moved from treating biological components as passive building materials—scaffolds, actuators, sensors—toward leveraging the self-organizing properties that make living systems fundamentally different from engineered ones. A neuron does not need to be told how to form a synapse; given the right biochemical environment, it seeks connections on its own. Harnessing that capacity, rather than overriding it, is the core engineering insight behind the neurobot program.
Implications Beyond the Petri Dish
The practical applications most frequently cited for living machines—precision drug delivery, tissue repair, environmental cleanup—remain speculative at this stage. No neurobot has operated inside a living organism or been deployed in an open environment. The gap between a self-swimming cell cluster in a laboratory dish and a therapeutic agent navigating the human bloodstream is vast, involving challenges of biocompatibility, immune response, scalability, and regulatory approval that have stalled far more mature biotechnologies.
What the neurobot does offer, with more immediacy, is a research platform. Observing how minimal neural circuits give rise to coordinated behavior in a system simple enough to be fully characterized could yield insights that are difficult to extract from the staggering complexity of even the simplest natural organisms. The nematode C. elegans, with its 302 neurons, has been studied for decades and its connectome fully mapped, yet the relationship between its wiring diagram and its behavioral repertoire remains incompletely understood. Neurobots, with even fewer neurons and an architecture shaped partly by design, may offer a more tractable model for probing the link between circuit structure and function.
There is also a conceptual tension worth tracking. As living machines acquire more biological sophistication—nervous systems, adaptive behavior, perhaps eventually sensory feedback loops—the ethical and definitional frameworks surrounding them will face pressure. The boundary between a biological tool and a rudimentary organism is not a bright line, and the history of bioethics suggests that regulatory and philosophical discourse tends to lag behind laboratory capability.
The neurobot sits at an early but consequential point on that curve. Whether it proves to be a stepping stone toward functional cyborg architectures or primarily a window into developmental neuroscience, the underlying achievement is clear: engineering has moved from copying biology's outputs to recruiting its processes. The question now is less whether living circuits can be built and more what happens when they start making decisions their designers did not explicitly program.
With reporting from IEEE Spectrum Robotics.
Source · IEEE Spectrum Robotics
