Robotic.orgFOR THE ROBOT-CURIOUS
← Edition 001Real · sourced researchFIELD NOTES / 11 SEPT 2026
RESEARCH, WITH THE SCALE TURNED DOWN

Small robots.
Enormous questions.

A tiny machine is never the whole system.
Follow the power. Find the steering. Ask what was tested.

An imagined spiral swimmer in a droplet beside the eye of a needle.
AI illustration / imagined scaleAI concept illustration, not a microscope image, a depiction of the reported robots, or a drawing to scale. The research below describes what was actually demonstrated.
CHANGE YOUR FRAME OF REFERENCE01 / SCALE LAB
100mm

The familiar world.

A hand-scale reference. A useful place to begin, before almost everything gets smaller.

100 mm reference width. The bar shows relative size, not a physical measurement of your screen.

REFERENCE / 100 MM

Where is the intelligence?

A robot can include a body, sensors, a controller and supporting equipment elsewhere. Smaller bodies do not necessarily mean smaller complete systems.

Explore the research below ↓
THREE RESEARCH STORIES. DISTINCT CLAIMS.

What the evidence says.

Sources checked 11 September 2026
Research dates shown separately
01
PENN + UNIVERSITY OF MICHIGAN · 17 DEC 2025

A swimmer with its own instructions.

Researchers reported light-powered swimming robots about 200 × 300 × 50 micrometers in size. Onboard computers let them sense temperature and change their paths without a tether or external magnetic steering.

Demonstrated

Programmable motion and responses to local temperature.

Still a possibility

Monitoring individual cells. This report does not establish medical treatment.

Source & experiment boundary

University research report, published 17 December 2025. Robotic.org has checked the report, not independently replicated the experiment.

Penn Today: original report
02
ETH ZURICH · 13 NOV 2025

A delivery system, not a tiny doctor.

Drug-loaded gel capsules were guided magnetically, then dissolved through magnetic heating to release their cargo. A catheter, external navigation equipment and X-ray imaging are part of the system.

Demonstrated

Targeting and clot dissolution in a vessel model. Navigation and visibility in pigs; navigation through cerebrospinal fluid in a sheep.

Not established by this report

Treatment in humans. Human trials were a future goal at publication.

Source & experiment boundary

The model and animal experiments establish different things. They should not be collapsed into a claim that a robot has treated a human patient.

ETH Zurich: original report
03
MAX PLANCK / AUGSBURG · PREPRINT, 30 SEPT 2025

Maybe the palm deserves a promotion.

ISyHand explores dexterity with an articulated palm and 18 degrees of freedom. The researchers trained cube reorientation in simulation, then tested it on the physical hand.

Demonstrated

The reported physical evaluation averaged 6.1 cube reorientations in the included episodes.

What the headline leaves out

Ten of 30 trials were excluded for pose-tracking failures. The experiment does not establish general household competence.

Source & experiment boundary

Research preprint. Read the evaluation, including excluded trials and failure cases, before comparing headline results.

ISyHand: paper and authors
Read methods and evaluation
Possible · thought experiment

What if medicine
had an address?

Imagine delivery to a chosen location. Now ask who steers, where energy comes from, how you see the machine, and what happens after the delivery.

Who is driving?

In ETH’s reported system, magnetic navigation comes from outside the body. An imagined autonomous courier would need another way to decide where it is and where to go.

A design question, not a medical recommendation. The Penn swimmer and ETH capsule are separate systems.

NOW TAKE THE IDEA OUTSIDE

Design an expedition.
At garden scale.

Your export

If downloading is blocked, copy these contents into a local file. Keep source labels and assumptions with anything you share.

Review your journal