Scientists Build a Tiny Robot That Can Change Medical Science Forever

Picture a robot so small it could slip through the wall of a living cell — not sci-fi, but something a lab team at Peking University has been quietly building for years. The material? DNA. In a major review published in SmartBot this January, researchers led by Lifeng Zhou map just how far DNA-based machines have come — and what they think comes next.

The Unlikely Building Material Behind Today's Smallest Machines

DNA has one job in most people's knows that: carry genetic instructions. But Nadrian Seeman noticed something in the 1980s that the field had mostly overlooked — DNA is also a precision building material. Its four chemical letters pair in predictable ways. You can design that pairing, control it, and fold DNA into almost any shape you want.

The real turning point came in 2006, when Paul Rothemund at Caltech published the DNA origami technique — folding a long strand into flat shapes using hundreds of shorter "staple" strands. Smiley faces. Triangles. Cute, but also a proof of concept. If you can fold DNA into a smiley face, maybe you can fold it into a hinge, a gripper, or something that opens only when it detects a cancer cell.

Why Steel, Silicon, and Chemicals All Hit a Wall

Building small machines is not new. Chemists earned the 2016 Nobel Prize doing it with synthetic compounds. Chip-makers have been shrinking transistors for decades. But those approaches hit a ceiling — you cannot program a silicon chip to rearrange itself when it detects a protein, and synthetic molecular machines struggled with reliability in biological environments.

DNA sidesteps both problems. A short double-stranded bundle behaves like a rigid rod. A single-stranded section acts like a flexible hinge. Rigid rod plus flexible hinge equals the basic grammar of a mechanism — and unlike proteins, which evolution took billions of years to optimize, DNA structures can be designed on a laptop in an afternoon.

How Do You Actually Build a Robot Out of DNA?

The Peking University team describes DNA machines that replicate classical engineering mechanisms. Not metaphorically — literally. Researchers have built revolute joints (like a door hinge), prismatic joints (like a sliding drawer), and even crank-slider mechanisms, all at nanometer scale, all from DNA strands.

Actuation comes from several sources. Strand displacement — introducing a new DNA strand that peels away an existing one — works like tripping a molecular latch. Electric and magnetic fields can spin entire DNA structures. Light can trigger folding or unfolding. Each joint can be programmed to respond to a different trigger. A single DNA robot can, in principle, have a dozen independently controllable moving parts.

From Lab Curiosity to Tumors, Viruses, and Computer Chips

Zhou's team built a four-fingered DNA nanogripper — inspired by the human hand — that detected SARS-CoV-2 in human saliva with sensitivity matching PCR tests, and showed early signs of physically blocking viral entry into cells. That was published in 2024. Not a simulation.

A separate DNA nanorobot selectively delivered anticoagulant drugs to tumor blood vessels in animal models, cutting off blood supply to tumors while sparing healthy tissue. Beyond medicine, DNA nanostructures are being used as chip-manufacturing templates, positioning gold nanoparticles at sub-5 nanometer spacing. One group recreated Van Gogh's Starry Night as a nanophotonic image with 65,536 addressable pixels — strange, beautiful, and a genuine demonstration of atomic-level spatial control.

What These Machines Still Cannot Do

The review is candid about limitations. The core problem is what the authors call "structural floppiness" — DNA joints jitter. Brownian motion, the random thermal kicks every nanoscale object experiences, cannot be switched off. String enough joints together and the uncertainty compounds badly. Building a complex DNA robot is less like assembling a watch and more like assembling one inside a blender on low.

Fabrication is another unsolved problem. Most DNA machines today are made in small lab batches over hours or days. A 2017 method using E. coli fermentation offered a path to scale, but complex DNA sequences tend to mutate during bacterial propagation. Simulation tools cannot yet model a DNA robot across its full duty cycle — the computational cost is still prohibitive. The authors are betting on AI to close that gap. That bet might pay off. It is not a sure thing.

• DNA as a construction material: — Unlike metals or plastics, DNA strands can be programmed to self-assemble into precise 3D shapes, making them uniquely suited for building machines at scales where conventional manufacturing fails.
• Motion without motors: — DNA machines move using strand displacement reactions, electric fields, or light — no mechanical motor required, which is the only approach that works reliably at one-thousandth the width of a human hair.
• The commercialization gap is real: — Despite two decades of research and genuinely impressive demonstrations, no DNA-based machine has reached commercial production — scale-up, stability in biological environments, and manufacturing cost remain unsolved problems.

"DNA machines are poised to play a transformative role in next-generation precision medicine, atomic-scale manufacturing, and the integration of biological and information technologies — with continued innovation and interdisciplinary collaboration, this field holds the potential to catalyze a new biotechnology-driven industry." — An, Wu, Xiong, Zhang, Dai, Zhou, SmartBot, 2026.