RoboSpeak — Robots 'grow' their own sensors. Mimicking moth 'veins', UV light triggers photosensitivity

“

Sensors can also be grown on-site.

01.

Sensory capabilities that were not there at the time of leaving the factory are created by themselves during work.

To give a robot an additional skill, there are generally a few ways: opening the body to add parts, hanging a module on the outside, or letting software find new uses for old hardware. The Curiosity rover used the last type of thinking,

taking the accelerometer originally used for navigation to measure Mars' gravity.

However, these few paths have a common point: new capabilities either come from the recombination of existing components or from things attached from the outside.

Recently, the University of Tartu in Estonia and the Italian Institute of Technology (IIT) collaborated to take a very different path:

Letting the robot synthesize a brand-new sensor on-site inside its own body, relying on chemical reactions.

The team named this process receptogenesis, which can be understood as “receptor genesis.”

They made a moth-mimicking robot. When it was first made, it had no reaction to ultraviolet light. After being exposed to strong ultraviolet light, a photosensitive area would be synthesized in the thorax. From then on, when it encountered ultraviolet light again, it would flap its wings and flash LEDs.

This research has already been published in 《Advanced Science》.

Researchers also compared several existing types of solutions: self-healing materials can heal damage, but ultimately only return to the original state; growth-type robots mimicking plant root systems do indeed add new matter, but the increment is accumulated at the growth end on the outer edge of the body; in-situ 3D printing adds components on the outside. According to the research team, previously no solution could

create hardware inside the body that is integrated with the body structure and also brings new functions.

02.

Inspiration comes from the “blood” in the wings of moths

The material basis of all this is a set of “vascular systems” that mimic insects.

Insects have an open circulatory system: the heart pumps out hemolymph, which flows throughout the body and also flows into the wings. The wings of butterflies and moths are actually living tissue; hemolymph participates in regulating wing hardness and temperature, and also participates in signal transmission.

The reference the team used is a type of moth called

Blue-winged moth

moth. The specimen was taken from the Estonian National Biodiversity Monitoring Project, collected using an insect trap in the summer of 2023.

The robot's “body” is 3D printed using transparent PETG material, and the internal transport network is divided into two layers.

The first layer is the “main artery,” printing channels 1 millimeter wide and 1.4 millimeters high, responsible for transporting liquid to the entire body. The second layer is the “capillaries”: when printing the internal filling, gaps are left between adjacent printed lines; the team intentionally utilized these gaps, letting them connect into a dense network of microchannels.

After the liquid enters the microchannel, it will slowly seep into the PETG material itself. In about 20 seconds, the penetration depth can reach around 10 micrometers.

In this way, by using only one printing process, the transport scale from tens of centimeters to several micrometers is opened up.

The robot actually made has a vascular coverage area of about 86 square centimeters on a single wing, which is about 25 times that of the chest, and the total vascular volume of the entire system is about 3 milliliters. Only one pump is used for the whole body, with no valves or other flow control components, and the common bubble problems in the pipelines do not have a large impact on it.

03.

With a blast of ultraviolet light, the transparent wings turned black.

The “blood” flowing inside the robot is a prepared chemical raw material: pyrrole (which is both a monomer and acts as a solvent), a type of iodonium salt photoinitiator, plus cellulose acetate used to improve mechanical properties.

PETG can pass ultraviolet light with wavelengths above 310 nanometers, so the 365 nanometer ultraviolet LED used by the team can shine directly into the interior of the material.

The photoinitiator decomposes after absorbing ultraviolet light, driving the polymerization of pyrrole into

polypyrrole. Polypyrrole is dark-colored, and the originally transparent areas will slowly turn black visibly to the naked eye, making the progress of the reaction clear at a glance.

The key is that the reaction only occurs in the areas that are illuminated.

The team performed a “lithography” test using a light-shielding mask, with a pattern resolution reaching 0.3 millimeters, primarily limited by light scattering caused by surface roughness.

In other words, the raw material can fill the entire body, but only the part that receives light truly becomes a sensor. Researchers envision that in a real environment, body posture, self-occlusion, and the distribution of surrounding light would jointly determine where the sensor is located, and in this experiment, this role is played by the mask.

Then why is it able to sense light?

After the newly generated polypyrrole is irradiated by ultraviolet light, the interior will briefly form

polarons and bipolarons, and the conductivity will increase accordingly, while the impedance decreases; the magnitude of the decrease is proportional to the ultraviolet dose. After the light irradiation stops, the signal recovers in about 300 seconds.

The comparative experiment is very intuitive: when only the raw materials were poured and not yet polymerized, shining a 0.9W LED at 20 mm resulted in no reaction; after shining a 1.6W LED at 43 mm once and completing the synthesis, it could sense both types of light sources.

Next is the closed loop. Two gold wires with a diameter of 30 micrometers were pre-embedded in the chest as electrodes, with a spacing of about 130 micrometers, and a reading is handled by an ATtiny microcontroller paired with a 10 milliampere-hour small battery. If the impedance drop exceeds the threshold (about 1%), a red LED flashes; if a drop is detected five consecutive times, it enters a 65-second “reaction mode”:

Pass an electric current through the nickel-titanium shape memory alloy wire for 5 seconds to fold the wings together, then let it cool naturally for 60 seconds, and then have the spring help reset it.

04.

Each sensor is unique and will also “accumulate experience”

There is a particularly interesting point in the research.

Since polypyrrole always grows into clusters randomly within the material, the resistance differences between different samples are very large, ranging from 200,000 ohms to 2,000,000 ohms.

The team did not treat this as a defect to be eliminated, but instead switched to using the relative rate of change of impedance as the signal. This is somewhat similar to biological sensors, focusing on “how much it changed” rather than absolute values.

More specifically, this sensor will “gain experience.” During the repeated testing process, the surrounding residual pyrrole continues to polymerize, the sensitivity becomes higher with use, and the base resistance also gradually rises.

Researchers described it as a “leaky integrator,” believing that the material structure itself became

Physical records of the robot's personal experiences.

After being stored for four months, the system can still work normally.

Of course, this is a

proof of concept, currently only demonstrating one pre-designed chemical route, and cannot be considered a truly autonomously evolving system; what raw materials are pumped into the blood vessels was decided by humans this time; where the raw materials come from and how they are replenished are also not within the scope of this research.

But the expansion space of this framework is clear. Just for polypyrrole, changing the synthesis conditions could point to different functions such as bioelectronic interfaces, ion sensing, anti-corrosion, and electromagnetic shielding, not to mention changing to other raw materials.

A vast toolbox of chemical synthesis, expected to become an “upgrade pack” for robot bodies.

It is worth mentioning that the methodology section of this study specifically notes that the team used Microsoft Copilot (GPT-5.2), as well as Anthropic's Claude Sonnet 4.6 and Opus 4.7, when writing microcontroller programs and Python data analysis scripts.

Future robots may no longer remain the same as they were when they left the factory.

Depending on what environment they encounter, they may grow what kind of bodies.

Paper link: https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.76999

END

Industrial robot companies

Eston Automation

|

Efte Robot

|

Fao Robot

|

Yuexiang Robot

|

Jieka Robot

|

Songling Robot

|

Luoshi Robotics

|

A Tongmu Robot

|

Geek+

|

Hikvision Robot

|

Yifei Technology

|

Ailite Robot

Service and special robot companies

Yijiahe

|

Jingpin Tieshang

|

Qitem Robotics

|

Shihe Robot

|

Pudu Robot

|

Shiluo Robot

|

Kuma Technology MAMMOTION

Humanoid robot companies

UBTech Robotics

|

Unitree

|

Yunchedu

|

Xingdongjinyuan

|

Weijing Robot

|

Zhujidongli

|

Leju Robot

|

Daxiang Robot

|

Magic Atom

|

Zhongqing Robot

|

Pasi Ni Perception

|

Cyborg Robot

|

Digital Huaxia

|

Fourier Intelligent

|

Tianlian Robot

|

Kepler Humanoid Robot

|

Zhongke Huiling

|

Qingbao Robot

|

Zhejiang Humanoid Robot Innovation Center

|

Dongyi Technology

|

Zhishi Technology

|

PNDbotics

|

Zhuoyide Robot

|

Qinglang Intelligent

|

Galileo GALILEO

|

Songyan Power

|

Tianji Intelligent

|

Woaan Robot

|

LiGong Huahui

|

Jiace Jinhua

|

Graphui Intelligence

Embodied intelligence companies

Cross Dimension Intelligent

|

Galaxy Universal

|

Qianxun Intelligent

|

Lingxin Qiaoshou

|

Ruierman Intelligent

|

Microintelligence

|

Tuisheng Technology

|

Zhongke Silicium Records

|

Shutu Technology

|

Lingqiao Intelligent

|

Xingchen Intelligent

|

Qiongche Intelligent

|

Fangzhou Infinite

| iFLYTEK |

Beijing Humanoid Robot Innovation Center

|

National and Local Co-built Humanoid Robot Innovation Center

|

Daimeng Robot

|

Shibitu Robot

|

Xinghaitu

|

Yuequan Bionics

|

Zeroth Power Robot

|

Zhongke ShenGu

|

Zhi Fangping

|

Daka Robot

|

Haocun Technology

|

Ju Shi Intelligent

|

Xynova Xiner Future

|

Feixi Technology

|

Weilu Power

|

Boden Intelligent

|

Qianjue Technology

|

Lingsheng Technology

|

Ji Cui Intelligent Manufacturing

|

Xinbeit Technology

|

Chen Hunxian Technology

|

Dexmal Original Power

|

Youliqi

|

Self-variable

|

Ruiyan Zhikong Lingqiao Hand

|

Qiwu Technology

|

RoboScience Machine Science

|

Zhongke Fifth Dynasty

|

Critical Point

|

Danghong Technology

|

Qiaojie Shu Wu

|

Vbot Veta Power

|

Tashan Technology

|

Ju Brain Pan Shi

|

Youai Zhijhe Robot

|

Zhixing Jian

|

Aimiao Robot

|

Lu Ming Robot

|

Quantum Intelligent

|

Ruoyu Technology

|

Tiehu Robot

|

Zhi Wuji

|

Origin Infinite

|

Blackman Technology

|

Qingyan Precision

Medical robot companies

Yuanhua Intelligent

|

Tianzhihang

|

Sizhe Rui Intelligent Medical

|

Jingfeng Medical

|

Tuo Dao Medical

|

Zhenyida

|

Shuru® Robot

|

Luosenbot

|

Shuimu Dongfang

｜

Kangnuo Sistem

|

Disight Medical

Upstream industry chain companies

Lvde Harmonic

|

Yinshi Robotics

|

Kunwei Technology

|

Muta Intelligent

|

Qingpou Vision

|

Benmo Technology

|

Lan Dian Touch Control

| Xinjingcheng Sensor

|

BrainCo Strong Brain Technology

|

Yuli Instrument

|

Jiya Jingji

|

Sila Technology

|

Shenyuan Sheng

|

Feipu Navigation Technology

|

Inks

|

Juxi Intelligent Drive

|

Lingyun Guang Yuanke Vision

|

Xuanji Power

|

Yiyou Technology

| Ruiyuan Precision |

Lingzhu Era

|

HIT Huawei Ke

|

Xinghui Sensor

|

Lingdi Technology

|

Quan Zhi Bo

|

CubeMars Robot Power

|

Wanglong Robot Elevator

|

Hangkai Microelectronics

|

Wutong Sensing and Control